How Good is Family Therapy? A Reassessment 9781487580377

Citation preview

HOW GOOD IS FAMILY THERAPY? A REASSESSMENT Family therapy is one of the most widely practised psychotherapies in North America. Roy and Frankel here provide a comprehensive and critical reassessment of the research literature regarding the efficacy of this form of treatment. The main thesis of the book is that this research is still in its infancy and that, although important contributions have been made, much work remains to be done. The book is divided into three parts. The first offers an overview of the current state of the family therapy field. The second assesses outcome studies of family therapy on the basis oflife-stage issues. It examines the literature on family treatment with children, adolescents, and adults. The third part reviews the outcome of a host of problems treated by this method: psychosomatic and medical conditions, alcoholism, anorexia nervosa, drug addiction, and placement prevention in child welfare. The authors conclude by reviewing the state of the art in the field and defining future directions for research. is a professor in the Faculty of Social Work and the Department of Clinical Health Psychology, Faculty of Medicine, at the U niversity of Manitoba. His recent publications include The Social Context of the Chronic Pain Sufferer and Chronic Pain in Old Age: An Integrated Biopsychosocial Perspective, an edited collection of original essays. RAN JAN ROY

HARVY FRANKEL is an associate professor in the Faculty of Social Work and diretor of the Child and Family Services Research Group at the University of Manitoba. He is the author of several articles on familycentred practice.

RANJAN ROY AND HARVY FRANKEL

How Good Is Family Therapy? A REASSESSMENT

UNIVERSITY OF TORONTO PRESS Toronto Buffalo London

© University of Toronto Press Incorporated 1995

Toronto Buffalo London Printed in Canada Reprinted in 2018

ISBN 0-8020-2926-4 (cloth) ISBN 978-0-8020-7427-0 (paper)

(§ Printed on acid-free paper

Canadian Cataloguing In Publlcadon Data

Roy, Ranjan. How good is family therapy? Includes bibliographical references and index. ISBN 0-8020-2926-4 (bound). - ISBN 978-0-8020-7427-0 (paper) 1. Family psychotherapy. I. Frankel. Harvy. II. Title.

RC488.5.R681995

616.89'156

C95-93547-0

University of Toronto Press acknowledges the financial assistance to its publishing program of the Canada Council and the Ontario Arts Council.

Contents

vii

FOREWORD PREFACE

xi

1 Family Therapy: An Overview 3 2 Children and Family Therapy 22 3 Juvenile Delinquency and Conduct Disorders

37

4 Adult Psychiatric Problems 59 5 Psychosomatic and Medical Disorders

81

6 Alcoholism and Family Therapy 98 7 Adolescent Drug Abuse and Mixed Emotional and Behavioural Disorders 115 8 Anorexia Nervosa

128

9 Family-Based Approaches to Placement Prevention in Child Welfare 151 10 Future Directions REFERENCES INDEX

209

185

175

Foreword NATHAN B. EPSTEIN

The practice of psychotherapy seems to be capable of generating an unusually high degree of enthusiasm and even blind faith among those in the field. Perhaps this phenomenon is necessary in order to attract and keep people working in this most difficult of endeavours - that of attempting to bring about change in human behaviour. The practice of family therapy is no exception. As one of the earliest workers in family therapy I can attest to the tremendous enthusiasm and excitement and the high level of commitment generated by the work carried out when the field was in its infancy. We all felt like pioneers discovering a new world or like revolutionaries about to succeed in overturning the old order, having found the ultimate solution to human behaviour problems. Indeed, some of the early workers began to talk seriously about solving such problems as war, poverty, racism - you name it - by the application of family therapy techniques. As the field has matured there has been a trend towards objective self-examination, which has resulted in therapy outcome research and in multiple reviews of work in the field. This book is a very welcome example of this trend. Ranjan Roy and Harvy Frankel, committed scholars, teachers, and practitioners of family therapy, present here a prodigiously exhaustive critical review of the literature. Their approach is one of scientific detachment in reporting their findings, yet at the same time they maintain their strong commitment to the field. They are very knowledgeable and skilled teachers who are interested in the development of the field of family therapy, rather than in perpetuating the mythologies that have accumulated over the years. The message of the book is sobering with regard to family therapy. Although the authors state that good outcome literature does exist,

viii Foreword and that this gives cause for optimism, they find that the amount of methodologically acceptable outcome research on family therapy is woefully inadequate. In their view, family therapy has yet to make a strong and persuasive case for its effectiveness based on hard data rather than only on claims of clinical success. As with previous, less comprehensive, reviews, this study indicates that (1) in certain clinical situations family therapy may be of some benefit - and is at least better than no therapy at all; (2) there are still no indications as to the kinds of situations family therapy is most effective in; (3) there appears to be no difference in effectiveness among different techniques or styles of family therapy; and (4) family therapy does not appear to be more effective than other forms of psychotherapy. The findings reported here are similar to the findings reported in reviews of all other types of psychotherapy; that is, the quantity and methodology of outcome research and the data concerning effectiveness were often found to be inadequate. Although these findings are sobering, neither the authors nor I feel that discouragement is justified. What these findings indicate is that the field of treatment outcome research is actually quite young and the technology of such research is at an early stage of development. As for myself, I am at present involved in a fourth generation of outcome research in family therapy. Each generation has become more sophisticated and has enabled us to look at and consider more factors in the therapeutic process. It is hoped that as this field progresses workers will be able to study satisfactorily many important factors and variables. This optimism does not deny the complex, laborious, and timeconsuming nature of this research. My research colleagues and I have been doing outcome research in family therapy for thirty-five years; although we still feel we are just at the beginning, we are becoming more comfortable and confident with our efforts. Along with the authors I feel that family therapy is a useful endeavour. But like all the major psychotherapies now practised, it hasn't quite lived up to its original promise. Serious reflection and ongoing research such as that reported here will prove useful to practitioners in their attempts at healing human illness and improving human relationships. This book is not only for practitioners and researchers of family therapy, but also for all those concerned with the current issues involved in the rapid developments and changes in the larger field of health care delivery. As the health care field becomes more entrepreneurial and

Foreword ix profit-driven there has been a strong movement to cut back on treatment, thin out the treatment process, 'dumb down' treatment teams, and in general skimp on care in order to generate profit. While some changes have been overdue and indeed welcome, there has been concern that the health care field is moving in a dangerous direction, to where profit rather than quality and effectiveness are the primary goals. To counter this concern the entrepreneurs in the field are in the forefront advocating (cynically, in my opinion) the necessity of 'quality measures' as guarantees of treatment effectiveness. While many of the measures currently in use by corporate bodies are nothing more than 'customer satisfaction surveys,' like those used in the marketing of commercial products, some corporations are now moving in a more sophisticated direction in response to pressures from knowledgeable experts in the field of treatment outcome research. We have seen many press releases, advertisements, and statements by health care corporation executives that glibly proclaim the advent of the development of satisfactory quality indicators to satisfy their 'customers.' However, anyone who carefully reads this book will realize that such statements are not yet based on fact. The field of measuring treatment outcome is extremely complex, difficult, and timeconsuming. It is a science at its earliest beginnings. This knowledge should protect us from the sales pitches of the hucksters of health care and yet give us hope for a future when we can look forward to a more mature science which will be capable of giving us meaningful and useful information on an important area of our lives. NATHAN EPSTEIN is Professor Emeritus in the Department of Psychiatry and Human Behavior at Brown University, and Psychiatrist-in-Chief at St Luke's Hospital in New Bedford, Massachusetts.

Preface

Jerome Frank and Pal Halmos, two eminent scholars of psychotherapy, respectively described psychotherapy as a myth and an act of faith. We choose our method, be it with doctoring or some Western model of psychotherapy, based on our own indoctrination and belief in the effectiveness of that particular approach. This indoctrination, insofar as choice of psychotherapy in the West is concerned, is in significant measure a function of training. Psychotherapy, unlike some aspects of medicine, has little basis in science. The disdain of many psychotherapists for good science may require a book in itself. Suffice it to say that modern science, rooted in reductionism, is considered antithetical to family therapy. Medicine, with its deep scientific tradition, and humanistic psychotherapy, of which family therapy is a part, are somehow viewed as antagonistic. The history of family therapy is interesting in that the early proponents of this approach found justification in the biological science of systems theory for treating the whole family instead of individuals. Having proclaimed that systems theory, or a semblance of it, was the basis of new-found solutions to a multitude of human vicissitudes, they basically ignored the need to test its efficacy. Pioneers are not expected to prove the effectiveness of their discoveries. And in the true tradition of psychotherapy, disciples flocked to this and that temple of family therapy with the same passion with which therapists of bygone eras embraced Freudian, Adlerian, Jungian, and Kleinian theories. These disciples, and we among them, did not need any proof of efficacy. In fact, in the early days of family therapy, the prevailing wisdom was that this endeavour was so novel and so complex that existing methods of outcome research were simply inadequate - or even irrelevant. Again, the faith of the early proponents of this novel treatment was absolute. It was almost sacrilege to raise questions or express doubts about family

xii

Preface

therapy as the treatment of choice. Even today literature on the efficacy of this treatment is scarce. Indeed, family therapy was offered and accepted as a panacea for an incredibly wide range of psychiatric and emotional problems. It would be erroneous to think that the practitioners of this ' art form' are any more in doubt about its efficacy today than they were twenty years ago. As practising family therapists and university teachers, we are constantly amazed by the fervour with which our students become new disciples of family therapy. Despite the fact that over the past two decades a discernible body of family therapy outcome literature has emerged, many students continue to show an almost instinctive distrust of research findings. Yet, that literature, without a doubt, has begun to come of age. Although family therapy is far from the panacea that it was once thought to be, it certainly deserves a respectable place among the major psychotherapies. The motivation to write this book is directly attributable to the questions and views expressed over the years by our family therapy students. In general, the students fall into two groups. First, those who show a remarkable critical faculty for any new information. They pose questions that challenge many of the basic assumptions held by family therapists. The students in the second group not only show an uncritical acceptance of family therapy, they become its staunchest defenders. We have been intrigued and influenced by both perspectives. Our quest in this book is straightforward. We set out to find for ourselves and for our students how effective family therapy is. Does it really do all the things our elders and betters have been telling us for so long (for instance, the clinical literature is almost completely devoid of reports on unsuccessful family therapy)? Our answer to that question is rather long and somewhat laborious. This book may not be an ideal bed-time read. In general, as an endeavour, family therapy has proven to be remarkably effective where we least expected it, such as with schizophrenia (not the double-bind perspective, but the work of those like Leff and Falloon), alcoholism (Mccrady), and juvenile status offenders. In another respect, while the outcome research may be wanting in areas such as anorexia nervosa or placement prevention in child welfare, the influence of family therapy on the overall treatment and management of these complex situations is very impressive. On the negative side, we are disappointed that family therapy outcome research is only in its infancy in the treatment of medical problems and

Preface xiii adolescent drug abuse. It is hoped that this book will reveal the complex nature of our findings. In reviewing the literature we have taken certain liberties. In the main we adopted a pragmatic approach. Our focus was to examine the outcome of systems-based family therapy. By and large we succeeded in adhering to that principle, but not strictly. In part we were not able to do so because many investigators simply did not describe the conceptual basis of the therapy under investigation, and we had to decide whether the family therapy approach was systems based. In others we deliberately incorporated studies that were not strictly systems-based family therapy, because of their intrinsic merit. We made a conscious effort, with minor slippages, to exclude purely behavioural approaches to family therapy. While we were cognizant of the value of the methodological soundness of the studies, and we used conventional yardsticks for making that assessment, we did include several poorly designed early studies to show the historical antecedents to methodologically superior later investigations. Many, if not most, of the early studies were uncontrolled and could be more aptly described as clinical reports with rudimentary evaluative components. Nevertheless, in a real sense, they laid the foundation for outcome research in family therapy . We also took liberties with the definition of family therapy. Just to demonstrate the scope of family involvement, we found compelling reasons to include a few studies involving, for example, family groups. By and large, however, in assessing outcome we subscribed to the conventional notion of family therapy. Despite the digressions the focus of this book is a critical assessment of the outcome of systemic family therapy. While we included uncontrolled studies, in drawing our conclusions we placed considerably more weight on controlled studies. We did not confine the literature review to any particular period. Our goal was to include the early as well as the most recent literature available. It is, however, possible that we have failed to include all relevant literature. We excluded nonEnglish publications because of the prohibitive cost of translation. We apologize for any other omission. This book has ten chapters divided into three parts. The first part, background issues, is a single chapter that provides an overview of the various schools of family therapy. Part 2, comprising three chapters, is concerned with assessing outcome studies of family therapy on the basis of life-stage issues. These chapters examine the outcome litera-

xiv Preface ture of family treatment with children, adolescents, and adults. The third part has six chapters and is a review of the outcome of a whole host of problems treated by family therapy. This section covers the following topics: psychosomatic and medical conditions, alcoholism, anorexia nervosa, drug abuse, and placement prevention. The final chapter provides an opportunity for us to take an overview of the matters we have discussed throughout this volume and offer our thoughts on future directions for outcome studies. Our hope is that this book will be especially useful for practitioners and students of family therapy. Our goal is to bridge the gap between practitioners and researchers and heighten practitioners' awareness of the strengths and weaknesses of this method of treatment. We do so in the belief that family therapy should not be just theory driven, and practitioners should have empirical support for what they do. Researchers may find the book useful as a source of reference. We have tried to keep to a minimum the use of research jargon and write in plain English to make the information more accessible to nonresearchers. We received a grant to support our research for this book from the office of Dr Terence Hogan, Vice-President of Research, University of Manitoba. Dean Harry Specht and Professor Richard Barth at the School of Social Welfare, University of California, Berkeley, made it possible for H.F. to complete portions of this book while enjoying Berkeley as a visiting scholar. They have our deep appreciation. Ms. Sharon Tully, librarian at the University of Manitoba, conducted much of the literature search for this project. This was a very complex and time-consuming process, and we thank her for her understanding and infinite patience. Our friends Brian Minty of the Department of Psychiatry, University of Manchester, England, Diane Hiebert-Murphy of the Faculty of Social Work, University of Manitoba, and Richard Barth, School of Social Welfare, University of California, Berkeley, read parts of the manuscript and gave us valuable advice, and we thank them for their support. Margaret Roy read the manuscript and gave us the benefit of her wisdom. We reserve our deepest gratitude for the students and clients we have had the privilege of working with for something like thirty-five years between the two ofus, who compel us to examine and re-examine what we know, and who, in the final analysis, are our best teachers. This book is dedicated to them. This book would not have been possible without the unfailing support of our families. R.R.'s two children are grown up and have their

Preface xv own homes; his wife Margaret was, as with all his projects, his principal adviser. Sandra, Julian, and Jessie patiently (mostly) allowed H.F. the time to work on this project, even when perhaps his attentions should have been focused on other pursuits. We would like to express our deep gratitude to Professor Nathan B. Epstein, who counts among the pioneers of family therapy, for agreeing to write the Foreword. This is especially gratifying for R.R. who had the privilege of being a member of Professor Epstein's Department at McMaster University, Hamilton, Canada. Professor Epstein was the founding Head of the Department of Psychiatry at McMaster University. Finally, it will be a gross oversight ifwe failed to express our gratitude to Virgil Duff, Executive Editor of the University of Toronto Press, whose friendship, advice, and guidance in this and previous projects have proved to be invaluable.

HOW GOOD IS FAMILY THERAPY?

I

Family Therapy: An Overview

Family therapy is a widely used method of psychotherapy that has gained enormous popularity over the past two decades or so. It is not that working with families was unknown among professional mental health workers before the emergence of systems-based family therapy. Social work has a time-honoured tradition of making the family and its environment the focus of intervention (Broderick and Schrader 1992). The new element introduced by the 'pioneers' of the family therapy field was a radically new theory of etiology to interpret and treat psychological, psychiatric, and interpersonal problems. Simply put, the contention was that symptoms were the products of family dysfunction in one way or another, and, therefore, intervention had to be with the family rather than the symptom carrier alone. The other radical departure lay in the rejection of a Cartesian formulation of a reductionistic (linear cause-effect) approach to science. Systems theory, at the heart of which is circular rather than linear causality, is the scientific underpinning of family therapy. The definition of system most widely accepted is Miller's (1965) : a system is a series of elements organized in enduring and consistent relationship with each other. Steinglass, after a thorough analysis of the pros and cons of the relevance of systems theory to family therapy, concluded that 'in the end, researchers and clinicians work with families because they believe that these naturally occurring behavioural systems cannot be ignored. Families clearly fit the most popular working definition of a system - even if, at times, it is hard to identify all the units that should be included within a particular family' (1987: 32). Wynne (1988) put forth a somewhat different view on the question of the application of 'systems' to family. He acknowledged the viability of

4 How Good Is Family Therapy? the concept, but added that 'a great deal of conceptual clarification was still needed. Systems concepts have all too often been used so carelessly that communication among theorists, clinicians, and researchers has been seriously impaired'(1988: 270). Furthermore, in the same vein as Dell (1986), who noted that 'clinicians know (experientially) that A does lead to B: that mothers do control families and Dr. Minuchin does change families, ' Wynne articulated the problems inherent in the notion of circularity, especially from a research point of view. His main objection to blind adherence to circular causality arose from the recognition that it ignored the 'time line' factor, 'which prevents truly circular processes from ever occurring' (ibid.: 271). Wynne's observations are of great import for researchers who struggle with methods that may truly incorporate 'circular' causality and, in the main, find the task hazardous and elusive. The major ramification for adopting the systems perspective was that without the focus on cause and effect (dependent and independent variables) new research methodologies had to be found to explain the symptomatic behaviour in the context of a 'circular' pattern of interaction between family members that created and maintained such behaviours. The conceptualization of psychosomatic families by Minuchin, Rosman, and Baker (1978) is an example of such a methodology, which continues to come under the close scrutiny ofresearchers (it is noteworthy, however, that Minuchin is directive and lineal in therapeutic situations and seemingly effective). Second, the relevance of applying traditional research tools to evaluate the outcome of treatment (family therapy) had to be re-examined. The debate that surrounds the selection of appropriate statistical techniques to assess outcome, which will be discussed presently, is furious indeed. These two issues continue to draw much attention from family therapy researchers. While researchers continue to struggle with these highly critical issues, practitioners remain mainly unconcerned. Complex reasons account for their indifference. Family therapy in this respect shares the same reality with psychoanalyses from another era. Practitioners have historically demonstrated a propensity for 'buying' into new and innovative methods of understanding and resolving psychological and psychiatric maladies. This led the British sociologist Paul Halmos (1965) and American psychiatrist Jerome Frank (1979) to observe that psychotherapists tended to operate on unshakeable faith. They believed in the infallibility of their approach. The foundation of their

Family Therapy: An Overview 5 faith was rooted in their unconditional acceptance of the words and preaching of their 'gurus.' Family therapy is no exception to that phenomenon, but is it where it stands? Werry (1989), in a somewhat overstated case against family therapy, proclaimed family therapy 'as a rather sad sack relying for its status largely on assertion, selfcongratulation, guruism and denigration of alternatives. Like psychoanalysis, family therapy resembles a religion rather than a proper professional endeavour.' This is an ungenerous view of the current state of family therapy research, but Werry's scepticism is not entirely devoid of merit. Family therapy establishments conduct 'master' classes, the purpose of which is to impress the novice and the believers with the 'dazzling moves' and use of 'incredible metaphors' by the 'masters.' This kind of activity more appropriately belongs in the arena of the performing arts than as a treatment method rooted in the scientific assumptions of systems theory. Werry, however, commits the common error of taking a static view of the field. Indeed, family therapy has its share of 'fathers and mothers.' Their task was to develop and propagate a radically new way of therapy, which they achieved with incredible success. The task of the next generation of researchers (believers and non-believers alike) and practitioners is to begin to ask the hard questions about why family therapy works and how effective it is. As this book purports to demonstrate, such questions are being asked with increasing regularity. The question of etiology (which is somewhat outside the scope of this book) in family therapy has been ' resolved' by assigning low priority to it (Roy 1987). Questions, however, persist. Is circular etiology (note that the term 'circularity' has given way to 'recursive' by some) subject to disconfirmation? What constitutes a family? Who is to be included and who is to be left out of therapy? How definable an act is family therapy? Several schools of therapy have emerged. Is one school superior to another? How much of family therapy is 'art' and how much is science? What is to be considered a favourable outcome, removal of the symptom, which curiously resurrects the question of etiology, or 'improved' family functioning, or both? What are the selection criteria for family therapy, or is it more of a panacea for all conditions ranging from childhood asthma to chronic pain to drug addiction to suicidal adolescents to physical and sexual abuse of women and children? If the scientific underpinning for family therapy is systemic, should other systems (besides the family) be included in treatment and, hence, in measuring outcome? How critical are therapist-related

6 How Good Is Family Therapy? variables in outcome research? How does family therapy compare with other psychotherapies or pharmacotherapy in certain disorders? In the remainder of the chapter we examine the conclusions of some of the major reviews of family therapy outcome research. Conceptual and methodological problems as well as omissions identified by the reviewers themselves will be highlighted. Against that broad background, the rest of the book will consider the current state of outcome research in selected areas. Models of Family Therapy- Myth or Reality

The late 1960s and 1970s witnessed an enormous proliferation of schools of family therapy. Indeed, history may judge those years as the heyday of family therapy, when the field was crowded with innovators, who each created their own particular brand of family therapy. Bowen, Haley, Minuchin, Palazzoli, and others laid claims to their version of family therapy. Those differing approaches became the 'schools' and 'models' within their own training programs and created legions of strict adherents. Minuchin (1982), in his autobiographical ruminations, described this era as one of building castles by family therapists. But maintaining the castles was proving too expensive, and soon they would be empty, he concluded. In the meantime, schools are merging and eclecticism and common sense is replacing, albeit slowly and grudgingly, rigid subscription to this school or that. Minuchin's structural and Haley's strategic schools probably gained more popularity than all the others. How similar or different are these schools? This is an important research question because of the rigour necessary to operationally define family therapy. Are these two schools totally differentiated, or are there important theoretical commonalities? Are the differences merely in appearances, or are they substantive? ls it possible, in practice, to be strictly one or the other, or is it even desirable? To pursue the question of the operational definition(s) used by researchers, we shall examine the review literature on family therapy outcome to gain a broad understanding of how this thorny issue has been tackled. In their most exhaustive review, Gurman and Kniskern (1978) made an interesting and meaningful distinction between behavioural and non-behavioural methods of marital therapy. It is an interesting distinction, as it provides a simple way of grouping divergent nonbehavioural methods. It should be noted, however, that most non-

Family Therapy: An Overview 7 behavioural family and couple therapies are directive as opposed to non-directive, and behavioural techniques are embedded in them. In a subsequent article, Gurman (1988) noted that 'there is a strong movement afoot in the family therapy field toward integration and rapprochement.' Some of the common 'ingredients' shared generally by the field of family therapy, he noted, were: (1) The process of reframing, whereby the presenting individual problem is reframed in the systemic context; (2) focus on communication; (3) alternative modes of problem solving; (4) modification of hierarchical incongruities and generational boundaries; and (5) modification of aversive interpersonal behaviours. In short, at least at the conceptual level, the divergent methods of family therapy shared some common elements. These methods employed very similar therapeutic (direct and indirect) moves and defined therapeutic success in identical terms. DeWitt (1978) resolved the dilemma of divergent techniques by drawing a distinction between conjoint and non-conjoint family therapy in her review of four previously published reviews of family therapy outcome studies. Conjoint family therapy meant that 'all relevant members of the family must have been treated together as a unit for all or major portion of treatment.' This tendency to group together under a broad rubric various 'schools' or 'models' of family therapy found justification in a review of ten previous reviews of family therapy outcome literature by Hazelrigg, Harris, and Boruin (1987). The basis of their argument was that despite differences between schools, family therapies were rooted in systems theory (which is at the heart of Gurman's viewpoint) and defined measures of successful outcome in a relatively uniform way. Hence, they argued that 'there is common ground that justifies combining the results of studies from different perspectives ... The methods of intervening may differ among family therapists, but the overall goal is one of systemic change' (ibid.: 430). In fact, these authors did not provide any information about the different methods of family therapy employed by researchers. Russell and her colleagues (1983), in reviewing family therapy outcome studies, did not directly raise the question of the divergent family therapy techniques, but addressed a very fundamental problem, namely, that regardless of the school, the goal of treatment was the removal of the presenting problem at the expense of correcting faulty family dynamics. The point of note is that outcome measure(s) was independent of the models or schools of family therapy, again provid-

8 How Good Is Family Therapy? ing a common ground for the varied schools. Epstein and Volk (1981), in an otherwise excellent review of outcomes of psychotherapy, repeated the distinction drawn by Gurman and Kniskem (1978) between behavioural and non-behavioural family therapies. All 'nonbehavioural' family therapies were assumed (since the point was not made) to encompass the same, if not, similar processes. LeBow (1981) took exception to the trend and stated that 'most of the outcome literature, however, has failed to discriminate between family therapy approaches.' He was unconvinced that Bowen's, Haley's, Guerney's, and Whittaker's approaches could be generalized as a single approach. Epstein (1988), in contrast to his earlier opinion, echoed Lebow's objection and stated that 'the most immediate priority is the development of rigorous, highly specific, and operationalized methods of family therapy' (ibid.: 121). Even the early reviewers were conscious of the challenge posed by different approaches of family therapy and established arbitrary definitions for the purpose of the review (Wells, Dilkes, and Trivelli 1972). Wells and Dezen (1978) acknowledged a marked increase in the number of outcome research studies between their two reviews, but their concern that 'schools' of family therapy could not be described as 'an entity' persisted. Wynne (1988) furthered the viewpoint (discounting Lebow's and Epstein's objection) of many reviewers by claiming that family therapy was more than a 'modality.' In fact, it was an approach to gather and organize information to determine and apply intervention in the context of the family system. Modalities in themselves were not as important as the theoretical underpinning upon which the models rested. The systems approach made the schools more similar than disparate, and, hence, in research the actual model of therapy was not as relevant as its scientific root(s). Family therapy is perhaps moving in the direction of uniformity, away from its exciting and multiple orientations. Or, is it possible that the distinctions between the models, only possible at a high level of abstraction, cannot be readily translated into operational terms, and, thus, some sort of methodological compromise is unavoidable? Difficulties associated with treatment variables were outlined by Guerney (1985), who noted that factors such as one family therapy versus multiple-family therapy, limited number of sessions versus openended treatment, marathon versus fifty-minute sessions, and, Guerney might have added, one model of family therapy vs another presented enormous challenges to the researchers.

Family Therapy: An Overview 9 As will become evident in the course of this book, researchers continue to employ and test the efficacy of specific methods of family therapy and compare the relative merit of different schools. Clinicians continue to promote and refine their particular schools and methods of family therapy, as a quick glance at the latest edition of the Handbook of Family Therapy will bear out (Gurman and Kniskern 1992). In their comprehensive review Gurman, Kniskern, and Pinsoff (1986) identified a clear trend towards, what they described as, theoretical and technical integration. Yet, they noted an absence of research on such integrative models. Their review is unusual in the sense that they examined the relative merit of 'schools' of family therapy to treat a variety of disorders. Any systematic drive towards integration of divergent methods of family therapy is still in its infancy. Nevertheless, wisdom gleaned from practice suggests that even the purists, in reality, tend to be eclectic and do whatever is in the best interests of the client-system. At an annual meeting of the American Association of Family Therapists, someone noted that Minuchin had made a paradoxical intervention in his live case demonstration, clearly a 'strategic' move, not usually associated with his structural school. Minuchin's response was characteristically unequivocal. In essence, he said, 'Use whatever works.' Kolevzon and Green (1983), in their analysis of models of family therapy, asked, 'How different are, in fact, these models of therapy in their actual practice? More important, are the differences between these models better predictors of therapeutic effectiveness than are the similarities?' (ibid.: 164). Reviews of outcome studies to date provide a partial answer to the first question, that is, researchers have found justification on conceptual grounds for adopting an integrative view. Epstein's (1988: 120) definition of family therapy is worth quoting as it offers a way of linking seemingly varied approaches to family therapy: family therapy is 'a therapeutic approach to working with the family as a system for the purpose of aiding the family members to achieve solutions to problems that interfere with their satisfactory functioning as individuals and as a family unit.' Again, subscription to the notion of the family as a'system' is the unifying theme. Outcome Measures

Family therapy outcome research has had special difficulty in defining and therefore measuring outcome. This problem has many sources,

10 How Good Is Family Therapy? the principal one being the opposition or reluctance to use traditional and therefore reductionistic measures of outcome. Symptom removal which is inevitably individually focused (Epstein resolves this problem by incorporating satisfactory function of the individual in his definition of family therapy), while the avowed measure of success in several of the therapeutic models of family therapy, raises the ire of many for the narrowness of the outcome measure(s). On the other hand, if families seek help with this problem or that, resolution of that problem, be it ketoacidosis for diabetic children or an alcoholic husband, is precisely what the families desire. This focus on symptom removal, in the minds of many, undermines the key objective of family therapy which is to rectify the dysfunctional relationships in a family system. Russell and her colleagues (1983), for example, were critical of Minuchin and his colleagues' choice of outcome measure in their investigation and treatment of anorectic families. Their main objection was that the researchers failed to report 'interaction within marital, parental, and sibling subsystems after therapy.' How valid is this criticism? The central point is that without these changes, improvements they chose to report presumably could not have happened. The issue is one of detail, because the therapy is predicated on a particular theory of etiology. In Minuchin's case the family attributes of psychosomatic families and physiological vulnerability of the identified patient, and remission or eradication of symptoms can only occur when the relationship issues have been successfully treated. Whether one chooses to re"port them is a moot issue. Jacobson (1988), while agreeing that multiple measures must be used in family therapy outcome research, finds himself in agreement with the view that 'ultimately a successful outcome means that the presenting (or emergent problem) has been eliminated' (ibid.: 153). He contends that 'changes in family interaction are often the hypothesised means whereby the family therapist proposes to solve the problems that brought the family in' (ibid.) . He goes further when he states that 'except in those instances where dysfunctional family interaction is the presenting problem, measures of family interaction are at best indirect measures of treatment outcome' (ibid.). The means and the ends are not to be confused. In keeping with Jacobson's proposition, Olson (1988) proposed multiple-level goals for second order change determined by the nature of the presenting problem. For an individual psychiatric symptom, the desirable change is elimination of the symptom. In contrast, for a family

Family Therapy: An Overview 11 confronted with family and extended family issues, the goal of treatment is changing the type of family system. The solution offered by Olson as well as Jacobson is unsatisfactory to many because of the prevailing view that a symptom is not a symptom, but a mere outward sign of deeper family conflicts. Guerney (1985), for example, preferred a very inclusive set of outcome variables that would include symptom eradication and factors such as adequacy of communication, distribution and mechanism of power, and adequacy of family conflict resolution. We shall now briefly examine the measures of outcome as reported by reviewers of outcome studies in family therapy. Hazelrigg, Harris, and Boruin (1987) identified several problems with outcome measures (dependent variables) in their very careful analysis of outcome studies. The concept of change, they contended, was often narrowly based on measures such as parent ratings of patient behaviour or experimenter ratings of patient behaviour or recividism. They felt that 'considerable progress could be made in evaluating family therapies through the increased use of multiple change indices.' They believed that broadly based measures by ensuring overlap would facilitate comparison between studies. The problem is not just one of overinclusiveness of outcome measures, but rather some agreement among researchers and clinicians about what constitutes 'success' in family therapy. Wells, Dilkes, and Trivelli (1972) divided their review of outcome studies into inadequate, borderline, and adequate based on some methodological considerations. Only two studies met their criteria of adequate (inclusion of control group and specifit outcome measures). They were impressed by the outcome measures employed by the investigators who included hospitalization rates and days lost for functioning for psychiatric patients following crisis-oriented family therapy. These outcome measures were patient focused. In fifteen studies that were deemed inadequate, outcome measures, in the main, were self-reports by clients and therapist evaluation. To what extent outcome measures have been refined might be assessed by a review conducted by Wells and Dezen (1978) several years later. The 1978 review covered the period between 1950 and 1970. The 1978 review was for the six-year period 1971 to 1976, and it examined uncontrolled reports, comparisons between treatment and no treatment, and comparisons between alternative treatments. Some of the old criticisms were still applicable to the uncontrolled studies, such as poorly defined outcome measures. However, a few uncontrolled studies - for example, Minuchin and his associates' study of psychoso-

12 How Good Is Family Therapy? matic families and a report from the McMaster group - despite lack of controls, were better designed as multiple and standard measures were used to assess outcome, a phenomenon almost totally absent in the 1972 review. Wells and Dezen (1978) found nine studies in the category of treatment versus no treatment and seventeen studies comparing family therapy with alternative treatments. These studies were clear signs of significant improvements in the methodology of outcome research. The outcome measures included pre-post measures of family functioning; improvement in presenting symptoms; improvement ratings by therapists, mothers, and independent clinicians; and the follow-up rate of recividism. In short, despite persisting methodological shortcomings, significant progress in refining outcome measures was evident in the early 1970s. DeWitt (1978) combined the results of four previous reviews of family therapy outcome studies and added studies that had been missed by previous reviewers. While a wide variety of outcome measures were used in the studies reviewed by DeWitt, she remained concerned that too much attention was paid to the characteristics of the identified patient and not enough to those of the family. Studies that investigated treatment efficacy in making changes at the level of the family system did not always purport to measure such changes. In other respects, DeWitt's conclusions about the continuing problems in measuring outcome were not much different from Wells and Dezen's. In a very thoughtful analysis of methodological problems confronting outcome research in family therapy, Lebow (1981) concluded that the adoption of widely accepted packages (for evaluation of outcome) was not very likely in the immediate future, yet some corrective measures could be taken to deal with some of the glaring gaps. For instance, Lebow advocated use of multiple perspectives and multiple criteria by which to assess change: 'In a thorough assessment, some measures should be included from all three perspectives (patient, therapist, rater), and measures should focus upon multiple aspects of change at both the identified patient and the system level.' The more standardized the measures are, the more feasible it is to conduct comparisons between studies. In assessing outcome Lebow emphasized the timing aspect of follow-up assessment, as many short-term gains may not last or effects of treatment may reveal themselves only after some time. His main criticism was that follow-up periods had been very short.

Family Therapy: An Overview 13 Russell and her associates (1983) also expressed dissatisfaction with the focus of the outcome research being on removal of presenting symptoms. They bemoaned the fact that 'few research teams have gone beyond the traditional focus on presenting problems to link system dynamics to treatment goals and subsequent outcome.' Their suggestions for improving the quality of research were in line with Lebow's. They proposed a single-case design with each family serving as its own control and repeated measures of family interaction which would eliminate some of the problems of traditional pre-post design. In essence, however, Russell and associates were proposing that outcome had to be examined in the context of the family rather than the individual's symptoms. As was noted earlier, these two perspectives are not mutually exclusive. Hazelrigg, Harris, and Borduin (1987) departed from the traditional approach to literature review and conducted an integrated statistical analysis of outcome studies in family therapy. Their findings confirmed some of the observations of earlier reviewers that in the first place outcome tended to be measured on some narrow parameters such as recidivism or experimenter ratings of patient behaviour, and, second, that use of multiple change indices was overdue. They found that only eight studies out of twenty included in their review used multiple measures. They reiterated the point made by others that without the benefit of multiple indices, comparisons between studies would remain onerous tasks. There appears to be some degree of agreement among the reviewers that the measures of outcome of family therapy, while gaining sophistication over time, remain problematic. The narrowness of outcome measures is a constant theme in almost all the reviews. The argument in favour of multiple indices for measuring change emanates from several sources. One is ideological, namely, that without the benefit of the measures for changes at the systemic level, and not just at the level of the identified patient, there is the ever present risk of betraying the systemic orientation of family therapy, a view that is at variance with that of Wynne, Jacobson, and Olson. Second, on a more positive note, multiple measures and packages of standardized instruments would facilitate more meaningful comparisons of outcome in family therapy. Reviewers seem to suggest, without resorting to categorical statements, that many researchers in the field continue to experience incongruity between the goals of their investigations and the tools they use to measure outcome.

14 How Good Is Family Therapy? Therapist Variables

The centrality of therapist variables was noted by Gurman and Kniskern (1981) now over a decade ago. How much attention and what kind of consideration has this factor received in outcome assessments of family therapy? • Measuring the impact of therapist characteristics remains severely neglected in family therapy outcome research. Epstein (1988) proposed that only the most highly trained therapists should participate in outcome studies, as that will maximize the probability of positive outcome. Jacobson (1988) had a different view on the value of 'experience' and level of training, as he contended that 'although it would be nice if clinical experience were directly related to positive outcome, it is an important but empirical question that can only be answered by directly comparing experienced and inexperienced therapists in the same studies' (ibid.: 151). Furthermore, Jacobson (1988) stated that on the basis of very limited empirical evidence in marital therapy, the hypothesis that experienced therapists produced better results was, at best, equivocal. We now examine whether and how much our reviewers paid attention to the question of therapist variables. Gurman and Kniskern (1978), in the most exhaustive review of the time, found that the key therapist variables that had positive association with outcome were the capacity to build and maintain relationships and male gender. Epstein (1988) in his enumeration of criteria for good outcome research in psychotherapy identified the requirement that the therapist be trained in the appropriate model being investigated, but they were silent on other attributes of the therapists. In Wells's two reviews (1972, 1978) the term 'therapist variables' did not appear for the reason that these variables were, in the main, ignored by researchers. This omission was not lost on Gurman and Kniskern (1978), who in a measured criticism of Wells's 1978 review noted that 'there are many family therapists, we among them, who believe, on clinical grounds, that therapist variables are at least as salient as technique variables.' They supported their belief by pointing out 'the growing body of empirical literature.' Despite their claim of a growing body of literature, the therapist variables continue to receive, at best, scanty attention from researchers. For instance, Russell et al. (1983) and Hazelrigg, Harris, and Borduin (1987) ignored this variable in their otherwise important reviews. This point was not lost on DeWitt

Family Therapy: An Overview

15

(1978). She stated that 'only minimal information is available on the characteristics of the therapists who participated in the conjoint family therapy reported in the studies reported here.' The simple truth is that the studies she reviewed did not generally consider the therapist variables of sufficient significance to merit analysis. Lebow (1981) recognized that in the broader psychotherapy literature the significance of the therapist variables in outcome research was gaining increasing prominence. He noted that factors such as the therapist's level of experience, age, sex, race, and relationship skills have been examined by researchers. In family therapy, he noted, existing data provided limited support for the proposition that more empathic therapists experienced more success. However, Lebow urged that future outcome research should make a more determined effort to delineate the significance of therapist variables. Jacobson's (1988) observations about the difficulties associated with treating 'therapists' as a 'randomized factor' demand attention from researchers. He acknowledged that he had no easy solution to this dilemma. He suggested that 'intramodel comparisons and studies conducted with relatively inexperienced therapists may lend themselves to using the 'therapist' as a between-subject factor, while intermodal comparisons with seasoned therapists may require adherence to current conventional wisdom' (ibid.: 152). In the course of our review, we explore the acceptance of Jacobson's proposition as well as the rigour that researchers have applied to the question of therapist variables.

Data Analysis

Dissatisfaction with statistical methods employed for data analysis is not hard to detect. The criticisms range from absence of any meaningful analysis to the use of inappropriate statistics, absence of descriptive statistics other than group means and standard deviation, and heavy reliance on statistical significance Uacobson 1988). Gurman and Knis kern also observed that 'evidence beyond the statistical significance of changes in group means of the clinical importance of changes achieved by families and individual family members' is necessary (1981: 770-1). For the outcome data to be meaningful, it is important that standard deviations of the change measure(s) used and the frequency distribution of the actual number of improved cases be reported.

16

How Good ls Family Therapy?

This issue has been further complicated by questions about the relevance of conventional analysis, mainly designed to explore linear associations, to situations where there is at least a debate about the absence of such relationships (Steir 1988). Pinsoff noted that 'a correlational analysis is not designed to elucidate non-linear relationship' (1988: 166) and proposed the use of a multivariate correlational strategy which permits the application of multiple regression to assess the 'extent to which combinations of variables, as opposed to single variables, predict outcome' (ibid.: 166). Some of the reviews examined here simply did not address the q uestion of data analysis. Epstein and Volk (1981), for example, recommended that 'adequate size of research sample [be] used. Statistical expertise is now a necessary component of all modem research groups.' In Wells's two reviews (Wells, Dilkes, and Trivelli 1972; Wells and Dezen 1978) problems associated with data analysis were not mentioned. The use of appropriate statistics to measure outcome was not discussed by DeWitt (1978) or Russell et al. (1983) either. LeBow (1981) urged the use of stepwise and multivariate models. Hazelrigg; Harris, and Borduin (1987) reported on the kind of statistical analysis employed by some of the researchers, but they did not provide a critique. They bemoaned the fact that statistical analyses performed my many researchers were not adequately reported. This oversight by the reviewers, who were very focused on the methodological shortcomings of the many studies they reviewed, is not easily explained. Gurman, Kniskern, and Pinsoff (1986) uncovered a conspicuous absence of descriptive statistics in reporting analysis of results. Nevertheless, this omission may reflect the current state of research in family therapy, namely, more immediate and fundamental methodological issues (to be examined in the rest of the book) demand attention, and, mistakenly, the precision with which data analysis needs to be dealt is relegated to a low level of priority. Key Conclusions of the Reviews

Despite all the major methodological shortcomings (including lack of control groups, small and ill-defined samples, lack of attention to therapist variables, inadequate instrumentation, poorly defined outcome measures, confusion over dependent and independent variables, multitude of outcome measures, problems of operational

Family Therapy: An Overview 17 definition of family treatments, and inappropriate and inadequate data analysis) of family therapy outcome research, the consensus among reviewers is that family therapy, with some qualifications, works. Wells, Dilkes, and Trivelli (1972) estimated, based on six uncontrolled studies, that combining the outcomes improvement and some improvement 69 per cent was the overall rate of success for family therapy with adults. With children the rate was 79 per cent. These authors were impressed by the paucity of adequate research design and the glaring omission in family therapy research of control groups, as well as several other methodological limitations. Wells and Dezen (1978) were basically unimpressed in their subsequent review. Only in uncontrolled single-group studies was positive outcome evident. Controlled and comparative studies, in the main, were found equivalent with non-formal and alternative treatments. The authors, however, remained sceptical. They noted that 'numerous methodological and practical difficulties beset the current body of family therapy research.' DeWitt (1978) came to similar conclusions as Wells and Dezen (1978) in that out of thirty-one studies they reviewed, twenty-three uncontrolled studies reported results that were similar to non-conjoint methods, and in the remaining studies, each with a comparison group, five were superior to there being no treatment. DeWitt was almost apologetic for the unimpressive results and pleaded that methodological deficits in almost all the studies could be explained on the basis that the research in this area was just beginning. Gurman and Kniskern (1981) repeated their calculation from a previous review (1978) of positive outcome for 73 per cent of family cases and 65 per cent of marital cases. They adopted a rather generous attitude towards flawed methodology (while at all times acknowledging such flaws), claiming that while methodological problems facing family therapy outcome research were serious, indeed, they went on to say, 'Nevertheless, we believe that, like love, methodological adequacy is not enough.' Prima facie, the association between love and research methods may not be self-evident, the point being love is not always 'logical' nor is it that love alone is enough. Thus, they invoked the notion of 'logical adequacy' to justify their conclusion that the results of poorly designed studies 'are entirely consistent with those of better quality' (1981: 752), and on that basis they posited an optimistic and perhaps exaggerated view of the efficacy of family therapy. Minimally, their ambivalence about methodological rigour (despite their claim to

18 How Good Is Family Therapy? the contrary) runs the risk of contributing to the family therapist's scanty regard for outcome research. Would they have found the same justification for flawed research design if the findings of these studies were unfavourable to family therapy? Gurman, Kniskern, and Pinsoff (1986) found much to praise in the state of outcome research in family therapy. They were rather noncritical of the research methods adopted by many of the studies, which led Raffa, Sypek, and Vogel (1990) to write that 'the reviews by Gurman and colleagues are scholarly, well reasoned, well written, and the most comprehensive in the marital and family literature. We do not believe, however, that they would convince anyone not already convinced of the 'efficacy' of family and marital therapy, given the methodological flaws in the studies as, for example, the general failure to employ control groups.' Russell and her colleagues (1983) also assumed a very uncritical stance of the research methodology in reporting the outcome of various studies related to alcoholism, drug abuse, juvenile offences, and a variety of childhood problems. Measuring outcome in terms of resolution of the presenting problems (used by the majority of the studies reviewed by them) rather than altering systems dynamics was at the heart of their criticism. They proposed, in agreement with Gurman and Kniskern, that 'much can be learned by considering the trends that appear across all imperfect projects.' There is no clear way of judging the extent to which statements of that kind may have hindered methodological advances in family therapy outcome research. Yet, these statements are there for anyone to see. The message seems to be 'if we cannot produce good research, let's make the most of bad research that seems to provide justification for our activities.' Hazelrigg, Harris, and Borduin (1987) reported the relative superiority of family therapy over situations of no therapy and alternative types of treatment. However, their conclusion was that the positive findings were tentative at best. They noted that most of the reports they reviewed involved families of children with behavioural problems despite the claim that family therapy was beneficial for nearly any client groups and that family therapies were 'slightly' superior to alternative treatments. It is noteworthy that only twenty out of 290 outcome studies met their criteria for inclusion which among other factors included a matched control group of either no treatment or alternative treatments. The other criteria included definition of a family as a minimum of a child and a parent, minimum of five families in each group,

Family Therapy: An Overview 19 complete reporting of statistical analyses and results, and the therapy process clearly meeting the definition of family therapy. Discussion This brief review is encouraging and discouraging. Indeed, the quality and quantity of research has improved significantly over the past two decades. Yet, there is considerable dismay expressed by many reviewers over the poor research design. The relative newness of the field of family therapy research was legitimately proffered by early reviewers as a viable reason for poor quality. Unfortunately, as recently as 1987 Hazelrigg and his associates resurrected the same reason in the face of almost insurmountable problems they faced trying to find a few studies that would meet the inclusion criteria of their review. Perhaps, time is upon us to look beyond the 'newness' issue. Raffa, Sypek, and Vogel (1990) have done just that. They summarized the criticisms of the outcome literature: (1) the paucity of controlled studies; (2) failure to describe the treatment techniques; (3) inadequate research design, beyond the issue on controls; (4) poor data analyses; (5) incongruity between treatment techniques employed and theoretical framework; (6) use of outcome measures lacking in reliability and validity; and (7) the total absence of effort to replicate positive outcome. This is a comprehensive list, but it fails to address the underlying reasons for the perpetuation of poor quality outcome research. The notion that somehow the uniqueness of family therapy has rendered the established methods of outcome research meaningless has, to some extent, contributed to sloppy designs, and the same reason may account for the publication of inadequate studies. Prominent investigators have found justification for poor research design and, as we pointed out, justified such research on rather questionable grounds. Raffa, Sypek, and Vogel (1990) did try to explain some of the shortcomings so commonly found in family outcome research. First, they claimed that the current state of the field of family therapy 'makes it impossible to construct meaningful control groups.' They attribute this difficulty to an absence of taxonomy for family pathology. In our view this is a serious problem, but not an insurmountable one. Minuchin's research on psychosomatic families, with a relatively precise description of family characteristics and very precise diagnosis, such as childhood asthma, makes it possible to establish adequate control groups for alternate treatments (since asthmatic children cannot be

20

How Good Is Family Therapy?

left untreated); we shall examine this question fully in a subsequent chapter. While the question of 'controls' is a serious omission, there are other equally serious problems. To pursue those questions we turn to Wynne (1988), who edited a volume based on the reports presented at a joint meeting of Family Process and the National Institute of Mental Health, in the United States. The scope of the research problems related to outcome research in family therapy was very substantial. That seemed to be the general sentiment of the participants in that conference. The catalogue of methodological problems recognized by the early critics was still to be easily found in much of current outcome literature. Family therapy outcome research has been beset by real and imagined issues, the central one being the relevance of established research methods to family therapy research. That is no excuse for a plethora of poor quality research which is the halhnark of much of family therapy outcome research. The point is that in the name of inapplicability of existing methods, great deficiencies have crept into the design of family research studies. Attention to simple design issues such as the clear establishment of dependent variables, definition of independent variables with some degree of clarity, careful instrumentation, control and comparison groups, and appropriate data analysis would vastly improve the quality of research and in the process reveal the shortcomings of conventional methods when applied to family therapy. The point is that to date relatively few studies are sound, judged against the criteria of existing methodology. In other words, the usefulness of current research techniques has been, to some extent, prejudged. In essence, the participants of that conference expressed collective concern with the state of family therapy outcome research and began the hard process of searching for solutions. Conclusion

Our effort in this chapter has been to paint the field with a broad brush. To that end we used, in the main, reports of reviews in family therapy outcome research. There is little doubt that the sheer volume of such research has increased at a phenomenal rate since the early 1970s. The quality has improved, but not nearly as phenomenally. There are early signs that resolution to some of the problems is in sight, problems such as grouping together a variety of schools and models of family therapy, tailoring outcome to the nature of the difficulties, using multiple mea-

Family Therapy: An Overview 21 sures, increasing use of control groups, and improved techniques of data analysis. There is also an emerging consensus that while the findings are encouraging, and indeed very promising in some instances, rigorous methodology must be applied if family therapy is to go beyond the stage of being a mere act of faith.

2

Children and Family Therapy

The child guidance movement in the United Kingdom and the United States was the precursor to family therapy. Ever since Freud there has been a rising awareness and mounting knowledge that children's emotional and psychological problems are deeply rooted in their relationship with their parents in general and mothers in particular. The mother-child relationship, from Sigmund Freud to Michael Rutter, has been at the centre of research and clinical exploration to explain childhood emotional problems. Systemic family therapy has extended the focus on the mother-child dyad to include all intimates in the lives of children. Logic would dictate that a family-based approach to treat children's psychological and emotional problems would be superior to any other types of treatment. In this chapter we explore the success of family therapy in reducing or ameliorating emotional conflicts and their behavioural manifestations in children. The effectiveness of family therapy with children has come under periodic review; it should be noted that we cover some of this literature under specific topics, such as psychosomatic problems in children (Chapter 5). Masten (1979) in her review concluded that 'there are major shortcomings in most of the available data, with only two well-controlled studies [in our judgment these two studies were far from well controlled). If the value of family therapy as a treatment alternative or, ideally, as the treatment of choice for a referred individual child is to be established, more and better controlled comparative outcome studies will be necessary.' Kirkby and Smyrnios (1990) reached a similar conclusion. They observed that 'carefully designed studies of psychotherapy with children and families are rare.' This is despite the fact that they .based their review on nine studies that

Children and Family Therapy 23 compared brief family therapy with an alternative method of intervention. We have excluded parent training type programs from our review. That topic merits consideration on its own. A sample of early studies, studies with patients on a waiting list or with no treatment groups as controls, and studies comparing family therapy with other methods of intervention are the focus for this chapter.

Early Studies In 1967 Martin discussed four cases. This study was very much in the nature of a pilot. Two 'experimental' families were compared with two cases on the waiting list. No formal comparisons were made and statistical analysis was altogether absent. Martin reported improvement in the children's school behaviour. The focus of intervention was to improve the family 's communication skills. This report was so preliminary that it would be best looked upon as a clinical report with a nominal evaluative component. In a retrospective study of thirty-five children, ten of whom were adolescents, fifteen retarded, and the remaining ten described as antisocial, Masters (1978) found that family therapy, by and large, proved ineffective. Seven of the ten families , despite careful screening, dropped out. Masters offered several reasons for the drop-out rate, one being the families ' poor socioeconomic status. Second, 'very poor interpersonal relationships proved a deterrent to the individual family members to cope with their problems in an open and objective manner.' Again, it would be erroneous to read more into this report than the experience of one practitioner. Santa-Barbara and associates (1979) reported a complex outcome study involving 279 families and eighty family therapists including staff and students. The average number of family sessions was six, ranging from two to thirteen or more. The children's problems were academic and/or behavioural, as identified by the family members. A battery of outcome measures was used to assess intellectual functioning, academic achievement, and disruptive school behaviour of the children. Studies of outcome at treatment closure showed that 79 per cent of families registered moderate or great improvement on the basis of therapist evaluation, but 45 per cent of families were rated as having a good or excellent prognosis. Seventy-six per cent of the families closed satisfactorily on the basis of the outcome measures. At the six-month

24

How Good Is Family Therapy?

follow-up 79 per cent of the subjects' original problems were better or much better. This was a complicated large-scale study, and, despite lack of a control group and a large number of therapists with varied levels of training, it was conducted with a great deal of care for methodological soundness. The results left little doubt about the benefit of treating the whole family in cases with children manifesting academic and/or behavioural problems. This study, at the very least, provided much incentive for the adoption of family therapy and was an early contrbution to family therapy outcome research. The final uncontrolled study we wish to report involved forty-three families with forty-five children who were involved in stealing (Seymour and Epston 1989). The families were seen an average of 3.3 times. A few of the children were in their teens, but over 80 per cent were under thirteen years. Outcome was assessed strictly on the basis of recurrence of stealing behaviour. At a two-month telephone follow-up interview by the therapist with a family member there were no instances of stealing in 54 per cent of the children. At the eight- to twelve-month follow-up that figure rose to 62 per cent; this information, too, was obtained by telephone from a family member. Seymour and Epston's study had several basic methodological problems. Families were not formally assessed for their functioning. Children were not assessed for their emotional or academic problems. The method of family therapy was not described, and therapy was conducted by the investigators. Outcome was reported in global terms, with no statistical analyses performed. Comments: It should be noted that some reports of family therapy with children are discussed in the chapter on family therapy with psychosomatic disorders. Dulcan (1984) and Smyrnios and Kirkby (1989) in their reviews of the family therapy literature included several reports of anecdotal accounts and uncontrolled studies of family therapy. The reports discussed in this section quite clearly convey the breadth of application of family therapy to treat children's problems. Despite the fact that the findings taken together were very encouraging, they did not truly reflect the efficacy of family therapy. Uncontrolled studies are, at best, suspect and at worst, misleading. Other than in the study by Santa-Barbara and colleagues, very meagre attention was given to even basic methodological considerations. Outcome measures were rarely objective. Therapy itself was often described in very general terms, and frequently the nature of involvement of the participants in

Children and Family Therapy 25 the therapeutic process was unclear. For the most part even rudimentary statistical analyses were absent. Under these conditions this group of uncontrolled reports were presumably biased and the findings somewhat overly optimistic. This last point receives much credence in the light of the findings of controlled studies. Nevertheless, these early uncontrolled reports probably contributed to greater use of family therapy with children and to vastly improved methods in future studies. Controlled Studies

Treatment versus No Treatment In the early literature controlled studies were virtually non-existent. Some were described as controlled, but fell far short of basic methodological requirements. The only study that compared family therapy with a no treatment group was that by Black and Urbanowicz (1987). They reported on thirty-eight bereaved children from twenty-one families who received family therapy in comparison with a no treatment control group of forty-five bereaved children from twenty-five families. Whether the groups were closely matched remained unclear because no statistical data were provided. At the time of bereavement the children were between the ages of under five years and twelve years or older. Outcome measures included parental depression and a variety of measures of the children's home and school behaviour as well as their health. Unfortunately, the measures used were not described by Black and Urbanowicz and were reported only in the follow-up data. Family therapy was provided by psychiatric social workers. Six therapy sessions were offered, spaced at two- to three-week intervals. The sessions took place at home two to three months after bereavement. The family therapy 'method' was not stated, although the description would suggest a systemic approach. Outcome at one year of follow-up, which involved twenty-one children, revealed that the treatment group did slightly better than the control group in terms of behaviour, mood, and health. At the two-year follow-up, which involved twenty-one children, the findings were in the same direction as earlier. The conclusion was that the treatment group had benefited from short-term family therapy. The major problem with the study was with the format of the report. While references were made to the Rutter A Scale, there was no men-

26

How Good ls Family Therapy?

tion of it in the text, let alone a description. Similarly, the method used to measure parental depression was not stated. The whole question of instrumentation was ignored, and family functioning issues were not addressed. The statistical analysis was kept to a minimum. For example, no attempt was made to locate the factors that could predict outcome, and there was no description of the statistical methods used for data analysis. This study had considerable scope, but through incomplete reporting and inadequate analysis its quality was compromised. Yet, the key finding was that family therapy with bereaved children was beneficial. This study lent support to the findings of the uncontrolled reports. The conclusion was that providing family therapy appears to be superior to doing nothing. In the following section the question of superiority of family therapy over other treatment methods will be examined. Family Therapy versus Other Treatments

Under this heading two main types of studies can be reported. First, comparative studies involving either structural family therapy (SFTl) or strategic family therapy (SFT2) versus another kind of therapy. Structural Family Therapy versus Other Treatments Ritterman (1978) conducted a complicated double-blind study, as her doctoral dissertation, to test the value of family therapy to treat hyperactivity in children. Forty children, with a confirmed diagnosis of hyperactivity and a mean age of eight years and four months were randomly assigned to four treatment conditions: (a) family therapy (based on a structural paradigm) alone; (b) combined family therapy and Ritalin; (c) family therapy and placebo; and (d) Ritalin alone. Children were given an exhaustive battery of pretreatment tests and the family completed a questionnaire devised for the study which was pretested for validity and reliability. Subjects in the treatment conditions involving family therapy received three sessions of family therapy over a maximum of seven weeks. As with all doctoral dissertations multiple analysis were conducted, but for the present purpose only the global findings are reported. Ritterman's central hypothesis that Ritalin alone would fail to produce a significant change on any single dimension of hyperactivity compared with the other treatment conditions was supported. As a result of a series of complex statistical analysis, several other conclu-

Children and Family Therapy 27 sions were reached. All study subjects showed significant improvement posttreatment, virtually on all measures. Ritalin treatment alone, however, consistently showed least improvement. Another key finding was that family therapy alone had either a neutral or negative effect, whereas family therapy and placebo and family therapy plus Ritalin showed a trend towards improvement. Many pages would be required to provide even a reasonable overview of this carefully designed and well-executed study. The subjects were carefully selected; outcome measures were clearly defined; analyses were well articulated; therapeutic processes were clearly described; and the results were intelligently reported. As with almost all dissertations, its major limitation was the absence of follow-up , which, given the many practical constraints, is rarely possible. Nevertheless, the study pointed out and, indeed, confirmed that treatment of complex biopsychoneurological problems requires a multifaceted approach to be effective. Family therapy appears to have a key place in the treatment regimen of complex pediatric disorders, although its value as the only treatment is discouraging. Szapocznik and associates (1989) studied the effectiveness of structural family therapy (SFTl) versus individual psychodynamic child therapy (IPCT) in problematic Hispanic boys. Of the sixty-nine children between the ages of six and twelve years who participated in the study, twenty-six were in SFTl, twenty-six in IPCT, and seventeen in a recreational control program. Descriptions were provided for the treatment and control conditions. Children were randomly assigned to the different groups. Contact hours for all subjects varied between a minimum of twelve and a maximum of twenty-four. Outcome measures consisted of standardized behavioural and self-report measures. Data were obtained at pretreatment, posttreatment, and one-year follow-up. Results were complex. The control condition was significantly less effective than treatment. Both treatment conditions were almost equivalent in their effectiveness on almost all measures. However, there were some inexplicable findings. While the children in SFTl showed symptomatic improvement, there was no parallel change in the family functioning. With IPCT the children also showed improvement, but their families showed deterioration in their general functioning, especially at follow-up. There was no apparent explanation for either the children's improvement in the SFTl group without major change in family functioning or the deterioration in family

28

How Good Is Family Therapy?

functioning in the IPCT group. In other words, one of the basic assumptions of family therapy that improved family functioning would cause elimination of the presenting symptoms failed to be validated in this study. On the other hand, the axiom that to treat the child to the exclusion of the family may be less than ideal did receive partial support. General support of the findings of the studies discussed so far and confirmation of the general superiority of social learning based parent training (SLBPT) over systems family therapy (SFT2), based on Minuchin's structural family therapy model, was provided by Wells and Egan (1988). Families of nineteen children with oppositional disorder were randomly assigned to the two treatment conditions. Treatment was provided by eleven clinical psychology doctoral candidates who were trained in these methods of treatment. They were supervised by three clinical psychologists. Outcome was assessed on the basis of a coding system that measured oppositional behaviours and other dimensions of parent-child interaction. Other measures included promoting appropriate behaviours in the parents and children. The Beck Depression Inventory and Spielberger State-Trait Anxiety Inventory were used with the mothers and the Locke-Wallace Marriage Inventory was employed to assess marital adjustment. Observations and completions of the instruments were made at pretreatment and immediately following termination of treatment. Both treatment conditions were described. A minimum of eight to a maximum of twelve therapy sessions were given. Membership in the treatment family group was determined by the therapists. Appropriate statistical analysis were performed. Outcome on the self-report measures failed to show any significant differences between the groups at pre- and posttreatment. However, for the behavioural observational measures some significant differences emerged in the areas of compliance with commands and compliance with total commands posttreatment in favour of SLBPT. In the treatment of oppositional disorder in children SLBPT was found to be superior over SFT2. Wells and Egan conducted a carefully designed and well-executed study. One of its deficits was the absence of the follow-up component. A factor that may have influenced outcome was the sheer number of therapists employed to carry out treatment. The investigators were careful to control for the therapist variables. Yet, the risk of contamination by therapist variables are ever present.

Children and Family Therapy 29 Strategic Family Therapy versus Other Treatments Simpson (1990) reported a study comparing the efficacy of Milan family therapy (which is considered a variation on the strategic model) with other methods of treatment for children with psychological problems. This project was based in the Department of Child Psychiatry at the Hospital for Sick Children in Edinburgh, Scotland. Initially 110 children were recruited, but through attrition eighty-seven children and their families participated in the study. The children received a battery of tests to determine their psychological and behavioural functioning. Family therapy activities were recorded in a standardized form at the end of each session. Milan family therapy was administered by two social workers and two child psychiatrists. Other treatments, which included psychodynamic psychotherapy, cognitive therapy, and also non-Milan family therapy, was offered by ten mental health professionals. Family therapy consisted of five sessions delivered over several weeks. Outcome was assessed at the end of treatment and at the six-month follow-up. Standard statistical analysis were performed. Results were mixed in some areas. Non-significant differences emerged in the areas of children's behaviour on the basis of Rutter Teacher's Scale or on the basis of the therapist's assessment of the child's presenting symptoms. Significant differences emerged in the parental rating of posttreatment improvement in family life. A strong association was also found between the improvement in the children's symptoms and improvement in family relations. A point of note was that the family therapy group on average attended three sessions as opposed to five for the group receiving other treatment. A key conclusion offered by Simpson was that Milan family therapy was superior to the other treatments in that comparable improvement was achieved with fewer sessions and fewer missed appointments. In other words, the cost effectiveness of the family therapy regime was superior. Simpson's was a complex and well-designed study. Comprehensive baseline data were collected, and treatment strategies were clearly stated. Statistical analysis was adequate. The point of note, however, was that the actual efficacy of Milan family therapy was only marginally superior to the other treatments. In that situation the argument of cost effectiveness assumes greater significance. Unfortunately, that aspect was not part of the analysis. In a comparative investigation of strategic family therapy (SFTl) versus behavioural therapy (BT) for children with psychiatric disorders

30 How Good Is Family Therapy? ranging from oppositional disorder to functional encropesis. Szykula, Morris, and Sudweeks (1987) randomly assigned forty-nine children to treatment groups. Twenty-two children were in SFTl and twentyseven children were in BT. The mean age of the children in SFTl was 9.54 years and in BT 8.04 years. The age differences were not statistically significant. The outcome measures were clearly stated. Pretreatment measures were completed, and posttreatment measures were obtained at between six and twelve weeks following the last session. Appropriate statistical analysis were performed. The treatment approaches were not described, nor was any information provided about the length of treatment. The results were unequivocal. Only marginal and non-significant differences emerged between the two approaches. However, clients with higher severity scores experienced more improvement with BT than did their counterparts in the SFT group. Although Szykula and colleagues described their study as preliminary, nonetheless, some basic data were omitted. Apart from any information about the treatment protocol, the authors failed to describe the actual participants in the family sessions. Nor was any information provided about the families. Family functioning was not objectively assessed. For a study that purports to report on the relative merit of two approaches, the failure to include such data must be seen as a major shortcoming. Sayger, Szykula, and Sudweeks (1992) compared strategic family therapy (SFT2) with behaviour therapy. They randomly assigned to the two groups forty-nine children referred to an out-patient child psychiatry unit with problems of over-anxiety, attention deficit hyperactivity, and oppositional defiance. Twenty-seven families participated in BT and twenty-two in SFT2. Outcome was assessed on the basis of the Sayger-Szykula Side-Effects Survey, which requires parents to make self-reports at the termination of treatment of positive or negative changes in areas other than the main target symptoms. This information was obtained by a research assistant who was blind to the treatment approaches. The treatment approaches were not described. T tests were performed for statistical analysis. The results showed that 100 per cent of the mothers in BT and 68 per cent in SFT2 registered satisfaction with the outcome. There were no negative side effects of the two treatments. Though novel in some respects, Sayger and colleagues' study had some major shortcomings. Three areas of concern were: (1) Reliance

Children and Family Therapy 31 solely on parent satisfaction to measure outcome. This could have been supplemented by instruments to measure improvement in the children's presenting symptoms with perhaps another measure to assess improvement in family functioning. (2) There was a lack of information about the therapeutic process. The actual participants in the treatment sessions, length and number of sessions, and issues of compliance were not reported. (3) No information on posttreatment was incorporated into the design. Comments: The inescapable fact that emerges from the group of comparative studies just discussed is that family therapy based on specific models for children does not, on the whole, appear to be superior to other forms of intervention, and especially behavioural interventions. In only two studies (Ritterman 1978; Simpson 1990) was family therapy shown to be either superior to or more cost effective than other treatments. While the methodological deficits of these comparative studies are in some instances quite serious, that ought not to minimize the general observation that family therapy for children, which had appeared so promising in an earlier period, was not found very impressive. This was a somewhat unexpected finding for us in view of the long and time-honoured linkage between childhood issues and family dynamics.

Non-Specific Family Therapy versus Other Treatments Studies considered in this section are hard to categorize. They share two common elements, namely, all families received a non-specific type of family therapy, and, second, comparison groups varied. An early study in this category was reported by Love, Kaswan, and Bugental (1972). Child therapy (thirty-three children) , parent counselling (twenty-eight children; both based on psychodynamic theory), and information feedback (thirty children), which involved both parents and teachers looking at the child's interpersonal environment and presumably making environmental modifications, if need be, for different socioeconomic groupings were assessed using a sample of ninety-one children aged between eight and thirteen years referred for behavioural or emotional problems by ten schools. Children were assigned to the treatment groups in rotation. Outcome was measured against improvement in school grades and school behaviours. The number of therapy sessions varied considerably between the three methods of treatment. Follow-up was conducted at six, twelve, and

32 How Good Is Family Therapy? twenty-four months after termination of therapy. Standard statistical analysis were performed. As far as grades were concerned both information feedback and parent counselling stopped the downward trend, and the effects of child therapy were mixed. The involvement of parents in the treatment was found to be more effective in improving the children's school performance. All interventions improved, albeit slightly, peer relations. The socioeconomic level of the family was found to be important in that children in upper social classes responded better with information feedback; children in lower socioeconomic groups responded better with counselling; and child therapy resulted in lower grades across the socioeconomic groups. Follow-up at twelve months showed that improvement in children's behaviours seemed temporally related to their clinical experience. In other words, improvement gained at the point of termination did not continue to improve over time. Parent counselling in this project by Love et al. was not derived from the principles of systems theory. Rather, it was predicated on psychodynamic theory. The amount of information provided for each treatment condition makes it difficult to have a clear picture of what actually transpired in therapy. Nevertheless, having some parental involvement was clearly shown to be beneficial in this study. No obvious reasons emerged for the relative ineffectiveness of child therapy. Perhaps it confirmed an old belief that a child's emotional and behavioural problems are more often than not rooted in the family. It should be stated that this was a complicated investigation with a sophisticated design involving three treatment modalities and their interaction with the socioeconomic status of the children's families. Hence, the key finding that family involvement in the treatment of the child is beneficial must be taken seriously. Yet, parental counselling was not inherently superior to information feedback. In sum, this study failed to furnish an unambiguous endorsement for family therapy. Pevsner (1982) randomly assigned to group parent training and family therapy fifteen families referred for behavioural problems with children. Group parent training was found superior to behaviourally oriented family therapy. Improvements were maintained at the six-month follow-up. Only twelve subjects completed treatment. This study was conducted with a small number of subjects, and although it was carefully designed, the results could best serve as a testable hypothesis for a study with a larger number of subjects. Fisher (1989) tested the efficacy of brief time-limited family therapy.

Children and Family Therapy 33 Four treatment conditions were established: (1) time-limited therapy of six sessions within eight weeks; (2) time-limited therapy of twelve sessions within sixteen weeks; (3) time-unlimited therapy; and (4) a waiting list control group. Thirty-seven families who had sought treatment at a child guidance clinic were randomly assigned to the four groups. Treatment was carried out by five therapists who were social workers and psychologists by training. They had varied theoretical orientations. A wide range of outcome measures were used. Appropriate statistical analysis were performed. The findings were unequivocal. The treatment groups fared better than the group from the waiting list. In terms of length of therapy, outcome seemed unrelated to the length of therapy. The last two reports were given summary treatment because they were not central to the issue of this chapter, namely, the merit of systemic family therapy. In keeping with reports discussed earlier in this section, Pevsner (1982) also found family therapy less effective than another method of treatment. Fisher's (1989) study was concerned with a very important dimension of family therapy which is the relationship between the length of treatment and outcome. Her findings suggest the need for further investigation of this question. Discussion

Masten (1979), cited at the beginning of this chapter, observed that family therapy outcome research was fraught with methodological flaws and that there was no way to arrive at a firm conclusion about the efficacy of family therapy with children. Fifteen years later, the situation has definitely improved. Much of this improvement, as far as methodological issues are concerned, can be found just in the number and quality of studies comparing family therapy with other interventions. We now address the key question of efficacy. Our conclusions are based on a review of those family therapy outcome studies where the method of family therapy was based explicitly on systems orientation or enough information about the therapeutic process was provided for us to judge that such was the case. Hence, the focus of family intervention with children here is narrow. We have omitted discussion of a vast amount of literature on topics such as parent education and behavioural family therapy with children. Undoubtedly, the early studies held much promise for family therapy for children's problems. In particular, the Santa-Barbara study,

34

How Good Is Family Therapy?

though uncontrolled, just by its sheer scope provided justification for optimism regarding the outcome of family therapy with children. That optimism, it must be sated, was not entirely groundless. A similar positive outlook was also reported by Minuchin's group in the treatment of a variety of pediatric disorders (see Chapter 5). However, it must be stated unequivocally that on the basis of the present review it is virtually impossible to arrive at an estimate of the proportion of children who benefit from family therapy. Part of this problem can be explained simply on the basis of the number of variables on which improvement is judged. Very rarely is a global score of improvement reported. Uncontrolled studies, for the most part, and one study comparing family therapy with a no treatment group were generally supportive of family therapy. If we are asking whether family therapy is better than nothing for treating children's behavioural and emotional problems, the answer has to be in the affirmative. However, the answer takes on a high level of complexity when the results of the comparative studies are examined. In general the findings of the comparative studies were mixed. We failed to find a single study that demonstrated a clear superiority of family therapy over other treatments. Yet Ritterman's study, for example, pointed out the complex role of family therapy in conjunction with drug therapy in the treatment of hyperactivity. Family therapy together with drugs rather than family therapy alone or drugs alone resulted in the most effective outcome. This finding lends credence to the axiom that complex medical problems are multifaceted and that a multimodal approach is generally superior to any individual treatment. When family therapy is judged against other psychological therapies, the efficacy of family therapy weakens. A point of caution here: superiority of other treatments for children over family therapy is far from absolute. Sometimes other treatments fare significantly better than family therapy, at other times only nominally so, and at yet other times the two modes of intervention are equally effective. There is no study to our knowledge that has examined the predictors of positive outcome of family therapy with children. If some children do well in family therapy irrespective of the superior outcome of other methods of treatment, the general assumption is that the latter is superior to the former. The truth, however, is that some children were indeed the beneficiaries of family therapy. What might have been the factors that contributed to their positive response to family therapy? This question is generally ignored for statistical reasons and because

Children and Family Therapy 35 subjects are randomly assigned and the groups are closely matched on many critical variables. That explanation provides a partial answer. An analogy from medicine might clarify the issue.· If patients with clinical depression, for example, respond differently to two drugs, the important question that follows is related to the specific mechanisms of the disease which might explain the different responses to the drugs. This level of scientific sophistication is yet to make its way into family therapy research. Our review failed to establish clear evidence for the superiority of any particular school of systemic family therapy. Part of the explanation can be found in the almost total absence of such studies. The only study that had an implicit goal of comparing the Milan approach with a variety of other psychological approaches including non-Milan family therapy did demonstrate the superiority of the Milan approach (Simpson 1990). However, this study was not designed to exclusively compare two methods of systems-based family therapy. Three comparative studies between structural family therapy and other interventions were examined. Not one showed clear superiority of this form of family therapy over others. Ritterman did provide some evidence that family therapy in conjunction with drug treatment was the most effective. Beyond that, the other two studies (Szapocznik et al. 1989; Wells and Egan 1988) failed to establish superior outcome of family therapy over psychodynamic therapy or a social learning based parent-training program. Three studies comparing strategic family therapy with other interventions were also reviewed (Szykula, Morris, and Sudweeks 1987; Sayger, Szykula, and Sudweeks 1992; Simpson 1990). Again, only one (Simpson 1990) showed that Milan (a variation of strategic) family therapy was more effective than a variety of other treatments. The two other (Szykula, Morris, and Sudweeks 1987; Sayger, Szykula, and Sudweeks 1992) failed to replicate this finding. From these sets of studies, involving two schools of family therapy vis-a-vis other interventions, it might be concluded, albeit indirectly, that one school of family therapy does not seem to hold sway over the other. Obviously, more direct evidence has to emerge from future research. What appears to be incontrovertible is that research to date has produced substantial evidence to suggest that systemic family therapy for children is not necessarily the treatment of choice. Yet a study here and there, such as Ritterman's (1978), Black and Urbanowicz's (1987), and Simpson's (1990) does hold out some promise for

36 How Good Is Family Therapy? systemic family therapy for children. It is also noteworthy that these three studies dealt with rather divergent problems. The present state of knowledge is inadequate to prescribe systemic family therapy for any specific childhood problem. Hence, the future task will be to ascertain whether certain childhood problems are indeed more amenable to the systemic approach.

3

Juvenile Delinquency and Conduct Disorders

Ideas about psychotherapy and adolescents have changed drastically over the past fifty years. Montalvo (Foreword in Fishman, 1988) observes that for the first half of the century clinicians focused on 'inner hydraulics in disarray, or social upheaval generated within a context of peers.' Treatment most often took the form of individual or group psychotherapy, and families were seen as passive observers to the unfolding of adolescent development. The emergence of family therapy, with its emphasis on interactional contexts, introduced an important alternative for dealing with problems associated with adolescence. The literature on family therapy abounds with applications to families with adolescents. The seminal ideas for structural family therapy were developed with this population (Minuchin et al. 1967). An illustration of the centrality of adolescence to family therapy practice can be found in Ackerman's (1980) observation that eight of eleven leading family therapists chose families with adolescents to demonstrate their respective approaches in Papp's (1977) volume of case studies. Fishman (1988) has argued that family therapy is the treatment of choice for working with troubled adolescents, and Breunlin et al. (1988) echoed that sentiment by declaring family therapy with adolescents as a subspeciality within the field of family therapy. Family therapy with adolescents has been described in relation to a diverse range of problems. One survey of the literature cited ninetynine family therapy articles that dealt with forty-four problem areas, ranging from abandonment to suicide (Breunlin et al. 1988). The literature related to the effectiveness of family therapy with adolescents, however, yields far less diversity. Gurman, Kniskern, and Pinsof (1986)

38 How Good ls Family Therapy? noted that this literature clusters in three areas: psychosomatic illnesses, delinquency and conduct disorders and mixed emotional and behavioural disorders. The methodological quality of these studies, not surprisingly, ranges from poor to excellent (Mann and Borduin 1991). This chapter deals with the broad issue of delinquency and conduct disorders (i.e., status offenders). All of the studies reviewed here involve some level of control or comparison. Psychosomatic illnesses are covered in Chapter 5 and adolescent drug abuse and mixed emotional and behavioural disorders are dealt with in Chapter 3. Chapter 4 includes adolescent psychiatric disorders, to the extent that they are identified in what are generally developmentally heterogeneous samples. We have excluded an examination of purely behavioural approaches (such as parent training or contingency contracting), except when these are employed for purposes of comparison. Delinquency Family interaction research has established, at least in general terms, that there is an important relationship between family functioning and delinquency (Tolan, Cromwell, and Brasswell, 1986). The first bridge between this research and family therapy was described in Families of the Slums (Minuchin et al. 1967). Although this study lacked methodological rigour, it was the first to examine delinquency as a function of family organization and then to propose related intervention strategies. Over the past thirty years systems-based therapists have extended this bridge by studying the effectiveness of family therapy with a range of delinquent behaviours. Status Offences

Status offences are acts that are considered to be unlawful because they are committed by minors. In other words, the status of the actor and not the act itself is the main determinant of defining the adolescent as delinquent. Examples of behaviours that qualify as status offences include running away, truancy, drinking, smoking, incorrigibility, curfew violation, and promiscuity. The most commonly cited status offences have to do with non-cooperation with parents and with running away (Beal and Duckro 1977; Druckman 1979). Responding to such behaviours is clearly not the sole domain of the juvenile justice system. Adolescents exhibiting the same behaviours might be consid-

Juvenile Delinquency and Conduct Disorders 39 erect to be conduct disordered, behaviour disordered, or emotionally disordered by the mental health system. In other contexts these behaviours might simply be described as an element of parent-adolescent conflict. Nevertheless, there is a discernible empirical literature on family therapy with juvenile status offenders. The first three studies we discuss are quite weak in terms of methodological rigour and, therefore, provide limited support for the effectiveness of family therapy with this population. Beal and Duckro (1977) report on a program that provided family therapy as an alternative to traditional procedures for dealing with juvenile status offenders. Families who chose this option could participate in short-term counselling (usually six to eight sessions) prior to formal adjudication of their adolescent. The treatment combined the principles of family crisis intervention (Langsley et al. 1968) and communication-oriented family therapy (Satir 1967). Forty-four randomly selected participants from one month of the program's operation were compared with a retrospective comparison group of fifty-four families served during the same month one year prior to the program's initiation. Avoidance of court proceedings constituted the sole criterion for establishing effectiveness. The analysis revealed that significantly fewer families who took part in the prehearing program proceeded to court (17 per cent), when compared with families served during the year prior to the program's existence (35 per cent). The majority of the families who did not proceed to court, however, were referred to external agencies for continued counselling. The utility of the study is limited by the lack of information on client and service characteristics. Perhaps most serious is the lack of basic and comparative information about families who chose to participate in the program and those who did not. It may be that the availability of counselling is simply a screening mechanism for identifying the most functional families. A related problem is apparent in the measure of effectiveness. The findings indicate that the program was relatively effective at diverting away from the court process. This may or may not be related to improved family functioning. In fact, we are told that about 50 per cent of the families in the overall program were referred for further counselling, and only 30 per cent terminated without seeking such a referral. Follow-up information is also lacking. This study provides minimal support for the assertion that family therapy is an effective intervention. Michaels and Green (1979) also evaluated the effects of family

40 How Good Is Family Therapy? therapy for families of status offenders. Sixty-four families who were treated with a combination of problem-solving (Haley 1976) and structural (Minuchin 1974) family therapy were compared with a similar group of seventy-five families who were served prior to the implementation of family therapy services. The family therapy group dramatically outperformed the comparison group in terms of placement in institutions or foster care (4 per cent vs 44 per cent), placement in detention homes (4 per cent vs 31 per cent), and processing through juvenile court (4 per cent vs 33 per cent). Comparisons between a randomly selected sample of twenty families from each group also showed that children in the family therapy group not only remained at home at a greater rate, but also experienced fewer placement episodes than those in the comparison group. This study shares many of the limitations cited above. There is no information about selection criteria, client characteristics, or service characteristics. While it appears that family therapy reduced placements and court processing, there are not sufficient controls to justify ruling out alternative hypotheses. Moreover, a link between outcomes and improved family functioning was not explored. Ostensen (1981) investigated the effects of family therapy with a subpopulation of status offenders. Twenty-eight runaways and their families who participated in at least three family therapy sessions (with or without temporary foster placement) were compared with forty-five clients who refused to participate, or withdrew prior to completing three sessions. It is not clear whether the use of foster care was part of a separate, integrated treatment model or if the distinction was made post hoc. Regardless, participants received an average of three to six sessions of communication-oriented therapy based on Wood and Schwartz's (1977) brief family intervention. The services received by non-participants were not specified. The total sample was predominantly female, with more than half of the subjects being between fourteen and fifteen years of age. Families were mainly Caucasian, and most included two adults. Approximately 35 per cent of the adolescents were first-time runaways, and an additional 35 per cent had run away more than twice. Unfortunately, there were no statistical comparisons between participants and nonparticipants. Follow-up data were collected in telephone interviews with the runaways. The analysis was limited to basic comparisons through t tests and chi squares. The findings indicated that family therapy partici-

Juvenile Delinquency and Conduct Disorders 41 pants reported significantly fewer subsequent runaway episodes (25 per cent), during a three-month follow-up period, as compared with non-participant families (62 per cent) This relationship held when family therapy with foster care (38 per cent) and family therapy without foster care (8 per cent) were separately compared with nonparticipation. A more rigorous analysis of the three groups simultaneously was not conducted. While this study provides general support for a family therapy approach for a subpopulation of status offenders, the possibility that participants and non-participants differed in some systematic way remains a viable alternative hypothesis. Recidivism, as reported by the offending adolescent, constitutes an inadequate measure of effectiveness that requires some means of verification. Moreover, no measures of family functioning were introduced into the study. This is especially problematic given that the observed differences in outcome were attributed to a specific model of intervention which was aimed at changing family interaction. The next two studies include minor methodological improvements, but demonstrate findings that do not support the effectiveness of family therapy with families of status offenders. Gruber (1979) compared fifty-four families of status offenders who received family counselling with seventy families who refused the service and were therefore provided with traditional services. The nature of the family counselling and the traditional service was not specified. Subsequent juvenile justice referrals were the sole criterion for effectiveness, and the time period covered by the study is not stated. The analysis was limited to comparisons of percentages as tested by chi square. The results indicated that there was no significant difference in subsequent referrals between the family counselling group (39 per cent) and the traditional services group (49 per cent). There was some indication that prior referrals, runaways, law enforcement referrals, and rural residence were each significantly related to increased recidivism. In contrast to the previously cited studies, Gruber (1979) did consider the possibility that voluntary family counselling operated as a screening mechanism. Data on families served during the year prior to the availability of family counselling were examined as a means of determining if the counselling option was systematically selected by more functional families who might resolve their difficulties, regardless of the intervention they received. The percentage of subsequent refer-

42

How Good Is Family Therapy?

rals for the year prior to the counselling service (43 per cent) was not significantly different from the subsequent year (44 per cent). This may indicate that family counselling had no discernible effects on subsequent recidivism, assuming that these families shared similar characteristics. This study was not very sophisticated in its methodology or analysis. It also lacked sufficient descriptions of service and client characteristics. Nevertheless, important contributions were made in attempting to ascertain the screening effects of a voluntary counselling program, and in beginning to consider some of the interacting factors that might influence recidivism. Druckman (1979) provides a more comprehensive, theoretically grounded study of a family treatment program for female status offenders. Families with adolescents whose delinquent behaviour appeared to stem from family relationship problems were included, if both the juvenile and her parents agreed to participate. The program consisted of a two-week residential phase, which included educational services and skill development, and an eight to twelve weeks' follow-up phase, which included family therapy with the juvenile at home. The family therapy approach was inadequately described as emphasizing 'here and now situations' (ibid.: 630). Participant families were mostly Caucasian (89 per cent), and their offending youth averaged fifteen years of age. Further descriptions of client characteristics were not provided. It is important to note that the program was severely underutilized (31 per cent of capacity) with a total sample of sixty-four families having agreed to participate. Moreover, only fourteen families (22 per cent) actually completed the program. Data from fifteen of the fifty drop-outs was used for comparative purposes. These families withdrew within the first five days of the program, with one-third opting for alternative non-family treatments. There was no reported attempt to establish the representativeness of this subsample in relation to the total group of families who withdrew. More important, information on families who chose not to participate was completely lacking. The functioning of these families was conceptualized according to the circumplex model (Olson, Russell, and Sprenkle 1979). The study was driven by six hypotheses which were aimed at both verifying the model and assessing outcome. The Family Environment Scale (Moos 1974) was administered to all available family members at intake and at completion (or after twelve weeks for drop-outs). This instrument

Juvenile Delinquency and Conduct Disorders 43 was used because it was thought to be conceptually compatible with the circumplex model. Recidivism was operationalized as any new court involvements following termination in the program. The duration of this monitoring was not specified. Four hypotheses proposed that the relevant Family Environment Scale (FES) subscale scores would be consistent with formulations based on the circumplex model. None of these hypotheses was supported by the data. It is impossible to determine if these findings can be attributed to a failure of the circumplex model, the specific application of the model, or its operationalization through the Family Environment Scale. The next set of hypotheses were directly related to program effectiveness. One hypothesis proposed that families who completed the program would show greater improvement in family functioning (FES), as compared with those families who withdrew. While both groups of families made significant gains, the family therapy group did not significantly outperform the comparison group. A sixth hypothesis speculated that families who completed treatment would have significantly lower recidivism rates than those who withdrew. In fact, a significant difference was found,but in the opposite direction to what was expected. Fifty per cent of the program participants were involved in further offences, as compared with 33 per cent of the comparison group. The design of this study makes it impossible to generalize beyond its subjects or judge the separate effects of residential treatment and family therapy. It is clear, however, that this particular program was not uniquely successful at improving family functioning or reducing recidivism. While the relationship between increased recidivism and program participation is worrisome, it is not clear whether the finding can be attributed to the program, a selection bias, or some external variable. Taken as a group, these investigations shed little light on the question of the effectiveness of family therapy with families of status offender adolescents. The mixed results are as easily attributable to weaknesses in methodology as to the intervention under study. While all of the studies employed a comparison group, none was constructed through random assignment or careful matching. Overall, questions related to the comparability of treatment and comparison groups were ignored. Only one study (Gruher) even entertained the possibility of systematic bias in treatment selection. Specific treatment approaches, service characteristics, and client characteristics (especially family

44

How Good Is Family Therapy?

functioning) are generally inadequately reported. Finally, measures of effectiveness tend to be extremely narrow, poorly operationalized, and often lack an explicit relationship to the treatment which is presumed to be responsible for the effects. The most impressive evidence for the effectiveness of family therapy with status offenders can be attributed to a series of studies associated with the development and evaluation of functional family therapy (Alexander and Parsons 1982). This approach views delinquent behaviour as a function of family organization and relies, primarily, on behavioural interventions to change interactional patterns. This hybrid of systemic and behavioural approaches has been the subject of a sequential set of studies designed to evaluate its conceptual bases, intervention technologies, efficacy, and impact (Barton and Alexander 1981). Alexander (1973) compared the interactional characteristics of twenty families of status offenders with twenty-two non-distressed families, through the use of videotaped interactional tasks. Based on an adaptation of Gibb's (1961) work with small groups, status offender families were found to demonstrate higher rates of defensiveness within all dyads and lower rates of supportiveness in parent-child dyads. The study also found that delinquent families reciprocated defensiveness, while non-distressed families reciprocated supportiveness. Having identified potentially changeable interactional patterns, the next step was to develop a treatment that would modify the interactions of delinquent families to approximate those of non-distressed families (Alexander and Parsons 1973; Parsons and Alexander 1973). Families of status offenders were randomly assigned to functional family therapy (n = 46), client-centred family groups (n = 19), eclectic psychodynamic family counselling (n = 11), or no treatment (n = 10). This study is covered over three reports in the literature. Parsons and Alexander (1973) reported on data from the first twenty experimental treatment families, along with the first ten recipients of group treatment and the ten families who served as no treatment controls. This well-designed substudy used a Solomon (1949) four-group design, but employed an attention-placebo condition (family groups) instead of a posttest-only control group. Assuming the correct statistical treatment of data, this design suggests that changes cannot be attributed to attention, maturation, or other intervening experiences, and, therefore, it enhances the probability that such changes are related to treatment. The family therapy was provided by graduate and undergraduate

Juvenile Delinquency and Conduct Disorders 45 students in clinical psychology who were given method-specific training and ongoing supervision. The treatment, which was well elaborated and included a manual, emphasized reciprocity and positive renforcement; equality and clarity of communication; and the development of adaptive, solution-oriented communication patterns. The program lasted ten to twelve weeks. Consistent with the emphasis on interactional patterns and the related explicit goals for intervention, the study was driven by the following outcome hypotheses: (1) treatment families would display less silent time than control families; (2) treatment families would display more equality in their distribution of communication time than control families; and (3) treatment families would display more simultaneous speech than control families. The results of the analysis, which was rigorous and well described, supported each hypothesis. These findings, which can be treated with considerable confidence, indicate that the intervention was effective at helping families modify their interactional patterns in the direction of non-distressed families. The second report of this project (Alexander and Parsons 1973) was concerned with exploring the link between changes in interaction and a reduction of problematic behaviours. Juvenile court records were examined six to eighteen months after termination for the entire sample of eighty-six families (including drop-outs), along with those of forty-six cases randomly selected from several hundred referred but not treated cases (post hoc yoked controls), and 2,800 cases treated by a variety of modalities throughout the county. While the first four groups were judged to be equivalent on demographics and prior recidivism, it was not possible to determine the characteristics of the latter two groups. Families receiving functional family therapy experienced significantly less recidivism of status offences (26 per cent), in comparison with the other groups. Eclectic psychodynamic family counselling was associated with the highest rates of recidivism (73 per cent), while the remaining groups showed rates of approximately 50 per cent. The results for subsequent criminal offences were less clear. While functional family therapy was associated with the fewest subsequent offences (17 per cent vs 21 per cent to 27 per cent), the differences were not statistically significant. A direct test of the relationship between interactional variables and recidivism was also conducted, and in all cases , the interactional variables distinguished between the two groups, in the directions that were expected.

46 How Good Is Family Therapy? In the final report related to this project, Klein, Alexander, and Parsons (1977) focused on the primary prevention effects of functional family therapy. They hypothesized that functional family therapy would decrease the rate of sibling involvement with delinquency. The findings , based on juvenile court records 2.5 to 3.5 years after termination, were remarkably similar to those for the earlier follow-up of targeted adolescents (Alexander and Parsons 1973). Functional family therapy was associated with significantly lower rates of sibling court involvement (20 per cent) as compared with the randomly assigned no treatment control group (40 per cent) , client-centred family groups (59 per cent), and eclectic psychodynamic family therapy (63 per cent) . When the four posttest interactional measures were examined for families with sibling delinquents and families without sibling delinquents, two of the measures showed significant differences. While the authors do not discuss this finding, it may indicate that some interactional changes are more durable than others. These findings provide strong support for the conceptual underpinnings of functional family therapy and its comparative effectiveness for reducing recidivism. It is especially important to note that comparisons were provided that reflected both traditional services and other family-focused treatments. As Alexander and Parsons (1973) argue, it appears that a focus on families is not sufficient, rather it is necessary to target specific family interactional qualities for change in a particular direction. The next study (Alexander et al. 1976) shifted the focus from comparative outcomes to an intratherapy examination of contributors to outcome. Twenty-one trainees, ranging from second-year psychology students to doctoral graduates with practice experience were given a ten-week course in functional family therapy. After the training, the therapists were rated on eight dimensions that reflected their ability to perform the skills associated with the model. One family was then randomly assigned to each therapist. The focus and structure of treatment was the same as described in the earlier reports. At termination each family was categorized according to outcome status. The four categories ranged from very poor outcome (termination against therapist's judgment after the first session, no movement on the complaint, and no interactional change) to very good outcome (joint termination, based on complaint cessation and effective interaction). Juvenile court and community mental health records were examined

Juvenile Delinquency and Conduct Disorders 47 for subsequent referrals, twelve to fifteen months after the intervention. Consistent with their outcome categorization, families in the poor outcome conditions (n = 9) showed the highest rates of recidivism (60 per cent and 50 per cent), while families in the good outcome conditions (n = 12) showed no recidivism at all. Moreover, the recidivism rate for the entire group was 23.8 per cent, which approximates the rate of 26 per cent found in Alexander and Parson's (1973) earlier study. Samples of family interaction were selected from the first and last portions of the initial session, and the first portion of the next-to-last session (except for families that completed only one session). There were significant differences between the first and second-last sessions, with families in the best outcome condition, reflecting the most adaptive communication. The posttraining ratings of therapist skills were explored for their contribution to the observed outcomes. Six of the eight skills were significantly correlated with client outcome and were ultimately collapsed into a relationship dimension and a structuring dimension (three scales each). Each dimension significantly distinguished between good-outcome and poor-outcome therapists. While both dimensions were predictive of outcome, the relationship dimension was a much more powerful predictor than the structuring dimension (45 per cent vs 15 per cent). This most likely indicates thanhe acquisition of structuring skills related to functional family therapy enhances the more generic relationship skills of the therapist. Overall, this study was well designed and properly implemented. It adds to the accumulated evidence of the effectiveness of functional family therapy and begins to identify contributors to successful outcome. Comment: The research related to functional family therapy with sta-

tus offenders is, in many ways, exemplary. Several limitations, however, should be noted. Throughout this research, status offenders are assumed to be a homogeneous population. There are no reported attempts to differentiate among families by offence or by the number of offences. Consequently, important information about predictors of effectiveness may be lacking. It should also be noted that information on behavioural outcomes was obtained through official records. A wider scan may have produced information about adolescents who subsequently came to the attention of other systems. In addition, subjects have been predominantly middle-class, Protestant (mostly Mormon) families residing in Utah. As Gurman, Kniskern, and Pinsof

48 How Good Is Family Therapy? (1986) note, many of these families may have stronger emotional ties and clearer organization than families in more heterogeneous communities or in more urbanized settings. Alexander and Barton (1976) argue, however, that functional family therapy focuses on process and, therefore, should generalize quite easily. For the most part functional family therapy was provided by specifically trained clinicians (usually students) within the context of a university-based training setting and a treatment project. At this point it is difficult to determine what effects full caseloads in a more traditional public service setting might generate. In this regard, it should be noted that Barton and colleagues (1985) report the successful generalization of treatment effects of therapy provided by undergraduates and child welfare workers to similar populations (two replications with other populations are reported later in this chapter). These studies, while far less rigorous than those reported above, make a beginning contribution towards answering questions related to generalization. Based on the findings to date, further replications aimed at broadening service contexts and more carefully specifying treatment populations are warranted. Serious Delinquency and Mixed Offences

This section examines eight studies that explore the effectiveness of family interventions with families of adolescents who have committed more serious offences or who are treated in a mixed sample of status offenders and more serious offenders. The studies described below are quite diverse in that they deal with first-time offenders and chronic offenders who were treated in both institutional and community contexts. The first group of studies involves comparisons between family therapy and probation services or other community based services typically provided for juvenile offenders. Johnson (1977) reported on the treatment of 190 families of delinquents who were directed to family treatment as a condition of probation. The adolescents had all committed status offences, crimes against persons, or crimes against property prior to their most recent adjudication. These families were matched (offence, age, sex, parental marital status, socioeconomic level, and ethnicity) with an equivalent number of comparison families who were assigned regular probation services. The family treatment was based on the early work of Erickson (see Haley 1967), Haley (1963), Minuchin (1965) , and Whitaker (1975). Pro-

Juvenile Delinquency and Conduct Disorders 49 bation included casework conferences and referrals to community resources. It should be noted that these resources were also available to clients receiving family therapy. Specific service utilization statistics were not reported for either group. Subsequent offences were noted at three-month intervals for the first year following termination and at the end of the second year. During the service period both groups showed a decline in delinquent activity, with the family therapy group significantly outperforming the probation group. By the end of the first year the groups showed an equal reduction in subsequent offences. During the second year, however, the family therapy group remained unchanged, while the probation group showed a significant increase in new offences. Based on these findings, Johnson concluded that family therapy yielded superior outcomes when compared with probation, even when probation goes beyond simple reporting to include active casework and referral. He also contends, somewhat erroneously, that the fact that adolescents receiving family treatment may have concurrently benefited from referrals to community resources does not minimize the importance of the findings. In fact , this might be true only if these external services were equivalent for both groups. This report does not provide adequate information about services and clients. It relies on matching to achieve equivalency, but does not statistically verify that it has been accomplished. Its most serious limitation, however, lies in the possibility that the families who were directed to family therapy rather than traditional probation services were selected precisely because the judges and probation officers determined that they would be more amenable to such an intervention. It is quite possible, therefore, that the two groups were not equivalent on salient interactional characteristics. McPherson, McDonald, and Ryer (1983) also compared family therapy with probation casework. Families of status, misdemeanour, or felony offenders whose offences were serious enough to warrant court supervision were systematically assigned to family therapy (n = 15) or probation services (n = 60) . The family therapy, which was provided by a single co-therapist team, lasted three to four months. The treatment appears to have been based on a combination of systemic and behavioural assumptions and techniques. Individual (parents only and delinquent only) and conjoint sessions were aimed at enhancing members' understanding and appreciation of one another, improving communication, improving discipline methods, promoting self-

50

How Good Is Family Therapy?

management skills, and examining expectations. Probation services, which were not described, were offered by fifteen counsellors. The sample was determined to be representative of children supervised by the county juvenile court, and, while basic demographics were reported, there was no information on the equivalency between the experimental and comparison groups. Outcome was operationalized according to recidivism (subsequent referrals and stays in detention) and family functioning (child-rearing attitudes, parent-adolescent communication, child behaviour problems, and the child's social competence) . Recidivism was examined four months after assignment and again three months later. While these intervals capture data at the end of family therapy and three months after termination, there is no information about the treatment status of the probation group. Notwithstanding this lack of information, the family therapy group significantly outperformed the probation group on all recidivism dimensions, with the exception of stays in detention at four months. The findings related to family functioning were far less conclusive. Family functioning data (standardized self-reports) were collected only at a point three months after assignment. The lack of pretreatment measures makes it impossible to judge pretreatment equivalency and whether or not either group changed over time. Regardless, the analysis indicated that the only significant difference between groups was that parents in the experimental group judged their child to be more socially competent that did their comparison group counterparts. This study clearly indicates that the experimental and comparison families had different experiences related to recidivism. The weak design, however, makes it impossible to attribute this difference to the treatment, especially since there is no information about the intensity and duration of the probation services. Similarly, the family functioning findings offer little information, beyond the fact that the two groups differed on one dimension at one point in time. Based on their findings, the researchers state that there is no support for the assumption that improved family functioning was connected to lower recidivism. In actuality, the study was not designed to adequately explore this question . Gordon and colleagues (1988) tested an adaptation of functional family therapy for a sample of fifty-four lower socioeconomic multiple offenders. In contrast to the original model, the treatment was home based and open ended. Twenty-seven families were ordered to family

Juvenile Delinquency and Conduct Disorders 51 therapy on the basis of serious family conflict evidenced by the fact that a parent had requested the child's removal. The treatment was provided by graduate students who were trained in the model. The twenty-seven comparison families were randomly selected from the population that came to the courts during the same period as the experimental group. Both groups participated in monthly or bimonthly probation monitoring meetings. The comparison group received no other services. Recidivism was examined between twenty-seven and thirty-two months after assignment. Males and females were analysed separately, probably because the latter were exclusively status offenders. Significantly fewer males in the family therapy group committed subsequent offences (20 per cent), when compared with males who received probation only (65 per cent) . None of the females in the comparison group were recidivists, while 75 per cent of their female comparison group counterparts committed additional offences. This study provides general support for the application of ft.mctional family therapy with a more difficult population. It should be noted, however, that the design of the study ensured that the experimental and comparison groups were not equivalent. While the experimental group was deemed to be at higher risk because of family conflict and parental requests for child placement, it is possible that the group was also more appropriate for a family-centred intervention. The absence of statistical comparisons between the groups on demographic and family functioning variables makes it difficult to determine what other factors might account for the differences in outcome. Two additional gaps in information are problematic. The authors contend that an open-ended treatment model is indicated for the population, but we are given no information regarding the duration and intensity of the experimental treatment. Similarly, we are told that the experimental group was selected because of parental requests for child placement, but placement is not examined as an outcome. Henggeler et al. (1986) and Henggeler, Melton, and Smith (1992) report on the effectiveness of multisystemic therapy (also termed 'family-ecological treatment') with juvenile offenders. Multisystemic therapy is conceptually based in family systems theory and social ecology (Bronfenbrenner 1979) . The model includes problem-focused interventions with families, peers, schools, and other systems, and also emphasizes child development variables. Interventions are individual-

52

How Good Is Family Therapy?

ized, based on comprehensive assessments, and they often include behavioural as well as systemic techniques. The first study (Henggeler et al. 1986) compared multisystemic treatment (n =57) with an alternative treatment (n =23) and a sample of non-clinical adolescents (n = 44) who served as developmental controls. The alternative treatment group was matched with the external group on demographic variables and histories of juvenile arrest. The developmental control group was matched on demographic variables. Subjects were predominantly black lower class males with an average of two previous arrests, including for relatively serious crimes. The multisystemic treatment was provided by fourteen doctoral psychology students who received training in the model and participated in ongoing supervision. The style, intensity, and duration of treatment varied considerably. An average of twenty hours of treatment was provided to families in this group, with a range from two to forty-seven hours. Termination occurred when presenting problems were eliminated or when further progress was unlikely. The authors are not specific about how these judgments were made. The alternative treatment group received services that reflected the range of interventions that were typically provided for juvenile offenders and their families in the locale of the study. These services, which were provided by community agencies, included family counselling, individual counselling, recreation programs, and alternative educational programs. Clients receiving alternative treatments averaged twenty-four hours of intervention over three months. Families in the developmental control group were recruited from local high schools. The control families had no history of arrest or psychiatric referrals, and adolescents' scores conformed to the nonclinical norms on a standardized behavioural problem checklist. Outcome variables included measures of personality, behaviour, family relationships, and family interaction. Curiously, recidivism was not explored. Treatment families were pretested upon referral and posttested three weeks after termination. Control families were evaluated at similar intervals. Overall, the analysis was detailed and well documented. Adolescents who participated in multisystemic therapy showed significant decreases in problematic behaviour related to conduct problems, anxiety-withdrawal, immaturity, and socialized aggression, while the alternative treatment and developmental control groups showed no significant change. No significant findings emerged on the

Juvenile Delinquency and Conduct Disorders 53 personality measure. While self-reports of family relations yielded no significant findings, the observational measures yielded several results that indicated improved family functioning for the multisystemic group and no significant changes in the alternative treatment group. It should be noted that the developmental control group also exhibited some improvement. In the second study, Henggeler, Melton, and Smith (1992) provide a rigorous comparison of multisystemic therapy with usual probation services. Multiple offending delinquents with serious and often violent records were randomly assigned to multisystemic treatment or probation services. The experimental treatment was provided by experienced therapists who carried four families at a time. Treatment lasted an average of thirteen weeks, was primarily home based, and varied in intensity. Fidelity to the model was monitored throughout. Probation services included a minimum of one meeting per month and focused on the delinquent's compliance with court-imposed conditions. Outcomes were assessed at fifty-nine weeks after referral and included measures of criminal behaviour and incarceration, as well as standardized measures of family functioning, peer relations, and adolescent symptomology and social competence. The analysis was appropriate, well documented, and included attrition analysis. Multisystemic treatment showed significant effects over probation services for arrests, recidivism, and days incarcerated. In addition, two of the psychosocial measures revealed significant differences. Families in the experimental group showed increased cohesion, while the comparison group experienced a decrease. Peer aggression decreased in the multisystemic group, but remained the same in the comparison group. Multisystemic treatment was found to be equally effective for families with different demographic and psychosocial characteristics. These studies provide preliminary support for the effectiveness of multisystemic treatment over probation services and other services that are typically used. While both studies highlight significant findings, there is little discussion of the variables that did not yield statistically significant differences. For example, it would be important to understand the lack of change in self-reports of family functioning, in light of significant improvements assessed through observational measures. As the researchers note, there is a need for more follow-ups, comparisons with specified alternative treatments, and specification of the multisystemic treatment process. The latter is especially important in light of the flexible and individualized nature of the model. Until

54 How Good Is Family Therapy? such specification is accomplished, it is difficult to determine whether multisystemic treatment is a cogent treatment model or simply a means of more broadly conceptualizing the problem combined with the extra efforts expended by enthusiastic therapists in the context of a demonstration project. The final study involving community-based treatment (Bank et al. 1991) compares parent training with family therapy in the treatment of families with chronic delinquents. Families of youths with multiple offences, at least one of which was a non-status offence, were randomly assigned to the Oregon Social Learning Center program in family management (n = 28) or a community control condition (n = 27) that included family therapy for all families and group counselling for about half of the families. Delinquents averaged fourteen years of age and had an average of eight prior offences. In general, parents had high school education and were skilled or semiskilled labourers. Thirty per cent of the families were headed by single parents. The parent training was conducted from a social learning (social interactional) perspective. Service was open-ended and involved an average of twenty-two hours of direct contact along with twenty-three hours of telephone contact. Families in the community control condition were not tracked directly, but were reported to receive an average of fifty hours of family therapy (behavioural and family systems) over five months. In addition, over half of the subjects participated in weekly group therapy focused on drug abuse. All families were followed for three years after intake on measures related to delinquent behaviour and incarceration. The parent-training group was assessed on family management variables through observations and parent reports. Both groups showed large and significant reductions in rates and prevalence of offences by the end of the threeyear follow-up period. The reductions for the parent-training group, although not significantly greater, did occur significantly sooner. Moreover, this was achieved while utilizing much less incarceration. There were no significant findings on family functioning for the parent-training group. This study was quite well conducted, although it was not able to adequately track the experiences of the community control group. In effect, it is a comparison between a well-specified parent-training model and a non-specified family therapy or group therapy model. Nevertheless, as the researchers note, both conditions appear to have

Juvenile Delinquency and Conduct Disorders 55 achieved the same results, although the parent-training intervention may operate more quickly and rely less on incarceration. On the other hand, they question the feasibility of implementing the approach on a large scale, because of the difficulty of the work. They maintain that the theoretical framework was proven to be sound, despite the lack of findings related to family functioning, and they recommend earlier and more intensive intervention. In general, there is considerable support for the efficacy of family therapy over probation, and to some extent over other communitybased services, for relatively serious and multiple offending delinquent youth. The evidence is most pronounced in relation to recidivism and far less convincing in the area of improved family functioning. While there is not sufficient information to declare one model to be more effective than another, many of the approaches reviewed here combine systemic and behavioural interventions to achieve relatively successful results. None of the studies, however, compared well-defined and articulated models with one another. At best a particular model is compared with a vaguely defined typical treatment. In other cases the experimental treatment is poorly defined or so encompassing that it is difficult to determine what is being compared. Perhaps most important, family therapy is usually compared with interventions of a different order. While both probation and family therapy seek to reduce recidivism, the latter operates on the assumption that this occurs, primarily, through alterations in family functioning. Future research must focus on family functioning as well as comparisons between well-defined family therapy models. The next two delinquency-related studies deal with attempts to reunify incarcerated adolescents with their families. Bogert and French (1978) compared the experiences of 239 adolescents served by a specialized unit with the randomly selected records of 239 incarcerated adolescents who did not participate in specialized treatment. The adolescents in the experimental group were deemed home placement failures and had been removed from their homes by court order to await long-term out-of-home placements. Offence histories and other basic information were not reported. The treatment, which was aimed at returning the child home, consisted of individual counselling, group therapy, and family therapy. Family therapy, which occurred once a week, was described as the core treatment. The family therapy varied by therapist and loosely conformed to what was termed

56 How Good Is Family Therapy? 'a short-term crisis intervention model.' Service statistics for the comparison group were not described. Cases served during one year of the unit's operation were compared with the records of the comparison group. Forty-one (17 per cent) of the youth in the experimental group returned home, as compared with 107 (45 per cent) of those in the comparison group. The majority (71 per cent) of those in the experimental group who were returned home were still at home one year later, as compared with 39 per cent of those in the comparison group. Moreover, probation was terminated for 29 per cent of the experimental home placements and 8 per cent of the comparison home placements. Similarly, the experimental group showed a lower rate of recidivism after home placement (44 per cent), than did the comparison group. This study is of limited value because it lacks information regarding treatment details and the equivalency of the groups it examined. While a greater number of the comparison group returned and remained at home, a greater proportion of those who were returned home through the experimental treatment remained at home. The experimental group also appeared to be more successful at reducing further involvement in the criminal justice system. The poor design of this study makes it impossible to attribute these effects to the experimental treatment or to family therapy as a component of that treatment. Minimally, it appears that the specialized unit which was aimed at achieving home placement may have achieved a means of more accurately identifying those who were most likely to remain at home. A somewhat more rigorous examination of an effort to return incarcerated adolescents to their families is offered in Barton et al.'s (1985) replication of functional family therapy. Thirty seriously offending incarcerated youth were referred to functional family therapy on the basis that they would be returning to environments that were at least minimally committed to maintaining them in their own homes. There was a temporally yoked comparison group of youth matched on demographics who had been incarcerated at similar times for similar offences. Subjects were primarily from lower middle and lower socioeconomic classes. Interventions began approximately three weeks prior to the delinquent's release from the institution. Families in the experimental group received an average of thirty hours of service which included functional family therapy, supplemented with supportive educational and employment services. Subjects in the comparison condi-

Juvenile Delinquency and Conduct Disorders 57 tion were in treatment group homes and received similar supportive services. Outcomes were examined fifteen months after treatment and focused only on criminal activity. Significantly fewer adolescents in the functional family therapy group (60 per cent) had been charged with an offence, in comparison with the alternative treatment group (93 per cent) . For those youth who did commit offences, those in the functional family therapy group did so at a significantly lower rate than those in the comparison group. There were, however, no significant group differences in the severity of offences. This study adds to the accumulated evidence in support of functional family therapy, assuming that receiving functional family therapy was the primary difference between the two groups. Unfortunately, there was not sufficient information to determine if the groups were equivalent in other respects. In addition, the absence of family functioning measures makes it difficult to confidently attribute the effects to the provision of functional family therapy. There should have been some consideration of the child's place of residence at the point of evaluation. Preventing recidivism and successfully reunifying delinquents with their families are separate, though related goals. Nevertheless, there is sufficient reason to explore the benefit of functional family therapy with seriously offending incarcerated youth. Family therapy would seem to be a logical choice for facilitating the return of incarcerated youth to their families. Curiously, there are few studies that examine the application of family therapy in this context. The two studies reviewed here provide tentative support for including family therapy in the treatment program for such youth, but much more extensive research is required before conclusions can be drawn. Conclusion

As stated at the outset, issues related to adolescents and adolescence have been a primary focus for family therapy since its inception. The particular label, descriptor, or diagnosis that is associated with problematic adolescent behaviour is typically more a function of the service system encountered than of the behaviour itself. This makes it exceedingly important for investigations into effectiveness to specify the characteristics of the samples involved. Unfortunately, this is usually not the case. Consequently, studies tend to be defined according to service

58

How Good Is Family Therapy?

sectors rather than the social, emotional, and behavioural attributes of the families involved. Considerable attention has been paid to the effectiveness of family therapy with adolescent status offenders. While a number of vaguely defined models have been tested with mixed results, the investigations related to functional family therapy stand out as carefully conducted research on a well-defined model. While replications with more varied samples are required, there is considerable empirical support for designating functional family therapy as the treatment of choice for status offending adolescents and their families. It is not possible, however, to be as definitive for other forms of juvenile delinquency. While family therapy, and especially those models of it that combine systemic and behavioural elements, appears to be more effective than the services typically provided for delinquent youth, no particular model stands out. This state of affairs is partly a consequence of selecting homogeneous samples, but also it stems from a lack of direct comparisons between well-elaborated approaches. Moreover, the studies reviewed here have not provided consistent evidence to support the assumption that there is a relationship between changes in family functioning and behavioural changes in the delinquent youth. Issues related to adolescence are among those most encountered by family therapy practitioners. Overall, there is cause for cautious optimism about the effectiveness of family therapy for this population. A more definitive endorsement, however, is not possible in light of the many limitations apparent in the research to date.

4

Adult Psychiatric Problems

This chapter will examine the efficacy of family therapy to treat schizophrenia and affective disorders, two very complex and multifaceted conditions. We have confined our review to selected controlled studies (with one exception) and have excluded not only other well-known studies, but reports of exclusive use of family-oriented educational programs which are known to be quite effective. This chapter is organized in a way that highlights the development in the focus and objectives of research almost decade by decade. The early history of family therapy and schizophrenia covering the decades of the 1950s and the 1960's shows a preoccupation with finding causal links between family dynamics and schizophrenia. This was followed by a series of studies that examined family crisis and exacerbation of the disease, and in more recent history family crisis has been replaced by more complex issues of family stress and its impact on schizophrenia with almost a total rejection of the etiological influence of family dynamics on schizophrenia. The goal of family therapy lately has been to reduce the level of family stress to prevent hospitalization as well as to foster better integration for the patient with the family and community. Schizophrenia and Family Therapy

Family therapy to treat schizophrenia has by now more than a thirtyyear history. From Frieda-Reichmann's concept of a schizophrenogenic mother, to Laing's 'mystifying parents' that could cause madness in the family, to the Palo Alto group's discovery and refinement of the double-bind theory and its role in the etiology of schizophrenia and its subsequent course there are several decades of preoccupation with the

60

How Good Is Family Therapy?

role of family factors in the causation of this perplexing syndrome. Lansky (1980) observed that 'while the search for the specific and convincing etiological role of family dynamics in any major psychiatric disorder has been strikingly unsuccessful, the anti-medical and antiindividual bias remains.' The question of the etiology of schizophrenia is uncertain and lies clearly outside the scope of this chapter. What is undeniable is that by now family therapy and schizophrenia have a time-honoured association. Massie and Beels (1972) observed that in the family therapy literature the definition of schizophrenia was so loose as to encompass a spectrum of illnesses, which meant that early research suffered from 'lack of diagnostic specificity.' Their second concern was that family therapy itself suffered from such lack of specificity. They reviewed eight early and almost pioneering studies which included the Galveston, Philadelphia, Palo Alto, National Institute of Mental Health (USA), and London investigations. Their conclusion was less than enthusiastic. The actual number of studies were few and limited in scope, the number of families treated was small with inadequate follow-up periods, and the studies were uncontrolled. Nonetheless, they concluded guardedly that 'family techniques do give indications of effectiveness in treating schizophrenia.' Mosher and Keith (1980), in their comprehensive review of psychosocial approaches which included family therapy to treat schizophrenia, concluded that 'questions [regarding outcome of family therapy] are easy to pose but notoriously difficult to answer. Indeed, questions about efficacy may be premature.' The main reason for the their despondency was that in 1980 they could only find three controlled studies of 'the outcome of family therapy that include a substantial proportion of schizophrenics.' In this chapter we will examine the extent to which this state of affairs has changed in the past decade. First, we review the three controlled studies that spanned a period of nearly fifteen years - from the early 1960s to the late 1970s. The three studies concerned were (1) a series of studies conducted by Langsley and colleagues in the mid- and late 1960s; (2) Ro-Trock, Wellisch, and Schooler (1977); and (3) Goldstein and associates (1978). Controlled Trials during the l 960s and l 970s

During the 1960s the Family Treatment Unit at Colorado Psychiatric Hospital implemented a family-oriented treatment program for

Adult Psychiatric Problems 61 schizophrenics. Several reports were published documenting what was and remains a most novel approach to the treatment and management of severely mentally ill persons. Space does not permit a detailed description of their family crisis therapy (FCT; see Langsley et al., 1968, for a detailed description of this approach) , but the key elements were family visits, home visits, telephone calls, and contact with social agencies. The mean duration of treatment was twenty-four days. The goal of FCT included, recompensation and readjustment to the usual environment and usual role performance.' Langsley, Flomenhaft, and Machotka (1969) compared 150 FCT patients with 150 hospitalized cases (mean duration of treatment, twenty-eight days) . The sample comprised acutely disturbed schizophrenics and other acute suicidal and behaviourally disturbed individuals. Outcome was assessed on rates of rehospitalization at the six- and eighteen-month follow-up. In addition, the Social Adjustment Inventory (SAi) and Personal Functioning Scale (PFS) were used as pre-post measures. Data analysis was confined to chi square. The two groups were well matched on demographic variables. The key findings were that (1) with regard to rehospitalization the FCT group reported significantly better outcome than the hospitalized cases; (2) on the SAi the trend was in the direction of better role performance for FCT than for the hospitalized group, but neither hospitalization nor FCT changed long-term maladaptive behaviour; and (3) on the PFS the outcome was more favourable at the six-month follow-up for FCT subjects who were 'out of commission ' for five days as opposed to twenty-three days for hospitalized subjects. Several years later Goldstein and associates (1978) conducted a controlled study of 104 young acute schizophrenics following a period of brief hospitalization. The patients were randomly assigned to four treatment conditions, namely, high and low levels of phenothiazine and presence or absence of social therapy. The latter was family 'oriented' and involved six sessions of crisis-oriented family therapy with significant others. The primary goal of this treatment was 'to help the patient and significant others use the events of the psychosis, rather than sealing it over and deflecting attention away from the psychotic episode.' The Brief Psychiatric Rating Scale was used to measure outcome. The research design was complex, and the analyses examined drug status, therapy status, premorbid level, and sex of patient in a 2 x 2 x 2 x 2 design. Analysis of covariance was used.

62 How Good Is Family Therapy? The high dose of drug and family therapy group of patients registered maximum benefits, the rate of relapse being O per cent at the end of six weeks, while for the low dose and no therapy the rate was 48 per cent. At the end of six months not a single relapse occurred, and the authors concluded that a 'combination of adequate dose level and a variant of crisis oriented family therapy seems to be the treatment combination of choice.' Unlike the previous two studies, which used crisis-oriented family interventions, the study by Ro-Trock, Wellisch, and Schooler (1977) acknowledged the developments in family therapy from psychodynamic to systems to behavioural approaches, and they described the actual process of family therapy. First, the therapy was directed at the nuclear family; second, the therapy was systems based; and third, the goal of the family therapy was to shift 'the balance of maladaptive relationships among all members of the nuclear family system so that new, adaptive forms of relating would become possible.' However, they failed to make explicit their assumption that family factors were either implicated in the etiology of this condition or contributed to the maintenance of pathological family homeostasis. This study involved twenty-eight hospitalized, mostly schizophrenic, patients, age thirteen to twenty-two. They were randomly assigned to family therapy and individual therapy groups. The two groups were closely matched on demographic variables. Both groups received ten sessions of family or individual therapy. Outcome was assessed on the basis of rehospitalization at three months, Family Interaction Coding System (FCIS), and an observational measure using the Revealed Differences Test (ROT) as the stimulus. The Parent-Adolescent Communication Inventory (PACI), Marital Communication Inventory (MCI), and the Interpersonal Checklist (ICL) were the other measures. Standard statistical methods, which included analyses of variance, chi square, and multiple t tests, were used. This study was one of the few that tried to control for therapist bias. A most unequivocal finding was that of readmission at the threemonth follow-up. None of the fourteen subjects in family therapy was readmitted. In contrast, six of the fourteen in individual therapy were. In addition, subjects in the family therapy group took less time to return to functioning and significantly improved family communication. The authors concluded that 'family therapy was more effective than individual therapy for hospitalized adolescents in terms of the adolescents'

Adult Psychiatric Problems 63 community adjustment and hospital recidivism.' On the negative side, the sample size was small, offsetting some of the benefits of sophisticated analyses. No information was reported on the use of psychotropic drugs, and the initial follow-up period was only three months. In their three-year follow-up, the results were completely reversed (Wellisch and Ro-Trock 1980). Twenty-four of the original twenty-eight families were contacted by telephone, and mothers of subjects were interviewed. In broad terms, 57 per cent (fifteen patients) of the family therapy group as opposed to 20 per cent (seven patients) of the individual therapy group were rehospitalized by the end of three years. Remission of a psychiatric patient is a complex phenomenon and perhaps no single factor can adequately explain it. The short-term nature of the effects of family therapy cannot be completely rejected, however, although the reasons for the superiority of individual therapy are not self-evident. The authors were silent on the possibility of other intervening variables that might have contributed to the dramatic reversal of fortune of family therapy. They simply reiterated the obvious that individual therapy was superior to family therapy in their study and offered little in the way of explanation or theoretically viable speculation for their remarkable finding. They found support for their finding in rather poorly conducted studies done in the past. In subsequent studies that we review below, evidence for the superiority of individual therapy over family therapy over the long term has not been found. Comment: First and foremost, it should be noted that the population reported in the Langsley and Ro-Trock studies, while predominantly schizophrenics, were not exclusively so. During the period of Langsley's studies, systems-based family therapy models or approaches were still in the early stages of development, and the literature was relatively sparse. Langsley, with a great deal of ingenuity and insight, anchored the family intervention to crisis theory, on the one hand, and emerging research on life-change events and their relationship to morbidity, on the other. Langsley's approach, though substantially effective in preventing recidivism, and evidently not predicated on any theories of family etiology of schizophrenia, was somewhat short on explaining the changes that occurred in the families to so fundamentally alter their way of dealing and living with a schizophrenic patient. Nevertheless, Langsley and his colleagues made a significant contribution by extending individually based crisis intervention to the family.

64

How Good Is Family Therapy?

Goldstein's study, despite its complex design, essentially replicated the Langsley findings. The outcome measures were more sophisticated, but issues pertaining to family dynamics were not addressed. Ro-Trock's family therapy approach was clearly predicated on the systems model, and one of the goals of family intervention was altering families' communication patterns. Some of the shortcomings of this study were the small sample size and lack of information regarding use of psychotropic drugs. The spectacular difference in the shortand long-term effects of family therapy was not adequately explained. This body of controlled studies yielded one consistent result, namely, that in the short run family intervention significantly reduced rehospitalization for schizophrenics and other severely disturbed psychiatric patients. Research during the 1980s

In this section we review the works of Falloon, Leff, and Glick. The current state of outcome research in this area is well represented by the combined effort of these three investigators and their partners. Their studies were controlled and included follow-up. Their methods were complex, and they employed state-of-the-art statistical methods for data analyses. These three investigators and their colleagues share a common view that family stress, while playing no known role in the etiology of schizophrenia, produced a worsening effect on the patient (either resulting in readmission or in the failure to function in the community), and the goal of treatment, therefore, was focused on amelioration of such stressors. Unlike many of their predecessors, these researchers adhered to very rigorous clinical diagnostic criteria for schizophrenia. Our decision to confine the review of the 1980s research to their work was also prompted by the simple observation that their individual contributions to the field were substantial and representative. During the 1980s a good deal of family therapy research in this field originated in the United Kingdom. The Medical Research Council Social Psychiatry Unit at London provided much of the impetus for this research. The implication of family dynamics in the etiology of schizophrenia had remained unproven. The focus of research shifted from earlier preoccupation with the contribution of family relationships (especially the perpetrator mother and the victim son) to the etiology of schizophrenia, to a consideration that the family members were

Adult Psychiatric Problems 65 engaged in complex interaction which could cause the patient to relapse. Leff (1989), on the basis of his extensive research, concluded that pathological family communication as the cause of schizophrenia (families do not cause schizophrenia) was untenable and proposed that while schizophrenia was a disease of the brain, 'the family representing the most intense emotional relationships in the patient's social milieu, has the capacity to augment stress or alleviate it.' Leff further observed that the 'central plank of this [his] research has been a measure of a relative's emotional attitude toward the patient' (1989). The Camberwell Family Interview was developed to elicit information on those attitudes and comprised the following scales: critical comments, hostility, overinvolvement, and warmth. Preliminary investigation demonstrated a discernible relationship between relatives' hostile attitude towards, overinvolvement with, and criticism of the patient, and relapse over a nine-month period following the patient's discharge from hospital. This conceptualization for maintenance and worsening of schizophrenia was not dissimilar to Minuchin's formulation of psychosomatic disorders (see Chapter 8), the essence of which was also that certain family characteristics created optimum conditions for the maintenance and deterioration of childhood asthma, anorexia nervosa, and certain types of pediatric diabetes. Leff, unlike Minuchin, departed from another conventional wisdom in family therapy circles in that he viewed the person with schizophrenia as ill (suffering from a brain disorder) without any metaphorical baggage. The task of the therapist was 'to maximise the success of the family in coping with the illness in their midst.' Falloon and associates (1986), in the same vein as Leff (1980), rejected any involvement of family dynamics in the etiology of schizophrenia. Falloon was, indeed, trenchant in his condemnation of the 'new epistemology,' whereby 'not only is the sufferer of the mental illness scapegoated as a functionless individual in the society, but his entire family is implicated as "pathogenic.'" He combined behavioural family therapy with educational approaches, which he termed a problem-solving approach, to treat families harbouring schizophrenic patients. Falloon proposed that 'the family is the basic unit of health, and it is a crucial determinant in the recovery and rehabilitation of the mentally ill members.' He also noted that families found the task of altering long-standing communication patterns hard, but were far more responsive 'to improve their problem-solving by adopting a more structured approach.'

66

How Good Is Family Therapy?

Review ofFalloon's Work In their reports over the past decade, Falloon and his associates (Falloon et al. 1982; Falloon et al. 1985a; Falloon et al. 1985b; Falloon and Pederson 1985; Falloon, Pederson, and AJ-Khyall 1986; Falloon et al. 1987) have consistently demonstrated the superiority of behavioural family treatment with its focus on problem solving, as opposed to individual management, designed to improve their problem-solving skills and reduction in excessive emotional responsiveness of family members close to the patients. In their 1982 study Falloon and his colleagues compared homebased family therapy with clinic-based individual supportive therapy for patients with a definite diagnosis of schizophrenia and a high family score on Expressed Emotion (EE) on the basis of the Camberwell Family Interview for most families. A few families scored low on the EE, but provided overwhelming evidence of high stress (Falloon et al. 1982). Thirty-six patients were randomly assigned to the two treatment groups. Outcome was measured on the severity of schizophrenic symptomatology, social functioning of the patient and family members, and community tenure, which translated into number of days spent in psychiatric facilities and number of hospital admissions. Family therapy was designed to reduce environmental stress and comprised techniques such as behavioural rehearsal, modelling, and social reinforcement. The first two sessions were used to educate the families and the patients about the disease. All patients in both treatment conditions were seen once a week for the first three months and biweekly the next six months, with monthly visits thereafter, for a total period of nine months. Three therapists provided the two treatments. Outcome data were based on the assessment at nine months. The results at the end of nine months clearly showed the superiority of the family over the individual intervention. Six per cent of the family group experienced a clinical relapse as opposed to 44 per cent in the individual group. As for days in the hospital, the family therapy patients spent 0.83 days and theindividual therapy patients 8.39 days. Finally, the blind ratings of schizophrenic symptoms supported the superiority of the family over the individual approach. In 1985 Falloon and associates reported on the results of their two years of follow-up on the above group of patients (Falloon et al. 1985a). At the fifteen-month follow -up six major episodes of psychopathology were reported in the family-managed group and thirty-one major episodes in the individually managed group. At the two-year follow-up

Adult Psychiatric Problems 67 eighteen subjects in the family group had experienced thirty-six florid episodes, while the eighteen subjects in the individual treatment group had experienced fifty-four episodes. The most telling finding was that fifteen of the individually treated subjects had a substantial episode of schizophrenia, and only three in the family-treated group had experienced such an episode. This was reflected in the admission rates of the two groups. The individual group had ten individuals admitted for a total of 408 days, but only four family-treated subjects had been hospitalized, and these for a total of sixty-six days. Maintenance of improvement from the the nine-month follow-up was also noted with target symptoms and blind rating of clinical state. In 1987 Falloon and associates further reported on the social functioning of the thirty-six schizophrenic patients discussed in their earlier reports. On the SAS-SR analysis of covariance showed the superiority of the family over the individual treatment approach. There was a significantly greater overall social adjustment in the family-treated group. On the Social Behavior Assessment Schedule structured interview with a key family member at the twenty-four-month assessment the family treatment group compared with the individual treatment group demonstrated significantly greater improvement in the performance of household chores, decision making, and leisure activities; this result approached statistically significant values. The essential findings were (1) significantly greater overall social adjustment by the family management cases; (2) family therapy patients had been functionally active for 12.6 months, compared with 7.2 months for the individually managed cases; and (3) major gains in social functioning were observed in the family group during the first nine months, but 50 per cent of the individually treated cases reported deterioration in their social functioning. The researchers concluded that 'social morbidity was reduced more efficiently by family than individual management during two years after an episode of schizophrenia.' In a rather novel pilot study, Lieberman, Falloon, and Aitchson (1984) treated fourteen young schizophrenic (mean age, twenty-six years) adults living with relatives who were high on the EE. All family members and the patients were invited to participate in familyoriented group treatment and social skills training in a store-front operation. Nine two-hour family group sessions were held. Each group consisted of three patients, their families, and three co-leaders. The first hour of the session was used for demonstrations and discussions, and during the second hour each leader worked separately with the

68

How Good Is Family Therapy?

specific problems of one family before the group reconvened to assign homework tasks. These sessions combined education as well as familyfocused problem solving. The Cambeiwell Family Interview (CFI) was used to assess the EE of family members, and it tapped into factors such as their attitudes and feelings towards the patient. This approach combined the essential elements of family group work and conventional family therapy. The key findings were (1) that families demonstrated a statistically significant increase in their knowledge about schizophrenia; (2) eleven high-EE parents attending the family group recorded a 60 per cent decline in their critical comments; (3) 21 per cent of patients relapsed during the first nine months. This was in contrast to two London studies that had reported a relapse rate of 58 per cent and 48 per cent at nine months following discharge. Comments: Falloon and colleagues, on the basis of their one study which spanned several years, have demonstrated the value of a particular type of family therapy which was a combination of education and information at the outset, followed by behavioural family therapy. This intervention was designed to improve the coping strategies of the patients and their intimates. The study had many strengths: sophisticated design, clear diagnostic group, explicit therapeutic strategies, a control group, multiple outcome measures, relevant statistical analyses, short- and medium-term follow-up, and recognition of some of the shortcomings of the study. On the negative side, the extent of participation and the actual membership of each family remained unanswered. For example, there was no mention of siblings or of who the actual participants in each family session were. That information is essential to correctly assess the intrinsic value of family treatment. Was it necessary to include all family members, or was it enough to have one parent participate? The implications may be of great import for clinicians who may wish to implement or replicate the treatment. The other point of note is that the actual number of subjects was relatively small. EE was used as a baseline measure but was not reported in the outcome (except in the 1984 pilot study) . That information would have demonstrated changes in the family dynamics - an important goal of treatment. Since three therapists treated both groups, it was quite likely that the results were somewhat confounded by therapist bias for family therapy. It should also be noted that at the two-year follow-up both groups showed improvement, which was not pronounced at the

Adult Psychiatric Problems 69 nine-month follow-up. A critical question arising out of the study and which requires further investigation is the relative benefit of each treatment condition for specific individuals. In other words, do some patients respond better to family treatment and others to individual modalities, and, if so, what might the distinguishing characteristics of these individuals be? Review ofLeff's Work Leff and his associates (Leff et al. 1982) reported a study of a controlled trial of social interventions with families of schizophrenic patients at high risk for relapse. The social intervention package included three elements, namely, an education program, a relatives' group, and family sessions. Twenty-four patients with Expressed Emotion relatives were assigned randomly to routine out-patient care and social interventions. The aim of the family therapy component of the package was to reduce the EE. The assessment package consisted of the Present State Examination, the Camberwell Family Interview, and the Knowledge Interview which was constructed to test the efficacy of the education program. Follow-up occurred at nine months after discharge; t tests were used for data analysis. The superiority of the social intervention package over the outpatient treatment was convincing. Significant gains were made in the experimental group by about 75 per cent of the families in reducing the critical comments and overinvolvement scores of the EE. There was only a non-significant change in EE in the control group. In addition, the relapse rate for the experimental group was 7 per cent (one patient) in contrast with the 50 per cent (six patients) in the control group. The overall objectives were obtained in 73 per cent of the experimental families, and no relapse occurred in this group of patients. At the two-year follow-up (Leff et al. 1985), for patients who remained on medication, social intervention significantly reduced the relapse rate (14 per cent compared with 78 per cent for the control group). However, two patients from the experimental group (one of them successfully) attempted suicide as did three from the control group. This study had two goals. The first goal was to demonstrate a causal relationship between high EE and relapse for schizophrenic patients, and the second was to test the effectiveness of social interventions to reduce relapses in schizophrenic patients. An obvious problem with this study was the wide variability in the number of sessions of family

70

How Good Is Family Therapy?

therapy, ranging from one to twenty-five. The analysis did not adequately control for this enormous anomaly. Yet another problem of some magnitude was that the membership in the family sessions varied from session to session. The process of family therapy or its theoretical orientation was not described. Furthermore, family therapy was only one component of the social intervention package. We shall presently elaborate these issues. In a subsequent study Leff and associates (1989) at least addressed one of the shortcomings of the previous study by comparing family therapy with social intervention strategy which included education plus a relatives' group. Biweekly family therapy sessions (median, seventeen sessions) were conducted at home, and, as before, other than to state that family therapists were experienced, no other information was provided about the therapy itself. The social intervention group also met every other week (median, 4.5 sessions). The schizophrenic patients were in high contact with their relatives who in turn had high EE scores. Eleven families were assigned to family therapy, and only six agreed to social intervention. The relapse rate at the nine-month follow-up for the family therapy group was 8 per cent, and for the social intervention group it was 17 per cent. These differences were far less dramatic than the differences reported in the previous Leff study. Besides, the number of subjects, especially in the social intervention group, was so small that any generalization of the results may be perilous. Another point of note is that the amount of involvement varied substantially between the two groups, raising questions about the uniformity of the effectiveness of the interventions. Did the family therapy group, in effect, receive a better deal? Nevertheless, family therapy was demonstrated to be superior on some hard measures of outcome. Comments: The Leff group's studies were carefully designed and executed. Appropriate statistical analyses were used and outcome measures, especially prevention of relapse, were of immense therapeutic value. Treatment conditions for the two groups in both studies were made explicit, although the theoretical underpinning and the specifics of the family therapy method used deserved more space. The relative smallness of the samples minimally raises a question of their generalizability despite the unequivocal nature of the results. Notwithstanding these reservations, the combined results of the two studies (1982, 1989) provide cause for optimism as far as the superiority of family therapy to

Adult Psychiatric Problems 71 achieve highly desirable goals over other psychological treatments is concerned. Furthermore, Leff left little doubt about the part played by family members with certain characteristics in effecting relapse in schizophrenic patients. Review ofGlick's Work Over the past several years Glick and his associates have reported on the efficacy of family therapy in the treatment of hospitalized psychiatric patients with schizophrenia. In a series of four articles they have reported their research in much detail (Glick et al. 1985; Hass et al. 1986; Spencer et al. 1988; Glick et al. 1990). We shall present an overview of their key findings as they relate specifically to the schizophrenic population. In their 1985 work Glick and colleagues reported on their six-month follow-up of a randomized trial of family therapy involving 184 patients, eighty with schizophrenic disorders and sixty-four with major depressive disorders. The two groups were randomly assigned to inpatient multimodal plus family intervention (IFI) and multimodal intervention only. The preliminary outcome measures included the Global Assessment Scale (GAS). which is a 100-point rating scale and measures the patient's overall level of functioning and severity of symptoms, and the Psychiatric Evaluation Form - overall Severity (PEF-OS) . Family measures included the Family Attitude Inventory, Patient Rejection Scale, and Family Evaluation Form. Four facets of family-patient and family-hospital relationships emerged as critical. They were (1) patient rejection, (2) critical attitudes, (3) family burden, and (4) family attitude towards treatment. The similarity of these characteristics with Leffs EE is noteworthy. Family-related findings were reported in 1986. For data analyses three-way analyses of covariance were performed. The goals of family therapy were lucidly described, and these included modification of stressors that might have precipitated the admission, future stressors that might do so again, family interaction that might contribute to stress, and strategies to deal with future stress. In short, promoting effective coping within the family was the principal objective of this study. In this respect, Glick shared the same objective as Falloon and Leff. Families were seen for a mean 8.95 sessions lasting for about forty-five minutes each. The process of family therapy or its theoretical orientation were not discussed. Outcome was mixed. Highly functioning schizophrenic patients,

72

How Good Is Family Therapy?

who had functioned relatively well prior to admission, and who had received IFI showed significantly greater improvement (both at the time of discharge and at the six-month follow-up) in global functioning than the comparison groups. In contrast, such improvement was not discernible in schizophrenic patients with poor prehospitalization functioning. The conclusion confirmed an old axiom that patients with better pre-morbid functioning tend to benefit most from psychological treatments. In their second report, Hass et al. (1986) provided a more complete set of family-related findings. Analyses of family data revealed a rather complex series of outcomes. First and foremost, on the family measures, family attitude to treatment and family openness to social support and, second, patient rejection and family burden accounted for 32 per cent and 24 per cent respectively of the variance in outcome. In terms of actual outcome of family intervention, a Gender x Treatment Interaction effect revealed that females improved more than males with family intervention, and for the schizophrenics with good prehospitalization functioning, families of both male and female patients showed improved attitudes. In brief, good prehospitalization functioning for both genders or female patients with poor pre hospitalization functioning were most likely to gain from family therapy. In their third report (Spencer et al. 1988), as well as their fourth and final report, Glick and colleagues (1990) presented the six-month and eighteen-month follow-up data and their major conclusions. The key findings were complex, but confirmed the usefulness of family therapy in conjunction with other treatments for female patients with a record of poor pre-admission functioning. Besides, the therapeutic effect did not manifest itself until eighteen months post-admission. The beneficial effects of family therapy were also evident in the altered attitudes of families of schizophrenic patients at the six- as well as eighteenmonth follow-up. Comments: First and foremost, the radical nature of the study must be recognized. The incorporation of family therapy in an in-patient program was the unique aspect of Glick's undertaking. Given the breadth of the study and the rigour of the design, the efficacy of family therapy was demonstrably limited. The reasons for IFI's relative success with females with poor pre-admission adjustment were far from selfevident and were not discussed.

Adult Psychiatric Problems 73 This very complex study is not without its shortcomings, which the authors themselves recognized. During the follow-up period, patients and families received various kinds of uncontrolled help, ranging from individual to family to group interventions. Although the researchers were careful in their analyses to demonstrate the effects of such interventions never to exceed chance, in individual cases the effects might have indeed been significant. Given the scope of the study, it was unfortunate that global measures of family functioning were not included. In fact, for reasons unknown, the investigators were rather stringent with relevant information pertaining to the families. For instance, information about the actual participants, their level of participation in individual sessions as well as over the entire family treatment program, the training of the family therapists, and some elaboration of the theoretical underpinnings on which their particular type of family therapy was predicated would have been enlightening. Since the purpose of the study was to test the efficacy of family therapy, an argument can be made that such factors may indeed be of some significance. Finally, the researchers, in a sense, demonstrated a certain amount of reluctance to explain their principal finding, namely, the gender effect. If this finding is valid (and it is on the basis of their study), then, of course, family therapy can be used sparingly with a preselected population. This limited use of family therapy is, however, disclaimed by their observation that 'families of patients with schizophrenia also show benefit from having received IFI,' and then there was the suggestion that postdischarge family therapy probably had a powerful effect on the final outcome. In the end, inclusion of family therapy in the treatment of hospitalized schizophrenics would appear to be only somewhat more prudent than not to do so. Before the topic of family therapy and schizophrenia is left two other studies must be acknowledged (Tarrier et al. 1988; Hogarty et al. 1986). Both had comparison groups, the goal of reducing high EE in families, and the two-year follow-up. Hogarty and associates (1986) reported a 34 per cent relapse in the family treatment group as opposed to 66. 7 per cent in the control group. Tarrier et al. (1988) reported 33 per cent relapse in the treatment group as opposed to 59 per cent in control high EE group. Both studies used some kind of family therapy in combination with a variety of other psychological and educational strategies.

74

How Good Is Family Therapy?

Family Therapy and Affective Disorders

Reasons for the relevance of family therapy in the treatment of affective disorders are varied and, on the whole, quite convincing. Anderson (1987) noted that in addition to genetic and other psychosocial forces, family dynamics were also intricately involved in the etiology, course, and outcome of these complex mental health problems. Gottlib, Wallace, and Colby (1990), in their extensive review of the literature, found empirical support for major disruptions in family relationships caused by depression. Birtchnell and Kennard (1983) clearly established the preponderance of marital and family problems in the 'neurotic' depressives compared with 'endogenously' depressed subjects. Vaughn and Leff (1976) found an association between family stress (Expressed Emotions) and relapse in patients with affective disorders. Despite the evidence the actual number of controlled family therapy outcomestudies is abysmal. Only four controlled studies, one of which is related to manic patients, exist to date (Clarkin et al. 1990; Davenport et al. 1977; Friedman 1975; McLean, Ogston, and Grauer 1975) . Explanation for this apparent lack of research interest eludes common sense. On the other hand, major breakthroughs in the pharmacological treatment of affective disorders must, in part, account for this state of affairs. Despite the very positive outcome of drug therapy, this brief review will show that family and couple therapies significantly enhance the well-being of the victims of mood disorders. All four studies, in different degrees, attest to that fact. Methods of family and couple therapy vary rather widely among the studies and range from a strictly behavioural approach to couple group therapy to marital therapy to family therapy. McLean and associates (1975) randomly assigned twenty depressed patients to an experimental group and ten patients to a control group. Each patient and spouse in the experimental group received one hour of conjoint therapy per week for eight weeks. Therapy was administered by a male and a female therapist. Patients in the control group had a mixture of interventions, and this included two subjects who did not receive any treatment. The treatment method consisted of training in social learning, feedback as to the perception of verbal interaction between patient and spouse, and training in the construction and use of behavioural contracts. Pre-post outcome measures included the Depression Adjective Check List, a half-hour tape-recording of problem-oriented discussion

Adult Psychiatric Problems 75 at home with the spouse, which was coded and listed five target behaviours considered to be problematic. These measures were also repeated three months after treatment. The experimental group showed significantly greater improvement on all outcome measures than the control group. This study, despite the impressive outcome, had serious shortcomings. The diagnosis of depression was determined by referring physicians and was not assessed independently. It is not at all clear on what variables the two groups were matched. Simple rules for setting up control groups were ignored. Demographic data were conspicuously absent. No information was given on the current treatment status of the patients in the experimental group. It would have been useful to know how many of these patients were on antidepressant drugs. The control group was small and very diverse and included two persons who were without treatment. The follow-up period was rather short. In a study of this kind, designed to demonstrate treatment superiority, maintenance of treatment effects over a longer period is a necessary measure. The study, at best, demonstrated that a variety of individuals with depression seemed to benefit from behavioural couple therapy over a three-month period. Friedman's (1975) investigation addressed many of the shortcomings of the previous study by adopting a sophisticated design to test the complicated interactional effects of drug and marital therapy in the treatment of depressive subjects. Patients were randomly assigned to four groups for treatment that lasted 12 weeks: drugs and marital therapy (n = 41); drugs and minimal therapy (n = 42); placebo and marital therapy (n = 43); and placebo and minimal therapy (n = 40). Apart from a number of psychiatric rating scales, the Family Role Task and Activity Scale and the Marital Relations Inventory were used. Multivariate analysis of variance was used to perform 2 x 2 analysis of variance. Results demonstrated considerable superiority of drugs and family therapy over their control conditions. Drugs and marital therapy produced different orders of benefit. Drugs were highly beneficial in reducing depressive symptomatology. Marital therapy was predictably more favoured in ameliorating family task-related problems and improving perceptions of marital relations. The study also pointed up a somewhat longer period for marital therapy to take effect as opposed to drug treatment. Drug therapy produced very fast changes in enhancing the perceptions of marital relations and also in reducing

76

How Good Is Family Therapy?

hostility. In contrast, marital therapy proved more efficient in producing lasting changes in both those areas by the end of treatment. This study, one of its kind, was carefully designed and conservatively analysed. A combination of marital and drug therapy was clearly shown to be more desirable than the control conditions. Those control conditions were also established with care and precision. The diagnosis of depression was objectively established. The age range of the patients was wide, from twenty-one to sixty-seven years, proving an important point that marital therapy could be applied with equal efficiency irrespective of age. The study was silent about the marital therapy itself, with no information on its theoretical orientation, or on the attributes and training of the therapists, thus making replication of the study impossible. An added shortcoming of the study was its failure to control for gender of the patient. This was a serious oversight as more women are affected by depression, and their family issues are likely to be different from those of males. Otherwise, this study provided convincing evidence of the benefits of short-term marital therapy as an adjunct in the treatment of depression. Longitudinal follow-up results of a study of this magnitude and complexity would have been very useful and would have thrown some light on the lasting effects of the interventions. Two key deficiencies of the previous study, namely, controlling for gender and longitudinal follow-up, were at the centre of an investigation by Clarkin and colleagues (1990) . This study was conducted with a group of in-patients with affective disorders. Diagnosis was based on DSM-III. Fifty subjects included twenty-nine unipolar and twenty-one bipolar individuals. Within two days of hospitalization, the subjects were randomly assigned to standardized hospital treatment and standardized treatment plus family therapy (IFI). Patients and their families received six family therapy sessions, each lasting forty-five minutes to an hour. Stress management and stress reduction were the primary goals ofIFI. Family-related measures included the Family Attitude Inventory. The patient's role functioning in the community was measured using the Role Performance Treatment Scale. Analysis of covariance was used for data analyses. Outcome was assessed at discharge, and at the six-month and eighteen-month follow-up. Results at both six and eighteen months showed that female bipolar depressives were the major beneficiaries of IFI, and the unipolar patients did better without the family component of treatment. Another important finding was that the families of female depressive

Adult Psychiatric Problems 77 patients in IFI had a better attitude towards the patient's treatment at both six and eighteen months' follow-up. However, the positive effect of IFI with all females was somewhat diminished over time. Males, on the whole, reacted negatively to IFI, which became evident at the eighteen-month follow-up. In short, IFI was at its most effective with bipolar female patients. This study, unlike any that preceded it, made a case for selective use of family therapy with major affective disorders. Undoubtedly, the study will require replication along with an improved understanding of the effectiveness of family therapy with one group as opposed to its negative repercussions with another. As with so many studies, this one was somewhat economical with information about the family therapy itself. Other than the mean number of sessions attended by the families, the report was silent on some very key issues, such as who were the participants, with what degree of regularity did the family members attend, who dropped out and why, and what kinds of changes occurred in the dynamics of the families? These factors, in addition to the ones discussed by the investigators, could have further clarified the findings or even offered alternative explanations for the findings. Retzer and associates (1991) reported their findings on the efficacy of systemic family therapy with twenty (ten males and ten females) manic-depressive subjects and ten (five males and five females) schizoaffective subjects. The outcome was assessed using a sevenitem, ten-point scale designed to measure various dimensions of family relations. They also calculated the actual number of hospitalizations for these patients prior to family therapy, and this was reassessed at follow-up. The mean number of family sessions was 5.95 for the manic-depressive group and 7.90 for the schizoaffective group. A variety of statistical techniques, such as the Yates corrected chi square, t tests, Mann-Whitney U-test, and Wilcoxon test for dependent samples were employed to offset the usual problems associated with small and unequal samples. The results at the three-year follow-up showed a 76.7 per cent reduction in relapse rates, 67.8 per cent for the manic-depressives and 89.8 per cent for the schizoaffectives. The authors claimed that these results were not in any way influenced by psychoactive drugs. In fact, 50 per cent of the schizoaffective and 30.8 per cent of the manic-depressives discontinued lithium after the completion of family therapy. Their explanation for this dramatic result was that these families were in a perpetual state of struggle with issues of autonomy on the one side and

78

How Good Is Family Therapy?

victimization and dependency on the other. The therapeutic process allowed these families to find a new balance between the two. This study, one of its kind, revealed that while systems family therapy was relatively effective with both diagnostic groups, it seemed more so with schizoaffectives. Given the sample size that finding has to be seen as hypothetical. The total sample for this study, while respectable, was quite small, especially for the schizoaffective group. The two groups were not exactly matched which does raise questions about the validity of the comparisons.The mean numbers of family therapy sessions were not, in fact, true indicators of the amount of family treatment received by individual families. For the whole group it ranged from one to nineteen sessions per family. The analysis did not take into account this enormous variability. Neither did the study reveal much about the participants in family therapy. Who were they, with what degree of regularity did they participate, did the same members of the family participate in every session, why did some families have so few sessions? Was that because they dropped out? This study must be seen as a preliminary one, and as such the validity of the findings must await replication with comparable samples of adequate size. In a follow-up study of sixty-five patients with a diagnosis of bipolar disease, eleven patients were managed on medication, forty-two patients went back to their own community for community care, and twelve received medication management together with weekly couples group therapy, which falls in the category of family-oriented therapy (Davenport et al. 1977). The groups were not comparable as there were significant differences in age, duration of illness, and duration of marriage between the treatment and the control groups. Outcome measures consisted of a four-point rating scale which assessed the patient's work status, social functioning, family adjustment, and mental status at follow-up. The treatment and follow-up period was twelve months. Chi square was used for data analysis. The results at follow-up showed that the couples group was doing significantly better in the areas of social functioning and family interaction than the community care group, and significantly better in the area of family interaction than the drug-maintenance group. The authors warned that the results were equivocal because of the between-group differences on a number of critical variables. Nonetheless, the couple therapy group did attain significant gains from their treatment experience. This study had a number of major methodological shortcomings.

Adult Psychiatric Problems 79 The failure to establish adequate control conditions rendered any meaningful comparison ineffectual. The outcome measures were rather loose, and valid instruments were not used. The absence of information about the couple therapy itself was pronounced. No information was provided on the participants, such as their level and frequency of participation. This study is best viewed as a pilot project. Conclusions How effective is family therapy in the treatment of major psychiatric disorders? First, family treatment appears to be an adjunct to drugs and other established therapies commonly used. This gives rise to a major methodological problem, namely, how to separate the treatment effects of all other modalities from those of family therapy? This becomes a function of the care with which control conditions are established, and a high level of sophistication is required for data analysis. Some studies adopted rigorous guidelines, and these were more evident with the schizophrenic studies than with the affective disorder ones. The general conclusion about the efficacy of family therapy with schizophrenics has to be that under some conditions family therapy is indeed beneficial. Lack of replication renders any larger claims premature, although the findings of the major studies taken together make a strong case for the adoption of family therapy as a necessary adjunct in the overall treatment of schizophrenia. The volume of family therapy research with affective disorder is small, and much of it is a decade or more old. The quality of research, compared with that for schizophrenics, could stand improvement. The outcome measures in several studies were less than rigorous. At this point the best that can be said is that family therapy as an adjunct in the treatment of affective disorders is not without promise. Retzer and associates (1991) have revived the notion of the merit of family therapy to the exclusion of the possible benefits from psychoactive drugs. In fact, they emphatically rule out such a possibility. Their findings would need verification in future replication studies. Glick and associates found family therapy to be selectively beneficial, such as for patients with good prehospitalization functioning and for female patients. This is in contrast to the more general benefits reported by other studies with schizophrenia and affective disorders. The respective findings cannot be readily reconciled, but they do raise the possibility that family therapy may be more effective for discharged

80

How Good ls Family Therapy?

patients whose conditions are in remission rather than for in-patients who presumably are still in the acute phase of the illness. This proposition is pure speculation since there does not exist a single study that has compared the relative merit of family therapy for hospitalized and discharged patients with major psychiatric disorders. One common criticism shared by all the studies was the investigators' general reluctance to share information about the therapy process. In some instances, the authors did report the details of the family therapy used in their research elsewhere. Even in those instances, information about the participants, frequency of participation, level of participation, drop-outs at various stages of treatment, changes in family relationships, and so forth was either not reported at all or only alluded to in passing. Therapist variables, methodologically a very challenging area, were generally ignored or compromised by the studies. Despite these criticisms, knowledge gained from these studies about the efficacy of family therapy in the treatment of major psychiatric illnesses is considerable as well as promising.

5

Psychosomatic and Medical Disorders

Medical and psychosomatic disorders enjoy a time-honoured association with family issues. Nearly fifty years ago, Henry B. Richardson (1945) observed, 'This [psychosomatic) movement was started by regarding the family as part of the individual and often ended unwittingly with a complimentary point of view as well, the individual is part of the family, in illnesses as well as in health. It is no longer easy to conceive of asthma or ulcer or obesity as a characteristic only of the individual.' These were far-fetched views in 1945. Another three decades passed before Minuchin and his colleagues claimed to articulate the centrality of family issues in conditions as divergent as pediatric asthma, anorexia nervosa, and a particular type of (brittle) diabetes (all described as psychosomatic conditions) in children (Minuchin, Rosman, and Baker 1978). Whether these conditions are truly 'psychosomatic' remains moot. Despite this long association between family dynamics and medical illness, this chapter will reveal that family therapy outcome studies with medical illnesses remain a novelty. Family therapy outcome literature on the adult patient is sufficiently paltry to compel us to incorporate in this chapter clinical reports with some evaluative component as well as family 'oriented' treatment outcome studies. This want of outcome research is made even less comprehensible in light of the enormous proliferation of clinical literature on the usefulness of family therapy with medical conditions over the past two decades. Leading the field is the voluminous literature on family intervention with childhood cancer. Among others, the benefits of family therapy for hemophilia (Ritterman 1982), trauma (Rosenthal and Young 1988), asthma (Gustaffson, Kjellman, and Cederbald 1986),

82

How Good Is Family Therapy?

diabetes (Weist, Finney, and Barnard 1993), coronary artery disease (Harding 1976; McGann 1976), renal diseases (Shambaugh and Kanter 1969), stroke (Evans et al. 1988) , terminal illness (Pettitt 1979), chronic pain syndrome (Roy 1989), and Huntington's disease (Phipps and Desplatt 1984; Murberg, Price, and Jalali 1988) have been reported. Weakland's (1977) plea to embody the 'interactional' viewpoint in the understanding and treatment of organic diseases has been, at the very least, not fully realized, partially, because the proof of the benefit of family therapy in the treatment of psychosomatic and organic diseases is yet to fully emerge. Nevertheless, clinical evidence for such benefit abounds. In a comprehensive review of the field, Sholevar and Perkel (1990) reported on the power and strength of family-systems intervention as adjunctive treatment for diseases as wide ranging as pancreatic cancer and AIDS (life-threatening diseases) to Huntington's and Alzheimer's (chronic diseases) and covering the pediatric to geriatric age span. They provided convincing case illustrations as proof of their claim. The obvious gap in their otherwise excellent report was the absence of empirical evidence to back up their clinical successes. Steinglass and his associates (1982) in discussing two cases of spinal cord injury in men in their fifties urged that embracing a family perspective was unavoidable, and training rehabilitation staff to appreciate the family dimension was a necessity. A similar stand was taken by Murberg and associates (1988) . They offered a powerful rationale for incorporating family therapy in the management of Huntington's disease. Six cases were reported to show the effectiveness of family therapy in 'solving problems related to specific family developmental crises and to shifts in family roles and hierarchies resulting from the progressive worsening of the Huntington's symptoms.' Apart from educating family members about the disease, family therapy was found useful to counteract some of the ill effects of the disease on family relationships. In a comprehensive discussion of the structural family approach to treat diseases causing disability, such as stroke, Wiley (1983) proffered a persuasive argument to implement that model of family therapy during the crisis phase of the disease. Within the systemic paradigm, Salvador Minuchin's structural family therapy has gained much support in the family treatment of medical illnesses (Griffith and Griffith 1987). Minuchin's pioneering work with diabetic and asthmatic children literally laid the foundation for family treatment of psychosomatic

Psychosomatic and Medical Disorders 83 disorders in children. His conceptual framework continues to serve as a model in the family treatment of psychosomatic and medical disorders. His research will be discussed in detail below (see also Chapter 8, on anorexia nervosa) . The application of a family-systems perspective to understand and treat complex medical illnesses has been the most significant development in the family treatment of medical illnesses. The systemic paradigm has contributed in a profound way to the understanding of the role of family dynamics in the etiology and/or perpetuation of several medical conditions. Furthermore, it has enhanced understanding of the impact of illnesses on the family (Bishop et al. 1986; Cobb, Miller, and Wieland 1959; Croog and Levine 1977; Deyo, Inui, and Sullivan 1982; Roy 1989). Hence, family therapy can alleviate or moderate certain conditions such as childhood asthma or in the face of a serious illness modify the negative repercussions on family functioning. However, the actual evidence for the part played by family dynamics in the etiology of medical and psychosomatic conditions is both sparse and controversial. Clinical evidence that family therapy can restore or improve family functioning in the face of severe or chronic medical conditions is more convincing. The weight of the empirical evidence is examined below. This chapter first reviews family therapy with pediatric conditions. Second, the adult literature is explored, and in the final section the current state of research in this field is examined. Family Therapy and Pediatric Conditions

Minuchin and his associates at the Philadelphia Child Guidance Clinic developed a blueprint for the family treatment of psychosomatic disorders in children. They identified flawed family dynamics that contributed to the development and perpetuation of such diverse disorders as asthma, 'psychosomatic' diabetes and abdominal pain. They established a clear distinction between primary and secondary psychosomatic disorders - primary being a condition where physiological dysfunction (for example, diabetes) was already present, and secondary being, where no predisposing physiological vulnerability (for example, anorexia nervosa) was evident. Regardless of physiological vulnerability, they asserted that families with psychosomatically sick children were (1) enmeshed, (2) rigid, (3) overprotective, and (4) showed an inability to resolve conflicts. The sick child played a major role in help-

84

How Good Is Family Therapy?

ing the family avoid conflict. It should be noted that there has been further refinement of their original model, and a certain amount of empirical validity for the model is also available (Gustaffson, Kjellman, and Cederbald 1986; Wood et al. 1989). A few studies have also produced evidence to the contrary (Burbeck 1979; Parker and Lipscombe 1979; Kog, Vandereycken and Vertommen 1985; Kog, Vertommen, and Vandereycken 1987). Abdominal Pain

Liebman and associates (1976) reported treating ten children with abdominal pain and their families. Key structural problems were evident in these families, and therapy, a combination of behavioural and structural family therapy, was instituted to remove them. In most cases one parent also presented with pain problems. Family therapy was reported to be effective in eliminating abdominal pain in a number of cases. These authors did not provide any statistical data for their analyses. Berger, Hoenig, and Liebman (1977) replicated the previous study with nineteen children with similar outcome. Unfortunately, this study, despite a slightly larger sample size, shared the same methodological shortcomings as the one before it. The lack of a control group and the absence of statistical analyses, combined with a high rate of spontaneous recovery and a small sample mitigated against any broad generalization of their claim about the effectiveness of family therapy. White (1979) reported on out-patient treatment of ten children with abdominal pain and their families using a combined structuralstrategic approach. Each step of the treatment was described and the general outcome reported in terms of remission of pain. Children ranged in ages from five to thirteen years. At the fifteen- and twentyfour-month follow-up recurrence of symptoms was observed in three children. This was a clinical report without any affectation of an experimental study. Hughes (1984) described psychogenic abdominal pain in twentythree children and concluded that these children, in the main, were depressed. The family aspect of treatment was mentioned in passing. Nevertheless, twenty-one of the twenty-three children improved markedly as a result of in-patient treatment of depression. This report is important, since the aforementioned studies were uncontrolled, to the extent that it offered an alternative explanation and treatment modality for abdominal pain in children. While recognizing several

Psychosomatic and Medical Disorders 85 issues in parent-child relationships as contributory factors in the etiology of pain, unlike the previous two studies, family therapy was not the central feature of treatment. In fact, Hughes and Zimin (1978) in an earlier report had recommended that if parents were confronted with serious problems they should be encouraged to find help for themselves. Comments: That family factors are implicated in the etiology and maintenance of non-organic abdominal pain in children is perhaps beyond question. The body of research, however, has only produced questionable findings about the efficacy of family therapy as the treatment of choice. Hughes's view of pediatric abdominal pain as a sign of depression was partly borne out by his report of the successful outcome of treatment and partly because his study was also uncontrolled. The value of family therapy to treat abdominal pain remains to be proven. Trauma in Children

In a clinical report involving seven male children between the ages of ten and eighteen years with head injuries, Hartman (1987) discussed the benefits of family therapy with four of them. The others were individually treated. Life-stage issues were identified as crucial. The therapeutic goal was to minimize or counteract the pernicious effects of trauma on the family. The Milan model of family therapy was the approach used. Various levels of success, based on observations by the therapist as well as by family members, were reported for all seven subjects. This was by no means an empirical investigation of outcome of family versus individual therapy. It was more in the way of providing guidelines for therapists, and yet a report of this kind has inherent value for generating testable hypotheses. Trauma of a very different order was reported by Wellisch (1985). Six adolescent children of parents with cancer were treated for actingout behaviour. These children were faced with (1) terminally ill or (2) slowly deteriorating parents or (3) situations where dying or severe disability and ultimately death were not the immediate problems, but rather the family system was strained by the very fact that there was one parent with a malignant disease. Intervention was predicated on systems-based family therapy. The developmental issues of the adolescent children combined with the disruption caused in the family

86

How Good Is Family Therapy?

system created serious conflictual situations. Positive outcome was reported on five of the six cases. This report was strictly clinical, and the family conflicts and their resolution were based on the author's personal observations. Rather limited information was provided on the treatment process. Comments: The sole purpose of mentioning the last two reports was to show the ingenuity and creativity with which family approaches are being incorporated to handle trauma in children. Their value is that they serve as a base for more formal investigations. The Diabetes Debate

Minuchin and colleagues presented their initial findings on the successful family treatment of 'brittle' diabetes, which, despite heated and not particularly enlightening debate, has basically remained untried and unproven. In their initial study Minuchin and colleagues (1975) discussed a special group of children with diabetes who were not only impervious to conventional treatment, but prone to ketoacidosis which generally resulted in readmission of these children to hospital. Investigation of family relationships yielded the same 'psychosomatic' patterns of family interactions as observed with asthmatic and anorexic children. Results of family therapy for thirteen diabetic children who were very prone to repeated hospitalization were, in the main, very positive. The further success of family therapy for children with psychosomatic diabetes was described in a subsequent book by Minuchin, Rosman, and Baker (1978). Based on a small sample size of children with psychosomatic diabetes and two control groups with other types of diabetes, they concluded that only family dynamics could account for the unresponsiveness to conventional treatment and proneness to episodes of ketoacidosis and the positive response to family intervention for the children in the experimental group. While physiological data were presented in support of the benefits of family therapy, even basic statistical analysis was not performed. This initial and perhaps only study of its kind (our very thorough and careful literature search failed to yield any other family investigation of diabetes based on the notion of psychosomatic families) generated a critical response almost a decade later. Coyne and Anderson (1988) found fault both at the conceptual and methodological levels. Their

Psychosomatic and Medical Disorders 87 central conclusion was that 'the model decontextualises the family and assigns to it characteristics that are more appropriately seen as reflections of the disease process, the family coping tasks that it entails, and the nature of the family's relationship with the health care system.' They also argued that the very notion of 'psychosomatic' diabetes was, at best, suspect. In a vigorous defence of the model, two proponents, Rosman and Baker (1988), provided the results of the statistical analyses, which in the main supported their claim of the existence of psychosomatic diabetes as well as of the benefits of family intervention. It is possible that the interactions between the physiological and family factors lack the specificity proposed by the psychosomatic model. What is not disputable is the wisdom of involving family members in the treatment of children with chronic illnesses. The reader might remember that the very notion of psychosomatic families has come under close scrutiny, and while the concept has been shown to have limitations, it has also received some degree of validation. The debate continues. A final thought on the debate is that outcome, however spectacular, based on a study of only seven subjects and controls and notwithstanding methodological issues cannot and should not be viewed as definitive in either direction, as proof positive of success or lack there of. It is merely a starting point. The influence of the family in the management of an insulin-dependent group of adolescent sufferers of a chronic type of diabetes was reported by Satin and colleagues (1989). They randomly assigned thirty-two families into three groups. The group interventions consisted of multifamily therapy, multifamily therapy plus parental simulation of diabetes, and control (no family treatment) . Six family group sessions were offered to judge their effects on the patients' metabolic control and family functioning. Standard measures were employed to monitor the metabolic status, and the Family Environment Scale was used to assess family functioning. Additionally, information was obtained on the perception of diabetes and estimates of patients' self-care. Standard statistical methods were employed for data analysis. Outcome at the six-month follow-up showed that adolescents in the multifamily therapy plus parental simulation group showed clinically significant improvements in the diabetes-related measure, and both treatment groups were improved on their perception of diabetes in relation to controls. On the family functioning scale, no significant

88

How Good Is Family Therapy?

changes were found. The authors speculated on the merit of global measures of family functioning in very specific situations such as helping family and patients cope better with diabetes. Comments: The weight of evidence for the effectiveness of family therapy for diabetes, psychosomatic or otherwise, is very limited. Rosman and Baker (1988) proclaimed that their work was pioneering, and it was up to the research community to further it. To date that wish has failed to materialize, and the reasons for this may be more complex than indifferent or 'lazy' researchers. While there is little debate about the value of family involvement in the treatment of chronic pediatric conditions, the real issue may very well be the tentativeness of Minuchin's concept of 'psychosomatic' families. There is indeed some evidence that other less complex methods of intervention, such as in a study reported by Golden and associates (1985), than those proposed by Minuchin and associates (1978) have obtained similar results in preventing recurrent ketoacidosis in diabetic children. Besides, as the study by Satin and colleagues (1989) clearly showed, family group intervention was highly effective in controlling diabetes in adolescents without any improvement on family functioning. The implication of this and the Golden study is that there is an inherent advantage in adopting a 'family' approach. The approach, however, does not have to be predicated on the assumptions of 'psychosomatic' families. Asthma and Family Therapy

Minuchin and associates (1975) provided a summary of treating ten cases of intractable asthma in children. Based on clinical assessment families of these children were found to be enmeshed, overprotective, rigid, and deficient in resolving family conflicts. The ages of the children ranged from six to sixteen years. All but two children were steroid dependent. The duration of treatment ranged from five to twenty-two months. At the time of reporting two children were still in treatment. Follow-up ranged from nine to thirty-eight months. Outcome, reported in terms of the control of asthma, was excellent (no school loss, mild attacks) in eight of the ten children. In the remainder, the outcome was good (some days off school, mild to moderate attacks) . This study shares some of the same criticisms as the reports on abdominal pain and anorexia nervosa- no control group, small sample size, absence of statistical analyses, and therapists reporting on the

Psychosomatic and Medical Disorders 89 outcome of their own work, among others. Nevertheless, this report was the first of its kind. At the very least, Minuchin's proposition presented an alternative view for conceptualizing and treating intractable asthma, even if at a hypothetical level, which could be put to rigorous test to determine its validity. Lask and Matthew (1979) put to test the value of family therapy to treat pediatric asthma in a controlled trial. The theoretical underpinning of their family therapy approach has some similarities with Minuchin's model in that overinvolvement, lack of conflict resolution, and overprotectiveness were recognized as family problems in need of resolution. In addition, sources of stress in the family such as family finances, parental attitudes to asthma, and promotion of ageappropriate behaviour in children were seen as essential elements of family therapy intervention. The elimination or reduction of sources of family stress, which contributed to the maintenance or exacerbation of asthma in the child, was seen as the central goal of family therapy. The approach was firmly rooted in the systemic perspective (Lask 1992). The only description of their approach was that the therapy involved the whole family 'with particular emphasis on understanding the individual's symptoms and behavior arising from and feeding back into the general family system of interaction.' Their goal was to alleviate 'some of the stresses which, in interaction with other physical factors, contribute to the recurrence of asthmatic attacks.' The therapist's task was to modify the family's attitude to the illness, fear of death of the child, and the experience of fearful emotions. Therapy would move the family away from the extreme and 'towards the more realistic.' Some of the principles of the systems perspective were evident, and yet, in the main, Lask's therapeutic approach lacked the specifity of Minuchin's psychosomatic family model. The sample consisted of thirty-three families with thirty-seven asthmatic children. This sample was randomly allocated to an experimental group with seventeen families and twenty-one children and the control group with sixteen families and sixteen asthmatic children. The two groups were closely matched for the severity of asthma, age range of the children, and social class of the parents. Both groups of children received standard care for asthma, but the children in the experimental group additionally attended six one-hour sessions of family therapy during a four-month period. Results were mixed. However, the children in the experimental group had significantly fewer wheeze-days and less thoracic gas volume. They also claimed alleviation of family

90

How Good Is Family Therapy?

psychopathology as a result of family therapy, but they failed to provide any supportive data. While noting the benefits of family therapy in the modification or alleviation of pediatric asthma, they urged similar studies with larger samples. This study had several limitations. The major shortcoming was the absence of any family-related data. Without benefit of baseline data pertaining to the nature of the family dysfunction and the changes that accrued from family therapy to deal more effectively with the asthmatic child, the study failed to fulfil its major mandate. Another flaw was the absence of any longitudinal follow-up . A controlled study of family therapy with asthmatic children by Gustaffson and colleagues (1986) tested the relationship between family dysfunction, based on Minuchin's concept of psychosomatic families, and efficacy of family therapy. They hypothesized that family enmeshment, rigidity, and lack of conflict-resolution abilities contributed to the maintenance and exacerbation of severe bronchial asthma in the children. Eighteen children with severe bronchial asthma were randomly assigned to two groups. The experimental group received family therapy, and the other group served as control. The mean age of the children was nine years. Extensive baseline data related to asthma were collected over eight months prior to and evaluated for eight months after the termination of family therapy. Family therapy was conducted by two therapists and the number of sessions ranged from two to twenty-one per family (mean, 8.8). There was no evidence that any objective measures of family functioning were used. The therapists tried to change 'dysfunctional patterns of family interaction, explored the role of asthma symptoms in the family system, tried to reveal hidden conflicts, strengthened boundaries between individuals and between parents and children, and tried to enhance communication about the emotional impact of the disease on the family members.' Twelve children who received family therapy showed significant improvement virtually on every measure of asthma symptoms. No such improvement was noted in the control group. This study raised a number of problematic issues. While the parameters related to asthma symptoms were carefully assessed, no such procedure was followed to establish baseline data for family functioning. The report was silent on even the clinical observations made by the family therapists. The number of family sessions ranged from two to twenty-one. Did therapy, in spite of the differences in the number of

Psychosomatic and Medical Disorders 91 treatment sessions, produce equivalent changes in family systems? Did the researchers measure the 'psychosomatic family' characteristics? Did they, in fact, find the family characteristics on which rested their central hypothesis? How dysfunctional were these families? Was treatment designed specifically to rectify the issues of enmeshment and so forth? Again, the authors were silent on that question. The results of this study leave no doubt about the effectiveness of family therapy to reduce asthmatic symptoms in children. Nevertheless, in the absence of satisfactory answers to the questions we have raised, the question remains whether the hypothesis was validated and empirical support provided for the concept of psychosomatic families. In the final study in this section, a group of Finnish investigators reported a controlled study involving thirty-seven (mean age, 13.9 years) children with chronic asthma (Backman et al. 1981). Children were assigned to two treatment groups. The first group of seventeen children were assessed for their psychological and social functioning, and the twenty children in the experimental group, in addition to other treatment, were referred for psychotherapy and family therapy. The mean duration of treatment was five years and nine months. Family functioning was based on Ackerman's conception of the psychodynamics of the family, the role of the family, and parental interaction. Family assessment was based on family interviews. Outcome measures were days in hospital per year and missed school hours. Long-term follow-up confirmed that the benefits of treatment were promising. Days in hospital and missed school hours were both reduced and maintained over time for the children in the experimental group. This was a complex study that started out with enormous possibilities. Instead, the study lacked rigour. Its methodological problems are almost too numerous to mention. Failure to adhere to principles of random assignment, poor instrumentation, and hence questionable information about family functioning, and the absence of statistical analyses were the most obvious shortcomings. Weingarten and associates (1985), in a pilot study, randomly assigned eighteen asthmatic children, aged between eight and eleven years, into a treatment and a no treatment control group. The treatment group had group sessions with a physiotherapist, individual sessions with a psychologist, and the mothers of the subjects had sessions with a social worker. A second control of twelve asthmatic children joined sessions with the psychologist. The family component of

92

How Good Is Family Therapy?

treatment comprised two one-hour sessions for ten weeks for the mothers with the social worker. Outcome measures were related to the asthmatic symptoms. This study did not report any family-related data. The effects of social work intervention with the mothers were not reported other than to suggest that the mothers tended to be overprotective of their children and the fact that the subjects slept in their parents' bed. Children in the treatment group recorded significant improvement in their peak expiratory flow rate (PEFR) compared with subjects in the two control groups. From a family therapy point of view this study is of limited value other than to suggest that the mothers' involvement in the treatment of the children somehow contributed to a positive outcome. Comments: Despite major methodological limitations, three controlled family therapy outcome studies (Backman et al. 1981; Gustaffson et al. 1986; Lask and Matthew 1979), taken together, begin to make a case for the efficacy of family therapy in the treatment of pediatric asthma. Psychological factors have a time-honoured association with pediatric asthma. Stress has been implicated in the exacerbation of this troublesome and at times fatal disease. All three studies implicated stress, in one way or another, in the deterioration of this disease. Only Lask was unequivocal about this fact. As with all other diseases, the relationship of Minuchin's 'psychosomatic family' factors with asthma awaits further replication and, thus, validation. The findings of the family-oriented studies were also positive. Any kind of family involvement seems to yield more favourable outcome than non-involvement of family in the treatment of pediatric asthma. Medical Illness in Adults and Family Therapy Family-Oriented Approaches

Herzoff (1979) discussed the merits of an open-ended group therapy program for cancer patients and their families. The treatment approach was derived from a marriage of psychodynamic and systems theories. Some 151 patients had the benefit of this group experience. The author was silent, other than through a case illustration, on the level of participation of family members and on how they gained from this experience. This report was strictly clinical. Plant and associates (1987) described a support group program for

Psychosomatic and Medical Disorders 93 oncology patients and their relatives and friends at a London hospital. Family group sessions were held once a month. Evaluation was based on a questionnaire mailed to the participants three months after the first session of the group. Of the 103 questionnaires sent, sixty (58 per cent) were returned. The composition of the respondents was patients 45 per cent, relatives 42 per cent, and friends 13 per cent. Statistical analyses were kept to a minimum. In the areas of information about cancer, understanding about it and happier feelings as a result of attending the group sessions were reported by a substantial number of participants. The patients reported a higher level of satisfaction than did the relatives. Theorell, Haggmark, and Eneroth (1987) investigated the effects of group intervention for the close female relatives of cancer patients. The treatment and control groups each had thirty-six subjects with the mean age of fifty-two and fifty-one years respectively. Treatment consisted of psychosocial intervention as well as information and medical treatment. Outcome measures included measures of psychiatric symptomatology, severity of patient's illness, and measures of cortico and prolactin levels. Blood samples were obtained monthly. Outcome was based on the relatives who experienced death of the family members. This sample comprised eighteen treatment and seventeen control subjects. At posttreatment, prolactin levels were lower in the treatment group. At the time of the patient's death cortisol levels were significantly higher in the treatment group. On the psychological measure the treatment group had significantly lower mental exhaustion following the death of the family member. A family counselling program, based on the principles of Shalit's 'coping wheel,' was conducted by Haggmark, Theorell, and Ek (1987). This instrument was developed for the measurement of coping strategies. Fifty-eight relatives of cancer patients in a Stockholm hospital agreed to participate. Eight relatives dropped out. A control group was established from relatives who were not offered the coping wheel, but who participated in the routine program. Twenty-two relatives in the experimental and nineteen in the control group followed for one and two months after the death of the patient. Outcome was based on the various aspects of functioning derived from the coping wheel. Appropriate statistical analysis was conducted. The results were unequivocal in the areas of activities concerning friends and relatives during treatment, ability to deal with feelings just

94 How Good Is Family Therapy? prior to the relative's death, and increased number of activities preceding and one month after death. The differences between the experimental and the control groups were significant. Goldberg and Wood (1985) reported a study involving the spouses of lung cancer patients. Spouses were treated with interpersonal psychotherapy. Outcome showed no significant differences on the psychological measures between the experimental and the control groups. This study raised the possibility that interpersonal problems may not be amenable to change if all parties involved are not participants in the therapeutic process. On the other hand, the following study showed the value of counselling of caregivers of stroke patients to maintain effective family functioning. In a controlled study of stroke patients comparing efficacy of counselling (n =63) and educational intervention (n =64) for caregivers with standard treatment of stroke patients (n = 61), results showed that six months and one year after the stroke, both interventions significantly improved caregiver knowledge and certain aspects of family functioning more than did standard care (Evans et al. 1988). One of the interesting finding was that while most families functioned effectively on the Family Assessment Device based on the McMaster Model of Family Functioning, one year after the stroke only the control families had deteriorated. This was a well-designed study with clear outcome measures. The samples were relatively large. The treatment approaches were clearly defined, and sufficient provision for follow-up was made. Also appropriate statistical analyses were used. Heinrich and Schag (1985) reported a couple group therapy program designed to enhance management of stress and activity. A sample of fifty-one patients made up the treatment and control groups. Twentysix patients and twenty-five spouses participated in the couple group program. Twenty-five patients were assigned to the control group. The treatment intervention was a six-week structured small group program. Each session lasted two hours. The goals of treatment were to educate the patients and spouses about cancer and specific skills for managing daily stresses. The method of intervention was cognitivebehavioural. A variety of outcome measures were employed, and appropriate statistics were used for data analyses. Patients and spouses were evaluated at pretreatment, posttreatment, and two months and four months posttreatment. The results indicated that the information aspect of treatment

Psychosomatic and Medical Disorders 95 increased the subjects' knowledge of cancer and of coping skills. The control group did not register any improvement. In contrast, on psychosocial adjustment both groups became better adjusted over time, and the treatment effects were non-significant. The improvements were maintained at the two-month follow-up. This particular finding, somewhat of an exception to the general rule, was not explained adequately by the researchers. The treatment group expressed a high level of satisfaction and continued to use the techniques they learned in the group. Outcome at the four-month follow-up was not reported. Comments: Four controlled studies discussed in this section confirm the value of a variety of family-oriented approaches to enhance coping skills in relatives. The results were unequivocal. The treatment subjects not only fared better than those in the control groups in many areas of coping, they, in fact, gave evidence of better psychological health, except in the last study. These studies taken together make a strong case for the adoption of group therapy and counselling programs for the significant others of medically ill patients. Family Therapy

Using an experimental design Christensen (1983) tested the efficacy of couple therapy for postmastectomy patients. Twenty mastectomy patients (mean age, 39.7 years) and their spouses received four weeks of structured couple intervention. The control group did not. Standardized instruments were used for data gathering. The Locke-Wallace Marital Adjustment Test was used to assess couple functioning, and standard statistical tests were performed. The goals of intervention were to improve the couple's relationship, providing information and discussion on postmastectomy adjustment, effects of concern on self-esteem, and improving communication and issues of long-term adjustment. The results were equivocal. No significant differences on the chi square test were found posttreatment between the experimental and the control groups. Analyses of covariance, however, showed that treatment significantly reduced depression in the patient, increased sexual satisfaction for both partners, and reduced emotional discomfort in both partners. Family dysfunction is a common occurrence among patients with chronic pain (Roy 1989). Unfortunately, outcome studies of family therapy with this population are few and methodologically wanting.

96 How Good Is Family Therapy? Hudgens (1979) treated twenty-four chronic pain families with combined operant conditioning and family therapy. She reported positive outcome in 75 per cent of the families treated. The study was uncontrolled and, hence, the actual effect of the family therapy could not be accurately assessed. Roy (1989) has described the lives of thirty-two patients (twenty with headache and twelve with back pain) with chronic pain and their families. Family therapy was found useful by sixteen couples in the headache group as opposed to four in the back pain group. Treatment was based on the McMaster model. This was by no means a scientific study of outcome of systemic family intervention, but it remains the most comprehensive account yet of family assessment and treatment of chronic pain sufferers and their family members. Comments: The point of note is that there is actually only one study of couple therapy for cancer patients. This study met some of the basic criteria for well-designed studies; hence, its findings could be construed as valid. However, there are several shortcomings in this area of research. Not a single study was found that involved the whole family. However well designed, a single study, by any measure, is simply inadequate to confirm or disconfirm the value of family therapy. To say that more well-designed studies are called for to provide further evidence of efficacy of family therapy is merely to state the obvious. The efficacy of family therapy with chronic pain patients remains uncertain. The Current State of Research

The actual quantity of outcome studies of systemic family interventions with medical and psychosomatic conditions is truly meagre. Family interventions with psychosomatic and medical disorders fall into two categories: first, where family relationship factors are implicated in the etiology and or maintenance of the disease such as childhood asthma, and second, where no such association is evident, and the purpose of therapy is to achieve a healthy level of adaptation to the disease such as cancer or coronary heart disease. Much of the debate in the literature has centred on the former. Minuchin's psychosomatic model, perhaps because it is the only articulated model based on family systems theory in the field, has attracted considerable debate. It is worth noting that most other major 'schools' of family therapy are conspicuous by their absence in this body of research. Our view is that the

Psychosomatic and Medical Disorders 97 model to date has not had the benefit of rigorous testing. Outcome studies using the model are few and not devoid of methodological concerns on both sides of the debate. The critical methodological problem of accurately measuring the characteristics of psychosomatic families has not been satisfactorily resolved. It is just too early to claim its unequivocal success or decry it as completely bogus. After all, Franz Alexander's proposition of an association of personality factors with a variety of illnesses served as a precursor to the development of the Type A personality and its association with coronary artery disease. Despite the debate, some of the findings such as Lask's and Gustaffson's, both controlled studies, leave little doubt about the general value of family therapy with asthma. While the psychosomatogenic model may be a matter of controversy, the overall effectiveness of family intervention with childhood diseases is less so. The value of family intervention, though still emerging, is more telling when the goal is to reverse the negative consequences of illness on the family system. The results of family interventions are powerful enough to confirm their efficacy. Gustaffson and Svedin (1988), in a retrospective study of forty-two children with psychosomatic and somatic illnesses who received family therapy treatment in a pediatric setting, found a greater decrease in number of hospitalization days than in matched controls. This was a very important study because it proved the cost effectiveness of family therapy. The theoretical underpinnings of some of these interventions were not always made explicit. The actual number of studies remains relatively small, and there are some contradictory findings. Yet the results tend to affirm the positive consequences of family interventions to restore healthy family relations and more effective coping with the negative fallout of illness or even death. The almost total absence of family therapy outcome studies with adult medically ill patients is not easily explainable. Considering the amount of family therapy literature, all claiming success, this absence of outcome studies is puzzling. The most that can be said about the efficacy of family therapy with adult medically ill patients is that the field is ripe for outcome studies, and if the clinical literature is any guide, a positive outcome is likely to be confirmed. In the meantime, however, practitioners do not have the benefit of empirical support for family therapy with this population.

6

Alcoholism and Family Therapy

Family issues have long been recognized as having great relevance to the treatment of alcoholism. The misconception about the role of the wife, who because of her own psychological needs married an alcoholic man, while losing all scientific credibility in recent years, unquestionably laid the foundation for subsequent family intervention with the families of alcoholics (Steinglass 1976). Jackson (1958) noted that 'once attention had been focused on the families of alcoholics, it became obvious that the relationships between the alcoholic and his family is not a one-way relationship. The family also affects the alcoholic and his illness.' The essence of this statement is simply that the alcoholic person cannot (or ideally should not) be treated to the exclusion of the family. Bowen (1974), in his brilliant report on the application of systems theory to explain and treat alcoholism, noted with optimism that 'family systems therapy offers no magic solution for the total problem [of alcoholism), but the theory does provide a different way to conceptualize the problem, and the therapy provides a number of approaches to the problem that are not available with conventional theory and therapy.' Steinglass (1976), in the first comprehensive review of family issues and family treatment related to alcoholism, noted that despite the magnitude of the problem of alcoholism, family therapy was rather a latecomer to the field. His conclusions, based on a comprehensive review of the efficacy of family therapy for alcoholism literature from 1950 through 1975, were sobering. He characterized most of the family therapy for alcoholism outcome literature as 'pilot in nature.' In addition to the usual shortcomings, such as small sample size, lack of control groups, poorly defined population, and inconsistent outcome

Alcoholism and Family Therapy 99 measures, Steinglass identified two critical factors that further accounted for the sorry state of this body of research: first, the absence of trained family therapists to work with this population and, second, the inclusion of a variety of outcome measures in the studies he reviewed. He recommended that alcohol consumption would be areasonable measure of outcome, because the goal of family therapy was to reduce the intake of alcohol. While there was much enthusiasm for family therapy practice, the quality of research had failed to achieve a reasonable standard. Many of the early outcome studies, while methodologically of questionable quality, raised critical questions and generated interesting hypotheses which proved to be of considerable value to future researchers. Steinglass's own work laid the foundation for the incorporation of the systemic perspective into the family approach for treating alcoholic families. This period also witnessed the birth and growth of couple-and-family-group approaches as well as behaviourally based couple and family interventions with alcoholic families. Some of these early studies will be examined here. We have adopted a more generous or a less rigorous approach for inclusion of studies that do not necessarily meet the criteria of 'good' research so as to include samples of earlier studies that have considerable historical merit. Such 'good' criteria for outcome studies were succinctly outlined by Mccrady (1989) in her review of the family-involved alcoholism treatment outcome literature. They are: (1) random assignments to experimental and control groups; (2) delivery of well-defined treatment within each condition; (3) adequate description of subjects in the study; (4) use of objective, reliable, and valid measures of treatment outcome; (5) collection of data by persons not delivering the treatment; (6) measurement of outcome across major areas of functioning; (7) adequate length of follow-up, with six months as the absolute minimum; (8) adequate rates of follow-up, usually 80 per cent or above, and use of reasonable strategies to account for missing data and missing subjects; (9) use of collateral data sources to validate subjects' self-reports; (10) well-defined criteria of success, improvement, no change, and deterioration; (11) appropriate statistical procedures to analyse the results. Out of eleven studies she reviewed, only two, which she conducted herself together with her colleagues, met her criteria. We intend to pay special attention to these. We have, however, adopted a more liberal approach in this review so that a more comprehensive understanding can be developed about the potential benefit of family

100 How Good Is Family Therapy? therapy. Another point of departure is that we have included behavioural family and marital therapy outcome studies in this chapter for the unavoidable reason that outcome studies based on systems perspectives are rare. First, we shall consider family therapy and couple therapy followed by family-oriented approaches, such as treatment of spouse alone, spousal groups, and patient-spouse groups. Family Systems Approaches

The principal proponent of this approach is Steinglass (1979). Others, of course, had used systemic family therapy, and clinical reports on the benefits of family therapy abound (for a review of that literature, see Kenward and Rissover 1980). Alas, such is not the case as far as the outcome literature is concerned. Outcome research based on this model and even conceptualizing the problem of alcoholism using systemic concepts was, in large measure, a singular contribution of Steinglass (1979). His treatment approach involved patient-spouse groups, and it will be discussed under the appropriate section. One of the earlier studies in this category was reported by Meeks and Kelly (1970) . Five families were recruited for the purpose of applying systems-based family therapy. The authors were careful to state that because of the very exploratory nature of this study, rigorous research procedures were not applied. The patients and their spouses were invohed in individual and group therapy during the intensive period of treatment. Only after their discharge from hospital, was family therapy, based on the work of Virginia Satir, implemented. At the time the patients were not receiving any other treatment. Family therapy lasted ten to twelve months, and each session was about an hour to an hour and a half in duration. Therapy was conducted by one of the authors. All members of the family, except very young children participated. Outcome was measured in terms of reduction in drinking and positive changes in family relationships. Positive outcome was reported on every measure. Statistical analyses were not used. The significance of this project was that alcoholism was redefined in the family context, and certain assumptions were made about the importance of family dynamics that may in some ways contribute and maintain alcoholic behaviour in one of its members. This basic idea was the cornerstone of Steinglass's work. Esser (1971) reported on fourteen alcoholic families treated with family therapy. The method of family therapy minimally involved two

Alcoholism and Family Therapy

101

generations in the therapeutic process. His sample consisted of thirteen male and one female alcoholic. The duration of therapy ranged from four months to two years. A mixed bag of outcome measures, mostly observational, were used, and abstinence and reduction in drinking and positive changes in family relationships were broadly the two main outcome measures. Twelve families reported positive changes, and two families failed to show any improvement. Like the Meek and Kelly study, Esser described this project as 'experimental' and demonstrated much creativity in reconceptualizing alcoholism as a metaphor for faulty family relationships. The value of the preceding two reports resides not at all in their methodological excellence, which was not there, but rather in their application of an emerging therapeutic method to an old problem. Comments: It remains a matter of some mystery that beyond these two very preliminary investigations, the outcome literature on family therapy for alcoholics involving the whole family is devoid of any further studies. Reasons for this lack of interest are far from self-evident, for there is much discussion in the clinical literature about the merit of family therapy for alcoholism (Winkel 1989). In a recent national survey of Veterans Affairs Treatment Programs for families of alcoholics in the United States, family therapy was used rather infrequently compared with other family-focused modalities. Of seventy alcoholism inpatient programs for married patients in the Veterans Hospital system 22.9 per cent did not offer any family therapy at all to any patients, and another 67.1 per cent offered it to less than half of the patients. Only 1.4 per cent of the in-patient programs provided family therapy to all of their patients (Salinas et al. 1991). A review of treatment programs in Massachusetts by Regan et al. (1983) also yielded rather discouraging results. Their central finding was that spouses of alcohol abusers were involved in treatment with moderate frequency, and children only infrequently. In an earlier report on the United States Navy Alcoholism Prevention Program, it was noted that family members were generally not included in treatment because of the budgetary constraints (Brownell 1978) . In yet another report on the merit of various treatment modalities for alcoholism 13 per cent of the successfully treated alcoholics and 14 per cent of the unsuccessful cases found family therapy most helpful and least helpful respectively (Hoffman, Noem, and Peterson 1976). The authors did not provide any explanation for this curious finding. It is conceivable that no single treatment modality, in

102 How Good Is Family Therapy? itself, could have been of much help for the treatment failure group. However, Janzen (1985-6), in a survey of alcoholic treatment programs, found that a majority of them were committed to family therapy. While systems-based family therapy for alcoholic families has gained currency, there appears to be a persistent absence of desire to test its efficacy using scientific methods. It is also a matter of interest that Steinglass, the principal proponent of the systems approach to treat alcoholism, did not implement a single family treatment program for alcoholics with an outcome study based on family treatment for alcoholics. His treatment approach was couple-groups in an in-patient setting. Marital Interventions

In this section we consider intervention that involved the patientspouse dyad as the unit of treatment. In a national review of the family treatment programs for alcoholics at Veterans Hospitals in the United States, Salinas et al. note that 'conjoint couples therapy, the most frequently used of all (family) modalities except for referral for family interventions to other agencies, was used with at least half the patients in 17 per cent of inpatient and 22 per cent of outpatient programs and at least for some patients in over 90 per cent of programs' (1991). Couple therapy, in the largest hospital system in the United States, would appear to be relatively common. This penchant for couple therapy is not visible in the outcome literature. Only a single study exists to assess the effectiveness of couple therapy. In Steinglass's (1976) review of the family treatment literature, not a single reference could be found related to couple therapy. The situation has hardly improved in the past decade. A point of note is that couple-group programs are frequently described in the literature either as 'couple' therapy or 'behavioural marital' therapy, and this could be a source of some confusion for the uninitiated. McCrady and colleagues (1986), in a unique study involving the couple dyad as the unit of intervention, examined the relative merit of three versions of family interventions. They were minimal spousal involvement (MSI; n = 21), alcohol focused spouse involvement (AFSI; n = 13), andAFSI plus behavioural marital therapy (ABMT; n= 19). Each treatment condition was carefully articulated and implemented by clinical psychologists or advanced doctoral level clinical psychology students. Therapists followed detailed instruction manuals

Alcoholism and Family Therapy

103

for the treatment conditions. Treatment consisted of fifteen ninetyminute sessions. The outcome measures consisted of pre- and posttreatment alcohol intake using the time-line follow-back interviewing technique. Other measures were used to obtain drinking-related data such as effects of drinking on physical health, occupation, liver functioning, and driving. The Marital Adjustment Test and Areas of Change Questionnaire were completed by both spouses at pre- and posttreatment and at the sixmonth follow-up. In addition, subjects completed the Spouse-Mate Role of the Psychological Functioning Inventory. During treatment both spouses completed daily ratings of marital satisfaction. Marital satisfaction was assessed on the basis of a ten-minute video-taped discussion of a problem selected by the couples. These tapes were coded using the Marital Interaction Coding System of the Oregon Marital Studies Program. Psychological and social functioning were also assessed using some of the subscales of the Psychological Functioning Inventory. Employment status was assessed in the preliminary interview and in monthly follow-up calls. Appropriate and sophisticated statistical analyses were visible in this study. Treatment compliance ranged from 16 per cent for MSI to 92 per cent for AFSI to 100 per cent for ABMT couples. Posttreatment, all subjects had noticeably reduced their alcohol intake. No main effects of treatment conditions were found. However, across the groups there was a significant increase in abstinent days from baseline to follow-up. After treatment the AFSI subjects increased their frequency of drinking more rapidly than the other two groups did, and the MSI subjects showed a significant increase in the frequency of heavy drinking. Positive treatment outcomes were observed, however, right across the board, but there were interesting differences between groups. Overall, the ABMT group registered the most improvement which was noticeable in the broad areas of marital relationship as well as in drinking-related behaviours. The results demonstrated the general superiority of ABMT. At the eighteen-month follow-up, Mccrady et al. (1991) studied forty-five couples: minimum spouse involvement (MSI), n = 14; alcohol-focused spouse involvement (AFSI), n = 12; and AFSI plus behavioural marital therapy (ABMT), n = 19. They had completed at least five sessions of treatment. It was found that the initial patterns of higher improvement in the ABMT group were maintained during the second nine months of follow-up. The authors concluded that ABMT held out

104 How Good Is Family Therapy? greater promise than the other two treatment conditions. Abstinence from alcohol and general improvement in marital satisfaction were the hallmarks of the ABMT group. In contrast, subjects in the other two treatment conditions experienced slow deterioration in their drinking habits. Comments: Considering the rise in the popularity of family therapy since the 1970s the relative absence of family and couple therapy outcome studies in the treatment of alcoholics remains a matter of some curiosity. Or, perhaps not. Alcoholism or alcoholic behaviour does not readily translate into metaphors for family conflicts, as do the socially more benign symptoms of headaches or even the symptoms of not so benign a condition as anorexia nervosa. Neither do headaches or anorexia nervosa carry any significant social stigma. Perhaps alcoholism is less attractive to family therapists because the stereotypical notion of the alcoholic continues to hold sway (Kahle and White 1991). French (1987) in a review of the family-alcohol literature noted, first, that many publications in the family and marriage literature regarded individual factors as the principal cause for alcoholism, and, second, that the cultural stereotype of the alcohol abuser as a skid-row bum might have contributed to this apparent lack of research interest. Regardless of the reasons, the two early uncontrolled family therapy reports and the superbly designed couple therapy study of Mccrady et al. (1986) held out some promise. The latter study met the key criterion of a well-designed study. The results were sufficiently powerful to demonstrate the efficacy of behavioural marital therapy when used in conjunction with other treatments. The authors themselves were cautious in interpreting some of the more subjective data related to marital satisfaction. Yet the decline in drinking, which was a more objective measure, did indicate significant gains by the ABMT group over the other two groups. Couple-Group Therapy

Unlike family and couple treatment, outcome studies related to couple-group interventions based on behavioural-marital and systemsoriented therapies have found mildly greater favour with researchers. Certainly, a handful of very high quality studies are to be found in this category. One of the very early studies involving couple-group therapy as

Alcoholism and Family Therapy

105

part of an alcoholic treatment program was reported by Gallant and associates (1970). One hundred and eighteen subjects and their spouses received couple-group therapy. There was no control group, and follow-up was not systematic. A marked improvement in drinking and an improved marital relationship was reported by fifty-three (45 per cent) , and the remaining sixty-five (55 per cent) were definite failures at follow-ups that ranged from a couple of months to nearly a year. While this report was extensive in scope, the research component was decidedly weak. The first controlled study was reported by Cadogan (1973). Forty patients and their spouses were recruited at the point of discharge from an in-patient alcohol treatment program. These couples were randomly assigned in equal numbers into a treatment group and a waiting list. Demographically the two groups were well matched. The groups sessions were open-ended and met weekly for ninety minutes. Outcome was measured primarily in terms of reduction in alcohol consumption. The results at the six-month follow-up showed that nine (45 per cent) patients in the experimental group were abstinent, four (20 per cent) were drinking at a reduced level, and seven (35 per cent) had major setbacks, mainly in the way of relapse. In contrast, only two (10 per cent) in the control group became abstinent, and the rest were either drinking moderately or had relapsed. Steinglass (1979) reported a study that involved conjoint hospitalization. This study had several unique features. Its theoretical orientation was derived from systems theory. He proposed a model based on that theory which postulated that 'entity labelled an "alcoholic system, " which was conceptualized as an interactional group in which behavior of family members had become organized around the acquisition and consumption of alcohol. A critical concept was that drinking and intoxicated behavior might, in certain circumstances, serve to "stabilize group interaction."' Later, Steinglass (1987a) , in his book The Alcoholic Family, described at length the theoretical and treatment issues that surrounded alcoholic family systems. The life-stage issues of these families were central to his premise for the development and maintenance of the alcoholic family. He noted that 'a potentially destabilizing chronic condition (alcoholism) becomes ... an organizing principle around which a number of critically important aspects of family behavior take shape.' Furthermore, he carefully articulated the regulatory mechanisms 'used

106 How Good Is Family Therapy? by family to shape and stabilize its life during middle-phase' (during this phase the activities of the alcoholic family are built around alcohol related behaviours as a central core). He described the 'alcoholic system' as one where 'the presence or absence of alcohol becomes the single most important variable determining the interactional behaviour not only between the identified drinker and other members of the family but among non-drinking members of the family as well.' Steinglass's 1979 study reported a complex investigation to test the efficacy of the family systems approach to the treatment of the alcoholic family (couples). The critical feature of this study was an inpatient treatment component for the alcoholic patient and the spouse who received multiple-couple group therapy. An added objective was to simulate the family environment of the patients in which drinking occurred. The treatment was administered in three phases: (1) an initial two-week out-patient phase during which the couples met for three sessions per week; (2) a ten-day in-patient phase, during which three couples were admitted simultaneously; (3) a posthospitalization out-patient phase for three weeks during which the couples met for two group meetings per week. The group reconvened at six-week intervals for follow-up for six months. Ten couples ranging in age from thirty to fifty-two years participated in this study. Steinglass described the treatment involved in each phase in some detail. The key objective was to destabilize the pattern of family functioning that maintained drinking at the core of its activities. The following measures were administered pre- and posttherapy and six months after the termination of treatment: (1) individual drinking behaviour: chronological drinking record; (2) individual psychiatric symptomatology (SCL-90); (3) individual psychosocial functioning: structured and scaled interview to assess maladjustment (SSIAM); (4) behavioural assessment of marital interaction: inventory of marital conflicts (IMC); and (5) self-report assessment of marital interaction: subject is marriage. Statistical analyses of the results were not presented. Eight of the ten couples completed therapy, and these eight couples included nine alcoholics. Five of the nine were drinking less alcohol at the six-month follow-up. Other aspects of the outcome were reported based on individual changes, and no single dominant pattern was discernible on the other measures. In broad terms, eight of the nine couples had changed for better the pattern and context of their drinking, and seven couples gave evidence of changes in their communication pattern and their resolution of marital conflicts.

Alcoholism and Family Therapy

107

Despite the sheer originality of this study, a major limitation was posed by the small number of subjects, which presumably, restricted adequate statistical analysis. Besides, this study, like most of its predecessors, did not have a control group and was thus 'pilot' in nature. The theoretical importance of this study is considerable, but because of its preliminary nature the results must be treated with caution. Steinglass had theorized about the complex nature of family dynamics of alcoholic families. He explicated those dynamic forces and intervened therapeutically with measurable success. One component of this type of intervention was the high cost. Judged against the cost to society of drinking, the cost of such treatment could only be modest. Perhaps because of the complexity of this study, to date there has not been a replication with a larger sample. Mccrady and colleagues (1982) rectified one of the major shortcomings of the Steinglass study by mounting a controlled investigation of three groups. Couples were randomly assigned to joint admission of patient and partner (eighteen couples), couple-group therapy with no joint admission (eight couples), or individual treatment (seven couples). Couple therapy in combination with joint admission was expected to produce the most favourable results. Outcome measures included a structured interview which obtained information about (1) marital status, (2) employment status, (3) residential status, and (4) drinking status - longest continuous abstinence. In addition, a detailed history of drinking patterns was obtained. Couples assigned to either of the couple therapy conditions joined a weekly couple group therapy program. Therapy focused on improving communication and other aspects of marital relationships. Outcome data were based on the follow-up of at least one member of each ofthe original thirty-three couples. Four years after treatment no significant differences were found between the three groups on any of the outcome measures. Nevertheless, the key finding was that significant improvements noted in all three groups at the end of treatment were not maintained over time (75 per cent showing improvement at the end of treatment and just under 33 per cent did so at the end of four years). Joint admission failed to show any superiority. This was a very well designed study. Yet there was virtually no information about the nature of the couples therapy, the therapist variables, or the in-patient program. The training and background of the therapists were not revealed. From the description of the study it would appear that the therapeutic orientation was systemic.

108 How Good Is Family Therapy? O'Farrell, Cutter, and Floyd (1985) also compared the efficacy of couple-group therapy based on behavioural principles. To that degree this study differed from the previous studies. Another point of departure was the absence of the in-patient component. This study was, at least, a partial replication of one by Hedberg and Campbell (1974) who demonstrated the superiority of 'behavioral family counselling' compared with three other individually oriented behavioural techniques. O'Farrell's group compared couple-group therapy based on behavioural principles, an interactional couple-group (perhaps predicated on systems principles?), and a no couple treatment control. Subjects were thirty-six male alcoholics and their wives. They were randomly assigned to the three groups. The first two groups received ten weekly sessions. Outcome measures included the Locke-Wallace Marital Adjustment Test, Marital Status Inventory, and observational data on marital interaction based on ten-minute video tapes of the sessions and interview data on the alcoholics' drinking adjustment obtained by using the time-line follow-back interview method. Multivariate analysis of covariance was used to determine any overall treatment effects. The results, though somewhat equivocal, pointed to the superiority of the behavioural couple group, which at the two- and six-month follow-up reported significant gains on the marital measures compared with the other two groups. On drinking measures all three groups reported significant improvement which was sustained at follow-up, but the behavioural couple group had significantly fewer drinking days than did the other two groups. The authors concluded that the behavioural couple therapy was superior to the interactional and no treatment conditions and was a useful adjunct to the treatment of alcoholics. Comment: A decade or more ago Kenward and Rissover (1980) noted

that 'at this time there is insufficient empirical evidence to provide agencies with guidelines for developing effective family systems treatment programs.' That pessimistic observation still holds. Empirical evidence in support of systems family therapy for the treatment of alcoholism is indeed meagre. Jacobson and associates (1989) commenting on Steinglass's work wrote that 'unfortunately, methodological problems produce ambiguity in the substantive conclusions. These methodological problems included ... the failure to test directly the fundamental notions in Steinglass' systemic model.' Assessing the efficacy of family therapy based on systems theory to

Alcoholism and Family Therapy 109 treat alcoholic families remains, to a surprising degree, an uncharted activity. Another inexplicable aspect of this body of research was the exclusion of children from the treatment. The fact that children of alcoholic parents are substantially ignored, not just in the research literature, but also in clinical practice was borne out by the national review of the alcohol treatment programs in the veterans hospital system in the United States (Salinas et al. 1991). Children appeared to have the lowest priority in the out-patient as well as the in-patient programs. For example, 92.2 per cent of the hospitals did not offer any group therapy for the children, and 90 per cent had no services for individual therapy for them in their out-patient services. Referral to Alateen was also less than a routine procedure. In contemporary family therapy, justification can be found for treating part of the family system. No such rationale was evident in the studies reviewed here. Perhaps for reasons of common sense or convenience, the marital dyad, and by implication marital discord, was seen as the most obvious target for intervention. Systems-based family therapy, the theoretical orientation most commonly associated with contemporary family therapy, was conspicuous by its relative absence in the alcohol literature. Preli, Protinsky, and Cross (1990) explored the 'structural' issues in alcoholic and control groups. They used the Family Adaptability and Cohesion Evaluation Scale (FACES) III to determine the boundary issues confronted by families with an alcoholic member. Their study found that the alcoholic families had disturbed interactional boundaries and rigid or chaotic patterns of adaptability. Their conclusion was that the findings supported Minuchin's structural model of family theory and that alcoholism was a family disease 'systematically impacting upon all members.' Unfortunately, to date no family therapy outcome study has been implemented to test the validity of the 'structural' formulation. Couple group therapy based on the principles of family-systems theory yielded equivocal results. Steinglass's study, while theoretically challenging, was uncontrolled and involved few subjects. Despite his success, the results cannot be generalized. McCrady and associates (1982) failed to show any inherent superiority over time of joint hospitalization over other couple therapy conditions. The systems aspect of the intervention in their study was not made explicit. Over two decades ago, a couple group approach was heralded as the 'treatment of choice for married alcoholics' by Gallant and colleagues (1970). Has it lived up to that prophecy? Couple-group therapy has

110 How Good Is Family Therapy? gained far less currency than conjoint couples therapy, according to Salinas et al. (1991). In their review of alcoholic treatment programs in hospitals they found that nearly 33 per cent of them did not offer couple-group therapy, and not a single one confined its total treatment to couple-group therapy. On the other hand, the McCrady and O'Farrell studies taken together hold out much promise. In his review of marital and family therapy, O'Farrell (1991) made the cautious observation that 'evidence is accumulating that MFT [marital and family treatment) helps stabilize marital and family relationships and supports improvement in alcoholic's drinking during the six-month period following treatment entry alcoholism.' That claim is essentially correct, albeit based on a small number of methodologically refined studies. As for the family-systems perspective, empirical evidence is sadly lacking. Hospitalization of the alcoholic and spouse in itself, contrary to Steinglass's early optimism and superb theoretical explication, was not borne out as superior to out-patient treatment. McCrady (1989) succinctly summarized the problem when she wrote that 'careful outcome studies that examine the process of family systems-oriented treatment and that compare the outcomes to appropriate control conditions are simply lacking.' Until such time as those gaps are filled, Steinglass's complex family dynamic issues implicated in the maintenance and perpetuation of alcoholic behaviour will remain a partially validated, and yet, tantalizing concept. Other Family-Oriented Approaches

The clinical literature is replete with reports of family-oriented treatment for alcoholism. Most of them are anecdotal, and they cover a very wide range of interventions. A partial list of such activities include wives' groups (Smith 1969), wives attending Al-Anon group therapy (Gorman and Rooney 1979), a very short-term (four days) intensive group treatment for alcoholic husbands and their wives (Corder, Corder, and Laidlaw 1972), behaviour modification training for wives (Cheek et al. 1971), and family members and significant others 'confronting' resistant alcoholics to seek treatment with positive outcome (Liepman, Nirenberg, and Begin 1989). In this section we consider a few studies which at the very least adopted an elementary evaluative component as well as follow-up. McNabb, Der-Karabetian, and Rhoads (1989) conducted a six-

Alcoholism and Family Therapy

111

month or more follow-up study on eighty adult in-patients who had received three types of 'family' interventions. The interventions consisted of (1) spouses who attended three or fewer group therapy sessions per week ( n = 30); (2) spouses who attended four or more sessions per week (n = 30) ; and (3) spouses who were treated for co-alcoholism (n = 20). A questionnaire was designed to obtain outcome data on (1) abstinence from alcohol; (2) improvement in family relations; (3) personal well-being; (4) physical well-being; and (5) beliefs about alcoholism as a disease. This questionnaire was administered over the phone. No data were collected at the point of discharge. On abstinence, while all three groups reported improvement, Group 3 registered 90 per cent improvement compared with 53 per cent for Group 1 and 73 per cent for Group 2. Similar patterns were evident in the area of family relations, personal well-being, and physical wellbeing. The conclusion was that a strong positive association was present between greater family involvement and outcome. The inpatient treatment for co-alcoholics was found to be most effective. This was especially significant as Group 3 comprised individuals with more severe alcohol problems. The findings of the McNab study are contrary to McCrady, who found that hospitalization of patient and spouse did not yield better outcome. However, McCrady's group therapy included both patients and spouses; McNab's did not. Judged against McCrady's criteria for a 'good' study, McNab fell far short in several ways. The study was retrospective, and no baseline data were obtained. Hence, the reports of improvement could not be objectively judged. This was a fundamental violation of a sound research strategy. Besides, how the patients were assigned to the three groups remained unclear. Subjects were not described in any length, treatment conditions were not clear, and outcome measures were not standardized. Given these deficits in methodology, the results of the study demand very careful interpretation. Thomas and associates (1987) reported a pilot study on the benefits of unilaterally treating spouses ofalcoholics, twenty-five of whom were recruited by newspaper advertisement. They were assigned to immediate and delayed treatment groups, the latter serving as control. The subjects were almost exclusively white and middle class. The key components of treatment were (1) clinical assessment; (2) abuserdirected interventions; (3) spouse-directed interventions; and (4) maintenance. A variety of outcome measures to assess pre- and postfunctioning in the area of marriage and drinking were used. The small

112 How Good Is Family Therapy? sample posed certain problems for conducting statistical analyses, but repeated measures analyses of variance were conducted to compare the effects of immediate treatment with those of delayed treatment. The treatment duration was between four and six months. Outcome was supportive of the hypothesis that unilateral treatment of the spouse of the alcoholic was effective in reducing alcohol consumption. Important gains were reported for the spouses and their alcoholic partners. The investigators urged that a unilateral approach should be subjected to experimental evaluation and, if found effective, used with spouses of uncooperative partners. Dittrich and Trapold (1984) offered a group-oriented treatment package to twenty-six women (twenty-three completed treatment) recruited through newspaper advertisements. These subjects were white and middle to upper-middle class. Their husbands were not receiving any treatment for alcoholism. A control group was drawn from the waiting list. The subjects were tested pre- and posttreatment for depression, anxiety, self-concept, and enabling (drinking) behaviours. The outcome was very encouraging. The treatment group showed significant improvement on all measures, as did the control group upon completion of treatment. Both groups maintained treatment gains over a one-year period. No data were provided on the effects of treatment on the drinking habits of the husbands. In a Japanese study, Nakamura and Takano (1991) reported on the benefits of family involvement in the treatment of female alcoholics. Family involvement was loosely defined and ranged from spouse to lover and could be no more than a visit during the first month of the rehabilitation program. This report is unclear about the nature of the help received by the family members. Only a passing reference was made in the discussion section that family members were 'counselled by trained rehabilitation counsellors, joined the group discussion program,' and took part in some other activities. The value of family visits, however, demonstrated a significantly greater improvement in the family-involved group than in those patients who had no family visits. This study is of considerable interest as it shows that visits from family members have a measurable impact on outcome. If these findings can be replicated then a simple policy to encourage family members to periodically visit their alcoholic relatives during the early stages of treatment can be of great benefit. One major shortcoming of this report is that while visiting by a loosely defined family member is emphasized, the fact that subjects received a great deal of additional help is

Alcoholism and Family Therapy

113

virtually ignored. It would appear that the visits were secondary to the significant amount of therapeutic involvement subjects received. Comment: We have provided a sampling of the variety of 'family oriented' approaches that prima facie appear to be beneficial. The methodology of all three studies described could be significantly improved. Nevertheless, spousal groups, unilateral treatment of the spouse, or visits by a family member in the early stages of treatment of the alcoholic patient seem to improve the odds in favour of positive outcome. Family therapy is not always feasible for practical and clinical reasons. Family-oriented approaches provide alternative strategies to have some involvement with significant others who seemingly play a key role in the treatment and recovery of their alcoholic family members (mostly spouses). Conclusion

This chapter might lead to the perception that family involvement in the treatment of alcoholism is relatively common. The United States Navy study in conjunction with the Veterans Hospitals report suggests that the reverse may, in fact, be the case. On the other hand, a recent attitudinal study of 132 members of the American Association for Marriage and Family Therapy (Zygarilicki and Smith 1992) revealed that 66.7 per cent used family therapy most often in the treatment of alcoholics. Perhaps this common approach among family therapists is slow in influencing some of the most important institutional programs for the treatment of alcoholism in the United States. If this review had a surprise for us, it was the absence of outcome studies based on systemic treatment of the alcoholic patient and his or her family. The 'family' is the other omission. Children were curiously absent from the vast majority of studies. The vulnerability and distress in children of alcoholic parents is well documented which makes this gap in the research literature even harder to comprehend. As for the question about the efficacy of a family approach to treat alcoholism, truly well designed studies were few. The Mccrady and O'Farrell investigations combined demonstrated the value of couple treatment, at least, for white middle-class Americans, with a reasonable prospect for long-term benefits. These studies were behaviourally based. The in-patient oriented treatments held out mixed promise. McCrady's study, which was by far the best designed, failed to show

114 How Good Is Family Therapy? any clear advantage of the in-patient treatment over other approaches. Family-oriented approaches, while numerous, suffered from serious methodological drawbacks. Yet, they held promise. The current state of family treatment research with alcoholics is not without major gaps. The socioeconomic bias is evident in almost every study. AfroAmericans are truly conspicuous by their absence from the samples. There is also a preference for middle-class subjects. Experimental designs for the outcome studies are relatively uncommon. Despite the proliferation over the past four decades of articles promoting various aspects of family dynamics related to alcoholism (Lawson and Lawson 1991), a major conclusion, at this point, has to be that the efficacy of a family-systems approach to treat alcoholism is unknown. A handful of studies using the systems perspective were seriously flawed and, at best, contributed to generating interesting hypotheses. Until such time as these hypotheses have been adequately tested, the question of the efficacy of a systemic family approach to treat alcoholism has to be postponed. The picture is more encouraging as far as the behavioural marital therapy (BMT) approaches are concerned. Studies using BMT consistently showed its superiority over a variety of control groups. A noteworthy exception in the family treatment of alcoholism literature is a comparison between family therapy approaches with other treatment approaches. Some years ago Kaufmann (1985) observed in his review of two decades of family treatment for substance abuse that 'we do not know with much certainty which family dysfunctions are specific to substance abuse, nor do we know which types of family dysfunctions are specifically associated with the abuse of different drugs or alcohol. We do not know that family treatment is superior to other forms of treatment for substance abuse.' The search continues.

7

Adolescent Drug Abuse and Mixed Emotional and Behavioural Disorders

This chapter considers two separate literatures. The first section focuses on adolescent drug abuse and the second on an eclectic range of emotional and/ or behavioural disorders. DrugAbuse

Adolescent drug abuse is a serious problem in terms of its prevalence and, especially, when one considers its implications for continued abuse into adulthood (loaning et al. 1992; Stanton 1979). While the role of the family in relation to drug abuse was recognized as early as the 1950s, systematic attempts to identify salient family factors and link them to treatment did not take hold until the early 1970s (Coleman and Davis, 1978; Stanton et al. 1982) . Stanton's (1979) comprehensive review of family treatment approaches to drug abuse identifies a relatively large descriptive literature, but only eighteen studies that attempt to quantify their results. Only six included comparison or control groups, and none of them dealt specifically with adolescence. Since that time it appears that Stanton's call for rigorous comparative studies has been heeded to a limited degree, at least in terms of family therapy and adolescent drug abuse. We review four studies related to adolescent drug abuse. It should be noted that they either excluded seriously addicted adolescents or did not distinguish between addiction and abuse. As will become apparent little attention is paid to the type of substances involved or the level of involvement. Friedman (1989) compared functional family therapy (described in

116 How Good Is Family Therapy? Chapter 9) and parent groups as treatments for adolescent drug abuse. The family therapy, which was delivered over twenty-four weeks, was provided by six experienced family therapists who received training specific to the model. Treatment fidelity was demonstrated by independent ratings of adherence to the model as outlined in the training manual created by its originators (Alexander and Barton 1983). Treatment for the parent groups was also delivered over twenty-four weeks and was based on a combination of concepts from parent effectiveness training (Gordon 1977), the Canadian Addiction Research Foundation (Shain, Suurvali, and Kitty 1980), and parent assertiveness training (Silberman and Wheelan 1980). Although random assignment was employed, the difference in the actual number of clients involved with their respective treatments (93 per cent with family therapy vs 67 per cent with parent groups) resulted in parent group adolescents being significantly younger and less educated. Individual and peer group counselling was available to adolescents in both groups. The extent to which these services were used is not stated, but there were no significant differences in outcome between the two groups. The families were predominantly of male, white, and Catholic subjects, with adolescents averaging almost eighteen years of age and nine years of education. The most commonly used drugs were alcohol (95 per cent) and marijuana (94 per cent), although various other drugs were consumed by the participants. A wide range of information including standardized scales were collected from the adolescents and their mothers at referral and fifteen months later. Ultimately, sixty-five variables were selected for the outcome analysis. When compared with their pretest scores, each group showed significant gains on most variables. The family therapy group showed significant improvement on fifty-eight of the sixty-five variables, while the parent group families improved on fifty-six variables. Although the findings are not fully detailed, they report reductions in substance abuse, improvement in parent-adolescent communication, improvement in family functioning, and reductions in symptoms. The two variables that were significant for the family therapy condition, but not the parent group, were not specified. Comparative outcomes were analysed using a total of fifteen factors derived from the original sixtyfive variables. The results of the analyses showed no significant between-group differences on any of the outcome factors. This indicates that the two treatments were equally effective. Although this report is presented as a comparison between func-

Adolescent Drug Abuse and Emotional/Behavioural Disorders

117

tional family therapy and parent groups, it would be more accurately described as a comparison between two treatment packages. The findings indicate that functional family therapy or parent groups combined with individual and peer counselling for the adolescent are both highly effective treatment packages for adolescent drug abuse. It is impossible, however, to determine the relative importance of each component. While parent groups are certainly more cost effective, it should be noted that the family therapy condition appears to have been more attractive to potential clients. In a later report Friedman, Tomko, and Utada (1991) focused specifically on the characteristics that predicted better outcomes for the family therapy condition. Their findings, in general, indicate that more positive views of the family led to better outcomes. While these data are important, the lack of comparison with those in the parent group seems to be a missed opportunity in terms of shedding light on who might benefit most from each modality. Joaning et al. (1992) compared the effects of family systems therapy, adolescent group therapy, and family drug education. The family therapy was described as an integration of structural (Minuchin 1974; Minuchin and Fishman 1981) and strategic (Haley 1976) approaches that was also influenced by the work of Stanton et al. (1982) and Selvini-Palazzoli et al. (1978) . A detailed manual was produced Ooaning et al. 1984), and families were seen weekly for a total of seven to fifteen sessions, with all sessions observed by a supervisor and at least one other therapist. The adolescent group therapy was designed to be representative of out-patient groups offered by hospitals and mental health centres. It was based on an integration of social skills training, cognitive development, and role theory and relied on group process as the primary mechanism of change. The group met weekly, for a total of twelve sessions. The family drug education condition consisted of groups of three to four families who met for six biweekly sessions. It consisted of formal presentations, films, and discussions related to drug abuse, adolescent behaviour, and family functioning. All conditions were staffed with doctoral students in family therapy and treatment fidelity was monitored by the investigators. The sample ranged from adolescents who were beginning to display behaviour indicative of drug use to those who were clearly dependent on drugs. A wide range of substance abuse was reported, although

118

How Good Is Family Therapy?

youth who were thought to be addicted to narcotics or solvents were excluded. Families were randomly assigned, with replacement, to one of the conditions until each condition contained at least twenty-three families who completed the pretest, treatment and posttest portions of the study. Families who dropped out prematurely were not significantly different on demographic variables from those who completed. Dependent measures included the Dyadic Adjustment scale (Spanier 1976), the Parent-Adolescent Communication Questionnaire (Olson et al. 1982), Family Coping Strategies (Olson et al. 1982), and the SelfReport Family Inventory (Beavers 1982). Drug use was evaluated according to a composite index which included periodic urinalysis, a drug involvement survey completed by all family members, analysis of videotapes, and therapist assessments. As the investigators indicated, this was not an objective measure of drug use, but an estimate based on patterns suggested by the data. At pretest there were no significant differences between groups on demographic variables or dependent measures. At posttest significantly fewer youth (40 per cent) in the family systems therapy condition were assessed as using drugs, in comparison with group therapy (68 per cent) and drug education conditions (66 per cent). It should be noted, however, that a considerable number of youth continued to use drugs, regardless or the treatment condition. There were no significant differences between groups on any of the other dependent measures. The total sample of adolescents, however, reported significant improvement in parent-adolescent communication, although data from the parents did not corroborate this finding. A six-month follow-up was planned, but the recession-related exodus from the area made it impossible to locate sufficient numbers of families. In general, this study was well designed and well reported. As the investigators point out, the measure of drug use was quite subjective and, therefore, vulnerable to bias. In addition, the sample contained a highly heterogeneous group of drug users, which may have served to dilute effects that might have been achieved for a more homogeneous group. The lack of findings related to family functioning is quite problematic. While family systems therapy appears to be more effective than the other treatments in relation to reducing drug use, this finding cannot be linked to any changes in family functioning. One possibility is that the reduction in drug use may be related to the fact that the family therapy condition was the only one in which a single family and a therapist had to form a relatively intense relationship. Moreover,

Adolescent Drug Abuse and Emotional/Behavioural Disorders

119

improvements in adolescents' perceptions of the quality of communication with parents appears to have been achieved regardless of the form of treatment and might occur without treatment. Overall, these results can only be regarded as tentative support for the effectiveness of family therapy with drug-abusing adolescents. While multisystemic therapy (described in Chapter 9) has not been applied exclusively to adolescent drug abusers, its effects on drug use in serious juvenile offenders has been highlighted (Henggeler et al. 1991) . In one study delinquents and their families were randomly assigned to either multisystemic therapy or to eclectic, psychodynamic individual therapy. About 13 per cent of the sample had previous arrests related to substance abuse. Four years later youth who had received multisystemic therapy had significantly lower rates of such offences, as compared with those who participated in individual counselling (4 per cent vs 16 per cent) . Another study compared multisystemic therapy with monthly probation meetings. Self-reported soft drug use was significantly lower for youth in the multisystemic treatment condition, at the point the posttest was administered. Hard drug use was evaluated, but an extremely low base rate was found and further analyses were not possible. These interventions were not targeted specifically at drug abusers, and the measures they employed were not adequate. The findings, however, suggest that multisystemic therapy may reduce drug use as part of the effort to reduce delinquency. The final studies related to family therapy and adolescent drug abuse compare family therapy conducted conjointly with family therapy conducted through one person (Szapocznik et al. 1983, 1986). In the first report (Szapocznik et al. 1983) sixty-two middle- and lower-class Hispanic families with drug-abusing adolescents were randomly assigned to one of two conditions. Both conditions, which were limited to twelve sessions, were based on an integration of structural (Minuchin 1974) and strategic (Haley 1976; Madanes 1981) family therapy. The conjoint condition specified a maximum of two individual sessions, and the one-person condition allowed a maximum of two family sessions. Treatment was provided by experienced, master's level family therapists. Families were assessed prior to treatment, at termination , and six to twelve months after termination . Measures included the Psychiatric Status Schedule (Spitzer et al. 1970), the Behavior Problem Checklist (Quay and Peterson 1979) , interactional ratings of standardized family tasks, and the Family Environment Scale (Moos 1974) .

120 How Good Is Family Therapy? Findings were reported for thirty-seven families who completed at least four sessions and two evaluations. Unfortunately, there was no information or analysis on the twenty-five families that did not meet these criteria. This is an important gap because comparisons between the two groups, in the reduced sample, revealed a small number of significant differences on the Psychiatric Status Schedule and the Behavior Problem Checklist. Overall, the details of the analysis are well described, although the presentation and interpretation of findings is often vague. Both treatment modalities appear to have been highly effective. At termination families in both groups showed significant gains on the Behavior Problem Checklist, Psychiatric Status Schedule, and interactional ratings. Adolescents showed significant gains on three of the Family Environment Scale subscales, and parents showed gains on three different subscales. Differences between groups were minimal and were limited to adolescents' scores on two Family Environment Scale subscales. Cohesion scores for adolescents in the conjoint group decreased, while for adolescents in the one-person group they increased (the desired direction). Similarly youth in the conjoint group perceived their families as more controlling after treatment, while those in the other condition perceived a significant decrease in control (the desired direction). Significant differences between posttest and follow-up were also quite minimal. Families in both groups showed continued improvement on the flexibility dimension of the interactional ratings. Families in the one-person condition, however, showed continued improvement on three Behavior Checklist Scales (socialized delinquency, inadequate development, and personality problems), while those in the conjoint condition showed deterioration. This relationship also held true for the drug abuse scale on the Psychiatric Status Schedule. In the second report of this study Szapocznik et al. (1983) describe findings on thirty-five different families. The basic design of this study is identical to the first. It appears that the Family Environment Scale was not administered to the second sample, and there is no explanation about why this instrument, which showed some differences in the first study, was excluded. An attrition analysis was completed, but the report is quite vague about what was described as general similarity. The findings of this second study replicate those of the first. Both groups showed significant increases in the vast majority of measures. Again, the one-person format appeared to be slightly more effective, but in different areas than were found in the first study.

Adolescent Drug Abuse and Emotional/Behavioural Disorders 121 These are interesting studies that suggest that structural/strategic family therapy was quite effective regardless of format. While both studies found minor differences by treatment modality, the lack of an attrition analysis (first study), significantly more one-person sessions (first study), and significance on different variables, make it impossible to assert the superiority of one format over the other. Finally, while the overall results are quite impressive, the lack of a control or true alternative treatment group makes it impossible to make statements of comparative efficacy. Discussion

Research on the effectiveness of family therapy with drug-abusing adolescents is minimal. It is interesting to note that all of the studies reviewed above utilized concepts and techniques derived from structural family therapy, at least to some extent. This probably speaks to the impact of Stanton et al.'s (1982) landmark investigation of family therapy with adult drug abusers. Overall, these studies pay surprisingly little attention to describing their samples. It appears that most of the research includes samples that are quite heterogeneous, in terms of the types of drugs used, the severity of the abuse, and the presence of other problems. There is no doubt that objective measures of drug use are difficult to obtain. Consequently, it appears that some investigators opt to collect data on a wide range of variables to obtain statistical significance somewhere. The result can often be a complicated and sometimes contradictory set of findings that are not grounded in a specific understanding of the problem. At this point it appears that family therapy may be an effective means of reducing drug abuse among adolescents. In addition, there is some evidence that these improvements occur along with changes in some aspects of family functioning. Whether family therapy is more effective than other modalities, however, remains an open question. Mixed Emotional and Behavioural Disorders

This final section focuses on a small group of studies that are particularly difficult to categorize. It includes investigations into family therapy with adolescents who are described as emotionally disturbed, behaviourally disordered, or simply as being in conflict with their fami-

122 How Good ls Family Therapy? lies. This heterogeneous group probably reflects the families most typically encountered, but least studied in the field of family therapy. First, we examine a series of studies utilizing Zuk's (1971) gobetween process with families of emotionally disturbed adolescents. This systems-based approach focuses on the identification and active alteration of triadic interactional patterns that are understood to be pathogenic. The emphasis on triads, therapist manoeuvrability, and therapeutic tasks has much in common with strategic approaches to family therapy (Stanton 1981). Garrigan and Bambrick (1975) studied nine emotionally disturbed adolescent boys (mean age, 12.4 years) and their two-parent families who received six weekly family therapy sessions, while a matched control group received no treatment. The sample, which was drawn from a school for emotionally disturbed children, consisted exclusively of white, middle-class families . Pretests were completed during the first weeks of treatment, and posttests were administered one week after termination. Given that treatment lasted a total of six weeks, this reflects a relatively short interval between administrations. Data were collected exclusively from the adolescent and his teacher and included measures of family functioning (Van der Veen et al. 1964), parental attitudes as perceived by the child (Barrett-Lennard 1962), classroom behaviour (Walker 1967), and self-concept (Piers and Harris 1969). The experimental group showed significantly greater gains than the control group on the family functioning measure. While there was no statistical significance on any of the other variables (not surprising, given the sample size), scores for the experimental group were consistently better than for the control group. While far from conclusive, this study provided preliminary support for the approach. A more comprehensive and rigorous study was reported by Garrigan and Bambrick (1977) two years later. Twenty-eight families of emotionally disturbed male and female adolescents (mean age, 14.4 years) were randomly assigned to a treatment or to a no treatment control group. As with the first study all families were intact, white, and middle class. Treatment consisted of ten sessions over sixteen weeks, and adherence to the model was assured by systematic training, supervision, and monitoring. Posttests were completed within two weeks of termination. Measures included the adjustment scale of the Family Concept Q Sort (Van der Veen et al. 1964), the Relationship Inventory (Barrett-

Adolescent Drug Abuse and Emotional/Behavioural Disorders

123

Lennard 1962), and the State-Trait Anxiety Inventory (Speilberger, Garsuch, and Lushene 1970), all of which were completed by family members. The Children's Self-Concept Scale (Piers and Harris 1969) was also completed by the adolescents, and the DevereuuxAdolescent Behavior Rating Scale (Spivack, Spotts, and Haimes 1967) was completed by parents and an independent classroom rater. To the investigators' credit the analysis (ANOVA and ANCOVA) considered the impact of gender and was organized according to Boolean concepts (Feinstein 1967) in order to identify the intersection of significant outcomes as perceived by various participants. In general, the findings indicated comparatively significant improvement in family functioning, marital functioning, and most aspects of adolescent behaviour at home and at school. There were also some differences according to the gender of the adolescent and the parent. There were no changes in adolescent self-concept or school-related state anxiety and parent trait or state anxiety. The adolescent behaviours that did not improve, according to parents and the observer, included bizarre cognition, emotional distance, and hyperactive behaviour. A third study (Garrigan and Bambrick 1979) sought to replicate earlier studies and to confirm Zuk's (1971) assertion that single-parent families were more difficult to treat. Twenty-four white, middle-class families (sixteen intact and four single-parent) with an emotionally disturbed adolescent were randomly assigned to a treatment or control group. Treatment consisted of ten family therapy sessions over a twenty-week period. The study design and dependent measures were identical to those in the previously cited study (Garrigan and Bambrick 1975). Adolescents in the treatment group made significant gains related to trait anxiety, regardless of family composition. State anxiety and several problematic behaviours showed reductions that approached significance, when compared with the control group. The hypothesis regarding single-parent families appears to have been supported in that they did not make significant gains compared with the controls on family adjustment, maternal congruence, and eight behavioural problems, while the treated two-parent families did. Despite the apparent difference between single-parent and two-parent families, Garrigan and Bambrick (1979) combined the samples of this study and the previous one in order to conduct a telephone followup. A total of seventy-two out of a possible ninety-six families were

124

How Good ls Family Therapy?

contacted between one and two years after treatment was completed. The findings indicated that significantly more adolescents in the treatment groups were employed or enroled in regular school (as opposed to the special program for emotionally disturbed youth) than were control group adolescents. In addition, the treatment group significantly outperformed the control group in terms of involvement with the juvenile justice system. There was no attempt, however, to account for the twenty-four families who did not participate in the follow-up. A second follow-up, reported in the same article, asked school personnel for thirty-two youths who had returned to regular programs to rate the subjects' school performance. Adolescents in the treatment group were rated as significantly more improved than those in the control group on academic progress and attention or concentration. Improvements in peer relations, group work, self-control, and prognosis for success approached significance, while no significant differences were found for emotional stability, relationships with adults, and general school adjustment. Taken as a group, these studies provide very tentative support for the effectiveness of family therapy based on the go-between process with emotionally disturbed youth. It should be recognized, however, that in each study the family therapy occurred within the context of a specialized educational program designed to meet the needs of this population. Moreover, sample sizes were quite small, analyses were limited, and the samples were homogeneous as to race and social class. Overall, these studies lacked meaningful interpretation, especially in terms of non-significant findings and findings that were contradictory between studies. Finally, there were no comparisons between family therapy based on the go-between process and any other form of formal treatment. Robin (1981) compared the effectiveness of problem-solving communication training with family therapy and with a waiting list control group in the treatment of parent-adolescent conflict. Thirty-three families recruited through community advertising and mental health clinics were matched according to sex of the adolescents, number of participating parents, and parents' perception of the difficulties as reported in the Conflict Behavior Questionnaire (Prinz, Kent, and O'Leary 1979). The families were then randomly assigned to one of three groups. Problem-solving communication training consisted of teaching the family a problem-solving model; correcting communication through

Adolescent Drug Abuse and Emotional/Behavioural Disorders

125

feedback, modelling, and rehearsal; and cognitive restructuring to alter unreasonable beliefs. The four therapists associated with this condition received specific training and supervision according to a detailed manual. The family therapy was provided by five therapists with varying orientations (two-family systems, two eclectic and one psychodynamic orientation) . Both conditions involved seven one-hour weekly sessions. A comprehensive battery of self-report and observational measures designed to assess parental control and decision making, specific disputes, communication conflict behaviour, and problemsolving skill was administered prior to assignment and after termination. Treated families also completed a ten-week follow-up. The analysis was comprehensive and well described. Families in both treatment groups made significant gains, in comparison with the control group, on behavioural measures of problem-solving communication skills, self-reports of negative affect, and communication conflict at home. These gains were maintained at follow-up. There were no significant differences between treatment groups, except that the problem-solving-communication training was superior to family therapy on the observational measure used to assess problem-solving behaviour. This is not a surprising finding, given the focus of the treatment. As Robin notes, the lack of any other differences between the two treatment groups leads one to question the necessity for explicit skills training in order to produce improvement in family functioning. This study is, primarily, a test of problem-solving communication training. The comparison of an explicit, well-specified communication training model with a varied and eclectic range of approaches to family therapy is not particularly meaningful. In essence, the findings suggest that any family-focused intervention is more effective than no intervention at all. A somewhat similar pattern of findings is apparent in a study that compares problem-solving communication training, behaviour management training, and structural family therapy for treating family conflicts in adolescents with attention-deficit hyperactivity disorder (Barkley et al. 1992). Sixty-four adolescents and their mothers were randomly assigned to one of the three treatment conditions mentioned above. No data were collected from fathers, and they were not required to participate in the treatment, although they were required to remain in treatment if they chose to participate. The rationale for the exclusion of fathers was not specified. Each treatment consisted of eight to ten weekly one-hour sessions. The parent(s) and the adolescent attended

126 How Good Is Family Therapy? all structural family therapy and problem-solving communication training sessions, and only the parent(s) attended the behaviour management training. The investigators went to considerable lengths to ensure treatment integrity and control for therapist-related confounds. All treatments were provided by the same two therapists. Each was trained in the treatments by an expert in the approach. Each therapy was guided by a session-by-session manual and an associated text. Therapists received six to eight hours of didactic training as well as weekly supervision on two pilot families. Therapists received ongoing supervision, and tapes of sessions were submitted to the trainers who rated them for adherence to the models. A wide range of measures were collected before and immediately after treatment, as well as at a three-month follow-up. These included the Child Behaviour Checklist for Parents and Youth (Achenbach and Edelbrock 1983, 1987), the Conflict Behaviour Questionnaire (Robin and Foster 1989), Locke-Wallace Marital Adjustment Test (Locke and Wallace 1959). the Beck Depression Inventory (Beck, Steer, and Garbin 1988), and the Family Beliefs Inventory (Vincent-Roehling and Robin 1986). In addition, videotaped interactional tasks were administered and scored using the Parent-Adolescent Interaction Coding System (Robin and Foster 1989). Therapists rated family cooperation for each session and families completed a consumer satisfaction survey at termination. The analysis was comprehensive, appropriate, and described in considerable detail. The mothers and adolescents in each of the three groups were found to be equivalent on all relevant sociodemographic variables. All treatments produced equally significant improvements in parent-adolescent communication, number of conflicts, intensity of anger, parent-reported school adjustment, parent-reported and selfreported adolescent externalizing and internalizing symptoms, and maternal depression. In addition, all three conditions yielded equally high consumer satisfaction ratings. All improvements were maintained at the three-month follow-up. As with Robin's study, however, ratings of family interaction did not corroborate these improvements. Recognizing that findings based on group mean differences do not necessarily represent clinical significance, the investigators followed Jacobson and Truax's (1991) recommendations for estimating clinically significant change in each case. Clinical improvement was apparent in 5 per cent to 30 per cent of sub-

Adolescent Drug Abuse and Emotional/Behavioural Disorders 127 jects, with no significant differences across groups. The proportion of families considered clinically recovered (i.e., fell in norms for nondistressed populations) was between 5 per cent and 20 per cent, with no significant differences across conditions. The Barkley group's study was well designed and implemented, although the exclusion of fathers in the data collection is questionable. The inclusion of a true control group would have strengthened the findings , but all three models have been shown to be more effective than no treatment in other contexts. The lack of significant differences in improvement rates between treatment groups would seem to indicate general equivalency among the family-oriented models presented. The low rates of clinical significance and the fact that findings from observational ratings did not corroborate self-reported improvements require further exploration. To some extent it appears that these models may not change family functioning as much as they relieve distress by altering attitudes and perceptions about family life. Discussion

As mentioned in Chapter 3 research on family therapy with adolescents is complicated by the fact that families dealing with similar issues tend to be categorized according to service sectors rather than shared characteristics. The research on mixed emotional and behavioural disorders is severely limited because of a lack of specificity regarding the issues under investigation. The studies reviewed here fail to adequately describe the interactional characteristics of the families and are also quite limited in their descriptions of interventions. In general, the link between family functioning and emotional and behavioural problems is quite tenuous. Based on an examination of these studies the best one can say is that family-focused approaches are probably better than nothing, and maybe they are better than other approaches when it comes to youth who are variably described as emotionally disturbed, hyperactive, or simply in conflict with their parents.

8

Anorexia Nervosa

Anorexia nervosa (AN) and, to a lesser extent, bulimia are eating disorders primarily affecting adolescent women. This chapter will focus on anorexia nervosa. Bulimia is excluded from this analysis because of the virtual absence of family therapy outcome studies on it (Hudson and Pope 1986). The major symptoms of anorexia nervosa are (1) weight loss of 25 per cent or more, (2) poor blood circulation, (3) slowed metabolism, and (4) amenorrhea. These patients demonstrate obsessive preoccupation with body weight, compulsive thinking about food, excessive physical activities (to keep weight down), and a highly distorted body image. AN is a 'psychogenic' disorder of refusal to eat rather than loss of appetite. A brief, yet elegant description of this condition is the following: 'Anorexia nervosa, a distorted biological solution to an existential problem for an adolescent and her (or occasionally his) family, is a crippling condition' (Crisp 1980). Conceptualization of AN has run the gamut of medical and psychiatric formulations - from pituitary to psychiatric to the modern view of AN as a complex biopsychosocial phenomenon. Treatment regimens include pharmacological, behavioural, psychoanalytic, group, and systemic-family therapy. Exclusive reliance on any one treatment is rare, and this fact has serious implications for this literature survey, which is solely concerned with the outcome of family therapy for AN. In their excellent review of drug treatment for AN, Rockwell, Nishita, and Ellinwood (1984) concluded that, 'there is no clear drug treatment of choice from which one can expect a consistent, even if small, beneficial effect, either on weight, attitudes, or behaviors.' Crisp (1980) and Garner, Garfinkel, and Bemis (1982) have emphasized the need for multi-axial psychotherapeutic approaches (psychodynamic,

Anorexia Nervosa

129

behavioural, family, group) that appear to be unavoidable in the treatment of anorexia nervosa. It is perhaps for that very reason that reports of the exclusive application of family therapy for the treatment of AN are very few. On the other hand, family therapy or familyoriented counselling is often one of the elements of the integrated treatment plan. Bruch (1974) combined psychoanalytic (deficits in ego functions) with interpersonal (overanxious mother eager to feed her child) and cognitive and perceptual defects to develop her complex view of AN. Dittmar and Bates (1987) viewed Bruch's interpretation of AN as the 'paradoxical and self-destructive expression of autonomy in the area of eating and body-control, while remaining a desperately dependent child in others.' Complex interpersonal conflicts are implicit in Bruch's analyses. According to Dittmar and Bates, 'The anorexic's search for identity is maladaptively pursued through the mind's control over body which is viewed as alien and exerting an independent will.' However, the centrality of the family dynamics in the etiology and perpetuation of AN was claimed by Selvini-Palazzoli (1974) in Milan, Italy, and by Minuchin and his colleagues (1978) in Philadelphia. Selvini-Palazzoli's proposition is that specific patterns of family relationships are regularly observed in anorectic families. These families display a common set of attributes, namely, parental inability to assume responsibility, a general tendency to recriminate, and a facade of marital harmony. Decisions are made usually for the 'good' of others thereby making self-sacrifice a prominent feature of family relationship, and this has implications for the development of autonomy in an adolescent. Parents, especially mothers who tend to be self-blaming, account for their conduct on the basis of their devotion to their children. Based on her study and treatment of twelve anorectic children and their families, Selvini-Palazzoli (1974) created the concept of 'three way matrimony' to describe the role of the child as a 'go-between' for the parents at the personal cost of lost opportunity for autonomy and independence. Children find themselves in a bind and constantly violate generational boundaries (the similarity of this view with Minuchin's structural perspective is apparent), although the treatment employed by Selvini-Palazzoli (1974) and the Milan group is predicated on strategic principles. In describing the 'psychosomatic' families, which included such diverse conditions as childhood asthma, diabetes, and anorexia ner-

130 How Good Is Family Therapy? vosa, Minuchin and associates (1978) identified certain patterns of family behaviour which created optimum conditions for these illnesses to flourish. Four intcractional characteristics, namely, (1) enmeshment, 2) rigidity, (3) overprotectiveness, and (4) lack of conflict resolution were uniformly observed. Minuchin and his colleagues claimed that 'a child who has learned to relate in highly enmeshed patterns can become an anorectic if certain other processes are also present in the family.' The 'other' processes consisted of dynamics embedded in the family characteristics, such as ill-defined boundaries, overprotectiveness impeding the development of autonomy in the child, and so forth. Bemporad and Ratey (1985), in a rather creative exploration into the past relationships of six recovered anorectic females, found significant evidence of family conflicts such as 'a particular mother-father match, mothers being depressed, bitter, and ... unfulfilled in terms of unfulfilled satisfaction' and the fathers 'ultramasculine' but beneath the facade, 'boyish, dependent and unreliable.' In congruence with Selvini-Palazzoli's observation, they also noted that the parents maintained 'a pretense of harmonious relationship.' Colapinto (1991) claimed that 'of the three syndromes treated by Minuchin and his team, anorexia nervosa was the best suited to a family focused approach. Unlike asthma and diabetes, in which the dysfunctional context plays just an exacerbating role on primarily physiological conditions, anorexia has no demonstrated physiological basis, which opens the possibility for a well designed family intervention to bring about a cure.' How effective Minuchin's formulation of and treatment for AN is indeed a matter of curiosity. Suffice it to say that at a fundamental level several problems remain unresolved. Yager (1982), based upon his reading of the 'psychosomatic' formulation of AN concluded that 'they [the family characteristics) do little to account specifically for the appearance of AN rather than any other illness.' He conceded, however, that while Minuchin's proposition may lack the power to explain the etiology of AN, the family systems formulation 'helps to explain how symptoms can be provoked and sustained once they appear.' Roy (1987) made similar observations about the general lack of specifity in the psychosomatic model to offer convincing evidence of etiology for any of the other psychosomatic disorders examined by Minuchin. Kog and colleagues (1987) in their careful re-evaluation of the family characteristics have found considerable evidence of 'overlapping' concepts in Minuchin's model. If, indeed, the concepts upon which the

Anorexia Nervosa

131

etiologic foundation for AN rests are flawed, can therapy be fully effective? Despite these criticisms family therapy is often an integral part of the therapeutic regimen for the treatment of AN, and the credit belongs to the pioneering works of Selvini-Palazzoli and Minuchin. Crisp (1980), whose research contribution to AN is enormous, has consistently emphasized the need for family involvement in the treatment of AN . He said, 'Central to the task of helping the anorectic is that of also involving or helping parents or others importantly concerned with them' (ibid.: 98) . Crisp, Harding, and McGuinness (1974), in an earlier study, had offered some evidence in support of the idea that AN in the family could serve a protective function for one or both parents. A few studies have demonstrated an interesting association between family factors and AN. Is AN familial? Only indirect evidence is available. Kalucy and colleagues (1977), in a sample of fifty-six anorectic patients, found that 23 per cent of fathers and 16 per cent of mothers presented a history of low adolescent weight or fear of overweight. Winokur and associates (1980), in a controlled study of twenty-five anorectic patients, found affective disorders in 22 per cent of their relatives, as opposed to 10 per cent in the control group. Cantwell and associates (1977) had reported seventeen parents and six siblings with a history of affective disorder in a group of twenty-five anorectic patients. On the basis of a very thorough review, Garfinkle and Garner (1982) concluded that 'the exact role of the family in the predisposition and perpetuation of anorexia nervosa is not clearly understood. This is partly due to methodological problems related to the few investigations that have been conducted in this area and partly due to the nature of anorexia nervosa, as a heterogeneous syndrome with multiple predispositions' (ibid.: 184). Reports on the outcome of family therapy for AN are varied and for that reason a simple method to classify them has been adopted: first, those studies where family therapy is the main (not the only) method of intervention; second, where family therapy is part of integrated treatment (family therapy plus); and third, treatments that may be described as having family orientations such as parents' group (familyoriented therapy) without formal family therapy. Within each category we shall carefully consider the methodological issues of the studies specifically related to family matters and the impact of family therapy on the final outcome. As will be seen the issues related to family therapy are only infrequently described at any length, but mostly passing

132

How Good Is Family Therapy?

reference is made to family therapy as one component in the treatment plan . Family Therapy as the Main Intervention

Minuchin and his associates (1978) reported the results of their treatment of the families of anoretic patients in great detail. It should be noted that their treatment regime was not exclusively confined to family therapy, but included a preliminary short-term behaviour modification program for their in-patients. In addition, the patients were under pediatric supervision. Nevertheless, the bulk of the treatment consisted of family therapy (lunch sessions) designed to 'transform the dysfunctional processes of mutual regulation of a psychosomatic family system' (ibid.: 125) . Minuchin et al. (1978) reported on the outcome of fifty-three subjects who were all treated within a systems model. The group consisted of forty-seven females and six males with a mean age of fourteen years. The mean duration of illness was 9.5 months. The families were rated for their functioning on enmeshment, rigidity, overprotectiveness, and conflict. A modified Wiltwyck Family Task questionnaire was used for observational assessment of the families. Fifty families completed the treatment of weekly family therapy. Outcome was judged as medical improvement regarding remission of anorexia symptoms and clinical assessment of psychosocial functioning. Patients were followed up for two years. An astonishing 86 per cent were reported as recovered, 4 per cent fair, 6 per cent unimproved, and 4 per cent relapsed. Minuchin's study of anorexia is unique, and the results are nothing short of spectacular. The most striking aspect of this study is simply the very detailed and meticulous reporting of every phase of the research. Second, the researchers themselves anticipated some of the criticisms, such as the relatively young age of the patients and the short duration of the illness. They acknowledge that 'probably it is true that younger anorectic patients are more amenable to treatment' (ibid.: 137), although Rosman et al. (1977) challenged the notion that the relatively young age of Minuchin's patients was a major contributing factor for positive outcome. The three unimproved cases were distributed over the entire age range (from nine to twenty-one). Third, Minuchin concluded that 'when anorexia nervosa patients are treated within a year of the beginning of the illness with a systems approach in the context of

Anorexia Nervosa

133

their family, they can be cured in a short period of time' (ibid. : 138). Another positive feature of the study was a clearly defined family therapy (structural) approach. Nevertheless, serious concerns remain about Minuchin's methodology. It is the very success of his approach that paradoxically raises very fundamental questions about it. Apart from the questions related to the sample characteristics, virtually no other data are provided on the family demographics or family functioning. Is it conceivable that certain types of families, such as those higher or lower socioeconomically, are more amenable to family treatment than are others? Does race or ethnicity feature in this equation? The absence of a control group is also problematic. Post hoc comparison of findings with previous studies is unsatisfactory and, indeed, that was their experience. Statistical analysis was kept to a minimum, and the reporting of outcome was anecdotal, largely because of the heavy reliance on observational techniques used for assessments; also the question of observer bias is not easy to overlook, and the influence of operant techniques on the outcome was ignored. It is undeniable, however, that structural family therapy was demonstrated to be enormously effective in treating young anorectics of recent onset. What made this treatment so successful? This is an important question because serious doubts have been raised about the very conceptualization of so-called psychosomatic families. Were there other factors beyond the family intervention that could account for such dramatic results? Have these results been replicated with improved research methodology? Minuchin's very novel systemic formulation of AN and the unmatched success of its treatment is best regarded as an experimental rather than authoritative conceptualization and therapy of this very complex disease. After nearly one and a half decades since Minuchin and associates published their work, replication has not occurred. Certainly no subsequent study has even come close to matching their success rate. In medicine a close to 90 per cent recovery rate is virtually a guarantee for a cure. This fact alone is cannon fodder for sceptics. Stierlin and Weber (1989), in a follow-up (mean, four years and five months) uncontrolled study of forty-two families also reported that 85 per cent of the anorectic families, following family therapy, 'changed in the direction of greater functionality.' Five families did not improve, and only one family deteriorated. The mean age of the patients at first show of symptom was fourteen years and eleven

134

How Good Is Family Therapy?

months and age at the beginning of therapy was eighteen years and two months. The mean number of family sessions was six. The therapy was based on systems theory. Russel and associates (1987) noted that neither Minuchin nor Selvini-Palazzoli 'verified their claims [of successful outcome of family therapy for AN] by undertaking controlled trials and systematically following up their patients, and yet there has evolved the impression that the efficacy of family therapy in anorexia nervosa has been demonstrated.' An example of such an ' impression' is a rather categorical statement by Gurman and Kniskern: 'At the moment, structural therapy should be considered the family therapy treatment of choice for these childhood psychosomatic [anorexia nervosa, diabetes, asthma] conditions' (1980: 750) . To address the shortcomings in the aforementioned studies, Russel and colleagues (1987) implemented a controlled trial comparing family therapy based on the structural model (focus on family organization) with individual supportive psychotherapy (educational, problemcentred, cognitive, interpretive and strategic). Their sample consisted of fifty-seven subjects with AN and twenty-three with bulimia nervosa. Seventy-three subjects participated in the study. They were divided into four groups on the basis of age of onset (eighteen years or less to nineteen years or more) and duration (less than three years to more than three years) of the disease. The first part of the treatment was inpatient to restore their weight. Just prior to discharge they were randomly allocated to supportive psychotherapy (thirty-nine patients) and family therapy (forty-one patients) groups. Outcome measures consisted of body weight, restoration of menstrual function , and ratings on the Morgan and Russell Outcome Schedule, which measures adjustment in five areas in AN patients: nutritional status, menstrual function , mental state, psychosexual adjustment, and social functioning. At one-year follow-up in the category of general outcome 23 per cent AN patients had good outcome, 16 per cent intermediate, and 61 per cent poor. Of the bulimia patients, 9 per cent showed good outcome, 13 per cent intermediate, and 78 per cent poor. However, in four of the five dimensions of the Morgan and Russell Outcome Schedule the family therapy group showed a significantly greater improvement over the individual therapy subjects. Their main finding confirmed Minuchin's data that family therapy seemed most beneficial for younger patients whose disease was of relatively recent onset. A secondary and some-

Anorexia Nervosa

135

what tentative finding was that individual supportive psychotherapy appeared to be more effective with patients whose illness commenced in late adolescence (age nineteen years). The authors themselves identified a series of methodological issues such as the risk of allocating more severely ill patients to one group or another. Randomized allocation might have averted that problem. They explained their low rate of recovery on the basis of the severity of disease in their subjects and that one year of follow-up was inadequate for the full benefits of the treatment to have taken effect. On the positive side, it was a controlled study, the therapeutic processes were clearly described, and information was provided about the training and background of the therapists. While undoubtedly meeting some of the key criteria of effective methodology, the Russel study still raises several issues. First, since the authors deny that 'pathological family processes were of causal importance,' their rationale for the adoption of family therapy was at best fuzzy because the efficacy of family therapy was under investigation. Why should altering 'family organization' have any effect on the behaviour of an anoretic unless the family is either implicated in the etiology of AN or plays a part in the maintenance and perpetuation of the disease? This is a major conceptual gap in this study. It is acknowledged by the authors that there was no attempt (other than, one presumes, clinical impressions, which they do not report) to assess and measure the family dysfunction of these patients. How dysfunctional were these families? What kinds of changes were introduced into the family system to effect positive outcome? Is it possible that those who failed to respond to family therapy in a positive way were more dysfunctional than those who did? It is curious that a study designed to demonstrate the efficacy of family therapy chose to be so parsimonious with family-related information and also rejected (perhaps correctly) the etiologic significance of family functioning in the genesis of AN. Or did it? The report went on to state that as a result of family therapy 'communication between the parents improved. Important issues, such as the gradual separation of adolescents from the family, were eventually contemplated with greater ease.' If the family dynamics lacked etiologic substance (on the basis of the author's previous statement), they certainly played a role in the perpetuation of the disease and dependency. The point of note about the Russel study is that family therapy proved far less effective than Minuchin et al. (1978) had previously

136 How Good Is Family Therapy? reported. The study design was an improvement on Minuchin's for it was controlled, relevant statistical methods were employed, and the sample size was more than adequate for statistical analyses. This study failed to even come close to matching Minuchin's remarkable success in treating AN. How different and how similar the two groups of families were on various aspects of family functioning was hard to ascertain. In a subsequent study the same group (Dare et al. 1990) elaborated on their earlier findings and concluded that AN patients eighteen years of age and younger with a duration of illness of less than three years were more likely to benefit from family therapy than from individual supportive psychotherapy at the end of the one year of treatment, thus registering clear disagreement with Minuchin et al.'s (1978) suggestion that age and duration of AN were not solely responsible for the very positive outcome of family therapy. For older patients or AN of longer duration or both the differences between the two therapies failed to reach statistical significance. In this report the authors further elaborated their method of family therapy which, in short, was a blend of Minuchin and Selvini-Palazzoli. Furthermore, they acknowledged the role of family dynamics in perpetuating and maintaining anorexic symptoms, while still emphasizing the absence of a causal link between family dynamics and AN. The findings of Morgan and Russell (1975) received further confirmation in their later study of seventy-eight patients with AN whose treatment was 'a pragmatic one and no single style [was) allowed doctrinaire precedence' (Morgan, Purgold, and Wellbourne 1983). On the predictive value of clinical factors, they suggested that a relatively late onset, a disturbed relationship between the patient and family members, and premorbid personality difficulties predicted poor treatment outcome for AN, and this, too, received further validation later (Russel et al. 1987). Nevertheless, in another study of the long-term outcome of treatment (one-to-one psychotherapy and weight restoration} for AN, parental marital relationship, relationship with mother and father or siblings and friends, and a host of family-related factors were found to be unrelated to outcome (Hall et al. 1984). Younger age of onset, however, was predictive of positive outcome, giving further credence to one of Russell's findings. Crisp (1980) also reported that early onset of AN following the start of puberty in conjunction with a shorter duration of illness (combined with other factors} was predictive of positive

Anorexia Nervosa

137

outcome. The contention that Minuchin's remarkable success could, in part, be attributed to patient characteristics (young age and short duration of illness), independent of treatment methods, appears to have some validity in the outcome research. Martin (1984) treated twenty-five adolescent anorectics, mean age 14.9 years and mean duration of illness 8.08 months, making this sample almost identical to Minuchin's. The principal method of assessment included individual and family interviews. All interviews and subsequent treatment were conducted by the author. The focus of treatment was the nuclear family system 'including individual subsystems' within it. The duration of treatment ranged from two to thirtynine months. Five patients also received 'mild behavior modification' treatment. Eighteen patients had 'some' in-patient experience, and four were admitted for psychiatric treatment. However, the bulk of the therapy was done on an out-patient basis. Outcome measures consisted of weight, menstrual function, eating patterns, and family and social life. At termination of treatment 45 per cent were at or over the ideal weight, and only three had established regular menses. At follow-up (mean sixty-one months, SD=l5), seventeen patients or 77 per cent of those who completed treatment received a rating of excellent, two or 9 per cent good, and one patient (4 per cent) fair. The rest were in the unsatisfactory category. The results were far less impressive than Minuchin's, but considerably better than Russell's. This study is fraught with problems, the most obvious being lack of a control group and a relatively small sample size. Although the outcome measures were comprehensive, sole reliance on a clinical interview for family assessment is, at best, suspect. Statistical analysis of the data was non-existent. The author remained silent about her findings related to family functions, and she gave no hint about the actual therapeutic process employed by her. The reasons for selecting 'therapy of the nuclear family system' were also elusive. On the positive side, the results suggested that long-term outcome of family therapy was likely to be more encouraging than the results at the point of termination of treatment. Summary: This body of research is essentially limited. Control studies are a rarity. The early promise of Minuchin's structural family therapy as a panacea for early onset AN in young patients appears to have suffered some setback. The true merit of that model still awaits careful

138 How Good Is Family Therapy? replication with improved research design. Until such time any final judgment on the merit of Minuchin's work should be postponed. To date only Russel's study is an approximation of such a replication, with far less encouraging results. There are so few studies in this category, each with its own set of methodological limitations, that it would be erroneous to draw any conclusions from them. Other than Minuchin's pioneering report the studies, including Russel's, provide very limited data on the families and the therapeutic process itself. Duration of therapy is often reported in global terms rather than the actual number of sessions. Standardized instruments are generally not employed to assess family function, and statistical analysis, with the exception of Russel's study, is confined to inefficient reporting of frequency data. Clinical 'impressions' are offered as substitutes for objective findings. The impact of therapist variables on outcome is ignored. The studies fail to describe the changes in the family system following treatment and at follow-up. On the basis of the reports presented, it is virtually impossible to make any estimates of outcome. There is, however, some agreement that family therapy appears to be most effective with young adolescents whose disease is of very recent onset. Garfinkle and Garner were emphatic that 'if the patient is 16 or younger and living at home, it has been our experience that family therapy must be considered as the primary mode of therapy' (1982: 293) . Young age of onset, however, is no guarantee for positive outcome. Bryant-Waugh and colleagues (1988) in their seven-year follow-up (in contrast to Minuchin's two years) of thirty anoretic children whose mean age at onset of the disease was 11.7 years (Minuchin's was fourteen years), found that eighteen children or 60 per cent reported good outcome as opposed to Minuchin's 86 per cent; ten children remained moderately to severely impaired, and two died. Treatment was a combination of weight restoration, behavioural therapy, individual psychotherapy, medication, a social skills development program, and family therapy based on Minuchin's structural model (Lask and Bryant-Waugh 1986). One of their startling findings was that onset of AN at age eleven years or earlier (combined with disturbed family life and other factors) was a poor predictor of outcome. Minuchin and his colleagues have unquestionably exerted profound influence in the management and treatment of AN. Dare and associates (1990) have raised some appropriate questions on this matter. They asked, 'ls it [relative failure of family therapy with older and

Anorexia Nervosa

139

chronic AN patients] to do with the extent to which symptoms and family structure are bound together; or is it due to the problem of engagement with older patients; or is it that we lack effective strategies to address "the paradox of the symptom" in older patients?' A definitive answer must await further study. The future direction of research must test in controlled situations the true efficacy of family therapy for anorexia nervosa. Family Therapy Plus Other Treatments

The studies in the previous section did not rely solely on family therapy to the exclusion of all other treatments. Medically, such exclusion is wholly undesirable. Yet, family therapy constituted the principal mode of intervention. In this section we are mainly concerned with studies where family therapy was only one of the elements of multimodal treatment. Walford and McCune (1991) reported on the outcome of treatment in an uncontrolled study which included a behavioural program for refeeding, individual psychotherapy with the child, and family therapy. Fifteen children under the age of thirteen years were involved in an inpatient treatment, with a minimum of three years of follow-up. The mean duration of treatment was 4.4 months. Outcome measures included the Morgan and Russell Outcome (Schedule) and the Present State Examination. The average age at follow-up was 17.7 years. There were fourteen surviving subjects. Outcome in terms of weight was good in ten subjects; twelve subjects reported regular menstruation; and occupational outcome was positive for twelve subjects. Psychosocial outcome was less satisfactory, as thirteen subjects were still living with parents, and six subjects expressing fear of or no interest in sexual relations. On the general outcome (combined measure of weight and menstrual functioning) portion of the Morgan-Russell Outcome Schedule, 47 per cent showed good outcome, 27 per cent intermediate, and another 27 per cent poor. On the 'average' score portion of the Morgan-Russell Outcome Schedule, which includes an aggregate of sexual, menstrual, mental, nutritional, and social functioning, ten subjects had good outcome, three intermediate, and two had a poor outcome. Walford and McCune did not provide any family or family-therapy related information other than the speculation that the poor prognosis 'hinted' at in this study might have been a function of parents feeling

140 How Good ls Family Therapy? demoralized and de-skilled because of the prolonged hospitalization of their children. The actual number of families who received family therapy was not declared. The sample size was small, and only very basic data analysis techniques were employed. Obviously early onset AN, unlike in some previous studies, notably those by Russel et al. (1987) and Minuchin et al. (1978), did not predict positive outcome for family therapy. The problem is that in the absence of a minimum of familyrelated data, even an elementary comparison with other studies is impossible. Jenkins (1987) described a three-year follow-up study of twenty-one AN adolescent girls treated in an in-patient program. The average age on admission was 15.05 years (SD=l.3). The duration of illness was three months to one year in twelve patients, and between one and two years in nine. The treatment was described as a 'strict refeeding program,' but included elements of other treatments such as antidepressant medication for 48 per cent of patients, electroconvulsive therapy (ECT) for one patient, individual supportive psychotherapy, and 'some family therapy/ counselling took place after satisfactory weight gain, for variable length of time.' An interesting family-related finding was 'lack of separation from parents [for the subjects), either through independent living or marriage.' Absence of marriage, given the average age of the subjects, was not surprising. The study did not report any further family-related information, other than that 'only two claimed family disharmony,' and it did not shed any light on the type or duration of family therapy, or for that matter on the number of families who received family therapy. At the three-year follow-up 48 per cent of the patients were found in the good category of outcome, 19 per cent intermediate, and the rest were in the poor category on Morgan and Russell's Outcome Schedule. The picture was basically unaltered at the four-year follow-up. Jenkins's study shares essentially the same set of methodological shortcomings so evident in most of the previous reports. Given the design it is impossible to make any judgment about the value of family therapy on its own or in conjunction with other treatments. It would be interesting to learn if the final outcome was in any way influenced by the families' response to family therapy. Given that nearly half the subjects did not meet the criteria of good outcome, the investigator had an opportunity to explore the factors that might have accounted for the variance. Unfortunately, the author steered clear of any statistical analyses.

Anorexia Nervosa

141

Marner and Westberg (1987) combined family therapy with group therapy for eight patients with AN and bulimia and separate group therapy for parents. The family therapy, predicated on the structural model and later changed to Selvini-Palazzoli's Milan model, was offered to nine families, and eight of them also participated in group treatments. Outcome was assessed on body weight and social functioning. At the point of termination of treatment three girls had gained normal weights and normal social functioning, four patients were underweight but had attained normal social functioning, and one had remained unchanged. This is best viewed as a clinical-anecdotal report because of the absence of a research design. The number of subjects was too small to draw any generalizable conclusions. The baseline information was confined to demographic data and weight loss. The family and socialfunctioning data were completely missing. How the families were assessed and how and what type of social functioning data were obtained did not feature in the report. Standard statistical analysis was not feasible given the small sample size, but the adoption of a singlecase design would have vastly improved the quality of this report. There was no comparison group. The authors did not present any data 'comparing the results of concomitant group therapy and family therapy.' The true value of this study is that it raised an interesting question about the merit of combining different models of psychotherapy to improve outcome. The findings of this report, of necessity, must be viewed as very soft. In a preliminary study Nagaraja (1974) also demonstrated the superiority of psychotherapy and family counselling over drugs plus symptomatic treatment and no treatment for anorectic children with cyclic vomiting. That study also failed to report even basic demographic and treatment-related data. Garfinkle, Moldofsky, and Garner (1977) conducted a comparative study between operant conditioning and psychotherapies which included an element of family therapy. Seventeen subjects received operant conditioning (this group of patients were lighter and younger at the onset of AN), 12 per cent (five patients) family therapy, 74 per cent (thirty-one patients) individual supportive psychotherapy, 21 per cent (nine patients) pharmacotherapy, 5 per cent (two patients) ECT, and 12 per cent (five patients) were seen for consultation and investigation. Seventeen patients treated with operant conditioning were compared with the remaining twenty-five. At an

142

How Good Is Family Therapy?

approximately thirty-one-month follow-up there were no differences between the operant conditioning group and the others. Regarding family therapy this report is of very limited value, because no details were provided for those 12 per cent who were so treated. Its most important finding, namely, that 'most patients with AN improve. This is true regardless of the method of treatment employed' is of considerable interest. Fosson and associates (1987) retrospectively investigated forty-eight children, thirteen boys and thirty-five girls, aged fourteen years or less, with 11. 7 years the mean age of onset of AN , who were treated in an inpatient program. The treatments included individual psychotherapy, behaviour modification, and conjoint family therapy. This is one of the few studies which reported the findings related to family dysfunction. Parental overinvolvement, communication problems, failure to resolve conflict, overprotective parents, and inconsistent or inadequate parenting were relatively common features of these families. Thirtyfour (71 per cent) of the families participated in family therapy. The results did not indicate the exact nature of the benefits accruing from family therapy on its own or in conjunction with other treatments. Sixty-eight per cent achieved all the therapeutic goals. The study was uncontrolled. It is surmised that family information was obtained by means of clinical interviews, and this creates problems of bias. Analysis was confined to descriptive statistics. Crisp and associates conducted two studies. The first one (Hall and Crisp 1987) randomly allocated thirty patients (fifteen in each group) with 'severe anorexia' to twelve sessions of dietary advice or twelve sessions of combined individual and family psychotherapy. The mean age of patients was 19.55 and 19.57 years for the psychotherapy and dietary groups respectively. The mean age of onset was just over seventeen years for both groups. All subjects had amenorrhea and had been ill for between six and seventy-two months. Each subject and her family were interviewed by a psychiatrist and a dietician to collect comprehensive medical and family information, which included family evaluation, mental state, history of AN, and psychosexual history. The family therapy was designed to 'change those aspects of relationships which tended to stifle the subject's development and maintain the anorexia, particularly overprotectiveness, conflict avoidance, enmeshment, and distancing within the family.' These family features are almost identical to Minuchin's description of 'psychosomatic' fam-

Anorexia Nervosa

143

ilies. The nature of the mix of family and individual therapy was not reported by Hall and Crisp, other than to state that willingness of the family to be involved in family therapy was a determining factor. The number of families participating in such treatment was not reported. The dietary treatment consisted of discussing diet, mood, and daily behavioural patterns. On some occasions families were also seen - how many and for what reasons were not reported. At one-year follow-up the assessment of all subjects and their families was conducted by an investigator who was blind to the treatment allocated. The compliance rate for the psychotherapy group was better with fourteen patients completing treatment, whereas in the dietary group eleven patients completed the entire treatment. No significant differences were found in the body weight. However, when allowance was made for a large standard deviation, the dietary group showed significant mean overall weight gain. The psychotherapy group showed significant improvement in the areas of sexual and social adjustments between presentation and follow-up. The dietary group showed mild non-significant improvement. Both groups registered significant improvement in global scores after one year. The global scores were derived from a calculation of six functions, namely, body weight, menstrual function, eating pattern, mental state, and sexual and social adjustment. At one-year follow-up, however, only four patients in the psychotherapy group had recovered. The rest accepted further psychotherapeutic help. All fifteen patients in the dietary group required further help, but only eight accepted it. These are not particularly encouraging outcomes and clearly suggest the necessity of longer term treatment and follow-up. Given the scope of the study it is regrettable that Hall and Crisp did not provide more details about family therapy. What kind of family dysfunctions were identified (the use of a standard family assessment instrument would have improved the quality of family data)? How many patients received family as well as individual psychotherapy? Who were the participants in family therapy? What was the role of the family therapist? Did patients who received both forms of psychotherapy perform better than those who did not or than the dietary group? Apparently only one therapist was involved in conducting individual psychotherapy as well as family therapy. What may be the implications of that fact alone, since the two therapies are grounded in very different theories and may even conflict with each other on matters of conceptualization of problems? Despite these gaps this study demonstrated

144

How Good Is Family Therapy?

the effectiveness of two short-term psychotherapeutic approaches to produce significant improvement, in the short run, in physical and psychological functioning. The study was well designed, comparing two types of treatments with well-matched groups of patients and using appropriate analyses. In their second study Crisp and associates (1991) expanded the scope of their previous work. Ninety patients with severe AN were randomly allocated to four groups: (1) In-patient comprehensive treatment involving weight restoration, weekly individual and family therapy, group therapy, dietary counselling, and occupational therapy. The in-patient phase was followed by twelve sessions of out-patient individual and family psychotherapy. (2) Twelve sessions, given over several months, of individual and family therapy on an out-patient basis. The therapist adopted an active role in the sessions, and the emphasis in family sessions was on setting appropriate family boundaries, dealing with conflict avoidance, and finding effective ways of problem solving (the structural approach appears to be the family therapy model used) . Individual psychotherapy was focused on issues related to the patient's status, difficulty in asking for help, low selfesteem, and so forth. (3) Out-patients involved in ten sessions of group therapy with other patients and ten sessions with their parents separately. Both Groups 2 and 3 also received dietary advice. (4) A control group of' no further treatment.' The Morgan and Russell Outcome Schedule, which provides a score of 1 to 12 on nutritional status, menstrual function, mental state, and sexual and socioeconomic adjustments, plus a global rating of O to 12 from individual ratings, was used. All three interventions were found effective at one-year follow-up. Weight gain and the Morgan-Russell scores were significantly improved from the baseline for all four groups (Groups 1 and 2 p < 0.01, and for Groups 3 and 4 p < 0.05). The authors cautioned that 'with an experimental design involving such restricted input from ourselves ... some patients from all treatment groups will also require treatment from other sources.' Such treatment for the 'no treatment' group during the first year seemingly had a negligible effect on outcome. This study, the most comprehensive of its kind and despite complex design and somewhat more inclusive treatments across the groups, virtually replicated findings of Hall and Crisp 's 1987 study. Some of the concerns we expressed about the 1987 study still applied. As family therapists we were disappointed that the investigators failed to report

Anorexia Nervosa

145

the changes in family structure and the relationship between those changes and outcome. Was their choice of family therapy based on a set of assumptions about the role of family dynamics in the etiology of AN or perhaps certain family characteristics were contributing to maintain anoretic behaviour? Theoretically, it is rather unsatisfactory not to have key family features included in the outcome measures. The true value of family therapy was impossible to assess in these two studies for the additional reason that family therapy was combined with individual psychotherapy. This problem can be resolved by clearly identifying the family dysfunction at the point of assessment and noting the changes posttreatment. Comment: The rate of recovery in this group of studies ranged from a high of 48 per cent (Jenkins 1987) after the three-year follow-up to a low of 26 per cent (Hall and Crisp 1987) at one-year follow-up. This group of studies had one particular problem in common. Collectively they provided very meagre information about family therapy. Not one employed any standardized measures of family functioning. Other than passing reference to 'family therapy' or a brief statement about the goal of family therapy, as can be found in Crisp's studies, the theoretical orientation or the model of the therapy was left to the reader's imagination. Factual information such as number of therapy sessions and who the participants were in those sessions was ignored. Only the studies by Crisp and associates and by Garfinkel were controlled and not riddled with methodological problems. As for the efficacy of family therapy, the two studies of Hall and Crisp and Crisp et al. shed, albeit, dim light. Because they combined individual and family therapies for two of the groups, one inescapable conclusion is that psychotherapy, predictably, is more effective than other treatments in dealing with sexual and social adjustment issues. Weight gain was not significantly influenced by combined psychotherapies. Both their studies demonstrated that longer term therapies were required for the outcome to be positive and lasting. This proposition challenged the belief among many family therapists that their method of treatment was essentially of a short-term nature. Garfinkel and colleagues' (1977) conclusion that AN patients tended to improve regardless of the method(s) of treatment has some validity in research and has, in fact, been substantially confirmed by a recent review of the outcome literature on AN (Steinhausen, Rauss-Mason, and Seidel 1991).

146 How Good Is Family Therapy? Family-Oriented Treatment

Studies in this section are few indeed. The problems with this group of studies are best illustrated by the following situation. Tolstrup and associates (1985), in their retrospective review of 151 patients with anorexia nervosa, made a passing reference, in describing treatment for one of the groups, to 'parental counselling,' and there was no further mention of it in the rest of their report. This is very characteristic of the way that the literature tends to handle treatment issues-a passing reference. Yet, Steinhausen and associates (1991), in their otherwise excellent review, categorized Tolstrup's treatment milieu as inclusive of family therapy. This is a rather generous, albeit erroneous, interpretation of the concept of' family therapy,' and it is misleading since in Tolstrup 's report 'parental counselling' was mentioned just once. Groen and Feldman-Tolenado (1966) described at length their attempt to work with parents of anorexic children. Parents were found to be unwilling or unable to provide the emotional support required by the patients. Parents were seen with the intention of modifying their attitudes towards their children and, in the process, to be educated about the disease. The parental treatment was a far cry from what is today considered systemic family therapy, but some of the elements of such therapy could still be detected. An in-patient treatment program and its outcome for 145 patients with a mean age of 20.5 (SD = 4.8) years and illness duration of 3.1 (SD= 3.1) years was reported by Pierloot and colleagues (1982). The treatment was offered in two stages: (1) a weight restoration program and behaviour therapy with a contingency management program, 'exploratory individual and family sessions,' and fortnightly (biweekly) group counselling of parents and spouses; (2) group psychotherapy, art and music therapy, and psychoanalytically oriented group therapy. Formal family therapy was excluded because most patients did not meet the criteria of age and duration (preadolescent and/ or adolescent with short duration of illness). Outcome data were based on seventyeight patients (an additional ten patients died) who were, on average, followed up for 4.2 years. Outcome was assessed using a Global Clinical Score. On anorectic symptoms nearly 78 per cent reported complete to moderate improvement. On psychosocial adjustment only 45 per cent were improved. Sixty-six patients had dropped out of treatment. The investigators noted that age, chronicity, and failure of previous treatment accounted for poor outcome.

Anorexia Nervosa

147

In a study of this nature, the purpose of which was to describe an inpatient program for older patients and with more chronic presentation of AN, the research component could be weak. The study was uncontrolled. It only made passing reference to family involvement. Family therapy per se was not implemented except as 'explorative.' The find ings of those explorations or the types of benefits experienced by parents and spouses from group counselling were not reported. It should be noted, however, that 'family' was not the focus, and the locus of treatment was very definitely the patient. Groen and Feldman-Tolenado provided an explanation for their decision to exclude formal family therapy on the grounds of age and chronicity factors, and thus they lent indirect support for the application of family therapy (by its exclusion to treat more chronic AN) with pre-adolescent or adolescent anorexics of recent onset. It is noteworthy that Engel and associates (1989) in a follow-up study of 208 AN patients with the mean age of twenty-three years, who also received inpatient treatment, excluded family therapy from their treatment on the ground that 'the method of treatment followed here ... follows the pathway of segregation from the pathogenic field of the family and thus stands in direct contradiction to family therapy that includes the family from the very beginning,' thus lending some support to Pierloot's proposition that older patients may be less amenable to family therapy. This observation is likely to remain somewhat contentious until controlled investigations of family therapy and other treatments are conducted with adequate sample size, proper instrumentation to assess family functioning, varied durations of illness, and different age groups of AN patients. Russel et al. (1987) have made a promising start in that direction. The main strength of Pierloot's study, which has major methodological deficits, was that it provided further evidence that poor outcome, regardless of a comprehensive approach to therapy, appeared to be associated with specific patient characteristics (older age, chronicity, and past treatment failure) . These patients, compared with Minuchin's, were at the other end of the continuum. Rose and Garfinkel (1980) described a group treatment program for parents of anorectic children. Ten couples participated. Five of these couples were dissatisfied with family therapy they had received previously. The group met biweekly for eighteen months. Outcome was based on a questionnaire completed by seven couples, of these six expressed positive feelings towards the group. They found the group

148

How Good Is Family Therapy?

support enabled them to cope with the stigma of the illness, and it provided an outlet for their bewildered feelings. Being educated about the disease was reassuring. As for the patients, the authors noted that 'the clinical course of the patients was not unlike others on our unit whose parents did not attend a group.' No follow-up data were reported. Comment: Our reason for including a sample of reports described in this segment was to emphasize the importance of the family in general, and the parents in particular, in the treatment of AN. These reports did not possess any particular research value, although the actual benefit of parental involvement either in parental group therapy or where it was designed to enhance parental sensitivity to and knowledge about AN certainly deserves the attention of researchers. Rose and Garfinkel (1980) did not find any clinical difference in the AN patients based on parental participation in group therapy. This finding was not supported by any objective measures. Thus, the true value of family- and parent-oriented programs remains unknown. Perhaps the therapeutic merit of such treatments can only be ascertained in long-term followup studies. The study by Engel and associates (1989) was the exception to the general rule for including family involvement on the ground that such therapy was contraindicated. Conclusion The place of family therapy in the treatment of AN remains to be fully ascertained. Admittedly, the frequency with which family therapy is incorporated in the overall integrated treatment program for AN is quite remarkable, especially measured over time. Prior to the 1970s family therapy was not a part of the therapeutic milieu for AN. In his review of forty-five outcome studies in AN between 1953 and 1981 Steinhausen (1983) found that only one had used family therapy. In contrast, Steinhausen and associates (1991) reported six out of twentythree studies of the past decade used family therapy as one element in an integrated treatment program for AN. The appearance of the works of Selvini-Palazzoli (1974) and a little later Minuchin and colleagues (1978) is in significant measure responsible for the increased use of family therapy in treating AN. This is despite several major methodological shortcomings in Minuchin's research such as lack of information about patient selection, socioeconomic status, absence of a face-to-face interview for most patients, the

Anorexia Nervosa

149

short duration of follow-up, the view of anorexia nervosa as a unitary concept, and the lack of a control group (Swift 1982; Maloney and Klykylo 1983) . Minuchin's contribution has to be seen in the context of his departure from the traditional biomedical approach to a radically new way of understanding and treating a complex disorder, and seemingly he has made his mark. Individuals such as Crisp in the United Kingdom and Garfinkel in Canada, who have devoted their professional life to the pursuit of research and treatment for AN, are emphatic about the key role of family therapy in its treatment and management. A literature review of outcome studies of the past two decades confirms that family therapy is frequently included in the treatment packages. The discouraging fact, however, is that the quality of research is such that Minuchin's early promise of family therapy as a breakthrough in the treatment of AN still remains just a promise. Other than Russel's study, which comes closest to a replication of the Minuchin study, with far less encouraging results, there is a marked absence of such efforts. The quality of research to date is such that any definitive claim of the efficacy of family therapy should be tempered with much caution. Family therapy is very rarely the only treatment or even the major component of integrated treatment. It is still possible from a methodological point of view to more accurately assess the impact of family therapy (for example, by measuring differences in outcome between those who accept family therapy as part of an integrated treatment program and those who refuse), but the literature reveals a clear poverty of that kind of approach. Family therapy researchers have abandoned this field of research which may also account for its currently unsatisfactory state. The evidence points in the direction that young adolescents whose onset of AN is close to the onset of menses and whose illness is of short duration (few months to a year or two) are more responsive to family therapy. It is to be noted that those two factors are predictors of positive outcome independent of the type of therapy. In this context Hsu (1980) concluded, based on his review of outcome studies of AN that 'good results have been obtained whether the principal method of treatment was nursing care, nursing care and psychotherapy, nursing care and chlorpromazine hydrochloride therapy, behavior modification or family therapy.' The point is that so far family therapy has failed to establish its superiority either by itself, despite claims to the contrary by Selvini-Palazzoli and Minuchin, or in conjunction with other therapies over alternative treatments.

150 How Good Is Family Therapy? The picture has somewhat altered since Hsu's review. Several studies discussed in this chapter have demonstrated that family therapy and psychotherapy in general is better suited for younger patients, and it is more effective in addressing psychosocial problems, whereas weight gain seems to be independent of the method of treatment. The true value of family therapy can only be borne out in carefully designed, controlled studies to affirm the efficacy of family therapy, showing that it has the therapeutic power to ameliorate suffering and grief in the anorectic patient and her family. Unfortunately, such studies are in desperately short supply.

9

Family-Based Approaches to Placement Prevention in Child Welfare

Historically, relationships between child welfare agencies and the families they serve have been ambivalent and, often, adversarial. Child welfare practice has been dominated by a child rescue philosophy along with a strong belief in the merits of personalistic psychology (Whittaker 1991). Not surprisingly, family therapy, as a discrete treatment modality, has played a very small role in formal child welfare practice (Whittington 1985; Reichertz and Frankel 1990). Because of a new drive for placement prevention, however, concepts from family therapy have increasingly influenced child welfare practice. While the principle of helping families to remain intact is evident in the rhetoric of child welfare practice, it has been less apparent at the level of direct service. In fact, child welfare agencies have consistently emphasized and utilized substitute care over efforts to improve family functioning (Frankel 1988; Giovannoni 1985; Kinney, Haapala, and Booth 1991; Laird and Hartman, 1985). Over the past several decades child welfare agencies have been criticized for expending massive resources while providing care of questionable quality (McGowan and Meezan 1983; Morton 1993). To some extent child welfare legislation has changed in response to these criticisms. Most legislation now reflects some level of commitment towards helping families to remain intact, before considering placement (Barthel 1992; Child Welfare Task Force 1991). These legislative changes have resulted in an increased interest and investment in developing child welfare services that support and strengthen families as a means of averting child placement.

152

How Good Is Family Therapy?

Family Preservation and Placement Prevention

Fuelled by the apparently spectacular results from early demonstration projects, along with the promise of cost savings, what has come to be termed 'the family preservation movement' has resulted in hundreds of diverse programs which opera 1e under both public and private auspices in several fields of practice (Biegel and Wells 1991). 'Family-based,' 'home-based,' family-centred, 'family support,' 'intensive family preservation,' and 'placemen1 prevention strategies' are all terms that have been used to describe similar efforts. Despite this confusion there is general agreement about the purpose and focus of these services. Family preservation programs serve vulnerable families in order to prevent the out-of-home placement of their children (Maluccio 1991). Family preservation services represent a broad and eclectic range of theoretical orientations including crisis theory, family-systems theory, social learning theory, ecological theory, functional theory, and social attachment theory (Barth 1988; Grigsby 1991). Attempts to differentiate among programs according to theoretical orientation have not yielded particularly meaningful results. It can be argued that all family preservation programs are ecological and systemic, given their emphasis on the interactional and contextual lives of the clients they serve. All could be described as crisis oriented, in light of their emphasis on relatively immediate, intensive, and time-limited service. Similarly, the principles of functional theory are apparent in the structure and philosophy of family preservation. Finally, social attachment theory is reflected in the very essence of family preservation, as a concept that values the permanency of parent-child relationships. As will become apparent, the studies reviewed here are far from pure applications of systemic family therapy. Rather they represent a broad range of models that assume that child placement and family functioning are inextricably tied. At the most general level any effort at supporting or strengthening families can be considered to be a family preservation program. This chapter, however, focuses on child welfare services that attempt to prevent placement by intervening with families who are considered to be at imminent risk of experiencing such an occurrence, usually because of child abuse, child neglect, child behaviour problems, or family conflict. The research literature is dominated by evaluations of single programs or statewide initiatives (in the United States) which include multiple service sites. For the most part these investigations are aimed

Family-Based Approaches to Placement Prevention

153

at assessing program performance, without regard to the experiences of a comparison or control group. While they may be of considerable value, especially at the local level, these studies do not answer the fundamental question of whether the intervention leads to more favourable outcomes than the status quo. This chapter selects from the most rigorous of the family preservation and placement prevention investigations reported in the literature. The studies discussed here all involve some level of control or comparison. Many also use multiple measures of outcome, include follow-ups, make some attempt to identify correlates of success or failure, and describe interventions. We will examine a total of nine investigations in detail. The first two fall under what the Child Welfare League of America (1989) broadly terms 'family-centred casework services.' The remaining studies examine what are termed 'intensive family-centred crisis services.' Family-Centred Casework Services

Family-centred casework services are aimed at families experiencing problems that threaten the well-being and safety of children and affect the stability of the family. They may encompass any or all of a broad range of activities including case management, counselling or therapy, skill building, advocacy, and the provision of concrete services. Ideally, caseloads are limited to fifteen families . Services are intended to be flexible, home-based, and easily accessible. Family-centred caseworkers maintain an ecological (family within its social context) view and endeavour to engage the family in evaluating strengths, problems, and service needs (Child Welfare League of America 1989). New York State Preventive Services Project

The New York State Preventive Services Project Oones, Neuman, and Shyne 1976; Jones 1985) is representative of many of the early responses to the realization that some children were entering care unnecessarily. The experimental intervention included a combination of counselling and concrete services. Caseloads were small, usually ten cases per worker. Placement prevention cases were selected on the basis of a referring worker's judgment that a child under the age of fourteen was at imminent risk of being placed within six months, in the absence of preventive services. A related criterion was the referring

154

How Good ls Family Therapy?

worker's determination that the family was likely to benefit from the experimental service. Families were randomly assigned to a prevention unit or to the referring unit for regular service. The total sample included 549 families, 373 in the experimental group and 176 in the comparison group. The sample was characterized by poor, multiproblem, female-headed, single-parent families, with an overrepresentation of minority groups. Outcomes were examined twelve months after the initiation of the project. This meant that cases had been open for variable durations with an average of nine months. A limited follow-up was conducted three months later. In general, the experimental group moderately, but consistently, outperformed the comparison group in relation to placement prevention, days in care, and worker ratings of child functioning, parental functioning, and improved environmental conditions. There were, however, several serious limitations to the study that made it impossible to treat the findings with confidence. Perhaps the most serious, was the fact that 76 per cent of the experimental cases and 82 per cent of the comparison cases were still open and receiving service at the point of evaluation. The majority of cases, therefore, were evaluated before the intervention was completed. A similar state of affairs was evident in two subsequent projects that were modelled after the New York State Preventive Services Project. The Hudson County Preventive Services Project (Magura 1981; Willems and DeRubies 1981) found that 52 per cent of experimental group cases and approximately 48 per cent of comparison group cases remained open three years after the project was initiated. Similarly, the Bronx Preventive Services Project (Halper and Jones 1981) found that 63 per cent of experimental cases and 45 per cent of comparison cases were still open at the point of evaluation. While each of these studies included some methodological improvements, the high proportion of open cases severely limited their usefulness. Nevertheless, these studies served to emphasize the importance of adopting a family orientation in child welfare practice. Fortunately, a follow-up study was conducted on the New York State Project (Jones 1985). This report was much stronger than the original evaluation because it focused exclusively on the placement prevention sample from New York City (70 per cent of the original sample) and was conducted five years later. This investigation included ninety-eight families in the experimental group and forty-four families in the comparison group. In all there were 243 children who were considered to

Family-Based Approaches to Placement Prevention

155

be at risk of placement. Comparisons of the services received by the two groups were not possible because service data on the comparison group were not collected beyond the first twelve months. Based on the existing data, Jones concluded that experimental cases received both a wider variety of services and greater intensity of service than their comparison group counterp,uts. The experimental cases remained open for an average of fourteen months, and the majority of cases (71 per cent) were closed by twenty-four months. The components of the experimental treatment were relatively well documented and generally consisted of counselling, along with the direct or indirect provision of financial assistance, medical services, and assistance with housing. The experimental services were not delivered with great intensity by today's standards, but were judged to be considerably more intensive than conventional services of the time. In-person contacts averaged 2.5 per month during the first year, 1.5 per month in the second year, and steadily decreased over time. The majority of these contacts (63 per cent) occurred in the clients' homes. By the end of the five-year follow-up period, significantly fewer (34 per cent) of the experimental group children entered foster care, as compared with the comparison group children (46 per cent). This was a relatively narrow definition of outcome because substitute care other than foster care was not examined. It should be emphasized that these results showed that the majority of children did not enter foster care, regardless of the services they received, even though they were all considered to be at imminent risk of being removed from their homes. This difficulty in accurately identifying children who are at imminent risk of being placed continues to plague family preservation researchers. There was a strong indication that the experimental treatment delayed entry into foster care. The benefits of delaying entry into care, however, are unclear. The delays may have been beneficial if they allowed children to enter care with parental support because of a perception that efforts to prevent placement had been extended. Presumably, this would result in more effective placement experiences. On the other hand, children may have had to endure problematic environments unnecessarily. The detailed data on family background for both groups and service provision for the experimental group made it possible to examine factors associated with entry to foster care. In general, the findings confirmed the existing practice wisdom. The major areas of predisposing

156 How Good Is Family Therapy? factors were the unavailability or incapacity of the mother, limited supports, problems in the child's functioning, and lateness of the intervention. When service-related variables were added to the analysis for the experimental group, five variables were predictive of a child not entering care. The important factors were related to the duration and completion of services and the use of experienced workers. Ramsey County Study Lyle and Nelson (1983) and Nelson (1985) reported on a demonstration project carried out at the Child Protection Section of the Ramsey County (Minnesota) Department of Human Services. The study used a posttest only design with seventy-four families who were systematically assigned to one of three existing child protection units (n = 40) or to an experimental home-based service unit (n = 34). Families where a child was 'extremely' likely to be placed in long-term care within three months and families with parents who had consistently failed in past therapeutic efforts were considered ineligible. A total of 161 out of 315 cases were considered eligible. Seventy-four per cent of the 154 ineligible cases were excluded because they were assessed to be likely candidates for long-term care within three months. This study reported on the seventy-four eligible cases that received at least nine months of treatment, by the point of evaluation. The sample was predominantly single-parent, Caucasian, and poor. Families were small, with young children and young primary caretakers. Consistent with the young age of targeted children, 82 per cent of the cases were opened for reasons related to parental functioning. Almost half of the cases involved primary caretakers with histories of violence, maltreatment, or substance abuse. The treatment and comparison groups were judged to be equivalent on all relevant variables. The home-based treatment was based on what was described as ecological family systems theory. The program was designed to be intensive (at least one contact per week) and open ended. Services were provided by social worker teams with small caseloads (average of eight cases per worker) and consisted of individual, family, and group counselling. The emphasis was on reducing isolation, restoring social networks, parental education, and communication training. The traditional child protection services which were delivered to the comparison group were grounded in a psychosocial medical model. These social workers carried an average of twenty-one cases. Their emphasis

Family-Based Approaches to Placement Prevention

157

was on assessing child risk, developing a case plan, and arranging for related services. Although workers in both conditions identified similar case objectives, differences in orientation were exemplified by the fact that the experimental workers described themselves as therapists, while the protection workers viewed themselves as case managers. Cases in the home-based unit remained open for an average of 313 days, as compared with '.l43 days for traditional services. This difference, however, was not statistically significant. The average number of client contacts per case was significantly higher in the home-based program (twenty-nine vs twelve). Workers from the home-based program also met with clients in their own homes much more frequently than comparison group workers did (91 per cent vs 45 per cent). According to social worker reports, individual or family counselling was the primary service in 100 per cent of the home-based cases and 43 per cent of the comparison cases. Home-based therapists never listed case management as a primary service, while 56 per cent of the comparison group workers did. At the point of evaluation, all families had received at least nine months of service. Unfortunately, there was no indication of how many cases had completed treatment at that point. Nevertheless, 76 per cent of the home-based treatment families and 55 per cent of the comparison group families remained intact during the study period. This difference approached statistical significance (p = 0.09). Interestingly, 56 per cent of the excluded cases also remained intact, even though they might be considered to be at higher risk of placement than either of the study groups. On average, children in the home-based treatment group spent significantly fewer days in care than those in the comparison group (sixteen vs fifty-four). There were no meaningful results regarding client or program characteristics related to placement. Comment: Family-centred casework placement prevention services represent an integration of family therapy with social casework. Such programs appear to be relatively long-term, open-ended interventions that are delivered with greater intensity than the conventional services. They emphasize treatment over case management and offer a broad array of clinical and concrete services to families with relatively young children. Consistent with the young age of targeted children, clinical services are directed primarily at parental functioning, with a secondary focus on social supports and the functioning of children. These programs appear to be slightly more successful than the comparison

158

How Good Is Family Therapy?

services at preventing placement. There is also an indication that these services delay entry into care, although findings related to duration of care are mixed. The New York State Project was able to identify some correlates of treatment outcome. It should come as no surprise to family therapists, however, that factors related to the primary caretaker and the family's environment played a key role in outcome. Similarly, the relationships between service intensity, service comprehensiveness, treatment completion, and outcome are all consistent with clinical wisdom. Perhaps the most obvious methodological gap apparent in these studies is the failure of the designs in not allowing sufficient time for treatment. This is extremely important when evaluating open-ended interventions. There were also some problems with client selection. The usefulness of these studies hinges, primarily, on their ability to accurately target children who were at imminent risk of being placed. The findings indicate that the inclusion criteria were not sufficient for identifying children who were at imminent risk of placement. Moreover, it appears that the exclusion criteria may have actually diverted those who were most likely to be placed. Despite these problems, the emergence of family-centred casework placement prevention programs signalled a new direction for child welfare practice and research. Current research on family-centred casework models is conspicuous by its absence. While it is likely such models have become commonplace, their long-term and relatively comprehensive approach may have rendered them less appealing for today's researchers and policy makers.

Intensive Family-Centred Crisis Services Intensive family-centred crisis services constitute the most extreme form of family preservation services. While family-centred casework services are aimed at families experiencing problems that threaten their stability, intensive family-centred crisis services are aimed at families who can no longer cope with such threats. The risk of immediate child placement is one indicator of this inability to cope. Interventions include home-based counselling, education, and supportive services delivered in an accessible, intensive, and flexible manner. Caseloads range from two to six families, and workers are available on a twentyfour-hour basis. Intensive family-centred crisis services are short term (four to twelve weeks), with families receiving an average of eight to

Family-Based Approaches to Placement Prevention

159

ten hours of direct contact per week. Ideally, workers spend at least 60 per cent of their time in face-to-face contact with family members. Intensive family-centred crisis workers maintain an ecological (family within its social context) orientation and are committed to family empowerment and the promotion of family strengths (Child Welfare League of America 1989). The major rationale for the brevity, intensity, and flexibility of intensive family-centred crisis services is derived from crisis intervention theory and the notion that people are most open to change during a time-limited period of crisis-induced disequilibrium (Nelson and Landsman 1992). Hennepin County Study

This study employed a quasi-experimental (overflow comparison group) design to evaluate a family preservation program implemented within the Child Welfare Division of the Hennepin County (Minnesota) Community Services Department (Auclaire and Schwartz 1986, 1987a, 1987b; Schwartz, Auclaire, and Harris 1991). The program was aimed exclusively at families with an adolescent who had already been approved for out-of-home placement. The strategy of targeting children whose placements had already been approved was a methodological improvement that goes a long way towards ensuring that the study population was truly at imminent risk of placement. The disadvantage, however, was that it also ensured a relatively late intervention that may not reasonably be expected to prevent at least a temporary initial placement. All families with an adolescent who was approved for placement were considered eligible for the treatment program. When there was an opening, the most recently approved placement case was selected. If there were no openings in the experimental program, the case received the usual placement services. This procedure resulted in a group of cases that was unsystematically assigned to the treatment unit (n = 58) and a group of cases that were eligible, but not assigned to the unit. Fifty-eight cases were randomly selected from the latter group in order to provide the comparison for this study. It was not clear why the researchers chose to randomly select the comparison group, rather than using the entire pool of eligible overflow cases. The sample was predominantly Caucasian and contained a similar proportion of single-parent and two-parent families. The comparison and treatment groups were generally equivalent on major sociodemo-

160

How Good Is Family Therapy?

graphic variables. The families exhibited multiple serious problems. Children were reported to have behavioural, family, school, health, and substance abuse problems. Many had a history of abuse, neglect, or involvement with the juvenile justice system. Parents were seen as lacking basic parenting skills. Many parents also reported histories of family violence and substance abuse. Not surprisingly, the majority (62 per cent) of the children in the treatment group were out of the home at the point of referral to the program. Most were in temporary shelter care. The experimental treatment used a structural family therapy approach (Minuchin 1971) . Treatment, which was provided by teams, was aimed at finding alternatives to out-of-home placement. Caseloads were limited to two families per worker, and the goal for length of services was four weeks. Workers were expected to be available to families beyond working hours, and much of the activity occurred in the family's home. Workers identified themselves, primarily, as family therapists, and their structural orientation was verified by a self-report survey of the techniques they employed. Service actually lasted an average of thirty-seven days. This included a mean of almost seventeen hours of contact, of which almost fourteen hours was direct. The details of service provision to the comparison group were not reported. Cases were followed for twelve to sixteen months, depending on when they were assigned. Given that treatment typically lasted a little over five weeks, this reflects a relatively long follow-up period. The major outcome measures included placement, days spent in care, and attainment of treatment goals. Three of the treatment group cases were excluded from parts of the analysis because they were in care throughout the entire study period. Of the fifty-five remaining treatment cases twenty-four (44 per cent) did not experience a placement episode. Forty-five per cent of those placed experienced only one episode, and 55 per cent experienced multiple placements. The findings for the comparison group were reported in two forms. The likelihood that the vast majority of comparison cases would experience a placement subsequent to their selection was inherent in the design of the study. Consequently, the researchers presented findings which included (unadjusted) and excluded (adjusted) the first placement, the latter being far more conservative. Interestingly, five of the comparison group children (9 per cent) did not enter care, even though they were approved for placement. Even approval for placement,

Family-Based Approaches to Placement Prevention

161

therefore, was not an entirely accurate indicator of imminent risk. When first placements were excluded, 41 per cent of the comparison group children (compared with 44 per cent of experimental group chil dren) remained in their own homes. Nineteen children (33 per cent) were placed once, and the remainder experienced multiple placements. Adjusting for the first placement yields similar findings in that seventeen (29 per cent) of those placed had only one placement. Based on the adjusted rates, there were only marginal differences in placement prevention for the two groups. Also of note was the fact that the majority of children entered care, regardless of the services they received. The types of placement utilized by the two groups, however, showed significant differences, with the treatment group using more shelter care, rather than treatment settings. If unadjusted data are considered, both groups used similar amounts of shelter care, while the comparison group utilized significantly more of the higher levels of care. This may be an indication that .the use of shelter care for respite may have been a treatment strategy employed by the placement prevention therapists. Placement as a treatment strategy is an important area that has not been explored by family preservation researchers. The treatment group also used significantly fewer placement days (4,777 vs 6,666) than the comparison group. Of course, the difference was even greater if first placements were not excluded for the comparison group (12,037 days). When comparisons were made between the expected number of days in care (from the original placement approval process) and the actual number of days spent in care, the treatment group significantly outperformed the comparison group (21 per cent of expected days vs 46 per cent) . The relationship held even when the first placement for the comparison group was included (32 per cent of expected days). Moreover, treatment group children (58 per cent) successfully completed placements (achieved the goals of the placement) at a higher rate than those in the comparison group (40 per cent adjusted; 49 per cent unadjusted) . It is worth noting here, that a large proportion of those placed did not successfully complete their placements. Overall, these findings suggest that the experimental treatment may have contributed to the more effective and efficient use of substitute care. A closer look at the treatment group yielded some interesting findings that emphasized the importance of engaging families in the treatment process. According to therapist ratings, only twenty-two of the

162 How Good Is Family Therapy? fifty-five families completed the treatment successfully. Twenty-three families participated, but did not achieve all of their goals, and ten families refused to participate at all. The proportion of placement days used was strongly related to the extent and quality of the family's participation in the program. Families who were sufficiently engaged in the process to set goals (n = 22) used a significantly smaller proportion of available placement days than did families who were unable to set goals (n = 33). While goal attainment achieved by children (as rated by the treatment team) was not related to time in care, goal attainment by parents was associated with using fewer placement days. Service intensity, however, was not directly related to treatment success. California Projects

Yuan et al. (1990) evaluated eight state-mandated family preservation demonstration projects in California. All of them accepted referrals exclusively from their respective county child protection services. The evaluation consisted of a before and after study with an experimental substudy in the final year. The enabling legislation specified that the target populations for the projects were families of abused or neglected children where it appeared that placement would occur if the services of the project were not provided. As with most placement prevention efforts, the operationalization of imminent risk presented some difficulty. Ultimately, imminent risk was determined to be present if the referring agency was planning to remove a child within a two-week period. The projects served a total of 709 families with 1,740 children, over a three-year period. The families were quite young. Adults averaged thirty-two years of age, and children averaged seven years. Almost half of primary caretakers and children were Caucasian, and about onequarter of primary caretakers and children were Hispanic. Almost half of the families were headed by single parents. Fifty-nine per cent of the families were receiving public assistance. One-third of adults were reported to have one or more impairments such as substance abuse or psychological dysfunction. Relatively few children (20 per cent) were reported to be impaired. Most of the children (43 per cent) were in physically abusive situations, 33 per cent were experiencing neglect, 12 per cent were at risk because of sexual abuse, and 6 per cent because of emotional maltreatment. Eighty-seven per cent of the children were at

Family-Based Approaches to Placement Prevention

163

home at the time of service. More than half (53 per cent) of the children had no prior history of placement The Child Well-Being Scales (Magura and Moses 1986) were used as an assessment device and as an outcome measure. Overall, the problems identified were consistent with what might be expected for families with young children. Therapists identified problems that related to the parental adequacy, parental relationships, support for the primary caretaker, and money management. Each therapist carried a maximum of three cases, although some projects used case aides in addition to therapists. Intervention was planned to be home-based, short-term (four to six weeks), and included twenty-four-hour availability. The actual methods of intervention depended on the specific program's philosophy and the therapist's orientation and training. The average durntion of service ranged from thirty-seven to sixtyseven days (somewhat longer than planned), with an average of three hours of service per day. Contrary to what was expected, only one-third of services occurred in the client's home. According to therapist reports, nine services were planned for over 80 per cent of the families. These included individual counselling, crisis intervention, parental skills training, family therapy, case planning with clients, case consultation, assessment, coordination with county workers, and coordination with other collaterals. On average, fourteen services were planned for each family. Surprisingly, concrete services were planned for less than 10 per cent of the cases. Families received an average of sixty hours of service, although the range across projects was from thirtythree to ninety-eight hours. A little over half of this time was spent in direct contact with clients. Three categories of service accounted for at least thirteen hours each: counselling or therapy, case planning and supervision, and other multiple services. Counselling constituted the major service in six of the eight projects. A major limitation to this study relates to the fact that at least 42 per cent of the families received some form of service after termination with their respective projects. Posttermination services are rarely tracked, so it is difficult to determine if the California experience holds for other studies. Outcome findings were based on data from 95 per cent of the 709 families served during a three-year period. Seventy-five per cent of the cases were closed because their goals were achieved, according to the therapist. Ten per cent were closed without full goal achievement, but because of the imposed time limit, and 15 per cent were closed

164

How Good Is Family Therapy?

because the family refused to continue or the child was placed. Data from the Child Well-Being Scales indicated that many of the families made important gains. The percentage of families who scored 'adequate' on all items at termination increased when compared with their own scores at intake. The greatest gains were on items related to parental recognition of problems, parental motivation, mental health care, continuity of parenting, and parental acceptance of children. For child-related items, the greatest gains were seen for abusive physical discipline, protection from abuse, threat of abuse, and children's family relations. Overall, only three items failed to show improvement: supervision of a teenager, support for the primary caretaker, and degree of child impairment. Placement outcomes were based on the child's status, according to county placement records, eight months after intake (approximately six months after service). Relative placements were included, but unofficial placements were not. Overall 138 (20 per cent) of the 675 families followed had at least one child enter care at some point during the evaluation period. Placement rates ranged from 15 per cent at one site to 36 per cent at another. Using children as the unit of analysis revealed that 247 (15 per cent) out of 1,661 children were placed. Seventy-five per cent of them were still in care eight months after initiation of service. Individual sites ranged from 9 per cent to 24 per cent. Children spent an average of ninety-nine days in care. When factors related to placement were explored, the following family-oriented variables were significant: a protection investigation subsequent to the completion of service, a lack of parental cooperation with case planning, a negative assessment of parental teaching and stimulation of a child, the presence of children ages six to ten years, poor money management, poor mental health care, poor continuity of parenting, service requiring less travel time, prior placements, and receiving public assistance. For child-related variables, the following factors were significant: prior placements, risk because of neglect, the presence of a threat of abuse, abusive discipline, child misconduct, and the child being out of the home at the point of referral. The findings of the overall evaluation were supported by an experimental substudy conducted at five sites, during the last year of the project. The substudy involved the random assignment of 304 cases to either an experimental project therapist or to traditional services; these 304 families, including 713 children, were followed for a period of eight months after referral. The random assignment was not effective at pro-

Family-Based Approaches to Placement Prevention

165

ducing equivalent groups. The experimental and comparison groups differed significantly on ethnicity, education of the primary caretaker, reasons for risk, legal status of children, and some Child Well-Being Scale scores. The extent to which the experimental and comparison groups received different services was not clear. All families in both groups received some services from the local child welfare agency. Analysis of variance on the percentage of families that received twenty services revealed significant differences in only three items. More experimental families received parental skills training, assistance with food, and assistance with rent than the comparison families did. Experimental families did receive slightly, but significantly more services than comparison families. The average number of contacts from county workers, however, did not differ between the two groups. Twenty-five per cent of the 143 project families and 20 per cent of the 150 comparison families experienced a placement during the eight months after referral, according to official records. The fact that these rates approximated those found in the non-experimental portion of the study adds credence to the findings. The difference between the groups was not statistically significant. Twenty-three per cent of the experimental families and 25 per cent of the comparison families had additional maltreatment investigations during the evaluation period. This difference was not statistically significant either. There were no significant differences in the average time spent in care (203 days for experimental group families vs 293 days for comparison families). Viewing the child as the unit of analysis yielded mixed results. The per centage of children placed in the experimental (18 per cent) and comparison (17 per cent) groups was not significantly different. There was some evidence that the experimental treatment delayed placement. Significantly more comparison children (13 per cent) were placed within the first sixty days of referral than children in the experimental group (7 per cent). Comparison group children also spent significantly more time in care (144 days vs 118 days).

New Jersey Study The Homebuilders model has been an extremely influential part of the family preservation movement. This long-standing and well-elaborated model includes a Rogerian approach to therapeutic relationships and strives to establish a cooperative, supportive, and non-judgmental context for treatment. Behavioural interventions and rational emotive

166 How Good Is Family Therapy? therapy form the core of the Homebuilders approach to change. Interventions related to self-management, parent training, child management, communication training, and general skills building are delivered both interactively and didactically. Finally, a broad range of concrete services are provided to alleviate crisis conditions, improve the family's environment, re-enforce therapeutic relationships, and connect clients to other needed resources (Kinney, Haapala, and Booth 1991; Lewis 1991) . One of the most rigorous examinations of the Homebuilders model was reported by Feldman (1991a, 1991b). In 1987 New Jersey's public child welfare agency contracted with five private agencies to provide family preservation services based on the Homebuilders model. Referrals of children at risk of placement were screened for the study after the referring agency determined that the child was safe at home and that at least one caretaker was willing to be involved. A 'risk of placement' protocol was used to determine eligibility. According to the protocol, referents had to document that at least one of the following conditions was present and serious enough to warrant imminent placement: abuse or neglect, risk of abuse or neglect, injury with unknown etiology, chronic runaway behaviour, out-of-control behaviour, poor parent-child relationships, delinquency, domestic violence, child's suicidal behaviour, parental incapacity, and parent or child requests for placement. Eligible families had to be former recipients of less intensive services, or there had to be a determination that such services would not be successful. Families were excluded if the primary problem was homelessness, or if the child at risk had prior placements or had been in their first placement for less than thirty days, at the point of referral. The latter exclusion was gradually relaxed and ultimately abandoned. In all 247 families were referred and screened. Thirty-three families were rejected because they did not meet the criteria, 117 families were randomly assigned to one of five family preservation sites, sixty-nine families were randomly assigned to a known comparison group, and twenty-eight families were randomly assigned to a blind comparison group. The comparison group families were returned to the referring agencies for usual services. The purpose of utilizing two comparison groups is unclear since differences in outcome between the two groups were not discussed. The random assignment appeared to be successful in that there were no significant differences in the characteristics of the study groups. In addition, the group of clients who were not accepted into the project

Family-Based Approaches to Placement Prevention

167

did not appear to be different from the families who were served. Data from both comparison groups were combined in the analysis, as there appeared to be no important demographic differences between them. Overall, the families lived in high-risk, impoverished neighbourhoods. Forty-six per cent of the families were headed by single parents, and 23 per cent contained both parents. They had an average of three children, with an average of one child targeted as being at risk of placement. Children at risk averaged thirteen years of age. Forty-four per cent of the families were Caucasian, 36 per cent were AfricanAmerican, and 19 per cent were Hispanic. Almost 87 per cent of the families were already known to the public child welfare agency, and almost 21 per cent had prior placements. According to referring workers the five most frequently cited primary secondary problems were a child's behaviour (64 per cent of cases) , a child's emotional problem (16 per cent), abuse or neglect (14 per cent), and a parental problem (5 per cent) . A retrospective survey of primary caretakers indicated an average of fourteen problems per family, prior to treatment. The study was driven by four major questions. The first question sought to verify that the Homebuilders model was actually delivered to the experimental group. This was the only study to formally pose this very important question. Consistent with the model, family preservation workers carried a maximum of two cases. While there was some variation across sites, cases remained open for an average of four weeks, with a range from less than one week to nine weeks. Therapists expended an average of fifty-eight hours per case, of which thirty-six hours involved face-to-face contact with clients. On average, families in the experimental group experienced nineteen contacts, during their time in the project. The relatively limited number of goals (an average of three) and their nature (for example, improve communication skills or anger management) also indicated adherence to the model. Finally, the fact that interventions related to child management, advocacy, emotional management, general clinical services, and interpersonal skills training were all provided to over 90 per cent of the experimental families can be understood as verification of the model. The only indication of some divergence lies in the observation that only 68 per cent of the families received concrete services. The services experienced by families in the comparison group were not documented. The second question asked if the family preservation program was more successful at placement prevention compared with usual services. A subquestion was concerned with whether gains, if any, were

168 How Good ls Family Therapy? maintained over time. Unlike other studies, Feldman examined placement statistics at termination and at six other intervals, over a one-year period. In addition, the definition of a placement included any type of placement of any length. By termination, 6 per cent (seven) of the families in the experimental group had a child in placement compared with 17 per cent (sixteen) of the comparison group families. This difference was statistically significant. Placement status was tracked at one, two, three, six, nine, and twelve months after termination. In each case the experimental group significantly outperformed the comparison group. By the end of a year 43 per cent (fifty) of the experimental group families and 57 per cent (fifty-five) of the comparison group families had experienced a placement. It was also clear, however, that the differences in the proportion of families experiencing placement between the experimental and comparison groups decreased considerably over time. This was demonstrated through the use of event-history analysis. The likelihood of placement (hazard rate) was higher for comparison group families at termination through to three months following termination. The situation reversed at the fourth month, while the risk of placement spiked for comparison group families at months five and six. After month six, however, families in the experimental group remained at higher risk of placement. Feldman observed that while the experimental treatment was 75 per cent more effective than comparison group interventions at termination, it was only 25 per cent more effective one year following termination. A related finding, was the fact that children from the experimental group entered placement at a significantly slower rate than those in the comparison group. On average families in the experimental group experienced placement after four months, while comparison group families experienced placement after only two months. There were no significant differences in the type, quantity, or duration of placements experienced by the control and experimental groups. A somewhat crude estimate of overall program impact was provided by examining county placement rates before and after the initiation of the family preservation services. The results showed little or no impact but were suggestive of declining future placement rates. The third question examined changes in family functioning between baseline assessment and follow-up. Overall, experimental group families made many significant gains over their own previous levels of functioning. Analyses of differences between the experimental and

Family-Based Approaches to Placement Prevention

169

comparison groups, however, yielded less dramatic results. Perceived social support was measured with the Interpersonal Support Evaluation List (Cohen and Hoberman 1983). This self-report instrument involves four subscales relating to appraisal support, self-esteem support, belonging support, and tangible support. The analysis revealed that families in the experimental group made significant gains on two of the four subscales (appraisal support and belonging support) between baseline assessment and three months following termination. There were no significant gains for families in the comparison group. When the data were converted to change scores, the experimental group made significantly more gains than the comparison group on only the belonging support subscale. The Child Well-Being Scales (Magura and Moses 1986) provided worker ratings on up to forty-three anchored child welfare-related scales. Change scores were calculated for a composite index, and three empirically derived factors (household adequacy, parental disposition, and child performance). Families in the experimental group made significant gains on the composite index, parental disposition, and child performance, but not on household adequacy, between baseline and termination. On average families in the comparison group made no significant gains. When the two groups were compared directly, however, the experimental group significantly outperformed the comparison group on only the parental disposition factor. The Family Environment Scale (Moos and Moos 1981) was used to capture family members' perceptions of their own families. This ninety-item self-report instrument is divided into ten subscales that assess three dimensions (relationships, personal growth, and system maintenance). The experimental group made significant gains on six subscales between baseline and three months following termination. Significant gains were made in all three of the relationship dimension subscales, two of the five personal growth subscales, and one of the two system maintenance subscales. The comparison group showed no significant changes over time on any of the subscales. When the change scores of the two groups were directly compared, the changes in the experimental group were significantly greater on only the cohesion and intellectual-cultural orientation subscales. The final study question sought to examine the relationships between differential placement outcomes and a variety of clientrelated and ecological variables. Bi-variate analyses indicated a significant relationship between referral reasons and placement. Specifically,

170 How Good Is Family Therapy? 53 per cent of those referred because of child behaviour, 37 per cent of abuse and neglect referrals, and 100 per cent of parental problem referrals experienced a placement. Family structure also appeared to be relevant with 58 per cent of single-parent families and only 37 per cent of two-parent families being involved in a placement. Socioeconomic status, prior placements, the child's age, family size, and race were not significantly related to outcome. Data from the Child Well-Being Scales indicated that placement was significantly related to judgments of lower composite and parental functioning at intake; and lower composite, parental, and child functioning at termination. Perceived social support was not related to outcome and data from the Family Environment Scale did not yield any meaningful results. A series of analyses were used to develop a statistical model of variables influencing placement. The final model with twenty-eight estimators and a sample of 170 families was analysed using logistic regression. The results were somewhat ambiguous. While the model correctly predicted placement 72.5 per cent of the time and no placement 77.8 per cent of the time, only five variables proved to be significant. Two of these variables were interactions with intervention (family structure and type of referral) , and the others included the worker's judgment of parental functioning at termination and two perceived social support subscales. Overall, the model demonstrated that a complex combination of ecological, behavioural, demographic, and service-related factors are related to placement. Comment: Based on the experiments and quasi-experiments reviewed here, intensive family-centred crisis services generally conform to the description provided by the Child Welfare League of America. It should be noted that most of the studies involved a preponderance of young families whose problems were typically associated with some aspect of parental functioning. The relative effectiveness of intensive familycentred crisis services remains an open question. While some studies reported that these services significantly outperform traditional services in terms of placement prevention, the most rigorously controlled studies tend to show smaller and sometimes negligible differences. The accumulated findings also suggest that intensive family-centred crisis services may be most effective over the short term, and that this effect deteriorates over time. Although it may not be reasonable to expect enduring changes from programs that are largely aimed at helping families regain their equilibrium to the extent that placement is no

Family-Based Approaches to Placement Prevention

171

longer imminent, maintenance and improvement would seem to depend on the availability and quality of less intensive services subsequent to the crisis programs. There is some evidence that intensive family-centred crisis services delay entry into care, but this is not a surprising result, especially over the short term, for programs that are vigorously committed to placement prevention. Findings related to effects on the duration and restrictiveness of care are mixed, although there is the possibility that such services contribute to a more effective and efficient use of substitute care. According to service providers the interventions associated with these programs created changes in many aspects of family functioning (especially parental functioning), even though there is little in the way of comparative findings or corroboration from other sources. Overall, the findings related to correlates of success and failure continue to point to the importance of parental functioning, prior placements, environmental conditions, and the ability of providers to engage families in the treatment process. Finally, there appears to be some basis for questioning the centrality of concrete service, other than the provision of transportation. Conclusion

Rigorous research regarding family-centred placement prevention is difficult to conduct. Researchers are faced with balancing the demands of funders, legislators, and administrators with the need to implement designs that have the potential for making meaningful contributions. Such investigations often involve the difficult task of gaining the cooperation of referring workers, client families, and those who provide services to comparison groups. In addition, researchers often have to rely on case records and information systems that were created for reasons other than research. All of this occurs within the context of the daily demands of child welfare practice and increasing pressure for immediate improvement in the system. Despite these difficulties, the studies reviewed in this chapter must be considered in light of the limitations that accompany them. Many of the studies involved newly formed demonstration projects. Such projects typically operate on a relatively small scale and are staffed by highly committed workers who contribute to an atmosphere of enthusiasm that might not endure in an ongoing, large-scale service.

172

How Good Is Family Therapy?

The fact that these projects were evaluated from their earliest days makes it likely that the interventions and other procedures were not fully developed, and that they may have changed over time. The evaluations of the family-centred casework services were carried out prematurely. Often a large proportion of the sample continued to be involved in treatment at the point of evaluation. The failure to analyse open and closed cases separately makes it very difficult to determine the effects of the intervention. Even closed cases were examined almost immediately after termination, which makes it difficult to determine what effects, if any, endure after the clients stop receiving the service. The very brief and circumscribed nature of intensive family -centred crisis services makes this less of an issue. A related problem, however, emerges because many families continue to receive services (usually from other sources) after their time in the program elapses. Some families also receive additional, external services (sometimes facilitated by intensive services workers), while they are actively intensive services clients. Most research to date has not dealt with this difficult issue. The selection criteria employed by many of the studies were quite ambiguous. In general, referring social workers were relied upon to identify children at risk of placement and, in some cases, families who were likely to make good use of the experimental services. It appears that many of the highest risk families were excluded from the familycentred casework studies because placement was too imminent. On the other hand, some intensive crisis services had trouble identifying clients at sufficiently imminent risk. Findings related to the placement experiences of the comparison groups are quite instructive. They clearly indicated that social workers were not adequate judges of which children would enter care in the absence of special services. Moreover, they indicated that many children would not enter care, regardless of the services they receive. We should not, however, assume that this is reflective of improvements in their situations. Some studies employed excessively narrow definitions of placement and relied on worker reports or agency information systems to determine the whereabouts of the children. It is likely that some informal placements, some formal placements outside the child welfare system, and some very brief placements were missed. Although this may not have affected comparative placement rates within a particular study, it might well contribute to conservative overall placement rates. While the inclusion of comparison groups has certainly strengthened these studies, the services provided to members of the compari-

Family-Based Approaches to Placement Prevention

173

son groups were usually not very well documented. In addition, only one study sought to verify that the intended experimental treatment model was actually implemented. This makes it impossible to determine what the salient differences between the two conditions might have been. At this point documentation of comparison and experimental treatments has been restricted to measures of intensity, duration, and the frequency of particular a priori interventions. This is not an adequate substitute for describing treatment processes. This may be due, in part, to the fact that family preservation services are fundamentally atheoretical. The studies with larger samples employed multiple service sites. Few studies have paid serious attention to the effect of variations between sites. Aggregating data in this manner may obscure some important findings. For example, the first published report of the New Jersey Homebuilders study was based on data from all but one site. The findings indicated that there were no significant differences in placement rates between the experimental and comparison groups at the final follow-up period. When the data from the additional site were added, however, the analysis revealed significant differences in aggregate placement rates. There is a difficult balance here between gaining sample sizes large enough for powerful analyses and ignoring important local differences in treatment programs. A related limitation has to do with the current preoccupation with placement as an outcome, almost to the exclusion of other indicators. This de-emphasizes the real possibility that family preservation programs are of considerable benefit, even if they do not prevent placements more effectively than the status quo. While preventing unnecessary placements has been viewed as the main purpose of these programs, it should not be presented as the only purpose. Generally, program effects on family, parental, and child functioning, as well as on qualitative elements of placement have been subjected to only superficial examination. This may become especially important if differences in placement rates between treatment and comparison groups are negligible over the long term. An overemphasis on comparative placement rates can be quite problematic because the number of children placed in a given locale is subject to a variety of factors. Family preservation programs implicity assume that some children who receive conventional services are placed unnecessarily. There is, however, the real possibility that most children, including those who should be placed, are not placed

174 How Good ls Family Therapy? because of unavailable or inaccessible services. Variations in placement also occur according to the quality of local conventional child welfare services. Presumably, jurisdictions with more effective services would place fewer children unnecessarily. Consequently, the differences between conventional services and placement prevention services would be quite small. On the other hand, poor quality conventional services would lead to large differences. There is no doubt that research on family preservation and placement prevention would benefit from increased standardization and systematization. There also appears to be considerable agreement that the field has to consider developing different practice models for families with different characteristics. Beyond that the literature abounds with conflicting recommendations for future family preservation research. For example, there are calls to both refine and abandon the use of imminent risk of placement as a primary selection criterion. Some argue that there is a need for large-scale, multivariate experiments, while others support the aggregation of small studies using less complex designs. Long-term, longitudinal designs are described as an essential means of determining the stability of program effects, but are also portrayed as somewhat irrelevant because programs with such limited objectives should not be expected to have direct longitudinal effects. The virtues of theory-based research are both extolled and discounted, while qualitative research is encouraged and discouraged (Barth 1992; Jones 1991; Nelson and Landsman 1992; Pecora, Fraser, and Haapala 1991; Rossi 1992; Schuerman et al. 1992; Thyer 1993; Wells and Biegel 1991; Yuan and Rivest 1990). One area of convergence, however, is the fact that research on family-centred placement prevention has failed to definitively establish that programs are effective at reducing placement or improving family functioning. Many programs, however, have demonstrated sufficiently promising results to have influenced child welfare practice and policy. The concept of family preservation is firmly entrenched in contemporary child welfare practice. Despite the somewhat disappointing findings, many people are committed to the general concept and to particular practice models. Placement prevention programs of several varieties have been widely implemented and continue to grow. The practice technology is far from perfect, but continues to develop along with the recognition that specialized placement prevention programs are but a small element in what should a broad continuum of services aimed at improving the lives of children and families.

10

Future Directions

The time is upon us to think out loud. We are certainly not discouraged by the current state of family therapy outcome research. Yet we feel strongly that such research is only just beginning. The tone of our deliberation may seem a bit sombre. That would be a correct assessment. Family therapy has yet to make a strong and persuasive case for its effectiveness based on hard data rather than only on claims of clinical success. The views we express in the remainder of this chapter are strong, but we do so in the knowledge that high quality outcome research can be done; we know that, for we have reviewed a number of such studies. Are we in a position to answer the key question we set out with: How effective is family therapy? Unhappily, perhaps the answer to this rather straightforward question is far from simple. The only certainty is the uncertainty that surrounds the question itself. First, we are somewhat surprised and disappointed by the actual number of welldesigned studies in this field, and even then the findings are often inconsistent or the methodological flaws so flagrant as to make the overall results very weak. The small number of controlled studies represent different types of investigations: controlled studies that compare a treatment group with no treatment or with placebo; comparisons (very few in number) between one type or model of family therapy with another; and comparisons of family therapy with other modes of treatment, such as drug or behavioural interventions. It is imperative to have all these kinds of studies, for obvious reasons. However, the results are not consistent - either across the studies or within the same type of studies - and outcome measures are far from standardized. This last point makes any across-the-board comparisons

176 How Good Is Family Therapy? almost meaningless. Despite this lack of consistency, or perhaps because of it, it is virtually impossible to arrive at a global conclusion about the efficacy of family therapy, certainly the studies do not provide confirmation that fomily therapy is the panacea for treating a wide variety of emotional and r('lationship disorders that it was once thought to be. Overall, the efficacy of family therapy is far from selfevident. More and better designed studies would correct these deficits; we found very few studies thnt met all the criteria of a well-designed study as propounded by f\1cCr •dy (see Chapter 6). A rather salient point that has been driven home for us is the absence of what might be termed 'fundamental research' to test some of the basic assumptions of family th~rapy predicated on systemic principles. An exception may be the double -bind proposition which was proffered as a cause for schizophrenia, and that notion has been rightly consigned to the pages of history. Minuchin's proposition of the etiological importance of the family structure in the etiology of a whole range of emotional and psychophysiological conditions has received, at best, very limited and cautious validation (Chapter 5). Minuchin, based on his observations, proposed that the root cause of a variety of family dysfunctions and pediatric psychosomatic disorders could be found in the faulty structure of the family. Although his own research in this area is riddled with methodological shortcomings, his clinical observations offer a set of hypotheses in need of testing. Empirical validations of Minuchin's observations are tenuous. However, a number of outcome studies do test the efficacy of Minuchin's model, and Minuchin's psychosomatogenic family may yet emerge as a flawed concept without any validity. Only good science can provide the proof one way or the other. Thus, the notion of symptom as a metaphor or the proposition that the individual's symptom is inevitably the product of faulty intrafamilial communication, to the exclusion of other factors, remains in the realm of conjecture. Research to prove this claim is virtually nonexistent. Nevertheless, major figures in family therapy have made extraordinary claims of success purely on the basis of their personal experience. Guruism continues to plague this field. A persistent problem in family therapy research, as well as practice, is that despite claims to the contrary, family therapy, in the main, remains an intrafamilial affair. We did find a couple of studies with adolescents (Chapter 3) that included systems other than the family. The concept of systems is generally not extended beyond the family boundary, and this tendency confirms recent concerns expressed in

· Future Directions

177

the family therapy literature of viewing the family as the repository of all ills. In a real sense, the practice of family therapy in significant measure remains theory (which is somewhat conjectural) driven because the empirical evidence in support of its efficacy continues to be uncertain. Assumptions upon which systems family therapy is predicated are generally taken for grantl'd by practitioners and researchers alike, and some key assumptions such as the influence of external systems on family functioning or the role of family dynamics in the formation of symptoms are doggedly igno, "d. Is it possible that in family therapy research the cart has bee,, put he fore the horse? A few investigators, such as Leff with schizophrenia (Chapter 4) and Alexander with delinquC'ncy (Chapter 3). revisit those assumptions. Before embarking on the outcome study, Leff and his associates empirically demonstrated the existence of undesirable relationship patterns in schizophrenic families (they firmly rejected the double-bind theory) that caused patients to be rehospitalized, and then they designed interventions to interdict those patterns. The psychopathology of those families based on prevailing assumptions was not accepted, and specificity was sought before a clearly defined intervention with stated objectives was implemented. Minuchin's early efforts to link psychosomatic disorders to family dysfunction held much promise, but the necessary research is yet to be carried out. Alexander studied interactional characteristics of status offenders and non-distressed families and proposed functional family therapy as a means of intervention with delinquency; as we noted in Chapter 3, his research stands out for its methodological superiority. Has family therapy come of age, as we proclaimed in the Preface? Our answer has to be equivocal. Good outcome literature exists, for instance, with alcoholism, some adult psychiatric disorders, and some problems of adolescence, and this gives us cause for optimism. Curiously, there was some departure from systems orientation with these problems, and systems-based interventions were not easy to find. On the down side, family therapy has claimed a very large territory as its own. As this review has shown such research, at best, is spotty, and there are major areas of omission such as outcome studies of family therapy with adult medical problems or with sexually abused children. It was our intention to review the literature on the latter, but that had to be abandoned because of the almost complete absence of outcome literature. On the other hand, family therapy with children's problems remains an open question. Since the genesis of family therapy can be

178 How Good Is Family Therapy? traced back to the child guidance movement, the fact that family therapy failed to show clear superiority over other treatments was a source of some surprise for us. Yet two controlled studies that do exist, with childhood asthma, are encouraging. There is a large body of research with family preservation programs, but the general state of outcome research with them is methodologically wanting. With anorexia nervosa the actual number of well-designed studies is very small, and while the results, on the whole, are discouraging, despite claims to the contrary by Minuchin and Salvini-Palazzoli, family therapy may indeed have a role in the treatment of this complex condition. However, the claims of some family therapy scholars that family therapy is the treatment of choke should be revisited. Conceptualization of anorexia nervosa as a product of family dysfunction awaits empirical validation. The trend in the general literature on anorexia nervosa is in the opposite direction, which is that this disorder has a complex multifactorial etiology, and the effectiveness of family therapy in treating it remains grossly undertested. The outcome of family therapy with adolescent problems is probably one of the most researched areas. Yet, even here we could only express guarded optimism about its efficacy. Very few studies decry the use of family therapy or find its application significantly less effective than ocher interventions, which is significant in itself. The simple truth is that the efficacy of family therapy remains untested in many areas or, as we have stated several times, the actual number of good outcome studies is so small that without the benefit of replication studies any claim of success should be tempered with caution. We think, unlike some other reviewers, that the results of good and poor studies combined do not serve as the true indicator of outcome. That kind of justification was perhaps acceptable when family therapy was a relatively novel therapy. Almost all early outcome studies were uncontrolled and methodologically unsound. We chose to report some of them mainly for historical reasons, as those studies helped lay the foundation for more sophisticated ones later on. The early studies, judged against any measure of acceptable methodology for outcome research, yielded very flawed results in the sense that almost without exception family therapy was shown to be a highly effective method of intervention. It would be folly to regard those findings as valid and only cause confusion if their results are artificially incorporated with the results of methodologically valid studies. Another related issue is that since almost all the early studies with dubious methodology yielded

Future Directions

179

positive results, which is certainly not the case with more recent studies (which yield varying outcomes), adding the results of successful outcome of those early studies to later well-designed studies with favourable outcome would only artificially inflate the success rate for family therapy, and thus distort the true picture. The prudent approach, albeit a conservative one, is to judge the efficacy of family therapy only on the basis of well-designed, controlled studies. We subscribe to the conventional view of what constitutes 'welldesigned' studies. In the course of our research for this book, we came to the conclusion that the doubts expressed by many concerning the merit of 'conventional' designs for family therapy outcome research were less than convincing. If anything, such doubts and even disdain for outcome research may have contributed to the evident disregard for research-driven practice among family therapists. Within the confines of current technology for controlled studies, we were impressed by the ingenuity used by many investigators to explore the relative merit of family therapy with this or that problem. We found McCrady's ten-point guide of good methodology (Chapter 6) very sound, and indeed necessary for credible outcome research. She herself successfully incorporated the ten points in designing her invaluable studies with alcoholism. We cannot accept the notion that systems-based family therapy does not lend itself to conventional design. That issue was briefly addressed in Chapter 1. Certainly many early proponents of family therapy did little to conceal their contempt for research. Book after book written by practitioners of family therapy reports the remarkable potency of the treatment, and anecdotal evidence is offered as sufficient justification for using family therapy. The reasons for our belief that comparative controlled studies are relevant to family therapy outcome research are the following: (1) such studies, albeit small in number, do exist, and (2) the notion that dependent and independent variables are inextricably confounded in systems family therapy requires careful re-examination. Problems to be ameliorated can be clearly stated as can be the means for doing so. The debate on 'circular' causality and a total denial of any cause-and-effect relationship, even in its most complex form, in family therapy circles has unduly hindered application of well-tried methodology for outcome research. Investigators in most studies tend to be more concerned with the amelioration of presenting symptoms than with effective family functioning as a result of therapy. The outcome literature can be strength-

180

How Good Is Family Therapy?

ened by reporting both results. If the proposition is that family dysfunction is the root cause for a certain kind of problem, it is instructive to show, in the first place, that such dysfunctions exist and that, indeed, those family dysfunctions when removed result in the resolution of the problem. Family therapists and family researchers need to be less defensive about the theory of etiology upon which they base their entire intervention. Etiologic issues are usually taken for granted, and rarely do they find explicit expression in the outcome literature. To reiterate, in this respect l'vlinuchin's model, which to our mind remains the most articulate in family therapy, of the psychosomatogenic family holds out much promise. We have noted throughgut this book the paucity of controlled studies. They are few and often far between. Even in areas where controlled studies do exist, they tend to be small in number, and replication of studies is a rarity in family therapy outcome research. But other gaps remain in this body of research. There is an almost complete absence of long-term follow-up to ascertain the effectiveness of family therapy. problems associated with implementing studies with a long-term follow-up component are many; however, without an accurate assessment of the lasting effects of treatment, claims about its success have to be tempered . Not many reports include data on family functioning. Since the rationale for family therapy is derived from the central notion that dysfunction in the family is at the heart of emotional, relationship, and even physical problems, this particular omission is difficult to comprehend. There appears to be considerable reluctance to address the question of the underlying family pathology or, to put it another way, the mechanisms that explain a state of disequilibrium. We urge that family issues that theoretically explain family pathology must be explicitly dealt with in the outcome literature. The relationship between family dysfunction and family pathology and the treatment of those dysfunctions and eradication of presenting symptoms must be demonstrated. It is not enough to show that family intervention ameliorated this or that presenting problem without showing improvement in family functioning. In the present state of knowledge, it would appear futile to claim superiority of one model of family therapy over another. Outside of behavioural family or marital therapy, the so-called models of family therapy have a great many elements in common, such as their systemic orientation, assumption of an intrafamilial genesis of the problem or

Future Directions

181

the symptom, centrality of communication, and family organization. To clearly delineate one model from another may present a considerable methodological challenge. Perhaps, for that reason alone, reliable and valid instruments for family functioning related to individual models are not easy to find. Hence, it is not surprising that, in the first place, there is not a sufficient body of research literature to confirm or deny the superiority of any particular model, and, second, and perhaps a more important point is that more often than not the theoretical orientation of the family therapy approach is not made explicit. Sometimes a passing reference is made to the orientation. Very rarely is any explicit relationship established between the modality, the intervention itself, and the instruments used to measure the efficacy of the model. Thus, problems may be explained invoking the 'structural' perspective, but how that was implemented in treatment, what the treatment protocol was, and how the structural organization of the family was measured preand posttreatment is not always self-evident. Standardizing outcomemeasures still remains a problem. The outcome literature is riddled with problems of, what amounts to, conceptual gaps and leaps. Several other problems have come to light about family therapy itself. A curious fact of many studies is the sheer parsimony with which the therapy itself is reported. There appears to be a naive notion that family therapy is a uniform intervention, and all families receive the same treatment. It is perhaps not enough to state that families on the average had so many sessions of family therapy which was systemic in orientation. Considering that the efficacy of family therapy was under investigation, this gap in the information provided was not easy to grasp. In some instances the modality of family therapy was mentioned in passing, in others reference was made to the systemic orientation, and in yet others nothing was said about it. Nevertheless, in a handful of studies not only was the therapy described, but so was the treatment protocol. The frequent absence of a treatment protocol made it impossible to assess if indeed the same treatment was received by all. This question concerning the uniformity of treatment-beyond the number of therapy sessions - is a serious one, but unfortunately it has not to any extent caught the attention of family therapy researchers. Related to the issue of treatment protocol and the larger one of uniformity of treatment, a most glaring omission in the literature is the lack of attention to the thorny question of therapist variables. This is a very complex undertaking and researchers' unwillingness to tackle it is easily understood. Yet, the simplest of factors such as gender and age

182

How Good Is Family Therapy?

and, especially in the United States, the race of the therapists, their length and modality of training, the provision for peer review or supervision of treatment, and the implications of the therapist-researcher model as opposed to therapists who had no involvement in the study were, by and large, ignored. Several studies even failed to report how many therapists were engaged in the project. Without some control of the therapist variables, there is considerable danger of sufficiently large variability in treatment that no generalized conclusions may be drawn. A great many studies failed to describe the actual participants in the treatment program. Demographic information on this group was rarely provided. Who were the participants? On what basis were the family members included or excluded? Did the original participants stay until termination of the treatment? Was the treatment outcome affected by members dropping out or attending sporadically? These questions may be critical in deciding outcome, and yet they remain unanswered. Technical problems were addressed in each chapter as we found them. Several studies were conducted with rather small sample sizes. When the number of subjects in a given study falls below a certain level, statistical analyses may become problematic. We also noted that often statistical analyses were either kept to a minimum, or only very basic tests were conducted. With a few exceptions issues related to clinical significance were ignored. To demonstrate the effects of treatment, the power of sophisticated statistical analyses must be brought to bear on outcome research. Finally, in many studies control and comparison groups were not carefully established. The necessary statistical tests were not always performed to establish equivalency between groups. All these problems are solvable, and it is hoped that future researchers will pay due attention to methodological issues. Despite serious methodological problems in the outcome literature we continue to believe that family therapy is a powerful tool to deal with many interpersonal, psychological, and emotional problems. What problems lend themselves best to family therapy remain open to debate. Family therapy may not be the panacea that it was once thought to be. Nevertheless, family therapy is unquestionably a worthwhile pursuit. At the clinical level its claims are large. However, given the current state of knowledge, such claims ought to be modified because in many areas there is such a clear paucity of well-designed outcome studies. Faced with that reality it is virtually impossible to make any acceptable claims one way or the other. Future researchers must carefully consider the methodological pros and cons and ensure

Future Directions

183

that their findings are less subject to the common criticisms that inevitably arise from poor methodology. We reiterate our plea for more research to test the fundamental assumptions of family therapy, such as symptom as a metaphor or symptom as the inevitable product of family dysfunction or the role of faulty hierarchy within the family system as the cause for exacerbation of several pediatric psychosomatic disorders. Freud's claim of the significance of childhood events in shaping adult personality was confirmed many, many years after his death by the work of Bowlby, Rutter, and others, and in recent years the vast amount of research related to the long-term effects of childhood sexual and physical abuse has added further credence to Freud's proposition. Similar research is called for in family therapy to begin to delineate the links between family dynamics and symptom formation. In this regard, family therapy still operates, to a very large extent, on belief and indirect evidence. The relationship between etiology and treatment must be demonstrated. That task has barely begun. Time is also upon us to shake loose from any antipathy to research. We must move family therapy from an act of faith to a valid and scientifically proven therapeutic intervention.

Recommendations To summarize the above discussion and address some of the major shortcomings of family therapy outcome research, we offer the following study design criteria. We believe that incorporation of these basic elements will substantially improve the quality of the outcome research. First and foremost, some of key assumptions upon which systemic family therapy is predicated need to be tested. The study should include: 1. A clear statement of the assumption to be tested; 2. A clearly defined subject group; 3. An adequate sample size with a statistically valid control group(s); 4. A clearly stated treatment protocol that must include the model of intervention, composition of the family, number of sessions, and number and qualifications of the therapists; 5. Use of approprite statistical techniques; 6. Multiple outcome measures that are to be assessed using valid and reliable instruments; 7. A minimum of two years' posttreatment follow-up; and 8. Attention to therapist variables.

References

Achenbach, T.M., and Edelbrock, C. (1983) . Manualfor the Child Behavior Checklist and Revised Child Behavior Profile. Burlington, VT: Thomas Achenbach. Achenbach, T.M., and Edelbrock, C. (1987). Manual for the Youth Self-Report Form and Youth Version of the Child Behavior Profile. Burlington, Vf: Thomas Achenbach. Ackerman, N.J. (1980). The family with adolescents.' In E.A. Carter and M. McGoldrick (eds.), The Family Life Cycle: A Framework for Family Therapy (pp. 147-69). NewYorlc Gardner. Alexander, J.F. (1973). 'Defensive and supportive communications in normal and deviant families.' Journal of Consulting and Clinical Psychology 40: 223-31.

Alexander, J.F., and Barton, C. (1976) . 'Behavioral systems therapy for families. ' In D.H. Olson (ed.) , Treating Relationships (pp. 168-88). Lake Mills, IO: Graphic. Alexander, J.F., and Barton, C. (1983) . Functional Family Therapy Training Manual Salt Lake City: Western States Family Institute. Alexander, J.F., and Parsons, B.V. (1973) . 'Short-term behavioral intervention with delinquent families: Impact on family process and recidivism.' Journal ofAbnormal Psychology 81: 219-25.

Alexander, J.F., and Parsons, B.V. (1982) . Functional Family Therapy: Principles and Procedures. Carmel, CA: Brooks/Cole. Alexander, J.F., Barton, C. , Schiavo, R.S., and Parsons, B.V. (1976) . 'Systemsbehavioral intervention with families of delinquents: Therapist characteristics, family behavior, and outcome.' Journal of Consulting and Clinical Psychology 44: 656-64.

Anderson, C. (1987). ' Depression and families.' Journal of Psychotherapy and the Family 3: 33-47.

186 References Auclaire, P. , and Schwartz, I.A. (1986). An Evaluation of the Effectiveness of

Intensive Home-Based Services as an Alternative to Placement for Ado/.escents and Their Families. Minneapolis, MN.: Hennepin County Commwiity Ser-

vices Department, and the University of Minnesota, Hubert H. Humphrey Institute of Public Affairs. Auclaire, P., and Schwartz, I.A. (1987a) . 'Are home based services effective? ' Children Today 16: 5-9. Auclaire, P., and Schwartz, I.A. (1987b). An Evaluation of the Effectiveness of Intensive Home-Based Services as an Alternative to Placement for Mo/.es-

cents and Their Families: A Follow-up Study of Placement Resource Utilization. Minneapolis, MN.: Hennepin County Community Services Depart-

ment, and the University of Minnesota, Hubert H. Humphrey Institute of Public Affairs. Backman, A., Haikonen, P., Hirvonen, E,. Rasanen, E., et al. (1981) . 'Chronic asthma in children: A medico psychosocial approach.' Psychatria Fennica. suppl., 85-95. Bank, L. , Marlowe, J.H., Reid, J.B., Patterson, G.R., and Weinrott, M.R. (1991). 'A comparative evaluation of parent-training interventions for families of chronic delinquents.' Journal ofAbnormal Chi/,d Psychology 19: 15-33. Barkley, R.A, Guevremont, D.C. , Anastopoulos, A.D., and Fletcher, KE. (1992). 'A comparison of three family therapy programs for treating family conflicts in adolescents with attention-deficit hyperactivity disorder.' Journal of Consulting and Clinical Psychology 60: 450-62. Barrett-Lennard, G.T. (1962). 'Dimensions of therapist response as causal factors in therapeutic change.' Psychologi,cal Monographs 73: 72. Barth, R.P. (1992). 'Visions for family preservation research. ' In Empowering Families: Papers from the Fifth Annual Conference on Family-Based Services. Riverdale, IL.: National Association for Family-Based Services. Barth, R.P. (1988). 'Theories guiding home-based intensive family preservation services.' InJ.K Whittaker, J. Kinney, E.M. Tracy, and C. Booth (eds.),

Improving Practice Technology for Work with High Risk Families: Lessons from the 'Homebuilders' Social Work Education Project (91-114). Seattle:

Center for Social Welfare Research , School of Social Work, University of Washington. Barthel. J. (1992). ForChi/,dren's Sake: The Promise of Family Preservation. (Available from the Edna McConnell Clark Foundation, 250 Park Ave. , New York, NY 10177-0026.) Barton, C. , and Alexander, J.F. (1981). 'Fwictional family therapy.' InA.S. Gurman and D.P. Kniskern (eds.), Handbook of Family Therapy (pp. 403--43). New York: Brwiner/Mazel.

References

187

Barton, C., Alexander, J.F., Waldron, H., Turner, C.W., and Warburton, J. (1985). 'Generalizing treatment effects offunctional family therapy: Three replications. ' American Journal of Family Therapy 13: 16-26. Beal, D., and Duckro, P. (1977). 'Family counselling as an alternative to legal action for the juvenile status offender. ' Journal of Marriage and Family

Counselling 3: 77-81.

Beavers, W.R. (1982) . 'Healthy, midrange, and severely dysfunctional families.' In F.W. Walsh (ed.). Normal Family Proc,esses (pp. 4~) . New York: Guilford. Beck, A.I. , Steer, R.A., and Garbin, M.G. (1988). 'Psychometric properties of the Beck Depression Inventory: 25 years of evaluation.' Clinical Psychology

Review 8: 77-100.

Bemporad, J., and Ratey, J. (1985). 'Intensive psychotherapy of former anorexic individuals.' American Journal of Psychotherapy 39: 454-66. Berger, H., Hoenig, P. , and Liebman, R. (1977) . 'Recurrent abdominal pain: Gaining control of the symptom.' American Journal of Diseases of Childhood 31: 134~.

Biegel, D.E., and Wells, K. (1991) . 'Introduction.' In K Wells and D.E. Biegel (eds.) , Family Preservation Services: Research and Evaluation (pp. xiii-xxiii). Newbury Parle, CA: Sage. Birtchnell, J., and Kennard, J. (1983) . 'Marriage and mental illness.' British Jour-

nal of Psychiatry 142: 193-8.

Bishop, D. , Epstein, N. , Keitner, G., Miller, I., and Srinivasan, S. (1986) 'Stroke: Morale, family functioning, health status, and function capacity.' Archives of

Physical Medicine and Rehabilitation 67: 84-7.

Black, D. , and Urbanowicz, M. (1987). 'Family intervention with bereaved children.' Journal of Child Psychology and Psychiatry andAIUed Disciplines 28: 467-76.

Bogert, A.J., and French, A.P. (1978). 'Successful short-term family therapy with incarcerated adolescents.' Journal ofJuvenile and Family Courts 29: 3-8.

Bowen, M. (1974) . 'Alcoholism as viewed through family systems theory and family psychotherapy. Annals of the New York Academy ofScienc,e 233: 11522.

Breunlin, D.C., Breunlin, C., Kearns, D.L., and Russell, W.P. (1988). 'A review of the literature on family therapy with adolescents, 1979-1987.' Journal ofAdo-

lescenc,e 11: 309-34.

Broderick, C., and Schrader, P. (1992). 'The history of professional marriage and family therapy.' In A. Gurman and D. Kiniskern (eds.), Handbook of Family Therapy, vol. 2. New York: Brunner Mazel.

188 References Bronfenbrenner, U. (1979). 'The ecology of human development: Experiments by nature and design.' Cambridge, MA: Harvard University Press. Brownell, S. (1978). The Navy alcoholism prevention program-worldwide.'

Alcolwlism, Clinical and Experimental Research 2: 362-5.

Bruch, H. (1974). 'Eating disorders, obesity, anorexia nervosa, and the person within.' London: Routledge and Kegan Paul. Bryant-Waugh, R., Knibbs, J., Fosson, A., Kaminski, Z., and Lask, B. (1988) . Long term follow-up of patients with early onset anorexia nervosa.' Archives of

Disease in Childhood 63: 5-9.

Burbeck, T. (1979). 'An empirical investigation ofthe psychosomatogenic family model.' Journal of Psychosomatic Research 23: 327-37. Cadogan, D. (1973). 'Marital group therapy in the treatment of alcoholism.'

Quarterly Journal of Studies in Alcoholism 34: 1187-94.

Cantwell, P., Stuizenberger, S., and Burroughs, J. (1977). 'Anorexia nervosa: An affective disorder?' Archives of General Psychiatry 34: 1087-93. Cheek, F., Franks, C., Laucius, J., and Burtle, V. (1971). 'Behavior-modification training for wives of alcoholics.' Quarterly Journal of Studies in Alcoholism 32: 456--61.

Child Welfare League of America. (1989). Standards for Seroice to Strengthen and Preserve Families with Children. Washington, DC: Child Welfare League of America Christensen, D. (1983). 'Postmastectomy couple counselling: An outcome study of a structured treatment protocol.' Sex and Marriage Therapy 9: 266--75.

Clarkin, J., Glick, I. , Hass, G., Spencer, J., Lewis, A., Peyser, A., DeMane, A., Good-Ellis, M., Harris, M., and Lestelle, V. (1990). 'A randomized clinical trial of in-patient family intervention, V: Results of affective disorders.' Journal of

Affective Disorders 18: 17-28.

Cobb, S., Miller, M., and Wieland, M. (1959). 'On the relationship between divorce and rheumatoid arthritis.' Arthritis and Rheumatism 8: 9 Cohen, S., and Hoberman, H.M. (1983). 'Positive events and social supports as buffers of life change stress.' Journal ofApplied Social Psychology 13: 99-125. Colapinto, J. (1991). 'Structural Family Therapy.' In A. Gurman and D. Kniskern (eds.), Handbook of Family Therapy(pp. 417-43). New York: Brunner/ Mazel. Coleman, S.B., and Davis, D.l. (1978). 'Family therapy and drug abuse: A national survey.' Family Process 17: 21-9. Corder, 8., Corder, R., and Laidlaw, N. (1972). 'An intensive treatment program for alcoholics and their wives.' Quarterly Journal of Studies in Alcolwlism 33: 1144-6.

References

189

Coyne, J., and Anderson, B. (1988). The "psychosomatic family" reconsidered: Diabetes in context' Journal ofMarital and Family Therapy 14: 113-23. Croog, S., and Levine, S. (1977). The Heart Patient Recovers. NewYorlc: Human Science Press. Crisp, A. (1980). Anorexia Nervosa: Let Me Be. London: Academic Press. Crisp, A., Harding, B., and McGuinness, B. (1974). 'Anorexia nervosa- Psychoneurotic characteristics of parents: A quantitative study.' Journal of Psyclwsomatic Research 18: 167-73. Crisp, A., Norton, K, Gowers, S., Halek, C., et al. (1991). 'A controlled study of the effect of therapies aimed at adolescent and family psychopathology in anorexia nervosa. ' British Journal of Psychiatry 159: 325-33. Dare, C. , Eisler, I., Russell, G., Gerald, F. , and George, I. (1990). The clinical and theoretical impact of a controlled trial of family therapy in anorexia nervosa. ' Journal ofMarital and Family Therapy 16: 39-57. Davenport, Y., Ebert, M., Adland, M. , and Goodwin, F. (1977). 'Couples group therapy as an adjunct to lithium maintenance of the manic patient.' American Journal of Orthopsychiatry 47: 495-502. Dell, P. (1986). 'In defense of"lineal causality."' Family Process 25: 513-20. Dewitt, K. (1978). The effectiveness of family therapy: A review of outcome research.' Archives of General Psychiatry 35: 549-61. Deyo, R., Inui, T., and Sullivan, B. (1982). Physical and psychosocial functions in rheumatoid arthritis: Clinical use of self-administered health instruments.' Archives ofInternal Medicine 142: 879-82. Dittmar, H., and Bates, B. (1987). 'Humanistic approaches to the understanding and treatment of anorexia nervosa.' Journal ofMolescence 10: 57-69. Dittrich, J., and Trapold, M. (1984). 'A treatment program for wives of alcoholics: An evaluation.' Bulletin of the Society of Psyclwlogists ofAddiction Behavior 3: 91-102. Druckman, J.M. (1979). 'A family oriented policy and treatment program for female juvenile offenders.' Journal ofMarriage and the Family 41: 627-36. Dulcan, M. (1984). ' Brief psychotherapy with children and their families: The state of the art. ' Journal of the American Academy of Child Psychiatry 23: 544-51. Engel, K, Wittem, M., Hentze, M., and Meyer, A. (1989). 'Long-term stability of anorexia nervosa treatments: Follow-up study of218 patients. ' Psychiatric Develnpments4: 39~07. Epstein, N. (1988). 'Dilemmas and choices in the design of family therapy research.' In L. Wynne (ed.), The State of the Art in Family Therapy Research: Controversies and Recommendations. New York: Family Process Press.

190

References

Epstein, N., and Volk, L. (1981) . ' Research on the results of psychotherapy: A summary of evidence.' American Journal of Psychiatry 138: 1027. Esser, P. (1971). 'Evaluation of Family Therapy with alcoholics.' British Journal ofAddiction to Alcohol Other Drugs 66: 251-5. Evans, R., Matlock, A., Bishop, D. , Stranahan, S., and Pederson, C. (1988) . 'Family intervention after stroke: Does counselling or education help?' Stroke 19: 1243-9. Falloon, I., and Pederson, J. (1985) . ' Family management in the prevention of morbidity of schizophrenia: The adjustment of the family unit.' British Journal of Psychiatry 147: 156-63. Falloon, I., Pederson, J., andAl-Khyall. M. (1986) . ' Enhancement of healthgiving family support versus treatment of family pathology.' Journal of Family Therapy 8: 339-50. Falloon, I., Boyd, J., McGill, C., Razani, J., Moss, H., and Gilderman, A. (1982). · Family management in the prevention in the exacerbation of schizophrenia: A controlled study.' New Eng/and Journal ofMedicine 306: 1437-40. Falloon, I., Boyd, J. , McGill, C., Williamson, M., et al. (1985a) . 'Family management in the prevention of morbidity of schizophrenia: Clinical outcome of a two-year longitudinal study. ' Archives of General Psychiatry 42: 887-96. Falloon, I., McGill, C. , Boyd, J. , and Pederson, J. (1985) . 'Family management in the prevention of morbidity of schizophrenia: Social outcome of a two-year longitudinal study.' PsychowgicalMedicine 17: 59-66. Falloon, I., McGill, C., Boyd, J. , and Pederson, J. (1987). Family management in the prevention of morbidity of schizophrenia: Social outcome of a two-year longitudinal study.' Psychowgical Medicine 17: 59-66. Feinstein, AR. (1967). Clinica[Judgemenl Baltimore, MD: Williams and Wilkins. Feldman, L.H. (1991a) . Assessing the Effectiveness of Family Preservation Services in New Jersey within an Ecowgical Context. Trenton, NJ: Department of Human Services, New jersey Division ofYouth and Family services, Bureau of Research, Evaluation, and Quality Assurance. Feldman, L.H. (1991 b). 'Evaluating the impact of intensive family preservation services in New Jersey.' In K Wells and D.E. Biegel (eds.), Family Preservation Services: Research and Evaluation (pp. 47-71). Newbury Park, CA: Sage. Fisher, S. (1989). The use of time-limits in brief psychotherapy: A comparison of six-session, twelve session, and unlimited treatment with families.' Family Process 23: 101-6. Fishman, H.C. (1988) . Treating Troubled Adolescents. New York: Basic Books. Fosson, A., Knibbs, J., Bryant-Waugh, R., and Las.le, B. (1987). 'Early onset of anorexia nervosa.' Archives of Disease in Children 62: 114-18.

References

191

Frank, J. (1979) . · Persuasion and healing: A comparative study.' Baltimore: Johns Hopkins Press. Frankel, H. (1988) . 'Family-centered, home-based services in child protection: A review ofthe Research.' Social Service Review 62: 137-57. French, S. (1987). 'Family approach to alcoholism: Why lack of interest among marriage and family professionals.' Journal of Drug Issues 17: 359-68. Friedman, A.S. (1975). 'Interaction of drug therapy with marital therapy in depressive patients.' Archives of General Psychiatry 32: 619-37. Friedman, A.S. (1989). 'Family therapy vs parent groups: Effects on adolescent drug abusers.' American Journal of Family Therapy 17: 335-47. Friedman, A.S., Tomko, L.A. , and Utada, A. (1991). 'Client and family characteristics that predict better therapy outcome for adolescent drug abusers.' Fam-

ily Dynamics ofAddiction Quarterly 1: 77-93.

Gallant, D., Rich, A., Bey, E., and Terranova, L. (1970). 'Group psychotherapy with married couples: A successful technique in New Orleans alcoholism clinic patients.' Journal of the Louisiana State Medical Society 122: 41-4. Garfinkle, P., and Gamer, D. (1982). AnorexiaNervosa: A Multidimensional Perspective. New York: Brunner/ Maze!. Garfinkle, P., Moldofsky, H., and Garner, D. (1977) . The outcome of anorexia nervosa.' In R. Vigersky (ed.). Anorexia Nervosa (pp. 315-29) . New York: Raven Press. Gamer, D., Garfinkel, P., and Bemis, K. (1982) . 'A multidimensional psychotherapy for anorexia nervosa. ' Journal of Eating Disorders 1: 3-46. Garrigan, J.J., and Bambrick, A.F. (1975) . 'Short-term family therapy with emotionally disturbed children.' Journal of Marriage and Family Counselling 1: 379-85.

Garrigan, J.J., and Bambrick, A.F. (1977) . 'Family therapy for disturbed children: Some experimental results in special education.' Journal of Marriage

and Family Counselling 3: 83-93.

Garrigan, J.J., and Bambrick, A.F. (1979). 'New findings in research on gobetween process. International]ournal of Family Therapy 1: 76-85. Gibb, J.R. (1961). 'Defensive communications.' Journal of Communication 3: 141-8.

Giovannoni, J.M. (1985) . 'Child abuse and neglect: An overview. ' In J. Laird and A. Hartman (eds.), Handbook of Child Welfare(pp. 193--212). NewYorlc Free Press. Glick, I., et al. (1985) . 'A controlled evaluation of inpatient intervention, I: Preliminary results of the six-month follow-up.' Archives of General Psychiatry 42: 882-6.

Glick, I., Spencer, J., Clarkin, J., Hass, G. , Lewis, A., Peyser, J., DeMane, N.,

192 References Good-Ellis, M., Harris, E., and Lestelle, V. (1990). 'A randomized clinical trial of in-patient family intervention, IV: Follow-up results for subjects with schizophrenics.' Schizophrenia Research 3: 187-200. Glick, I., Clarkin, J., Hass, G., Spencer, J., et al. (1991) . 'A randomized clinical trial of inpatient family intervention, VI: Mediating variables and outcome.'

Family Process 30: 85-99.

Goldberg, R., and Wood, M. (1985). 'Psychotherapy for the spouses of lung cancer patients: Assessment of an intervention. Psyclwtherapyand Psychoso-

matics 43: 141-50.

Golden, M., Herrod, A, and Orr, D. (1985) . 'An approach to prevention of recurrent diabetic ketoacidosis in the pediatric population.' Journal of Pedi-

atrics 107: 195-200.

Goldstein, M., Rodnick, E., Evans, J., May, P., and Steinberg, M. (1978). 'Drug and family therapy in the aftercare of acute schizophrenics.' Archives of Gen-

eral Psychiatry 35: 1169-77.

Gordon, D.A.,Arbuthnot, J. , Gustafson, K.E., and McGreen, P. (1988). 'Homebased behavioral-systems Family Therapy with disadvantaged juvenile delinquents.' American Journal of Family Therapy 16: 243-55. Gordon, T. (1977). 'Parent effectiveness training: A preventive program and its delivery system.' In G.W. Albee and J.M. Joffee (eds.), Primary Prevention of Psychopathology(pp. 175-86). Hanover, NH: University Press of New England. Gonnan, J., and Rooney, J. (1979). The influence of AL-Anon on the coping behaviour of wives of alcoholics.' Journal of Studies in Alcoholism 40: 1030-7.

Gottlib, I., Wallace, P., and Colby, C. (1990). 'Marital and family therapy for depression.' In B. Wolman and G. Stricker (eds.), Depressive Disorder: Facts, Theories, and Treatment Methods. pp. 396-424. New Yorlc Wiley. Griffith, J., and Griffith, M. (1987). 'Structural family therapy in chronic illness. '

Psychosomatics 28: 202-5.

Grigsby, R.K. (1991) . Theories that guide intensive family preservation services: A second look. ' In E.S. Morton and R.K. Grigsby (eds.), Advancing Family Preservation Practice (pp. 16-27). Newbury Park, CA:. Sage. Groen J., and Feldman-Toledano, Z. (1966). 'Educative treatment of patients and parents in anorexia nervosa.' British Journal of Psychiatry 112: 671-8. Gruher, M. (l 979) . 'Family counselling and the status offender. ' Juvenile and

Family Court Journal 30: 23-8.

Guerney, B. (1985). 'Family therapy research: What are the most important variables.' International]ournal of Family Therapy 7: 40-9. Gunnan, A. (1988) . 'Issues in the specification of family therapy interventions.'

References 193 In L Wynn (ed.), The State of the Art in Family Therapy Research: Controversies and Recommendations. New York: Family Process Press. Gunnan, A., and Kniskem, D. (1978). 'Research on marital and family therapy. In S. Garfield and A. Bergin (eds.), Handbook of Psychotherapy and Behavior Change. New York: Wiley. Gunnan, A., and Kniskem, D. (1980). 'Family therapy outcome research; Knowns and unknowns.' In A. Gurmanand D. Kniskern (eds.), Hand book of Family Therapy (pp. 742-76). New York: Brunner/Mazel. Gunnan, A., and Kniskern, D. (eds.) (1992). Handbook ofFamily T1ierapy, vol. 2. New York: Brunner/Mazel. Gunnan, A. , Kniskern, D., and Pinsoff, W. (1986). 'Research on the process and outcome of marital and family therapy.' In S. Garfield and A Bergin (eds.),

Handbook of Psychotherapy and Behavior Change: An Empirical Analyses

(3rd ed.). New Yorlc: Wiley. Gustaffson, P., and Svedin, C. (1988). 'Cost-effectiveness: Family therapy in a pediatric setting.' Family Systems Medicine 6: 162-75. Gustaffson, P., Kjellman, N., and Cederblad, M. (1986). 'Family therapy in the treatment of severe childhood asthma.' Journal of Psychosomatic Research 30: 369-74. Haggmark, C., Theorell, T., and Ek, B. (1987). 'Coping and social activity patterns among relatives of cancer patients.' Social Scienc:e and Medicine 25: 1021-5. Haley, J. (1963). Strategies of Psyclwtherapy. New York: Grune and Stratton. Haley, J. (1976) . Prob km-solving Therapy. San Francisco: Jossey-Bass. Haley, J. (ed.) , (1967). Advanced Techniques of Hypnosis and Therapy: Sel£cted Papers of Milton H. Erickson. New Yorlc: Grune and Stratton. Hall, A., and Crisp, A (1987). 'Brief psychotherapy in the treatment of anorexia nervosa: Outcome at one year.' BritishJournalof Psychiatry 151: 185-91. Hall, A., Slim, E., Hawker, F., and Salmond (1984) . 'Anorexia nervosa: Longterm outcome in 50 female patients.' British Journal of Psychiatry 145: 407-13. Hal mos, P. (1965). Faith of tire Counsellors. London: Constable. Halper, G., and Jones, M.A. (1981). 'Serving Families at Risk of Dissolution: Public Preventive Services in New York City. NewYorlc: Human Resources Administration. Harding, L (1976). 'Group intervention for wives of myocardial infarction patients.' Nursing Clinics of North America 11: 339-47. Hartman, S. (1987). 'Patterns of change in families following severe head injuries in children.' Australian and New Zealand Journal of Family Therapy 8: 125-30.

194

References

Hass, G., Glick, I., Clarkin, J. , Spencer, J., Lewis, A., Peyser, J., DeMane, N., Good-Ellis, M., Harris, E., and Lestelle, V. (1986). 'Inpatient family intetvention: A randomized clinical trial, II: Results at hospital discharge.' Archives of

General Psychiatry 45: 217-24.

Hazelrigg, M. , Harris, M., and Borduin, C. (1987). 'Evaluating the effectiveness of family therapies: An integrative review and analyses.' Psychology Bulletin 101: 421-42.

Hedberg, A., and Campbell, L. (1974) . 'A comparison of four behavioral treatment of alcoholism. ' Journal of Behavior Therapy and Experimental Psychol-

ogy 5: 251-6.

Heinrich, R., and Schag, C. (1985). 'Stress and activity management Group treatment for cancer patients and spouses.' Journal of Consulting and Clini-

cal Psychology 53: 439-46.

Henggeler, S.W., Rodick, J.D., Borduin, C.M., Hanson, C.L., Watson, S.M., and Urey, J.R. (1986). 'Multisystemic treatment of juvenile offenders: Effects on adolescent behavior and family interaction. ' Developmental Psychology 22: 132-41.

Henggeler, S.W., Borduin, C.M., Melton, G.B., Mann, B.J., Smith, L.A., Hall, J.A. , Cone, L., and Fucci, B.R. (1991). 'Effects of multisystemic therapy on drug use and abuse in serious juvenile offenders: A progress report from two outcome studies.' Family Dynamics ofAddiction Quarterly 1: 40-51. Henggeler, S.W., Melton, G.B., and Smith, L.A. (1992). 'Family preservation using multisystemic therapy: An effective alternative to incarcerating serious juvenile offenders.' Journal of Consulting and Clinica,l Psyclwlogy 60: 953-61.

Herzoff, N. (1979). 'A therapeutic group for cancer patients and their families.'

Cancer Nursing 2: 469-7 4.

Hoffman, H. , Noem, A., and Petersen, D. (1976) . 'Treatment effectiveness as judged by successfully and unsuccessfully treated alcoholics. ' Drug and

Alcolwl Dependency 1: 241-6.

Hogarty, G., Anderson, C., Reiss, D., Kornblith, S. , Greebwald, D. , Javana, C. , and Madonia, M. (1986). 'Family psycho-education, social skills training, and maintenance chemotherapy in the aftercare treatment of schizophrenia.'

Archives of General Psychiatry 43: 633-42.

Hsu, L. (1980). 'Outcome of anorexia nervosa.' Archives of General Psychiatry 37: 1041-6.

Hudgens, A. (1979). 'Family-oriented treatment of chronic pain.' Journal of

Marriage and the Family 5: 67-78.

Hudson, J., and Pope, H. (1986) . 'Treatment of bulimia: A review of current studies.' New Directions ofMental Health Service 31: 71-85.

References

195

Hughes, M. (1984). 'Recurrent abdominal pain and childhood depression.'

American Journal of Ortlwpsychiatry 54: 146-55.

Hughes, M. , and Zimin, R. (1978). 'Children with psychogenic abdominal pain and their families. ' Clinical Pediatrics 17: 560-73. Jackson, J. (1958). 'Alcoholism and the family.' Annals of theAmericanAmdemy

of Political and Social Science 315: 90-8.

Jacobson, N. (1988). 'Guidelines for the design of family therapy outcome research.' In L. Wynn (ed.), The State of the Art in Family Therapy Research: Controversies and Recommendations. New Yor.lc: Family Process Press. Jacobson, N.S., and Truax, P. (1991). 'Clinical significance: A statistical approach to defining meaningful change in psychotherapy research.' Jour-

nal of Consulting and Clinical Psychology 59: 12- 19.

Jacobson, N., Holtzworth-Munroe, A. , and Schmaling, K. (1989). 'Marital therapy and spouse involvement in the treatment of depression, agoraphobia, and alcoholism.' Journal of Clinical Counselling and Psychiatry 57: 5-10. Janzen, J. (1985-6). 'Use of family treatment methods by alcoholic treatment services.' Alrohol, Health, and Research World 10: 44-6. Jenkins, M. (1987). 'An outcome study of anorexia nervosa in an adolescent unit. ' Journal ofAdo/esc,ence 10: 71-81 . Joaning, H., Quinn, W., Fischer, J., and Arredondo, R. (1984). Family Therapy versus Traditional Therapy for Drug Abusers. (Grant no. ROl DA 03733) Rockville, MD: NIDA. Joaning, H., Quinn, W. , Thomas, F., and Mullen, R. (1992) . 'Treating adolescent drug abuse: A comparison of family systems therapy, group therapy, and family drug education.' Journal ofMarital and Family 11ierapy 18: 345-56. Johnson, T.F. (1977). 'The results offamily therapy with juvenile offenders.'

Juvenile Justice 28: 29-34.

Jones, M.A. (1985) . A Serond Cha.nee for Families: Five Years Later. New York: Child Welfare League of America. Jones, M.A. (1991). 'Measuring outcomes.' In K. Wells and D.E. Biegel (eds.), Family Preservation Services: Research and Evaluation (pp. 159-86) . Newbury Park, CA: Sage. Jones, M.A. , Neuman, R. , and Shyne, A.W. (1976) . A SecondChanc,efor Families: Evaluation ofa Program to Reduce Foster Care. New York: Child Welfare League of America Kahle, D., and White, R. (1991) . 'Attitudes toward alcoholism among psychologists, family and child coW1sellors.' Journal of Studies ofAlrolwlism 52: 321-4.

Kalucy, R., Crisp, A. , and Harding, B. (1977). 'A study of 56 families with anorexia nervosa.' British Journal of Medicine and Psychology 50: 24-30.

196 References Kaufmann, E. (1985). 'Family systems therapy and substance abuse: An overview of two decades of research and clinical experience.' lnternational]ournal ofAddictions 6-7: 897-916. Kenward, K. , and Rissover, J. (1980). 'Family systems approach to the treatment and prevention of alcoholism.' Family Therapy 7: 97-108. Kinney, J., Haapala, D., and Booth, C. (1991). Keeping Families Together: The Homebuilders Model New York: Aldine De Gruyter. Kirkby, R., and Smymios, K (1990). 'Child-oriented family therapy outcome research: Comparisons between brief family therapy and alternative treatment.' Australian and New Zea/and Journal of Family Therapy 11: 75-84. Klein, N.C., Alexander, J.F., and Parsons, B.V. (1977). 'Impact of family systems intervention on recidivism and sibling delinquency: A model of primary prevention an program evaluation.' Journal ofConsulting and Clinical Psychology 45: 469-74. Kog, E., Vandereycken, W., and Vertommen, H. (1985) . 'The psychosomatic family model: A critical analysis of family interaction concepts.' Journal of Family Therapy 7: 31-44. Kog, E., Vandereycken, W., and Vertommen, H. (1987) . 'Minuchin's psychosomatic family model revised: A concept validation study using a multitraitmultimethod approach.' Family Process 26: 235-53. Kolevzon, M., and Green, H. ( 1983) 'An experimentally based inductive approach to learning about family therapy.' American Journal of Family Therapy 11: 35-42. Laird, J., and Hartman, A. ( 1985). 'The context of child welfare.' In J. Laird and A. Hartman (eds.), HandbookofChUd Weifare(pp. 1-9). New York: Free Press. Langsley, D., Machotka, P., and Flomenhaft, K. (1968). 'Family crisis therapy: Results and implications.' Family process 7: 145-58. Langsley, D.G., Kaplan, D.M., Pittman, F.D., and Machotka, P. (1968). The Treatment of Families in Crisis. New York: GrWle and Stratton. Langsley, D., Flomenhaft, K., and Machotka, P. (1969). 'Follow-up evaluation of family crisis therapy.' American Journal of Orthopsychiatry39: 753-9. Langsley, D., Pittman, F., and Swank, G. (1969). 'Family crisis in schizophrenics and other mental patients.' Journal of Nervous and Mental Disease 149: 270-6. Lansky, M. (1980). 'Research in family therapy.' Advances in Family Psychiatry 2: 3-18. Lask, B. (1992). 'Psychological treatments for childhood asthma.' Archives of Diseases of Childhood 67: 891.

References

197

Lask, B., and Bryant-Waugh, R. (1986) . 'Childhood onset of anorexia nervosa.'

Recent Advances in Pediatrics 8: 21-31.

Lask, 8., and Matthew, D. (1979) . 'Childhood asthma: A controlled trial of family psychotherapy.' Archives of Diseases of Childhood. 54: 116-19. Lawson, A., and Lawson, G. (1991). 'Classic articles in the field of family dynamics of addiction: The early years, 1953-1980.' Family Dynamics and

Addiction Quarterly 1: 27-40.

LeBow, J. (1981). 'Issues in the assessment ofoutcome in family therapy.' Fam-

ily Process 20: 167-88. Leff, J. (1980). 'Developments in family treatment of schizophrenia. ' Advances in Family Psychiatry 2: 313-33. Leff, J. {1989). 'Fam.Hy factors in schizophrenia.' Psychiatric Annals 19: 542-7. Leff, J. , Kuipers, L., Berkowitz, R., Vries, R. , and Sturgeon, D. (1982) . 'A controlled trial of social intervention in the families of schizophrenic patients.'

British Journal of Psychiatry 141: 121-34.

Leff, J. , Kuipers, L., Berkowitz, R., Eberlin-Vries, R., and Sturgeon, D. (1985). 'A controlJed trial of social intervention in the families of schizophrenic patients: Two year follow-up.' British Journal of Psychiatry 146: 594--600. Leff, J., Berkowitz, R., Shavit, N. , Strachan, A., Glass, I. , and Vaughn, C. (1989) . 'A trial of family therapy vs a relatives group for schizophrenia.' British Jour-

nal of Psychiatry 154: 58--6.

Lewis, R.E. (1991 ). 'What are the characteristics of intensive family preservation services?' In M.W. Fraser, P.J. Pecora, and D.A. Haapala {eds.), Families in Crisis: The Impact ofIntensive Family Preservation Services {pp. 93-107) . New York: Aldine De Gruyter. Liebman, R., Honig, P., and Berger, H. (1976). 'An integrated treatment program for psychogenic pain.' Family Process 15: 397-406. Lieberman, R., Falloon, I., and Aitchson, R. (1984) . 'Multiple family therapy for schizophrenia: A behavioral, problem-solving approach.' Psychosocial Reha-

bilitation Journal 7: 60-77.

Liepman, M., Nirenberg, T., and Begin, A. (1989). 'Evaluation of a program designed to help family and significant others to motivate resistant alcoholics into recovery.' AmerimnJournalof Drugs and A/coho/Abuse 15: 209-21. Locke, H.J. , and Wallace, K.M. (1959). 'Short marital adjustment and prediction tests: Their reliability and validity.' Journal ofMarriage and Family Living 21: 251-5.

Love, L., Kaswan, J., and Bugental, D. (1972) . 'Differential effectiveness of three clinical interventions for different socioeconomic groupings. ' Journal of

Counselling and Clinical Psyclwlogy29: 347--60.

Lyle, C.G., and Nelson, J.P. (1983). 'Home-Based vs Traditional Child Protection

198 References Services: A Study of the Home-Based Servic.es Demonstration Project in the Ramsey County Human Servic.es Department. (Available from the National Resource Center on Family-Based Services, University of Iowa School of Social Worl