Elements Of Clinical Study Design, Biostatistics & Research is designed as a toolbox for biomedical researchers. The

*443*
*53*
*35MB*

*English*
*Pages 150
[152]*
*Year 2023*

*Table of contents : CoverTitleCopyrightEnd User License AgreementContentsForewordPreface CONSENT FOR PUBLICATION CONFLICT OF INTERESTAcknowledgementsStudy Designs in Clinical Research INTRODUCTION OBSERVATIONAL STUDIES Case Series Case-Control Studies Cross-Sectional Studies, Surveys Methods of Survey By Self-Administered Questionnaire Interviews Reliability Validity Pilot Testing Cohort Studies Prospective Cohort Historical Cohort or Retrospective Cohort Study Amphispective Cohort Studies INTERVENTIONAL STUDIES (EXPERIMENTAL STUDIES) Animal Studies Human Studies/Clinical Trials Controlled Trials Uncontrolled Studies META-ANALYSIS CONCLUSION Study Designs in medical research fall into three categories:Scales of Measurement, Descriptive Statistics & Data Presentation INTRODUCTION SCALES OF MEASUREMENT Nominal (Categorical Scale) Ordinal Scale Interval Scale Ratio Scale MEASUREMENT OF CENTRAL TENDENCY Calculation of Measures of Central Tendency Mean Median Mode Uses of Central Tendency GUIDELINES TO DECIDE WHICH MEASURE OF CENTRAL TENDENCY IS BEST TO BE USED Mean Median Mode Geometrical Mean MEASURES OF SPREAD The Range Standard Deviation Co-efficient of Variation (CV) Percentile GUIDELINES FOR USING DIFFERENT MEASURES OF DISPERSION PRESENTATION OF DATA IN TABLES AND GRAPHS Tables Simple Table Frequency Distribution Table Charts and Diagrams Simple Bar Chart Component Bar Chart Histogram Frequency Polygon Graphs and Line Diagram Pie Diagram or Pie Charts Pictogram CONCLUSION Scales of measurement are: Descriptive Statistics Data PresentationInferential Statistics INTRODUCTION PROBABILITY Objective Probability Subjective Probabilities EXPERIMENT AND EVENT Experiment Event Complementary Event Mutually Exclusive Event and the Rule of Addition Independent Events LIKELIHOOD SENSITIVITY SPECIFICITY BAYES' THEOREM POPULATION AND SAMPLES POWER OF STUDY METHOD OF SAMPLE SELECTION Non-probabilities Sampling Probability Sampling Simple Random Sample Systematic Random Sample Stratified Sample Random Cluster Sample Properties of Good Sampling Random Assignment RANDOM VARIABLES AND PROBABILITY DISTRIBUTION Poisson Distribution Binomial Distribution The Normal or Gaussian Distribution THE CENTRAL LIMIT THEORY SAMPLING DISTRIBUTION Features of The Sampling Distribution SAMPLE SIZE APPROACHES TO STATISTICAL INFERENCES A confidence interval Estimate and Estimation: Point Estimate: Interval Estimate: Hypothesis Testing Steps Making a statement (Stating) research question in terms of statistical Hypothesis Making the Decision on Suitable Test Statistics Chi-Square Test Selecting the level of Significance and Determination of the Value Determination of the value of Significance Performing the Calculations Making Conclusions ERRORS IN THE HYPOTHESIS TEST The Type One Error The Type Two Error POWER OF STUDY Significance of P-Value and Alpha PAIRED T TEST INTRA-RATER RELIABILITY PEARSON'S CORRELATION COEFFICIEN MEASUREMENT OF THE SAME VARIABLE BY TWO DIFFERENT PROCEDURES APPLICATION OF T-TEST Sample Size COMPARISON OF MEANS IN THREE OR MORE GROUPS Analysis of Variance (ANOVA) Example 1: Example 2: Pre-Condition (Assumptions in ANOVA) Multiple Comparison Procedure Post Hoc Comparisons COMPARISON OF FREQUENCIES OR PROPORTIONS IN MORE THAN TWO GROUPS CONCLUSIONStatistical Methods for Relationship Variables INTRODUCTION THE RELATIONSHIP BETWEEN TWO NUMERICAL OBSERVATIONS (CHARACTERISTICS) Correlation Assumptions in Correlation Regression THE RELATIONSHIP BETWEEN TWO ORDINAL CHARACTERISTICS THE RELATIONSHIP BETWEEN TWO NOMINAL CHARACTERISTICS Experimental Event Rate (EER) The Control Event Rate (CER) Relative Risk (RR) Relative Risk Reduction (RRR) Absolute Risk Reduction (ARR) Number Needed To Treat (NNT) Absolute Risk Increase (ARI) Number Needed to Harm (NNH) Odds Ratios Multiple Regression ANALYSIS OF COVARIANCE(ANCOVA) LOGISTIC REGRESSION MULTIVARIATE ANALYSIS OF VARIANCE (MANOVA) DIAGNOSTIC PROCEDURES Sensitivity Specificity Use of Sensitivity and Specificity for Revision of Probabilities: LIKELIHOOD RATIO Calculation of likelihood ratio BAYES' THEOREM Use of Sensitivity and Specificity in Making Clinical Diagnosis (Figs. 4.2 and 4.3) RECEIVER OPERATING CHARACTERISTIC (ROC) CURVE CONCLUSIONClinical Research INTRODUCTION CLINICAL RESEARCH TYPES OF CLINICAL TRIALS PHASES OF CLINICAL TRIALS Phase I Phase II Phase II(a) Phase II(b) (Pivotal Trial) Phase III(a) Phase-III (a) i) Superiority Clinical Trials Phase-III (a) ii)Equivalence Clinical Trial Phase-III (a) iii)Non-inferiority Clinical Trial Phase-III (b) Phase IV Clinical Trials RISK OF BIASES IN RANDOMIZED CONTROLLED TRIALS (RCTS) Cochrane Tool for Bias Risk Assessment CONSORT GUIDELINES CONSORT Checklist PRAGMATIC TRIALS (PRCT) (THERAPY EVALUATION) Main differences of (i) RCT and (ii) pRCT ELEMENTS OF PRAGMATIC TRIALS Formulation of Research Question Defining the patient group Identifying comparison group Defining the Treatment Protocol Sample Size Enrolment, Randomization, and Referral Outcome Analysis Reporting and Publication REVERSE PHARMACOLOGY IN DRUG DEVELOPMENT Reverse Pharmacology Concept THERE ARE FOUR STAGES OF REVERSE PHARMACOLOGY Stage-1:Selection of Herbal Remedy Literature Research Retrospective Treatment Outcome Study (RTO) Interview With Traditional Healer Stage-2: Dose Escalating Studies Dose Optimization Preservation of the Study Plant (Herb) Stage-3 Randomized Controlled Trial (RCT) Stage-4 Isolation And Testing Of Active Compounds CONCLUSIONSurvey Research INTRODUCTION DESIGNING SURVEY TOOL The Research Question (Framing) ADMINISTERING THE QUESTION Survey Methods Developing Survey Questionnaire Scales of Measurements Positive and negative categories Balancing the responses Balancing the Probes Revised balanced probe (questionnaire) Using the Likert Scale Suggestions for Writing Probes Layout Of Questionnaire RELIABILITY AND VALIDITY OF SURVEY INSTRUMENT Reliability Test Re-Test Internal Consistency: Alternative form: Intra-Observer Consistency: Inter Observer Consistency: Validity Face Validity: Content Validity: Criterion Validity: Construct Validity: ADMINISTRATION OF THE SURVEY INSTRUMENT Pilot testing Response Rate Methods to Increase Response Rates SELECTION OF REPRESENTATIVE SAMPLE AND DETERMINATION OF SAMPLE SIZE Determination of Sample Size ANALYSIS OF SURVEY OBSERVATIONS CONCLUSIONPlanning and Writing Research Projects INTRODUCTION PLANNING OF A RESEARCH PROJECT REVIEW OF LITERATURE AND IDENTIFICATION OF THE RESEARCH PROBLEM RESEARCH GAP ANALYSIS & FRAMING OF RESEARCH QUESTION CHOOSING THE APPROPRIATE STUDY DESIGN TO ANSWER THE RESEARCH QUESTION Experimental/Interventional studies Clinical trials Cohort Studies Case-Control Study Design Surveys & Cross-Sectional Studies Case Series Meta-Analysis Fishing Expedition LEVEL OF SIGNIFICANCE & SELECTION OF SAMPLE SIZE Power of the Study Variation in Standard Deviation BIASES IN THE STUDY Procedure Bias Recall Bias Insensitive Measurement bias Compliance Bias Selection bias Admission Rate Bias APPLICATION OF STATISTICAL PROCEDURE AND TEST OF SIGNIFICANCE Multiple Tests of Significance REPORT WRITING &DATA PRESENTATION SENDING RESEARCH WORK/ARTICLE FOR PUBLICATION Sources of guidelines for specific study design GUIDELINES FOR REPORTING STATISTICS Presentation of Results DISCUSSION AND CONCLUSION A researcher must remember that; CONCLUSIONAPPENDIX-AETHICAL GUIDELINES FOR CLINICAL RESEARCH INTRODUCTION PRINCIPALS OF ETHICS (CORE VALUES) Respecting the Autonomy Of Research Subjects Voluntary participation Autonomy and research involving minors Autonomy and Research in Schools Age Information to subjects Exceptions from informed consent Avoiding Harm Privacy and Data Protection Protecting research data and confidentiality Storing/ destroying research data Protecting privacy in research publications ETHICAL REVIEW Ethical Review Required Concerns of the Institutional Ethics Committee Information to the research subjects Privacy and data protectionGLOSSARYREFERENCES REFERENCESSubject IndexBack Cover*

Elements of Clinical Study Design, Biostatistics & Research Authored by S.S.Patel Centre of Excellence for Preclinical Safety & Efficacy Studies, Interdisciplinary Research and Therapy Evaluation, DattaMeghe Institute of Medical Sciences (Deemed to be University), NAAC Accredited Sawangi(Meghe),DMIMS (DU), Nagpur, India

& Aditya Patel Department of Conservative Dentistry & Endodontics, SharadPawar Dental College, DMIMS(DU), Sawangi (Meghe) Wardha Nagpur, India

Elements of Clinical Study Design, Biostatistics & Research Authors: Aditya Patel and S.S.Patel ISBN (Online): 978-981-5123-79-1 ISBN (Print): 978-981-5123-80-7 ISBN (Paperback): 978-981-5123-81-4 © 2023, Bentham Books imprint. Published by Bentham Science Publishers Pte. Ltd. Singapore. All Rights Reserved. First published in 2023.

BSP-EB-PRO-9789815123791-TP-133-TC-07-PD-20230320

BENTHAM SCIENCE PUBLISHERS LTD.

End User License Agreement (for non-institutional, personal use) This is an agreement between you and Bentham Science Publishers Ltd. Please read this License Agreement carefully before using the ebook/echapter/ejournal (“Work”). Your use of the Work constitutes your agreement to the terms and conditions set forth in this License Agreement. If you do not agree to these terms and conditions then you should not use the Work. Bentham Science Publishers agrees to grant you a non-exclusive, non-transferable limited license to use the Work subject to and in accordance with the following terms and conditions. This License Agreement is for non-library, personal use only. For a library / institutional / multi user license in respect of the Work, please contact: [email protected].

Usage Rules: 1. All rights reserved: The Work is the subject of copyright and Bentham Science Publishers either owns the Work (and the copyright in it) or is licensed to distribute the Work. You shall not copy, reproduce, modify, remove, delete, augment, add to, publish, transmit, sell, resell, create derivative works from, or in any way exploit the Work or make the Work available for others to do any of the same, in any form or by any means, in whole or in part, in each case without the prior written permission of Bentham Science Publishers, unless stated otherwise in this License Agreement. 2. You may download a copy of the Work on one occasion to one personal computer (including tablet, laptop, desktop, or other such devices). You may make one back-up copy of the Work to avoid losing it. 3. The unauthorised use or distribution of copyrighted or other proprietary content is illegal and could subject you to liability for substantial money damages. You will be liable for any damage resulting from your misuse of the Work or any violation of this License Agreement, including any infringement by you of copyrights or proprietary rights.

Disclaimer: Bentham Science Publishers does not guarantee that the information in the Work is error-free, or warrant that it will meet your requirements or that access to the Work will be uninterrupted or error-free. The Work is provided "as is" without warranty of any kind, either express or implied or statutory, including, without limitation, implied warranties of merchantability and fitness for a particular purpose. The entire risk as to the results and performance of the Work is assumed by you. No responsibility is assumed by Bentham Science Publishers, its staff, editors and/or authors for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products instruction, advertisements or ideas contained in the Work.

Limitation of Liability: In no event will Bentham Science Publishers, its staff, editors and/or authors, be liable for any damages, including, without limitation, special, incidental and/or consequential damages and/or damages for lost data and/or profits arising out of (whether directly or indirectly) the use or inability to use the Work. The entire liability of Bentham Science Publishers shall be limited to the amount actually paid by you for the Work.

General: 1. Any dispute or claim arising out of or in connection with this License Agreement or the Work (including non-contractual disputes or claims) will be governed by and construed in accordance with the laws of Singapore. Each party agrees that the courts of the state of Singapore shall have exclusive jurisdiction to settle any dispute or claim arising out of or in connection with this License Agreement or the Work (including non-contractual disputes or claims). 2. Your rights under this License Agreement will automatically terminate without notice and without the

need for a court order if at any point you breach any terms of this License Agreement. In no event will any delay or failure by Bentham Science Publishers in enforcing your compliance with this License Agreement constitute a waiver of any of its rights. 3. You acknowledge that you have read this License Agreement, and agree to be bound by its terms and conditions. To the extent that any other terms and conditions presented on any website of Bentham Science Publishers conflict with, or are inconsistent with, the terms and conditions set out in this License Agreement, you acknowledge that the terms and conditions set out in this License Agreement shall prevail. Bentham Science Publishers Pte. Ltd. 80 Robinson Road #02-00 Singapore 068898 Singapore Email: [email protected]

BSP-EB-PRO-9789815123791-TP-133-TC-07-PD-20230320

CONTENTS FOREWORD ........................................................................................................................................... i PREFACE ................................................................................................................................................ iv CONSENT FOR PUBLICATION ................................................................................................ v CONFLICT OF INTEREST ......................................................................................................... v ACKNOWLEDGEMENTS .................................................................................................................... vi CHAPTER 1 STUDY DESIGNS IN CLINICAL RESEARCH ........................................................ INTRODUCTION .......................................................................................................................... OBSERVATIONAL STUDIES ..................................................................................................... Case Series .............................................................................................................................. Case-Control Studies .............................................................................................................. Cross-Sectional Studies, Surveys ............................................................................................ Methods of Survey ......................................................................................................... By Self-Administered Questionnaire ............................................................................. Interviews ...................................................................................................................... Reliability ...................................................................................................................... Validity .......................................................................................................................... Pilot Testing .................................................................................................................. Cohort Studies ......................................................................................................................... Prospective Cohort ........................................................................................................ Historical Cohort or Retrospective Cohort Study ......................................................... Amphispective Cohort Studies ....................................................................................... INTERVENTIONAL STUDIES (EXPERIMENTAL STUDIES) ............................................. Animal Studies ........................................................................................................................ Human Studies/Clinical Trials ................................................................................................ Controlled Trials ........................................................................................................... Uncontrolled Studies ..................................................................................................... META-ANALYSIS ......................................................................................................................... CONCLUSION ............................................................................................................................... Study Designs in medical research fall into three categories: ................................................ CHAPTER 2 SCALES OF MEASUREMENT, DESCRIPTIVE STATISTICS & DATA PRESENTATION .................................................................................................................................... INTRODUCTION .......................................................................................................................... SCALES OF MEASUREMENT ................................................................................................... Nominal (Categorical Scale) ................................................................................................... Ordinal Scale ........................................................................................................................... Interval Scale .......................................................................................................................... Ratio Scale .............................................................................................................................. MEASUREMENT OF CENTRAL TENDENCY ........................................................................ Calculation of Measures of Central Tendency ........................................................................ Mean .............................................................................................................................. Median ........................................................................................................................... Mode .............................................................................................................................. Uses of Central Tendency ....................................................................................................... GUIDELINES TO DECIDE WHICH MEASURE OF CENTRAL TENDENCY IS BEST TO BE USED ................................................................................................................................... Mean ....................................................................................................................................... Median ....................................................................................................................................

1 1 2 2 3 4 4 4 4 5 6 6 7 7 7 8 9 9 9 9 11 12 12 12 14 14 14 15 15 15 16 16 16 16 17 18 18 19 19 19

Mode ....................................................................................................................................... Geometrical Mean ................................................................................................................... MEASURES OF SPREAD ............................................................................................................. The Range ............................................................................................................................... Standard Deviation .................................................................................................................. Co-efficient of Variation (CV) ................................................................................................ Percentile ...................................................................................................................... GUIDELINES FOR USING DIFFERENT MEASURES OF DISPERSION ........................... PRESENTATION OF DATA IN TABLES AND GRAPHS ...................................................... Tables ...................................................................................................................................... Simple Table .................................................................................................................. Frequency Distribution Table ....................................................................................... Charts and Diagrams ............................................................................................................... Simple Bar Chart ........................................................................................................... Component Bar Chart ................................................................................................... Histogram ................................................................................................................................ Frequency Polygon ....................................................................................................... Graphs and Line Diagram ....................................................................................................... Pie Diagram or Pie Charts ....................................................................................................... Pictogram ................................................................................................................................ CONCLUSION ............................................................................................................................... Scales of measurement are: ..................................................................................................... Descriptive Statistics ............................................................................................................... Data Presentation ....................................................................................................................

19 19 19 19 19 20 21 22 22 22 23 23 24 24 26 26 27 27 27 28 29 29 29 29

CHAPTER 3 INFERENTIAL STATISTICS ..................................................................................... INTRODUCTION .......................................................................................................................... PROBABILITY .............................................................................................................................. Objective Probability .............................................................................................................. Subjective Probabilities .......................................................................................................... EXPERIMENT AND EVENT ....................................................................................................... Experiment .............................................................................................................................. Event ....................................................................................................................................... Complementary Event ................................................................................................... Mutually Exclusive Event and the Rule of Addition ...................................................... Independent Events ....................................................................................................... LIKELIHOOD ................................................................................................................................ SENSITIVITY ................................................................................................................................. SPECIFICITY ................................................................................................................................. BAYES' THEOREM ...................................................................................................................... POPULATION AND SAMPLES .................................................................................................. POWER OF STUDY ...................................................................................................................... METHOD OF SAMPLE SELECTION ........................................................................................ Non-probabilities Sampling .................................................................................................... Probability Sampling .............................................................................................................. Simple Random Sample ................................................................................................. Systematic Random Sample ........................................................................................... Stratified Sample ........................................................................................................... Random Cluster Sample ................................................................................................ Properties of Good Sampling ........................................................................................ Random Assignment ......................................................................................................

31 31 31 31 32 32 32 33 33 33 34 34 35 35 35 36 36 37 37 37 37 37 38 38 38 38

RANDOM VARIABLES AND PROBABILITY DISTRIBUTION .......................................... Poisson Distribution ................................................................................................................ Binomial Distribution ............................................................................................................. The Normal or Gaussian Distribution ..................................................................................... THE CENTRAL LIMIT THEORY .............................................................................................. SAMPLING DISTRIBUTION ...................................................................................................... Features of The Sampling Distribution ................................................................................... SAMPLE SIZE ................................................................................................................................ APPROACHES TO STATISTICAL INFERENCES ................................................................. A confidence interval .............................................................................................................. Estimate and Estimation: .............................................................................................. Point Estimate: .............................................................................................................. Interval Estimate: .......................................................................................................... Hypothesis Testing Steps ........................................................................................................ Making a statement (Stating) research question in terms of statistical Hypothesis ..... Making the Decision on Suitable Test Statistics ........................................................... Chi-Square Test ............................................................................................................. Selecting the level of Significance and Determination of the Value ............................. Determination of the value of Significance ................................................................... Performing the Calculations ......................................................................................... Making Conclusions ...................................................................................................... ERRORS IN THE HYPOTHESIS TEST ..................................................................................... The Type One Error ................................................................................................................ The Type Two Error ............................................................................................................... POWER OF STUDY ...................................................................................................................... Significance of P-Value and Alpha ......................................................................................... PAIRED T TEST ................................................................................................................... INTRA-RATER RELIABILITY .................................................................................................. PEARSON'S CORRELATION COEFFICIEN ........................................................................... MEASUREMENT OF THE SAME VARIABLE BY TWO DIFFERENT PROCEDURES APPLICATION OF T-TEST ........................................................................................................ Sample Size ............................................................................................................................. COMPARISON OF MEANS IN THREE OR MORE GROUPS .............................................. Analysis of Variance (ANOVA) ............................................................................................. Example 1: .................................................................................................................... Example 2: .................................................................................................................... Pre-Condition (Assumptions in ANOVA) ...................................................................... Multiple Comparison Procedure ............................................................................................. Post Hoc Comparisons .................................................................................................. COMPARISON OF FREQUENCIES OR PROPORTIONS IN MORE THAN TWO GROUPS .......................................................................................................................................... CONCLUSION ...............................................................................................................................

39 39 40 40 41 42 42 42 43 43 44 45 45 45 45 46 46 47 47 47 47 48 48 48 48 48 49 49 49 50 50 50 51 51 51 52 52 53 53

CHAPTER 4 STATISTICAL METHODS FOR RELATIONSHIP VARIABLES ........................ INTRODUCTION .......................................................................................................................... THE RELATIONSHIP BETWEEN TWO NUMERICAL OBSERVATIONS (CHARACTERISTICS) ................................................................................................................. Correlation .............................................................................................................................. Assumptions in Correlation ........................................................................................... Regression ..................................................................................................................... THE RELATIONSHIP BETWEEN TWO ORDINAL CHARACTERISTICS ......................

55 55

53 54

55 55 56 58 58

THE RELATIONSHIP BETWEEN TWO NOMINAL CHARACTERISTICS ...................... Experimental Event Rate (EER) ............................................................................................. The Control Event Rate (CER) ............................................................................................... Relative Risk (RR) .................................................................................................................. Relative Risk Reduction (RRR) .............................................................................................. Absolute Risk Reduction (ARR) ............................................................................................ Number Needed To Treat (NNT) ........................................................................................... Absolute Risk Increase (ARI) ................................................................................................. Number Needed to Harm (NNH) ............................................................................................ Odds Ratios ............................................................................................................................. Multiple Regression ................................................................................................................ ANALYSIS OF COVARIANCE(ANCOVA) ............................................................................... LOGISTIC REGRESSION ........................................................................................................... MULTIVARIATE ANALYSIS OF VARIANCE (MANOVA) .................................................. DIAGNOSTIC PROCEDURES .................................................................................................... Sensitivity ............................................................................................................................... Specificity ............................................................................................................................... Use of Sensitivity and Specificity for Revision of Probabilities: ........................................... LIKELIHOOD RATIO .................................................................................................................. Calculation of likelihood ratio ................................................................................................ BAYES' THEOREM ...................................................................................................................... Use of Sensitivity and Specificity in Making Clinical Diagnosis (Figs. 4.2 and 4.3) ............ RECEIVER OPERATING CHARACTERISTIC (ROC) CURVE ........................................... CONCLUSION ...............................................................................................................................

58 59 59 60 60 60 61 61 62 62 62 63 63 64 64 64 64 65 65 65 66 67 68 69

CHAPTER 5 CLINICAL RESEARCH .............................................................................................. INTRODUCTION .......................................................................................................................... CLINICAL RESEARCH ............................................................................................................... TYPES OF CLINICAL TRIALS .................................................................................................. PHASES OF CLINICAL TRIALS ............................................................................................... Phase I ..................................................................................................................................... Phase II .................................................................................................................................... Phase II(a) ..................................................................................................................... Phase II(b) (Pivotal Trial) ............................................................................................ Phase III(a) .............................................................................................................................. Phase-III (a) i) Superiority Clinical Trials ................................................................... Phase-III (a) ii)Equivalence Clinical Trial ................................................................... Phase-III (a) iii)Non-inferiority Clinical Trial ............................................................. Phase-III (b) ............................................................................................................................ Phase IV Clinical Trials .......................................................................................................... RISK OF BIASES IN RANDOMIZED CONTROLLED TRIALS (RCTS) ............................ Cochrane Tool for Bias Risk Assessment ............................................................................... CONSORT GUIDELINES ..................................................................................................... CONSORT Checklist .............................................................................................................. PRAGMATIC TRIALS (PRCT) (THERAPY EVALUATION) ............................................... Main differences of (i) RCT and (ii) pRCT ............................................................................ ELEMENTS OF PRAGMATIC TRIALS .................................................................................... Formulation of Research Question ......................................................................................... Defining the patient group ...................................................................................................... Identifying comparison group ................................................................................................. Defining the Treatment Protocol ............................................................................................

71 71 71 72 72 72 72 72 73 73 73 75 76 76 77 77 77 78 78 79 79 79 80 80 80 81

Sample Size ............................................................................................................................. Enrolment, Randomization, and Referral ............................................................................... Outcome .................................................................................................................................. Analysis ................................................................................................................................... Reporting and Publication ....................................................................................................... REVERSE PHARMACOLOGY IN DRUG DEVELOPMENT ................................................ Reverse Pharmacology Concept ............................................................................................. THERE ARE FOUR STAGES OF REVERSE PHARMACOLOGY ....................................... Stage-1:Selection of Herbal Remedy ...................................................................................... Literature Research ....................................................................................................... Retrospective Treatment Outcome Study (RTO) ........................................................... Interview With Traditional Healer ................................................................................ Stage-2: Dose Escalating Studies ............................................................................................ Dose Optimization ......................................................................................................... Preservation of the Study Plant (Herb) ......................................................................... Stage-3 Randomized Controlled Trial (RCT) ......................................................................... Stage-4 Isolation And Testing Of Active Compounds ........................................................... CONCLUSION ...............................................................................................................................

81 81 81 82 82 82 83 83 83 83 84 85 85 86 86 87 87 87

CHAPTER 6 SURVEY RESEARCH .................................................................................................. INTRODUCTION .......................................................................................................................... DESIGNING SURVEY TOOL ..................................................................................................... The Research Question (Framing) .......................................................................................... ADMINISTERING THE QUESTION ......................................................................................... Survey Methods ...................................................................................................................... Developing Survey Questionnaire .......................................................................................... Scales of Measurements ................................................................................................ Positive and negative categories ................................................................................... Balancing the responses ................................................................................................ Balancing the Probes .................................................................................................... Revised balanced probe (questionnaire) ....................................................................... Using the Likert Scale ................................................................................................... Suggestions for Writing Probes .................................................................................... Layout Of Questionnaire ............................................................................................... RELIABILITY AND VALIDITY OF SURVEY INSTRUMENT ............................................. Reliability ................................................................................................................................ Test Re-Test ................................................................................................................... Internal Consistency: .................................................................................................... Alternative form: ........................................................................................................... Intra-Observer Consistency: ......................................................................................... Inter Observer Consistency: ......................................................................................... Validity ................................................................................................................................... Face Validity: ................................................................................................................ Content Validity: ........................................................................................................... Criterion Validity: ......................................................................................................... Construct Validity: ........................................................................................................ ADMINISTRATION OF THE SURVEY INSTRUMENT ........................................................ Pilot testing ............................................................................................................................. Response Rate ......................................................................................................................... Methods to Increase Response Rates ............................................................................

89 89 90 90 90 90 90 90 91 92 92 92 93 94 94 94 94 94 94 95 95 95 95 95 95 95 96 96 96 96 96

SELECTION OF REPRESENTATIVE SAMPLE AND DETERMINATION OF SAMPLE SIZE ................................................................................................................................................. Determination of Sample Size ................................................................................................ ANALYSIS OF SURVEY OBSERVATIONS ............................................................................. CONCLUSION ...............................................................................................................................

97 97 97 98

CHAPTER 7 PLANNING AND WRITING RESEARCH PROJECTS .......................................... INTRODUCTION .......................................................................................................................... PLANNING OF A RESEARCH PROJECT ................................................................................ REVIEW OF LITERATURE AND IDENTIFICATION OF THE RESEARCH PROBLEM RESEARCH GAP ANALYSIS & FRAMING OF RESEARCH QUESTION ......................... CHOOSING THE APPROPRIATE STUDY DESIGN TO ANSWER THE RESEARCH QUESTION ..................................................................................................................................... Experimental/Interventional studies ....................................................................................... Clinical trials ........................................................................................................................... Cohort Studies ............................................................................................................... Case-Control Study Design ........................................................................................... Surveys & Cross-Sectional Studies ............................................................................... Case Series .................................................................................................................... Meta-Analysis ................................................................................................................ Fishing Expedition ........................................................................................................ LEVEL OF SIGNIFICANCE & SELECTION OF SAMPLE SIZE ......................................... Power of the Study .................................................................................................................. Variation in Standard Deviation ............................................................................................. BIASES IN THE STUDY ............................................................................................................... Procedure Bias ........................................................................................................................ Recall Bias .............................................................................................................................. Insensitive Measurement bias ................................................................................................. Compliance Bias ..................................................................................................................... Selection bias .......................................................................................................................... Admission Rate Bias ............................................................................................................... APPLICATION OF STATISTICAL PROCEDURE AND TEST OF SIGNIFICANCE ........ Multiple Tests of Significance ................................................................................................ REPORT WRITING &DATA PRESENTATION ...................................................................... SENDING RESEARCH WORK/ARTICLE FOR PUBLICATION ......................................... Sources of guidelines for specific study design ...................................................................... GUIDELINES FOR REPORTING STATISTICS ...................................................................... Presentation of Results ............................................................................................................ DISCUSSION AND CONCLUSION ............................................................................................ A researcher must remember that; .......................................................................................... CONCLUSION ...............................................................................................................................

99 99 99 100 100

APPENDIX-A ETHICAL GUIDELINES FOR CLINICAL RESEARCH ....................................... INTRODUCTION .......................................................................................................................... PRINCIPALS OF ETHICS (CORE VALUES) ........................................................................... Respecting the Autonomy Of Research Subjects ................................................................... Voluntary participation ................................................................................................. Autonomy and research involving minors ..................................................................... Autonomy and Research in Schools .............................................................................. Age ................................................................................................................................. Information to subjects .................................................................................................. Exceptions from informed consent ................................................................................

108 108 108 108 108 109 110 110 110 111

101 101 101 101 101 102 102 102 102 102 103 103 103 103 103 104 104 104 104 104 104 105 105 105 106 106 106 106 107

Avoiding Harm ....................................................................................................................... Privacy and Data Protection .................................................................................................... Protecting research data and confidentiality ................................................................ Storing/ destroying research data ................................................................................. Protecting privacy in research publications ................................................................. ETHICAL REVIEW ...................................................................................................................... Ethical Review Required ........................................................................................................ Concerns of the Institutional Ethics Committee ..................................................................... Information to the research subjects ....................................................................................... Privacy and data protection .....................................................................................................

111 111 111 111 112 112 112 113 113 114

GLOSSARY ............................................................................................................................................. 115 REFERENCES ........................................................................................................................................ 127 SUBJECT INDEX ....................................................................................................................................

i

FOREWORD It is indeed very heartening and equally gratifying that Dr. S. S. Patel, Former Dean of Jawaharlal Nehru Medical College, SawangiMeghe, Wardha, and presently the Chief Coordinator of DattaMeghe Institute of Medical Sciences (Deemed to be University), has ventured in bringing out a creative, magnificent creation in the form of this Book titled “Elements of Study Designs, Clinical Biostatistics and Clinical Research.” It is imperative and inevitable to note that the objectives of medical education are very clearly crystallized globally in the international Charter that has been worked out under the joint signatures of all the countries across the Globe, which mandate that the trained Graduate and also the specialist generated thereto should turn out to be a good clinician, an effective medical teacher, and a keen researcher. The trinity of objectives, when assessed in terms of manifestation, categorically brings out that the keen researcher part of it remains in limbo for wide and varied reasons. It is in this very context one is required to take cognizance of a material reality that in spite of having the highest number of medical schools in India as compared to any other part of the world and thereby having the highest annual intake capacity for graduate and various broad specialties postgraduate courses whereby the highest quantum of trained, skilled health manpower is generated. Yet, the contribution of the said manpower in the domain of research output in the arena of medical sciences globally is minuscule in nature, bringing out the grossest possible mismatch. This research output deficit is not accidental in nature. On the contrary, it is primarily because of the impoverished orientation of the learner in the domain of research principles, research methodology, and research ethics. The emphasis which is laid down at the various levels of learning on training and orientation in respect of research principles and methodology is nonexistent in as much as it turns out to be an arena of least priority. Consequent to this prevailing situation, the learner as output is totally unequipped to dispense the research output expectations out of him in any manner whatsoever. In addition, another problem that plagues the educational scenario in medical colleges in the country is the grossest paucity of reading material in the said arena, which could be availed for teaching, training, and learning purposes. The material that is availed as of now falls short with respect to the orientation of the learner emphatically in the domain of clinical research, especially in the context of clinical Biostatistics and the arena of clarity in regard to the elements with reference to study designs. More often than not, the said arena is deciphered on the required occasions through conjectural means and modes in the absence of the required clarity. Therefore, the assistance turns out to be inconsistent with the material requirement. These vital lacunae that are operational in the system of medical education as of now need serious grappling at various levels in order to create an appropriate academic ambiance on the said count. However, the most important lacuna in regard to the paucity of the handy reading material on the said count is adequately and sumptuously dealt with by the elegant authorship that is brought out in this book. The topics compiled in the book are vital and significant in ‘general’ as a whole, but the significant aspect is the emphasis that is worked out in the embodied chapters, significantly facilitatory in an analytical understanding of the requirements thereto. The topic of study designs in ‘Medical research’ brings out an easily decipherable depiction of understanding the

ii

patterns of study design and the edifice of the nomenclature thereto, which is bound to go well with the learner in a huge way. Likewise, the chapter on scales of measurements, descriptive statistics, and research data presentation brings out handy workable operational modes and modalities on the required count that is bound to give the learner enriched with clarity on the vital areas of research methodology, enabling him to grapple with the said issues on his own and emerge out successfully in the said venture. The chapter on inferential statistics is brought out in an easy, handy, and free-flowing manner whereby the jargon depictions which otherwise baffle a learner on the said area are replaced by easily understandable modes and manners that bring out handy understanding in the learner capable of availing the same with responsibility and capacity to bring the desired outcome thereto. The depiction of statistical methods for a relationship of variables which otherwise is tough to correlate with reference to the required ambit thereto is invoked in such a lucid manner that it is bound to equip the learner with the required capacity to work out the same on his own and thereby making him free and immune from his compulsory dependence on statisticians in a blind manner. On this count, the chapter turns out to be a beacon’s light in the existing otherwise dominating darkness in the said arena of indulgence and the resultant operations thereof. The soul of inclusions in this notable book is the chapter on ‘Clinical Research’, which has brought out the expertise of the author in emphatic and self-speaking turns in a vociferous manner. He has poured the life long experience of his into this chapter to make any and every reader decipher the scope, meaning, and ambit of what is designated as ‘Clinical research’ including the manifestations that are expected out of it for the purposes of its translational accomplishments in larger societal interests. This opens genuine and wide vistas of focused understanding in the otherwise non-decipherable domain of clinical research. It is said ‘what is ethically wrong cannot be legally just and political right,’ hence the paramount need for incorporation of ethicality in research in modes, manners, operations, and outcomes. The governing ethical principles of research emanating from the ‘Helsinki Declaration’ till the guidelines notified by the Indian Council of Medical Research and their operational utilization is the hallmark of the embodied chapter on Ethics in Clinical Research. The vital aspect of research methodology and its resultant outcome in terms of publications and potentization also mandates that the research scholar has to be adequately equipped and armed with the much-needed armory of being an efficacious planner and writer of research projects seeking much-needed funding for the said purpose by the various statutory funding agencies including the non-governmental organizations that carter to the said cause. It categorically brings out that writing research projects is not ‘an Art’ but a definitive science mandating a required approach, commensurate structuring, needed inclusions, and depictions in regard to operations, outcome, expected translator component and the operational modes and manners thereto. Speaking for the special feature of the book from beginning to end, the only expression at my disposal is that it is a ‘treatise’ on ‘elements of study design, clinical biostatistics and clinical research’ which is made easy, handy, free-flowing decipherable and usable by one and all independent of the status or the level of the user. As such, the creative writing that the author has invoked is bound to be of immense use to the user from the point of view of clarity of understanding, utility, and purpose fulfillment, but it is bound to fill in a huge void that remains unfilled all these years for wide and varied reasons. The impact it generates is unending and would turn out to be an illuminating source of light devouring the prevalent

iii

darkness in its entirety in the true spirit of the Vedic hymn depicting ‘Tamaso Ma Jyotirgamay.’ I would be failing in my duty if I did not wholeheartedly salute the diligent effort of the author whom I have known for over three decades and have always found inspiring as a mentor to all concerned. But the mentorship that is bound to be generated out of this creative manifestation of his in this book leads him to a state of ‘immortality’ which otherwise is the toughest and most difficult of the attainments in human life.

Vedprakash Mishra Datta Meghe Institute of Higher Education and Research, Sawangi (Meghe), Wardha, Nagpur, Maharashtra, India

iv

PREFACE The word statistics is derived from the Latin word ‘Status’, meaning position or standing. As the skills of statistics were used by the tax assessor to assess the ‘Position’ or ‘Standing’ in society, i.e., the assessment of the assets of an individual. Presently the term statistics is used for analyzing data and taking out inferences. The theory of statistics was developed by mathematicians to predict something based on previous observations or information they proved through mathematical equations. Statistics are used even in our day-to-day activities by way of taking out averages, means, and ranges. In the field of sports, statistics are increasingly used for making predictions and in administrative decision-making. Cricket is one such example wherein statistical analysis provides strong support in team selection. Statistical formulae, if applied intelligently, can assist all types of decision-making. For example: Calculating the number of beds required in the Intensive Care Unit of a hospital or the number of ambulances needed in the Casualty section of a hospital. Health science researchers and clinicians who make clinical decisions based on available research inputs have got to have sound knowledge of what is known as Medical Biostatistics or Clinical Biostatistics. One of the most important uses of biostatics is in epidemiological studies. The term medical epidemiology refers to the study of health and diseases in the human population. Health professionals, especially clinicians and medical researchers, must know whether the published information is worth utilizing in decision-making. The editors of journals usually screen the research articles from the point of view of organization and analysis of the research material. Their focus is on the contents rather than methodology, including the suitability of applying the statistical tools. As such, expert statistical consultation should be sought during the planning of any research project with clarity of the research question to be answered. There is good documentation of problems with medical research articles. Williamson et al. (1992) remarked that out of 4200 medical studies published in 30 Journals, which included the British Medical Journal, Journal of the American Medical Association, New England Journal of Medicine, Canadian Medical Association Journal, and Lancet, only about 20% articles met the assessor's criteria for validity. Practitioners in medicine read the research articles to apply the results of research inpatient care for diagnostic and therapeutic purposes. The medical researchers review the research articles to find out the research gap and design a study to fill the research gap. A sound understanding of research methods and biostatics is indispensable for the interpretation of information about drugs, equipment, and diagnostic procedures, for evaluating study protocols and research articles and for participation in research projects. It is very important for a researcher, specifically in health sciences, to have skills to interpret to publish results and can make decisions on whether the statistical tests are used properly. This title is intended to be used by researchers in health sciences, academicians, and clinicians to make them understand the basic concept of research and biostatistics so that they can make effective and efficient use in their respective fields of their indulgence. The researcher is expected to know only which study design is best suited to answer his/her research question and which statistical test is appropriate to draw reliable inferences from the

v

research data. Although a good number of books are available on Biostatistics, most of them deal with core or classical parts of statistics and their application procedures in quite detail. The medical researcher finds it difficult to select the contents of his requirement suited for the application. In this work, details have been curtailed based on my experience as a learner, guide, and supervisor of research in Medicine, Dental Sciences, and the Indian System of Medicine. I hope this book will serve the intended purpose.

CONSENT FOR PUBLICATION Not applicable.

CONFLICT OF INTEREST The author declares no conflict of interest, financial or otherwise.

Satyawan Singh Patel Centre of Excellence for Preclinical Safety & Efficacy Studies Interdisciplinary Research and Therapy Evaluation DattaMeghe Institute of Medical Sciences, (Deemed to be University) NAAC Accredited Sawangi(Meghe),DMIMS (DU) Nagpur, India

vi

ACKNOWLEDGEMENTS At the outset, I would like to acknowledge all my Post Graduate Scholars and Clinical Researchers who inspired me to undertake this challenging task. I would like to thank Prof.Ved Prakash Mishra, Pro-Chancellor & Chief Advisor, DMIMS (DU), took painstaking efforts in the creation of an environment in DMIMS (DU), which stimulated and induced me to undertake this project. His constructive critical appraisal enhanced the quality of this work. I thank, from the core of my heart, my wife, Mrs.Sheela Patel, for the care, cooperation, and encouragement rendered by her and for staying with me through all thicks and thins of life. I will never forget the immense contribution and all the assistance I have received from Mr.Manish Deshmukh, Asst. Professor, in completion of this work. I would also like to express my gratefulness and thanks to Dr. Shraddha Patel, Assistant Professor, Department of Oral Diagnosis & Radiology, SPDC, Sawangi (Meghe), Wardha, whom I am fortunate to have as my Daughter-In-Law, for her incessant assistance and technological support at every stage of this work. I would like to acknowledge the kind cooperation and valuable help extended to me by Mr.AshishTambe, Mr.Suresh Kumar, Stenographers, Mr. AniketPhathade, Consultant, R&D, for his cooperation in the preparation of graphics and Mr. Vrushabh Jain, Personal Assistant for all the typographical assistance they provided during the process. Satyawan Singh Patel Date: 02.03.2020

Elements Of Clinical Study Design, Biostatistics & Research, 2023, 1-13

1

CHAPTER 1

Study Designs in Clinical Research Abstract: Study designs in medical research fall mainly into three categories: Observational studies, Interventional studies & meta-analyses. Further, each type of study comprises of subtypes. Each study design with its subtypes, and applicability of a specific study design, along with advantages & limitations, are discussed in this chapter.

Keywords: Blinding, Case series, Case-control studies, Cross-sectional studies, Cohort, Clinical studies, Controlled studies, Interventional study, Meta-analysis, Observational study. INTRODUCTION Research in general & medical research, in particular, is to be conducted to answer mainly three questions 1) Why do the researchers want to do the research, i.e., research gap? 2) What the researcher wants to do, i.e., the research question, 3) What will be the suitable study design necessary to answer the research question, and 4) the investigator has to decide how the research should be conducted? meaning thereby selecting the appropriate methodology to be used. Thus, the choice of study design is a pivotal factor in finding a correct answer. This chapter deals with commonly used study designs in medical research, as knowing or familiarizing with the design of the study is very important for understanding the conclusions drawn from the research. Understanding of Research/ Study designs will be facilitated by understanding the term “Research” from a statistical point of view. Research means a systemic way of Data Collection, analyzing, and drawing conclusions that create new knowledge. This process requires that the research plan (study design) developed by the researcher should answer the research question appropriately. Whereas Research Methodology refers to the detailed process of data collection, the tools, and the technique of Data collection [1].

S.S.Patel & Aditya Patel All rights reserved-© 2023 Bentham Science Publishers

2 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

FIG.1.1 STUDY DESIGNS IN CLINICAL RESEARCH

1

2

Observational Studies

3 Meta Analysis

Interventional Studies Animal Studies

Human Studies (Clinical Trials)

Descriptive or case series

Uncontrolled

Controlled

Case Control Studies (Retrospective)

Concurrent Control

Non Concurrent Control

Cross Sectional Studies Surveys Prevalence

External Control Randomized

Non-Randomized

Blind

Open

Cohort Studies

Historical Control

Use of result of another investigator Single Blind

Double Blind

Prospective Amphispective Historical Cohort [Retrospective)

...4..

Fig. (1.1). Study Design In Clinical Research.

Classification of Study Designs: Medical research can be classified broadly into three categories (Fig. 1.1). 1. The observational studies wherein the subjects are observed, and no specific intervention is provided. 2. The interventional studies where a specified intervention's effect is observed. 3. Meta-Analysis. OBSERVATIONAL STUDIES Case Series It is the simplest of study designs in which the researcher reports interesting and curiosity-provoking observations in a small number of patient cases. Case series

Study Designs

Elements Of Clinical Study Design, Biostatistics & Research 3

studies are suitable for the generation of hypotheses. The hypotheses generated may be investigated further in case-control, cross-sectional, or cohort study designs. The power of evidence in the Case series is weak. Case-Control Studies These are the commonly used observational retrospective studies wherein two groups, One having an outcome and the other without an outcome, are analyzed (Fig. 1.2). These studies differ from the case series by the presence or absence of a control group. The patient in case-control studies is selected based on some disease or outcome. These are the commonly used observational retrospective studies wherein two groups are analyzed, One having an outcome and the other without an outcome. The controls are the patients or individuals without the disease or outcome. These studies begin with the presence or absence of outcome and then look backward in time to detect the patient's possible causes or risk factors. The case-control studies are retrospective studies. Therefore, sometimes it becomes difficult to decide whether the reported study is a case-control study or a case series report. The confusion is because both types of studies are conceived and reported after the instances have happened [1, 2] (Table.1.1).

Fig. (1.2). Case Control Studies. Table 1.1. Case Contro studies Advantages & Disadvantages. Sr. No.

Advantages

Limitations

1)

Useful to study rare conditions or rare diseases that may not occur for many years.

Case-control studies, in general, have a lower power of evidence as compared to cohort studies.

2)

Can be completed in a shorter time than cohort studies

A good number of biasing factors can play a role

4 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

(Table 1.1) cont.....

Sr. No.

Advantages

Limitations

3)

Comparatively less expensive

Retrospective study and conclusions can be affected by the number of uncontrolled variables

Cross-Sectional Studies, Surveys Cross-sectional studies are also called epidemiology studies and prevalence studies. In general, a cross-sectional study is an observational study in which the data is collected from a sample at a specific point in time and analyzed. These studies are designed to answer the question “what is happening presently' whereas the case-control studies are designed to ask the question “what happened.” Methods of Survey There are two methods: By Self-Administered Questionnaire The administration of the questionnaire could be in person or via email. Interviews Interviews could be in person or over Multi-Activity Device (MAD) or cell phone. Advantages and Disadvantages of Survey Methods There are advantages and disadvantages to both methods. The selection of the method depends on the research question and the nature of the respondents for the survey. The research question should be developed appropriately. The question could be open-ended or close-ended. Then the researcher has to decide the scale of measurement regarding the format of the questionnaire; there are no hard and fast rules; however, some guidelines suggested are: a. The instructions should be placed wherever they are needed. b. Even on the cast of repetition, it is better to place these guidelines on the top of the continuation page. c. The branching question should be avoided. d. Directional arrow and other visual guides should be given for the assessments of the subject.

Study Designs

Elements Of Clinical Study Design, Biostatistics & Research 5

e. Usually, the easier question should be placed first, and the questions should be placed in a logical order so that the questions about the same topic are in continuation. f. It is better to place the demographic question at the end and a separate section mentioning this section. g. The next group of questions is about use. It is better to use an existing and pre-validated questionnaire in a survey. If a new survey instrument is developed, the survey instrument is to be pilot tested for its reliability and validity [3]. Reliability The researchers should use the existing pre-validated questionnaires or survey instruments as far as possible. This saves time and effort in developing the questionnaire and then establishing the reliability and validity of the survey instrument. The term reliability refers to the reproducibility of the finding of the instrument on repeated administration to the same subject. There are five types of Reliability: • Test Re-Test The responses are stable after repeated administration. This can be measured by administration twice or more to the same subjects. • Internal Consistency Internal consistency refers to an agreement among items, i.e., the various items measure the same thing. This can be measured by Cronbach's alpha test, i.e., average correlation. • Alternative form Alternative form means the measurement of the same topic by different items. This can be decided by the correlation of a square between the items. • Intra-Observer Consistency It is measured by the agreement between the different observations made by the same observer. It is read by McNemar Statistics.

6 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

• Inter Observer Consistency: It is measured by the agreement between the different observers. The significance can be decided by kappa (k) statistics. Validity Validity is defined as an instrument's property that indicates the instrument's capability and how well it measures the characteristic (variable). The measure of estimating the validity are face validity, Content validity, Criterion validity, and Construct validity (FC3). • Face Validity Face validity is the degree to which a questionnaire or a test appears to measure what is supposed to be measured. • Content Validity Content validity indicates the degree to which the items on the instrument represent the knowledge of the characteristic being investigated. • Criterion Validity Criterion validity is the instrument's capacity to predict an associated characteristic. For example, an instrument is developed to measure the quality of life; thus, the score should be comparable with the physical examination and the patient's subjective feelings. • Construct Validity It is very difficult to define Construct validity. It demonstrates that the instrument is related to other instruments assessed at the same time. It is not related to the instruments which measure the other characteristic. It requires the administration of many tests or instruments on the same group of individuals and then evaluating the pattern of the relationship. Pilot Testing Pilot testing of a new survey instrument is a very important part of the success of any survey. The pilot testing is carried out after the development of the question. It should be done before the questionnaire is prepared for large-scale administration. The ideal response rate in a survey has been prescribed that should

Study Designs

Elements Of Clinical Study Design, Biostatistics & Research 7

be between 52-85%. It has been mentioned by the authors of various surveys that pre-notification to the concerned sample increases the response rate by 7-8%. The provision of monitoring incentives increases the response rate by 16-30%. The most requirement is that confidentiality must be guaranteed. Estimation Of The Sample Size Depends On The Research Question: The sampling method could be simple random sampling or stratified random sampling. The sampling methods described later can be random sampling, stratified random sampling, or random cluster sampling. Analysis Of Survey Results Various authors have analyzed their reports. They have used, depending on their research question. The most commonly used statistical test is Logistic Regression. If the data of the observation is complex, the weightage system for the groups or individuals must be assigned. Cohort Studies Prospective Cohort A cohort is defined as a group of people (subjects) who have some common features and remain in the study group over a period of time (Table. 1.2). A cohort study is a prospective observational study involving two groups of subjects. One exposed to some agent or has some risk factor, and the second group of a subject who does not have that risk factor nor is exposed to the agent. Cohort studies are directed forward in a time direction as the question asked in the cohort study is “what will happen” (Fig. 1.3) [1, 2]. Historical Cohort or Retrospective Cohort Study In a historical cohort study, the information collected in the past and kept in records is utilized. This study design is only possible if the records of the patients are complete in all respect, including the follow-up of patients. In the historical cohort, information collected in the past and kept on record is analyzed to answer the question of what happened to the two groups. In India, as patient record is not maintained, in most of the hospitals and clinics, as it is expected to be maintained, this study design should be avoided.

8 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

Fig. (1.3). Cohort Flow Chart.

Amphispective Cohort Studies It is a mixed cohort study in which retrospective (Previous Patient Record) and prospective observations are utilized for analysis. Table 1.2. Cohort Studies Advantages and Limitations. Sr. No.

Advantages

Limitations

1)

Prospective studies have a comparatively higher power of evidence as compared to a case-control study

More time taking as compared to a casecontrol study

2)

Biasing factors can be minimized

More expensive as compared to a casecontrol study

Study Designs

Elements Of Clinical Study Design, Biostatistics & Research 9

(Table 1.2) cont.....

Sr. No. 3)

Advantages

Limitations

Not useful to study rare conditions or A cohort study can be undertaken for confirmation of diseases that may not occur over a period of the inferences that emerged from a case-control study time

INTERVENTIONAL STUDIES (EXPERIMENTAL STUDIES) Interventional studies can be animal studies or human studies. These could be with control or without control. Animal Studies Human Studies/Clinical Trials Interventional studies in medicine that are conducted in humans are called Clinical trials (Table 1.3). The clinical trials could be of two types, i.e., controlled and uncontrolled. Controlled Trials The experimental procedure, treatment, or drug is compared with another drug or procedure. The control could be a reference, standard control, or it could be a placebo. Concurrent Control In this study design, there are two groups of subjects, out of which one group receives the experimental drug and the other group receives the placebo, other drugs, or other standard procedures; this group is called the Control group. Both the groups, i.e., Interventional and Control, are treated alike in all respect except for the procedure or drug treatment so that any difference between the two groups, if found, would be due to the intervention and will not be on account of other factors as both the groups are treated at the same time. This study design is called Concurrent Control. Further, the concurrent control could be randomized or non-randomized.

10 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

Randomized and Non-randomized Control The group to subject or patient is allocated or assigned by a random allocation through a computer-generated random allocation table. Whereas in the nonrandomized study, the researcher allocates the subject to the control or experimental group. • Randomized clinical trials could be of two types: i.Blind ii.Open The blind study could be again of two sub-types-Single blind and Double-blind. • Single Blindrandomized Clinical The patient does not know whether he is receiving the experimental drug, standard drug, or a placebo. Thus this eliminates the psychodynamic part of the patient (Patient Bias), and the changes which are observed in clinical parameters are because of the experimental drug, not because of the psychological element of the patient. However, the single-blind study does not eliminate the chances that the researchers will see what they expect to see. • Randomized Double-Blind Clinical Trial Double-Blind studies are designed to eliminate the Researcher's Bias. In DoubleBlind studies, neither the researcher nor the subject knows which treatment is given to which group of patients. In this study, a third independent person records the observation of both groups of the subject,i.e., those of the control and international group. This study design can be called an ideal or model study design. It's a quality study design with high validity having level 1 power of evidence [3]. Non-randomized Control The studies that do not use randomized assignments are called non-randomized trials. Sometimes they are referred to as simply clinical trials or comparative studies. If there is no mention of randomization, it is to be taken as a nonrandomized trial. It is believed that studies with non-randomized control studies suffer from many sources of biases; therefore, conclusions drawn from them are questionable.

Study Designs

Elements Of Clinical Study Design, Biostatistics & Research 11

Non-Concurrent Control Study In this study design, the interventional group and the controlled group are not treated simultaneously. Non-concurrent control trials could be of two types: External Control or Historical Control In external control, the subjects/patients used as control are external to the study. The investigator may use the results of other investigators' research. Historical Control: The controls are the patients whom the researchers have treated in another manner. The historical controls are used to study diseases for which a cure does not exist. For example, many cancers are used as historical controls. The investigator should evaluate whether the other factors might have changed since the time historical controls were treated. Self Control or Self Control Study Design The same group of subjects is used for experimental and control options. This type of study uses patients as their own controls, and therefore it is called a selfcontrol study. The modified self-control study is called a crossover study. It uses a combination of concurrent and self-control groups. One group is assigned to the experimental treatment, and the other group is assigned to the control treatment. After a time, the experimental treatment and placebo are withdrawn from both groups for a washout period or for a time called the washout period. During the washout period, the patients and controls received no treatment. After the washout period, the groups are crossover, i.e., the Control group, which did not receive the experimental drug, is now assigned to the experimental drug/treatment, and the group that did receive the experimental drugs is now given the placebo. If these studies are used appropriately, they themselves are powerful study designs. Uncontrolled Studies The clinical studies in which the investigator has only one group, i.e., the interventional group, and there is no control group for comparison, are called uncontrolled studies. Strictly speaking, they are not experiments or trials. In such studies, a group of patients subjected to a procedure is followed for the long term. The major disadvantage of such studies is that the investigators assume that the

12 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

treatment/ procedure used has resulted in indifference. But there is no scientific evidence for the same [4]. Table 1.3. Advantages & Limitations of Clinical Trials. Sr. No. 1

2

Advantage

Limitations

The randomized, double-blind clinical trial is the A long duration of time is required for the gold standard against which other study designs conduction of the clinical trials, and the expensed are compared A clinical trial is the best study design wherein the objective has established the efficacy of treatments in a procedure.

Sometimes ethical issues arise of depriving the interventional group of the best standard care is established drug or treatment.

META-ANALYSIS Meta-analysis is a study design that does not fit either in the observational or experimental studies category. A meta-analysis combines published information from other studies for an overall conclusion. Meta-analysis is like review articles, but it also includes a quantitative assessment and a summary of findings. The meta-analysis can be done in observational studies as well as in interventional studies. However, the meta-analysis should report the findings of these two studies separately. The meta-analysis is suitable when the studies reported have small numbers of subjects or their reported conclusions are different [5, 6]. CONCLUSION Study Designs in medical research fall into three categories: 1) Observational studies in which the subjects are observed. They are further categorized into the following types: a. Case Series b. Case-control studies (Retrospective) c. Cross-sectional studies d. Cohort studies 2) Interventional studies in which the effect of a specific intervention is observed. They can be animal studies or Human studies.

Study Designs

Elements Of Clinical Study Design, Biostatistics & Research 13

Human studies can further be categorized as: a. Controlled studies: Generate a higher level of evidence. b. Non-Controlled studies: Although simpler to conduct, the level of evidence generated is lower. 3) Meta-Analysis: Data from other published eligible studies are combined, and overall conclusions are drawn.

14

Elements Of Clinical Study Design, Biostatistics & Research, 2023, 14-30

CHAPTER 2

Scales of Measurement, Descriptive Statistics & Data Presentation Abstract: The scales of measurement could be numerical, nominal or ordinal. The measurement of data requires measurement of its central tendency and discussion. Such data can be presented, depending on its type & nature, in the form of tables & figures. The different types of descriptive statistics & the suitable form of presentation are discussed in this chapter.

Keywords: Graphs, Histogram, Mean, Mode, Median, Nominal scale, Numerical scale, Ordinal scale, Pictogram, Proportion, Range, Standard deviation. INTRODUCTION It has been aptly said that in research, what you say if you can measure it & express it in numbers; your information has some value. If you cannot, your information is of meager value. This chapter deals with different kinds of data collected in medical research. What is important is the type of observation & the kind of scale on which it is measured. Depending on the data type and its appropriate measurement scale, deciding which statistics are to be used for the summarization of the data is called descriptive statistics. Further, the data is to be presented in suitable tables & figures to facilitate understanding the conclusions drawn from the research [7]. SCALES OF MEASUREMENT Commonly used scales of measurement in medicine are: I. II. III. IV.

Nominal Scale Ordinal Scale Interval Scale Ratio Scale S.S.Patel & Aditya Patel All rights reserved-© 2023 Bentham Science Publishers

Scales of Measurement

Elements Of Clinical Study Design, Biostatistics & Research 15

Nominal (Categorical Scale) Nominal scales are used for qualitative classification. They can only determine whether the individual items belong to distinct categories. Quantifying or ranking the categories in order is not possible on this scale. Performing arithmetic or logical operations on nominal data is also not possible. Nominal variables are also called (non-ranked) categorical. The number of occurrences in each category is referred to as the frequency count for that category. The category is often dichotomous, i.e., binary, where there are only two possibilities. Variables that have only two categories or levels, either the outcome occurs or does not occur. For example, a disease is cured by a drug or not cured. The evaluation of medical treatment or surgical procedure as well as the presence of possible risk factors, are often described as either occurring or not occurring. Ordinal Scale In the ordinal scale, inherent order occurs among the categories but is artificially converted into numbers. Ordinal data have order, but the intervals between scale points may be uneven. Due to the lack of equal distances, arithmetic operations are not possible, but logical operations can be performed. In clinical practice, ordinal scales are used to determine the number of risk factors or the appropriate type of therapy. For example, the stages of carcinoma cervix from stage 0 to 4 is an ordinal scale. Similarly, other carcinomas, like breast cancer, are also staged. Other examples of ordinal variables might include: determining the pain level of a patient (1-10 scale), Satisfaction level (very dissatisfied, dissatisfied, neutral, satisfied, very satisfied), etc. Interval Scale These metric scales have constant, equal distances between values, but the zero point is arbitrary. They can be measured on a Linear scale, and the intervals keep the same importance throughout the scale. This scale is also characterized by the fact that the number zero is an existing variable. In the ordinal scale, zero means that the data does not exist. In the interval scale, zero has meaning – for example, if you measure degrees, zero has a temperature. Interval scales can also be used to keep counts of publications or citations, Intelligence Quotient (IQ test score), Body Mass Index, age (years), etc.

16 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

Ratio Scale Ratio scales are metric scales and can be the most informative scales. Ratio scales also differ from interval scales in that the scale has a 'true zero'. The number zero means that the data has no value point. An example of this is height or weight, as someone cannot be zero inches tall or weigh zero pounds, or be of negative inches or negative pounds. Examples of Ratio scales include weight, pulse rate, respiratory rate, body height, etc. The data is nominal and is defined by identity. It can be classified in order, contains intervals, and can be broken down into exact values. MEASUREMENT OF CENTRAL TENDENCY The summary of data in a research study communicates a lot of information. One of the most useful summaries of numbers is an indicator of the center of distribution of the observation. This summary is called “Central Tendency.” Three measures of central tendency are used in medicine and epidemiology. These three measures are: ● ● ●

Mean Median Mode

Calculation of Measures of Central Tendency Mean The Mean is the arithmetic average of the observation. It is denoted by (called X Bar). It is calculated by adding the value of all observations and dividing it by the total number of observations. The formula for Mean is written as = ∑X/n, where Denotes mean, ∑(Sigma) means to add, “X” denotes individual observations, and 'n' denotes the number of observations. For example (Table 2.1). If the recorded heart rate of 18 patients is as under: Table 2.1. Systolic blood pressure. Subject ID

Systolic Blood Pressure (in mmHg)

A

139

B

151

C

201

Scales of Measurement

Elements Of Clinical Study Design, Biostatistics & Research 17

(Table 2.1) cont.....

Subject ID

Systolic Blood Pressure (in mmHg)

D

170

E

123

F

121

G

119

H

100

I

164

J

161

K

164

L

138

M

118

N

130

O

109

P

92

Q

126

R

139

̅ = ∑X = X n

139+151+201+⋯+139 18

= 136.94

The Mean is equal to 136.94. Another mean is known as the geometrical Mean. For the calculation of the geometrical Mean, each observation is squared, and the Mean is calculated. In medicine, the Geocentrical Mean is very rarely used and hence not described. Median If the observations are arranged in ascending or descending order, the Median is the middle observation,i.e., the point at which half the observations are smaller, and half are larger. The Median is indicated by M or Md. The procedure for calculating the Median is as follows: o Arrange the observation serially from smallest to largest or Vice Versa. o Count to find the middle value. The Median is the middle value of an odd number; if the observations are in an even number, then the Mean of the two middle values becomes the Median.

18 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

For example: 92, 100, 109, 118, 119, 121, 123, 126, 130, 138, 139, 139, 151, 161, 164, 164, 170, 201., the total observations are 18. The middle two are 130 and 138. The Median comes out to be 134. When the numbers are arranged serially as the total patients are even-numbered therefore mean of the two middles is to be calculated. The two middle numbers are 130 and 138; the sum is 268/2 = 134. Mode The Mode is called the most fashionable number,i.e., and the Mode is the value that occurs most frequently. If no single observation occurs most frequently, then the data set can have more than one Mode. In the above example, the Mode is 139 and 164. Uses of Central Tendency To decide which central tendency best suits a particular set of observations, two factors are important, namely: On the scale of measurement (numerical or ordinal), the two shapes of the observed distribution, i.e., .whether a distribution is symmetrical about the Mean or skewed to the left or right. A Symmetric distribution has the same shape on both sides of the Mean. In a skewed distribution, if the outlying values are smaller, then the distribution is said to be skewed to the left, and if the outlying values are larger, the distribution is skewed to the right. Sometimes Bimodal distributions are observed because of some underlying phenomena. For example, the number of patients with respiratory tract infections reporting to a hospital in a year follows a bimodal distribution since people tend to develop respiratory infections mostly in the rainy and winter season. The following facts help us to know the shape of distribution without plotting: • In the said example, as the Median and mean are different; therefore the data is not symmetric. • If the Mean is larger than the Median, then the distribution is skewed to the right. • If the Mean is smaller than the Median, the distribution is skewed to the left.

Scales of Measurement

Elements Of Clinical Study Design, Biostatistics & Research 19

• In the case of the above example data, as the Mean is larger (136) than the Median (134), the distribution is skewed to the right. GUIDELINES TO DECIDE WHICH TENDENCY IS BEST TO BE USED

MEASURE

OF

CENTRAL

Mean The Mean is used for numerical data, wherein the distribution is symmetric (not skewed). Median The Median is used for ordinal data or numerical data wherein the distribution is skewed. Mode The Mode is to be used wherein the distribution of data is Bimodal. Geometrical Mean The geometrical mean is generally used for observations measured on a logarithmic scale. MEASURES OF SPREAD The spread is also known as Dispersion. The information provided by the central tendency becomes more meaningful with the expression of the spread. The measures of spread are as under: The Range The range is the difference between the largest and the smallest observation. It is determined after arranging the data in rank order. Sometimes range may be expressed in terms of minimum and maximum values instead of using the term range. In the previous example, the systolic blood pressure ranges from 92 to 201, or a minimum of 92 and a maximum of 201. Standard Deviation The standard deviation measures the spread of data about their Mean. Standard

20 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

deviation is the most commonly used measure of Dispersion in medical and health data. The standard deviation is symbolized as SD. The Formula is̅ )𝟐 Ʃ(𝐗−𝐗

SD =√

𝐧−𝟏

∑- represents sum X- Individual Observed readings is the Mean of observations n- represents the number of observations The Standard Deviation is a very important statistic because: a)It is an essential part of many statistical tests. b) It is very useful in describing the spread of observations about the mean values. Low standard deviation means data are clustered around the Mean, and a high standard deviation indicates data are more spread out. A standard deviation close to zero indicates that data points are close to the Mean, whereas a high or low standard deviation indicates data points are respectively above or below the Mean. Rules of thumb when using the standard deviation are: I. If the distribution of observation is well shared, 67% of observations lie between the Mean plus/minus 1 of standard deviation. II. Regardless of the data distribution, 95% of the value always lies between the Mean plus/minus 2 of standard deviation. III. 99.7% of the observation lies between the mean plus/minus 3 standard deviations. The standard deviation can be calculated easily with the help of a computer using the computational formula. ❍

Co-efficient of Variation (CV) It is a useful measure of the relative spread of data. It is frequently used in biological sciences. The coefficient of variation can be defined as the standard deviation divided by the Mean multiplied by 100. If the standard deviation is

Scales of Measurement

Elements Of Clinical Study Design, Biostatistics & Research 21

represented by SD and the Mean by 'x'. The formula for the coefficient of variation is: 𝐶𝑉 =

SD X

x 100

Percentile Percentile denotes the location of data (Parameters) in terms of percentage. The data is arranged in ascending order to decide the percentile. The 50th percentile means that 50% of the data is above that value and 50% is below. 20th percentile denotes that 20% value is below 20th and 80% is above. Quartile means that the values are divided into four parts, wherein Q1 is known as the first quartile; the 25th percentile will fall below 25%, and 75% will be above. In the notation of quartiles Q2, 50th percentile reflects that the 50% value will fall below and 50% above. The third quartile is the 75th percentile means that 75% of values will fall below 75%, and 25% will lay above that value. The percentile can be presented in a graphical form also. Usually, the percentile is used in demographic studies (Fig. 2.1).

Fig. (2.1). Percentile graph.

An institution conducted an entrance examination for admission to the MBBS

22 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

course. Ten thousand candidates appeared in the examination; the annual intake capacity of the institution was 1000. The administration of the institution decided to call 4 and ½ times more candidates, then the intake capacity in the percentile graph was plated. The 50th percentile is the Median in the graph and comes out to be 40% marks according to the graph candidate scoring 40% in above or 4000 and 500; thereforth 50th percentile comes out to be 40% accordingly candidate scoring 40% in above marks are to be called for interview [4]. GUIDELINES FOR USING DIFFERENT MEASURES OF DISPERSION i. Standard deviation is used for symmetric numerical data along with Mean. ii. Percentile is used in two conditions: When the Median is used with ordinal data or with skewed numerical data. When the Mean is used, the objective is to compare individual observation with a set of norms. iii. The range is used with numerical data when the purpose is to emphasize extreme values. iv. The coefficient of variation is used for the comparison of distribution on a different scale. ❍ ❍

PRESENTATION OF DATA IN TABLES AND GRAPHS The commonly used methods of presenting data are as under: i. ii. iii. iv. v. vi.

Tables Charts and Diagrams Histogram Graphs Pie Diagram or Pie Charts Pictogram

Tables Tables could be of two types: a. Simple Table (Tables 2.2 & 2.3) b. Complex or Frequency Distribution Tables A table could be a simple or complex table, depending upon the number of observations of a single set or multiple sets. Irrespective of whether the tables are simple or complex, certain principles should be applied while designing a table:

Scales of Measurement

Elements Of Clinical Study Design, Biostatistics & Research 23

a. A table should be numbered, i.e., Table no. 1, Table no. 2, etc. b. Each table should be given a title, and the title should be self-explanatory and brief. c. The headings of the column and rows should be clear and concise. d. The data must be presented according to the size and importance; the order of arrangement can be chronological, alphabetical, or in any other way suitable to the data. e. The table should not be too large. f. Footnotes are to be given wherever necessary to provide an explanatory note or additional information. Simple Table Table 2.2. The population of some states in India. States

Population 1st March 2011

Andhra Pradesh

8,46,65,533

Bihar

10,38,04,637

Madhya Pradesh

7,25,97,565

Uttar Pradesh Source: Census of India, 2011.

19,95,81,477

Table 2.3. Population of IndiaYearwise. Year

Population

1901

2,38,396.000

1921

2,51,321.000

1981

6,85,185.000

1991

8,43,930.000

2001

10,27,015.247

2011

12,10,193.422

Frequency Distribution Table The frequency distribution table divides the data into class intervals (convenient groups) and the number of items (frequency) that occur. Each group is shown in the adjacent column. For example (Table 2.4) there is a group of patients suffering from poliomyelitis. The age of the admitted patients is stated below. 8,24,18,5,6,12,4,3,3,2,3,23,9,18,16,1,2,3,5,11,13,15,9,11,11,7,10,6,9,5,16,20,4,3,3 ,3,10,3,2,1,6,9,3,7,1,4,8,1,4,6,4,15,22,2,1,4,7,1,12,3,23,4,19,6,2,2,4,14,2,2,21,3,2, 9,3,2,1,7,19

24 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

The frequency table is constructed as under: Table 2.4.. Frequency Distribution Table. The data given above may be conveniently analyzed as shown below: Class Interval

Age Group (in years)

Frequency

1

0-4

35

2

5-9

18

3

10-14

11

4

15-19

8

5

20-24

6

In the above example, the age has been split into a group of five. This group of five is known as the class interval. The number of observations in each group is called frequency. Charts and Diagrams The charts and diagrams are used to present simple statistical data. Bar Charts: a. Simple bar chart b. Multiple Bar Chart or c. Component Bar Chart Simple Bar Chart The bar charts have a strong impact on the imagination. The length of the bar should be proportional to the magnitude of the information presented in the form of the bar. Charts are easy to prepare and easy for visual comparison of the information. The bars may be vertical or horizontal (Table 2.5). The bar should be separated by appropriate space. A suitable scale should be chosen to present the length of the bar.

Scales of Measurement

Elements Of Clinical Study Design, Biostatistics & Research 25

Multiple Bar Chart Table 2.5. Simple Bar Chart.

Mean age Marriage (Females) in some countries India

16

U.A.E.

18

Sri…

20

Israel

22

Japan

24 0

5

10

15

20

25

30

Age In multiple bar charts, two or more bars can be grouped together (Table 2.6). Table 2.6. Multiple Bar Chart Population and Land are by Region.

70 60 50 40 30 20 10 0

58

20

24 14 5

Asia

10

19

18 8

7

18 6

2

8

Latin North Oceani Europe Africa Americ USSR Americ a a a

Population

58

14

10

8

7

6

2

Land

20

5

24

18

19

18

8

Population Land

26 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

Component Bar Chart If the information to be presented is required to be divided into two or more parts. The component bar charts are chosen. Each part representing a certain item should be proportional to the magnitude of that particular item (Fig. 2.2). For example, the following component bar chart expresses a census of the Indian population (in millions) which was carried out over a 10-year interval from 1901 to 2011. The total height of the bar denotes the total population, and the red portion denotes growth.

Fig. (2.2). Component Bar Chart iii) Histogram.

Histogram The Histogram is a pictorial diagram of a frequency distribution. It is constructed by a series of blocks. The blocks are adjacent to each other. The class intervals are plotted along the “X” Axis (Horizontal Axis), and the frequency is plotted along the “Y” Axis (Vertical). The area of each block (the rectangle is proportional to frequency). The Histogram represents numerical data, whereas the bar graph represents categorical data. In Histogram, there is no gap between the bars.

Scales of Measurement

Elements Of Clinical Study Design, Biostatistics & Research 27

For example, a study is done on a sample population of age between 20 to 80 years. The whole study sample is divided into 7 class intervals (each of a 10-year span). The class interval is plotted against the X-axis, and the frequency of age distribution is plotted against Y-axis; we get a Histogram as follows. The Histogram can be symmetrical (if it's divided into two equal halves) or Asymmetrical (i.e., skewed towards the right or left) Frequency Polygon Another method of depicting frequency distribution is constructing a frequency polygon. It is constructed by joining the midpoints of the histogram block. An example is a frequency polygon of reading systolic blood pressure in the community. Graphs and Line Diagram Line diagrams and graphs are used to show the trend of the relationship between two variables. Out of these two variables, one is called an independent variable, and the other is called a dependent variable. For example, an administration of an antihypertensive drug, say, hydrochlorothiazide will bring about falling systolic and diastolic B.P. in a dose-dependent manner. In this example, the dose is an independent variable, whereas a fall in blood pressure is a dependent variable. In plotting the diagram or the graph, it is conventional to plot the independent variable along the “X” Axis, whereas the dependent variable is plotted along the “Y” Axis. In another example: the growth in the Indian population is depicted in a time-dependent manner. Here the time is an independent variable, whereas the population trend is a dependent variable, as per the data provided in Table 2.1. Pie Diagram or Pie Charts Pie diagrams or pie charts are useful to depict the distribution of some characteristics in a study population (Fig. 2.3). The proportions depicted in the pie chart are in the proportion expressed in terms of percentage, wherein 100% is equivalent to 3600. Accordingly, the diagram can be plotted. For example, in a research study pediatric population from 3 years to 14 years was included. In this study, 100 subjects were enrolled if we consider the 3-year class intervals. The numbers are 3 to 5 years 20 patients, 6 to 8 years 30 patients, 9 to 11 years 15 patients, and 12 to 14 years 35 patients.

28 Elements Of Clinical Study Design, Biostatistics & Research

20%

35%

Patel and Patel

3 to 5 Year 6 to 8 years 9 to 11 years

30% 15%

12 to 14 years

Fig. (2.3). Pie diagram.

Pictogram In a pictogram, small pictures or symbols are used to present the data. Pictograms are frequently used to demonstrate information about the rural population. For example, male to female ratio of patients in a hospital is studied, and we find 60% of males and 40% of females, i.e., the ratio of 3:2 [7, 8]. 1. Scales of measurement are • Nominal scale • Ordinal scale • Interval Scale • Ratio Scale 2. Descriptive Statistics a. Measurement of Central Tendency • Mean • Median

Scales of Measurement

Elements Of Clinical Study Design, Biostatistics & Research 29

• Mode b. Measurement of Dispersion • Standard deviation • Range 3. Data Presentation • Tables • Charts, Diagrams, and Graphs • Histogram • Special Curve • Pictogram CONCLUSION Scales of measurement are: ● ● ● ●

Nominal scale Ordinal scale Interval Scale Ratio Scale

Descriptive Statistics a. Measurement of Central Tendency ● ● ●

Mean Median Mode

b. Measurement of Dispersion ● ●

Standard deviation Range

Data Presentation ●

Tables

30 Elements Of Clinical Study Design, Biostatistics & Research ● ● ● ●

Charts, Diagrams, and Graphs Histogram Special Curve Pictogram

Patel and Patel

Elements Of Clinical Study Design, Biostatistics & Research, 2023, 31-54

31

CHAPTER 3

Inferential Statistics Abstract: Probability could be objective or subjective. Byestherom is a formula to calculate conditional Probability. The sample methods of random sampling are simple random sampling, systematic random sampling, Stratified random sampling, and cluster random sampling. Use of hypothesis testing, if used as inferential statistics, definite steps are required to be followed. Using confidence intervals as inferential statistics provide more useful information in clinical research. This chapter incorporates a discussion on these aspects.

Keywords: Annova, Confidence interval, Dunnet’s test, Hypothesistesting, Inferential statistics, Probability, Randomsampling, Random variables, Students test, Turkey’s HSD test. INTRODUCTION The statistical methods discussed in Chapter 2 are called descriptive statistics as they help the researchers to describe & summarize the research data. In this chapter, we deal with the basic probability concepts to evaluate the data using statistical methods & draw inferences. In statistical methods, it is assumed that the sample selected is representative of the larger population to which the inferences are to be made applicable & selected through appropriate randomization. The statistical approach depends on the research question. In medical research, only two methods are used as inferential statistics, namely confidence interval & hypothesis testing. Caution is required while using the student t-test, particularly when it is used for nonparametric data; a good number of errors may occur. The emphasis in this chapter is placed on concept generation rather than the convenience of calculation [8, 9]. PROBABILITY Objective Probability Suppose an experiment is repeated many times and assume that one or more outcomes can result from each trial (Experiment repetition). In this situation, the S.S.Patel & Aditya Patel All rights reserved-© 2023 Bentham Science Publishers

32 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

Probability of a given outcome is the number of times the given outcome occurs, divided by the total number of trials. If the outcomes occur every time in a trial, say in ten trials, the outcome occurs ten times, then P= 10/10 = 1. Thus, the Probability is one, i.e., the maximum. If the outcome does not occur even once, i.e., it occurs zero times, the P=0/10= 0. If the event occurs five times, the P=5/10=0.5, i.e., 50%. The outcome can take any number. The estimation (estimate) of Probability can be made based on the theoretical model. It can be determined empirically. For example, if we toss a coin ten times, the theoretical possibility (P) of the coming head is five times that P of Head =5/10=0.5, and that of the coming tail is also five times P of Tails =5/10=0.5. In other words, Probability of a coming head is 50%, and that of the tail is 50%, but if we repeat this experiment on ten different occasions, it may not happen regularly. In one trial, it could vary from 0 to10, taking any number on one occasion. However, in place of 10, if we repeat the experiment 100 times, the chances of obtaining 50: 50 percent are more likely. This Probability is called Objective Probability [9 - 11]. Subjective Probabilities It is the best guess depending on the person's previous experience, for example, Provisional diagnosis in a clinical setting. Thus it is the probability estimate reflecting the person's opinion or best guess whether an event will occur or not. Subjective probabilities are important in medicine because the subjective Probability is the basis of physicians' opinions. Physician's opinion about whether a patient has specific diseases or not (Provisional Diagnosis in Clinical Practice) [12]. Understanding the concept of Probability is important for: ●

●

●

Understanding and interpretation of published research data in the form of graphs and tables. Making predictions about the inferences of researchers' own data that how much confidence one can have an estimate like means proportions or relative risk. It is essential to understand the meaning of P values given in published research articles.

EXPERIMENT AND EVENT Experiment In Probability, an experiment is defined as any planned process of data collection.

Inferential Statistics

Elements Of Clinical Study Design, Biostatistics & Research 33

Event An experiment consists of a number of independent repetitions or trials under the same conditions. Complementary Event An event opposite to the event of interest is called a complimentary event, e.g., if we want to study the cases of vivax malaria in a patient with fever, vivax malaria is an event of interest, the Complementary event is the patient of fever, not having vivax malaria. The Probability of the complimentary event is calculated as under: P of having Vivax + P not having Vivax = 1, This means P not having Vivax = 1- P of having Vivax or P of having Vivax = 1- P of not having Vivax Mutually Exclusive Event and the Rule of Addition Mutually exclusive events are two or more events when the occurrence of one excludes the occurrence of the other for example, if a person has blood group A, he cannot have blood group O. Here, blood groups 'O' and 'A' are mutually exclusive events (Table 3.1). The Probability of two mutually exclusive events is the Probability of the occurrence of either one event or the other. This Probability is calculated by adding the probabilities of the two events. This is called the rule of addition [5, 6]. Table 3.1. Mutually exclusive event. PROBABILITIES Blood-Type

Males

Females

Total

O

0.21

0.21

0.42

A

0.215

0.215

0.43

B

0.055

0.055

0.11

AB

0.02

0.02

0.04

Total

0.50

0.50

1.00

34 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

Distribution of blood type by gender: In the above tabular example, the Probability that a randomly selected person has either blood group 'O' or 'A' [13]. In the above example, each outcome, i.e., type of blood group “O, A, B, and AB,” is a separate event. The Probability of two mutually exclusive events occurring is the Probability of the sum total of the two events is called the addition rule for Probability. In the above example, the Probability that a randomly selected person (randomly) has either blood group “O” or blood group “A” is P(O or A) = P(O)+P(A) = 0.42+0.43 = 0.85. Independent Events In the example of the blood groups, if we take into consideration the gender of the subjects, the male can have any type of blood group that is A, B, O, or AB. Similarly, the female can have any blood group A, B, O, or AB. In this example, the blood groups and gender are independent events. Thus, independent events can be defined as two different events if the outcome of one event does not affect the outcome of the other event. The Probability of the independent event is calculated by multiplying the probabilities of the two events. Some authors call this a rule of Multiplication. Let's say from the above-mentioned example; The probability of finding 2 random individuals with Blood group B = P (B) x P(B) = 0.055 x 0.055 = 0.003025 The terms Probability, Odds, and Likelihood are considered synonymously. However, for the purpose of statistical analysis, they are different. Odds are defined as the Probability that an event occurs divided by the Probability that the event does not occur. For example, the Odds that a person has a blood group “O” = P(O)/1-P(O) = 0.42/(1-0.42) = 0.72. LIKELIHOOD It is related to Bayes' theorem, which will be discussed along with Bayes'theorem.(Chapter 4) [10, 13].

Inferential Statistics

Elements Of Clinical Study Design, Biostatistics & Research 35

The conditional Probability is defined as “The probability of an event (A) given that another event (B) which is in relation to A has already occurred” it is denoted as P(A|B). The symbol “|” represents the word “given.” This means the event after “|,”i.e., B, has already occurred. As per Bayes Theorem, the conditional Probability can be calculated as follows; P(A|B) = P(A∩B)/P(B) Where, P(A ∩ B) is the Probability of event A and event B P(B) is the Probability of event B, provided P(B) is not equal to 0. SENSITIVITY The sensitivity of the diagnostic test can be defined as the ability of the test to detect the condition for which it is testing. This means it is the ability of the test to identify the subjects who are actually having the disease (True Positive Rate). A high Sensitivity test will give low false-negative results. A test with high sensitivity has a low fall negative rate. SPECIFICITY The term specificity is correctly identifying the subjects who do not have the disease(True Negative). If the specificity of the test is high, it means the test will give low false-positive results. In other words, we can say that specificity refers to specificity in health. BAYES' THEOREM Thomas Bayes was a British mathematician who developed the formula for calculating the conditional Probability of one event from the conditional Probability of other events. The formula was developed in the 18th century and was published after 200 years after his death. The formula is commonly used for calculating predictive values of diagnostic tests. Bayes' theorem in the term of sensitivity and specificity can be written as under [14]; Sensitivity x Prior Probability ____________________________________________________________Sensitivity x Prior probability+ [(False – positive rate x (1 – Prior Probability)]

36 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

Let us take an example to explain Bayes' theorem. Suppose a 55-year-old male patient comes with a history of back pain. After clinical examination, the clinician has to rule out or confirm the spinal malignancy diagnosis. If he knows the prior Probability of malignancy in the patient is 0.20, the sensitivity of E.S.R. is 78%, and the specificity is 67%. We have sensitivity x prior Probability in the numerator, i.e., 0.78x0.20. In the denomination, the quantity in the numerator is repeated and added to the false positive rate, .33, times 1 minus the Probability of malignancy 0.80. Thus we have 0.78 𝑥 0.20 0.78 𝑥 0.20 + (0.33 𝑥 0.80) 0.156 = 0.37 = 0.156 + 0.264

𝑃(𝐷+ /𝑇+ ) =

The predictive value of positive E.S.R. above 20 mm hg/1hr is 0.37. POPULATION AND SAMPLES The most important purpose of research in health science is to generalize findings from the study sample to a large population. This inference is drawn by using statistical methods based on Probability. In statistics, the term Population is used to describe a large set or collection of items that have something in common. A sample is a subset of the population selected to be representative of the large population. There are the following reasons to study samples instead of the population: i. ii. iii. iv. v.

Samples can be studied in a shorter duration of time. A study of the sample is less expensive. The samples' results are accurate if the sampling is done properly. Samples can be selected to reduce heterogeneity. In summary, in terms of sample size, we can say bigger does not always mean better.

The researcher must plan the size appropriately, as required by the study, before beginning the research. This process is called the determination of the power of the study [8, 10, 14]. POWER OF STUDY The power of the study is the process of estimating the number of subjects for a given study. The power of a study represents the Probability of finding a

Inferential Statistics

Elements Of Clinical Study Design, Biostatistics & Research 37

difference that exists in a population. It depends on the chosen level of Significance, the difference that we look for (effect size), variability of the measured variables, and sample size. METHOD OF SAMPLE SELECTION There are two methods for sample selection Non-probabilities Sampling Non-probabilities sampling could be a convenient sample or quoted sample. In non-probability samples, the Probability of a subject's selection is unknown. Normally, when the term samples are used in the context of an observational study, it is assumed that the sample has been randomly selected. In nonprobability sampling, the sample may suffer from the selection bias of the researcher [11]. Probability Sampling The purpose of sampling is to ensure that the findings of the sample or data will lead to reliable and valid inferences. This objective can be achieved by using probability samples. In medicine, four probability samplings are commonly used. They are: Simple random sample Systematic random sample Stratified random sample Cluster random sample Simple Random Sample In simple random sampling, every subject has an equal probability of being selected for the study. There are two methods of simple random sampling. It can be done by using a table of random numbers or a computer-generated list of random numbers. Systematic Random Sample In this method, every nth item is selected. Then it is determined by dividing the number of items in the sampling frame by the desired sample size. For example, there are 3000 available records of patients suffering from a particular disease. Out of which 200 is, the sample size 3000÷ 200 = 15, so each 15thcase paper is selected for analysis.

38 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

Stratified Sample In Stratified sampling, the population is first divided into selected subgroups, and then a random sample is selected from each subgroup. Random Cluster Sample In random cluster sampling, the population is divided into clusters, and then a subset of the cluster is randomly selected. Clusters are commonly based on geographical areas. This approach is used in epidemiological research rather than in clinical studies [10, 11]. Properties of Good Sampling i. It should be unbiased; that is, it should be free from all types of biases. ii. The variance in the two samples should be small. This is one of the reasons why the mean is used rather than the median as a measure of central tendency because the standard error of the median is near about 20% larger in comparison to the standard error of the mean when the distribution of observation is approximately normal. iii. However, if the observations are not normally distributed (the curve is not bell-shaped, and they are skewed). Statistics of choice are median, not mean. Random Assignment In an experimental study, such as randomized clinical trials, subjects are first selected for inclusion in the study depending on the appropriate inclusion criteria. After selection, subjects are then assigned to different treatment groups. Then the assignment of a subject to various groups is done by using random methods, and the process is called a random assignment. Furthermore, assignment to the treatment group could be single-blind or double-blind. As discussed earlier in chapter 2 (Study Design in Medicine Research), while interpreting the random samples, the term target population is the population to which the researcher wishes to generalize the research findings. The sampling population refers to the population from which the sample was actually drawn. Suppose a population or a sample is representative of the target population, and the distribution of important characteristics in the samples is the same as that is the target population. In that case, the sample is called a “Representative Sample.” [15]

Inferential Statistics

Elements Of Clinical Study Design, Biostatistics & Research 39

RANDOM VARIABLES AND PROBABILITY DISTRIBUTION The characteristic of interest in a study is known as a variable. The term variable refers to the characteristics value, which varies from one subject to another. This variation is an inherent biological variation among individuals. A random variable is a variable in a study in which subjects are randomly selected. The values of a random variable can be summarized in a frequency distribution called the probability distribution. In research of health sciences, three probabilities distributions are important. They are: ● ● ●

The Poisson Distribution Binomial Distribution The Normal or Gaussian Distribution

Poisson Distribution The Poisson distribution is a discrete distribution. This is applicable when the outcome is the number of times an event occurs. The Poisson distribution is used to determine the Probability of rare events. It gives the Probability that an outcome occurs at a specific number of times when the number of trials is large, and the Probability of any one occurrence is small. For example, the Poisson distribution may be used to plan the number of beds a hospital needs in its Intensive Care Unit. It can also be used to determine the number of bacterial colonies growing in the given amount of culture medium or determine the number of particular types of cells (malignant) cells in a given amount of body fluid, say, cerebrospinal fluid (CSF) or pleural fluid. It can also be used to calculate the number of ambulances required in the emergency department of a hospital. The Probability in Poisson Distribution can be calculated by the formula as under: Suppose X is the occurrence. Then P(X) = (e-λ λx) / X! The “X” is the occurrence of the event, the Greek letter Lambda (λ) is the value of parameter 'd' variance of the Poisson distribution, and “e” is the base of natural logarithms, which is equal to 2.718. The term Lambda is called the parameter of Poisson Distribution. Therefore only one piece of information,i.e., the Lambda, is required to characterize any given Poisson distribution.

40 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

Binomial Distribution In binomial distribution, an event either occurs or does not occur. Therefore it is known as a binary event. Hence, the event can have only binary outcomes, i.e., yes or no, or positive or negative, denoted by A and B. The Probability of A is denoted by π (pie) or P (A) = π. Therefore, the Probability of B must be 1-π; this is because B can only occur if A does not occur. If an experiment involving this event is repeated n times, and the outcome is independent of one trial to another, how the Probability of the event can be calculated? To understand this, let us take an example; 5-year survival of a prostate cancer patient, wherein the pre-treatment prostate antigen is not less than 10 international units, has been studied. The Probability of 5-year survival is 0.8. If 5-year survival is represented by PS and PD represents death before 5 years. Ādenotes the Probability of survival, then Ā=PS=0.8 and1-Ā=PD=0.2. The Normal or Gaussian Distribution The normal distribution is continuous. In the normal distribution, we get a smooth bell-shaped curve that is symmetric on both sides of the mean. The symbol for the normal distribution is the Greek letter 'mu' (Fig. 3.1).

Area Under a normal Curve between'a' and 'b' Fig. (3.1). Gaussian Distribution.

The curve is shown in Fig. (3.1). The standard deviation of the distribution is denoted by Greek Letter Sigma (∑). Sigma is the horizontal distance between the

Inferential Statistics

Elements Of Clinical Study Design, Biostatistics & Research 41

mean and the point of inflection on the curve. The point of inflection is the point where the curve changes from convex to concave. The mean and standard deviation are the two parameters of normal distribution, and they determine the location on the number line and the shape of the normal curve. Therefore many different normal curves are possible each one for every value of the mean and the standard deviation. As the normal distribution is a probability distribution, the area under the curve is equal to 1. As it is a symmetric distribution, half of the area lies on the left side of the mean, and half the area lies on the right side of the mean. In case the mean of a normal distribution is not zero (0), and the standard deviation is not one for statistical calculation, “Z” formation is required to be made for standard normal distribution. This exercise, what is called as “Z” term formation, denotes the deviation from the mean in terms of units of standard deviation. The formula for this is: ܈ൌ

െ σ

Where: -mu is the mean -Sigma is the standard deviation -X is the random variable THE CENTRAL LIMIT THEORY Mathematically it has been proved that if the mean is normally distributed in the population and the sample and target population are normally distributed, provided that the variable is Gaussian distributed. The central limit theory is used in statistics to estimate the sampling distribution of the mean without calculating the sampling distribution for the mean for a statistical calculation; a researcher wants to calculate each time. The following properties characterize the central limit theory: i. Suppose the distribution of a variable is normal in the population. In that case, the sampling distribution of the mean is also normal, if the central size of the study sample is more than 30. It is this reason why a sample size of 30 or more is used in clinical studies.

42 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

ii. The mean of sampling distribution or the mean of means is equal to the population mean. iii. The standard error of the mean is the standard deviation of the means in a sampling distribution. Therefore in the statistical calculation, the standard error is used instead of using Standard Deviation. The standard deviation tells how much variability can occur among the individual in a population. Whereas Standard Errors tell about the variability in the means of the population. SAMPLING DISTRIBUTION In any selected sample, the distribution of an individual variable, say, systolic blood pressure, is different from the distribution of the means. This is called a sampling distribution [15]. Features of The Sampling Distribution i. Choice of the statistic of interest, for example, the mean, standard deviation, or proportions ii. Random selection of sample iii. The size of a random sample is the most important feature of research iv. Specification of the population which is being sampled The sampling distribution in medical studies, the important question of how often the possibility of chances of outcome is there actually, there is no difference between the outcome of the controlled and interventional group. SAMPLE SIZE Sample Size, in simple terms, can be called the minimum number of subjects required in a study. Sample Size is one of the important factors on which the power of the study depends [5]. Power of study, in simple terms, is the ability of the study to detect the actual difference in the study groups. For the estimation of sample size, the following information is needed: ●

●

The effect size is the magnitude of the difference or the relationship. As the effect size increases, the sample size decreases. If the effect is doubled, the sample size will become one-fourth. Level of Significance is the Probability of incorrectly rejecting the Null

Inferential Statistics

● ●

Elements Of Clinical Study Design, Biostatistics & Research 43

Hypothesis. It is also known as the alpha value in hypothesis testing; usually, the value of Alpha is taken as 0.05 The disagreed level of power – It is usually 80% The Standard deviation– For determining the sample size, we require an estimate of the standard deviation in population and in the sample [10].

For the calculation of sample size, various methods are available in modern days. Various Statistical software are available, for example, PASS, Sample Power, and others. APPROACHES TO STATISTICAL INFERENCES There are two approaches to drawing a statistical inference: I. Confidence Interval II. Hypothesis Testing A confidence interval It is a special form of computed research data. This interval gives the Probability that the parameter, e.g., mean or proportion, is contained within that interval. The commonly used confidence intervals in health sciences are 90%, 95%, and 99%. a. For comparing the difference between two means, the formula is: Confidence interval(CI) = Mean Difference ± Number Related to Confidence Level (95%) × Standard Error of the Difference b. For comparing two proportions, the formula is: For comparing the confidence interval of two proportions, the general formula is the same CI=The Statistics ± Number Related to Confidence Level (95%) × Standard Error of the Statistics. To understand, let us take a hypothetical example (Table 3.2). A survey was conducted in three medical colleges to know the proportions of the teachers who were trained in the teaching of Evidence-Based Medicine (EBM) and were involved in the teaching of EBM and those teachers who were not trained in EBM teaching; the data was tabulated as under:

44 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

Table 3.2. EBM data from three medical colleges. .

Screen

College Name

Total

Yes

No

Jawaharlal Medical College

97

28

125

Govt. Medical College, Nagpur

199

99

298

Rural Medical College, Sewagram

34

11

45

Didactic Teaching

261

110

371

Interactive Teaching

40

12

52

Yes

154

62

216

No

175

76

251

Yes

175

27

202

No

155

111

266

Methods of Teaching

Teachers Teaching

Trained in EBM

The proportion is defined as a part divided by the total. The proportion of the teachers trained in EBM is 175/202= 0.866, and the proportion without EBM training is 155/266= 0.583. This can be written in a generalized formula as proportion = a/a+b. We want to know whether the proportion of the teachers who taught the students is different for those who were trained in EBM from those who were not trained in EBM. The confidence interval of 95% can be calculated for the two proportions by the earlier described formulae. However, the process is a little bit tedious. Therefore it is beyond the scope of a book of this nature. If required, the researcher can consult an expert statistician. The test of Hypothesis is applied to permit the generalization of the data from a study sample to the larger population from which the sample is drawn. It is very important to remember that estimation and hypothesis testing are based on the assumption that the random sample has been selected properly, is free from selection bias, and is representative of the population. Estimate and Estimation: The oxford dictionary meaning of the word estimate is the approximate judgment or making rough calculations. Estimation is the process of using information from a sample to draw conclusions about the values of parameters in a population.

Inferential Statistics

Elements Of Clinical Study Design, Biostatistics & Research 45

Point Estimate: The mean and proportions of the sample are point estimates as they are specific numbers as opposed to a range or interval. Interval Estimate: Suppose the variability is expressed in terms of range. It is called an interval estimate. A Confidence Interval (CI) is a range of estimates for an unknown parameter. Suppose we measure the weight of children between the age of 5 years to 14, and if the weight measurement comes out to be 7kg to 12kg. The confidence intervals would be 12kg upper and 7kg lower limits. If the confidence intervals are associated with Probability, the ends of the confidence interval, i.e., 7kg and 12kg, are called the upper and lower confidence limits. In other words, the confidence limits of the confidence interval are computed from sample data and have a given probability that the unknown parameter is located between them. A confidence interval is computed at a designated confidence level. The commonly used confidence levels in terms of percentage are 90%, 95%, and 99% [16]. Hypothesis Testing Steps i. Making a statement (stating) research question in terms of statistical Hypothesis Stating Null Hypothesis (Ho): Stating an alternative hypothesis (Ha): ii. Making a decision on the suitable test statistic iii. Selecting the level of Significance and determination of the values iv. Determination of the value of Significance v. Performing the calculations vi. Making conclusion ❍ ❍

Making a statement (Stating) research question in terms of statistical Hypothesis The statistical Hypothesis is: Stating the Null Hypothesis (Ho) The Null Hypothesis implies that there is no difference between the mean of the sample and the mean of the population. A Null Hypothesis is a type of statistical

46 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

Hypothesis that proposes that no statistical significance exists in a set of given observations. If any difference is observed, it's due to chance. This can be understood by the example of the legal system of India, which considers an accused person innocent till the judiciary declares him guilty based on evidence produced by the council of prosecutors. It means that the researcher has to prove that there is a difference in the means of sample, the mean of the population, and the Probability of occurrence by chance. Stating Alternative Hypothesis (Ha) Ha is an alternative hypothesis, in fact, the research hypothesis. It implies that there is a statically significant difference between the two groups. If the Null Hypothesis (Ho) is rejected based on the evidence of the data of the sample, then the alternative Hypothesis is accepted. In case the evidence is not sufficient to reject the Null Hypothesis; usually, the term used is that the Null Hypothesis cannot be rejected rather than using the term that the Null Hypothesis is accepted. It is because it may be that a better study may be designed wherein there may be a possibility that the Null Hypothesis may be rejected [5]. Making the Decision on Suitable Test Statistics Choosing the right test statistics is the most important part of answering a research question. Student's T-Test This is suitable for comparing the means in two groups. The assumption for the use of the t-test is that the observations in each group are normally distributed. Suppose this assumption is violated and the observations are not normally distributed in two groups. In that case, it gives the values lower than what they are, resulting in the rejection of the Null Hypothesis and reaching the conclusion that there is a difference, whereas no real difference exists. Chi-Square Test This test is applied for nonparametric data or a comparison of proportions. The Chi-square test is very easy to understand, and also the calculations are simple. Sometimes it is overused by researchers, wherein other tests may give better results. It is very important to remember that the application of unsuitable statistical tests will lead to the drawing of erroneous conclusions. Certain data (the situations) where the chi-square test should not be used are as under:

Inferential Statistics ● ●

●

Elements Of Clinical Study Design, Biostatistics & Research 47

If the data is numerical, a Student t-test should be used If the expected frequencies of nonparametric data are small, i.e., less than “5”. In this situation, the chi-square test should not be used. The indicated test is Fisher's exact test. If the objective is to estimate between two nominal measures, then the Relative risk or odds ratio is a suitable test rather than a chi-square test.

The McNemar's Test This test is used in place of the Paired t-test if the observations are nominal instead of numerical. The Sign Test It is used to test medians if the observations are skewed. Wilcoxon Signed Rank Test This test is used if the observations are not normally distributed. Selecting the level of Significance and Determination of the Value Selecting the level of Significance for a statistical test and determining the value the test statistic must attend to be declared as significant. The level of Significance should be chosen before the performance of the statistical test. This chosen value of the level of Significance is termed an alpha value. It is denoted by the Greek letter alpha (ɑ). The alpha value gives the Probability of rejecting the Null Hypothesis incorrectly and coming to the conclusion that there is a difference, wherein the actual difference does not exist. Therefore, this alpha value should be small. In medical research, the values used for Alpha are 0.05, 0.01, and 0.001. Determination of the value of Significance The value of Significance is also called the critical value. The critical value is the dividing line between the area of acceptance of the Hypothesis and the area of rejection. If the confidence interval is 90%, the acceptance area is central 95%, and the rejection area is 2.5% on either side in 2 tailed t distribution. Performing the Calculations As software's are easily available to perform calculations through computers, laptops, or palmtops or MAD (Multi-Activity Device), there is no need to discuss the details of these in a book of this nature. Making Conclusions In conclusion, either the Null Hypothesis is rejected, or consequently, the

48 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

alternate Hypothesis is accepted. If the Null Hypothesis is not rejected, usually it is not accepted; rather, a conclusion is drawn that, based on the data of the researcher and application of the statistical test, the Null Hypothesis cannot be rejected. ERRORS IN THE HYPOTHESIS TEST There can be two types of errors in testing a hypothesis. The Type One Error If the Null Hypothesis is rejected, when it is true, this is known as a type one error. This error occurs due to the selection of a lower level of Significance (lower level of ɑ for the test). This results in the conclusion that there is a difference between the two groups, whereas no difference exists. The Probability of making a type one error is denoted by Alpha (ɑ). The Type Two Error The type two error may be called the reverse of type one error, i.e., wrongly accepting the Null Hypothesis, meaning thereby that there is no difference, whereas the difference really exists. Type two error is denoted by the Greek letter beta (β). POWER OF STUDY The power of study has been defined as the ability of a study to detect a difference between study groups when a true difference exists. Power is the Probability of avoiding a Type II error. Evidently, the high power of any study is desirable. Mathematically it is calculated as 1-β, where β is the value of type II error. Powers lower than the value of 0.8 are generally considered too low for most areas of research [16]. Significance of P-Value and Alpha A p-value of a study tells us the Probability of obtaining an effect at least as large as the one we actually observed in the sample data. For example: if we are testing a new drug against an old one for treating a disease, we find that the p-value of the hypothesis test is 0.02. This means if there truly was no difference between the new drug and the old one, then 2% of the times that we perform this hypothesis test, we would obtain the effect observed in the sample data, or larger, simply due to random sample error.

Inferential Statistics

Elements Of Clinical Study Design, Biostatistics & Research 49

This means that obtaining the sample data we actually did would be pretty rare if there was no difference between the new and old drugs. Thus, we would be inclined to reject the Null Hypothesis and conclude that there is a difference between the new drug and the old one. The alpha level of a hypothesis test is the threshold we use to determine whether or not our p-value is low enough to reject the Null Hypothesis. In testing a Hypothesis, the value of Alpha, that is, the level of Significance, is decided by the researcher and is often set at 0.05, but it is sometimes set as low as 0.01 or as high as 0.10. Now, if the P-value is lesser than Alpha, the Null Hypothesis is rejected, and Ha is accepted. If P-value is more than Alpha, the Null Hypothesis cannot be rejected. The P-value should be calculated after performing the statistical test. It is advisable to calculate P-value by a computer program and the exact figure to be reported. The practice of reporting less than 0.5 or less than 0.01 is not that accurate from the point of view of testing of Hypothesis. PAIRED T TEST If the same subjects are measured on a numerical variable before and after an intervention, the paired t-test is applied. This paired t-test is also called a Matched group t-test. INTRA-RATER RELIABILITY When one person observes the same subject or reports on some image or some specimen or some slides and the observations are compared, the degree of agreement is called Intra-Ratter Reliability. If two or more persons record the same observations, their degree of agreement is called Inter Ratter reliability. If the measurement is on a nominal scale, Intra-ratter reliability is calculated by using kappa (k) statistics. PEARSON'S CORRELATION COEFFICIEN If the observations are on a numerical scale, the correlation between two numerical observations is measured by Pearson's Correlation Coefficient.

50 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

MEASUREMENT OF THE SAME VARIABLE BY TWO DIFFERENT PROCEDURES Such observations are compared against some set standards procedure which is called the “Gold Standard.” In that case, sensitivity and specificity are appropriately used. In case there is no gold standard, kappa statistics are used to measure the degree of agreement between the two procedures. APPLICATION OF T-TEST For the applicability of the t-test, one assumption is that the standard deviation is equal; therefore, the two standard deviations or two separate standard deviations are pooled. This is calculated by taking out the average of the two standard deviations. Sample Size Sample Size Calculation for Studies in Two Groups The formula for assuming the sample size is as under: ൫ݖఈ െ ݖఉ ൯ߪ ݊ ൌ ʹ ̴̴̴̴̴̴̴̴̴̴̴̴̴ ߤଵ െ ߤଶ Where μ1 – μ2 (Mu) is the difference between the standard deviation of the two groups, zɑ is the value of z (z tailed) related to Alpha (ɑ), and zβ is the lower onetailed value of z related to the beta (β). Comparing the two means by using a confidence interval, the formula for the confidence interval for the difference between the two means is as under: Confidence interval = mean difference ± number related to confidence interval x standard error of the difference.

The formula of sample size for proportions in two groups: n=

ଶ ௭ഀ ඥଶగభ ሺଵିగభ ሻି௭ഁ ඥగభ ሺଵିగభ ሻାగమ ሺଵିగమ గభ ିగమ

Inferential Statistics

Elements Of Clinical Study Design, Biostatistics & Research 51

Where n is the number of two groups, zα is the two-tailed z value related to the Null Hypothesis and zβ is the lower one-tailed z value related to the alternative Hypothesis. Where π1denotes the proportions in one group and π2 denotes the proportion in other groups. COMPARISON OF MEANS IN THREE OR MORE GROUPS Analysis of Variance (ANOVA) If the research project involves comparing more than two groups, applying multiple t-tests between different pairs of means will change the Significance (ɑ) level for the whole experiment; therefore, a global test is to be performed. One of the tests applied for the comparison of multiple groups of observation is called Analysis of Variance (ANOVA). If the result of the ANOVA is significant, then the data can be compared further, i.e., a comparison can be made among the pairs or combinations of groups. It is important to understand the terminology used in relation to the application of ANOVA. The term factor, in ANOVA, is used for the variable by which groups are formed (independent variable). The number of groups, as defined by a factor, is referred to as the number of levels of the factor in clinical studies. The different treatment groups are actually levels, with reference to ANOVA. It is to be remembered that the terms one-way ANOVA or two-way ANOVA, do not refer to the one or two groups which are to be analyzed. They refer to the factors. If the relationship of one factor is to be analyzed, one-way ANOVA is applied. If two factors are to be analyzed, two-way ANOVA is applied [16]. To Understand this Term Better, Let us Take Two Examples: Example 1: In patients suffering from hypothyroidism, Thyroid Hormone Replacement Therapy (THRT) is given. Most commonly, levothyroxine –T4 is given. The response is monitored by measuring the TSH levels in the blood. We know that in hypothyroidism, TSH levels are raised. After the administration of T4, the TSH level returns to normal in individuals with normal thyroid function. The thyroid gland secretes levothyroxine (T4) and levo-triodothyronine(T3). 20% of plasma T3 is secreted by the thyroid gland, and the remaining is produced by peripheral conversion of T4 to T3. Administration of T4 alone leads to normal levels of T3 and reserves the thyroid function to normal. However, some studies suggest that with the administration of a combination of T4 and T3, patients feel better as compared to the administration of T4 alone. To settle this controversy, suppose a clinical study is planned to evaluate the effect of T4 replacement therapy on plasma T4 and

52 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

T3 concentration. These three groups of patients are to be found: ● ● ●

Group one- patients with no thyroid dysfunction as control Group Two- patients suffering from hypothyroidism and taking T4 Group Three- patients diagnosed with thyroiditis based on the presence of thyroperoxidase antibodies, but have a normal serum TSH, value and is not taking T4 replacement.

In the above study, the patients are divided into three groups based on thyroid function: The relationship between T3, and T4 is to be studied; therefore, this study will need the application of one-factor ANOVA or one-way ANOVA. The number of groups in this example, three, is called the levels of the factor. Example 2: Let us take another example. Studies have demonstrated that in patients suffering from hyperthyroidism, there is Impaired Glucose Tolerance and hypersecretion of insulin. It is also reported that being overweight also increases insulin secretion. Suppose we want to undertake a study to see the effect of hyperthyroidism and being overweight on the secretion of insulin. This study will require the application of two-way ANOVA or two factors ANOVA. Weight is one factor, and hyperthyroidism is another factor. Pre-Condition (Assumptions in ANOVA) ANOVA is a parametric method; therefore, all the assumptions, like the t-test, are applicable. I. Values of the outcome variable (dependent variable) are to be normally distributed within each group or level of the factor. In an earlier example: the level of T4 is assumed to be normally distributed in each of the three groups. II. The variance of the population in each group should be the same. In our example, the squared standard deviation (variance) of T4 level is equal in the three groups. III. The observations in random samples are independent because the value of one observation is not related to the other. In our example: the value of one subject's T4 level has no effect/influence on that of any other subject. As such, the F test is robust with respect to a small violation of the assumption of normality. However, if the observations are extremely skewed, the Kruskal-Wallis

Inferential Statistics

Elements Of Clinical Study Design, Biostatistics & Research 53

nonparametric test is recommended. Multiple Comparison Procedure Post Hoc Comparisons Post Hoc is a Latin word that means “after this.” The Post Hoc comparisons are made after an ANOVA has resulted in a significant test. Although there are many Post Hoc tests described in books of statistics, the important Post Hoc test for clinical biostatistics are: Tukey Test or Tukey's HSD Procedure (Honestly Significant Difference): Tukey's HSD test is used to compare all pairs of means. Statistical research has demonstrated that this test is the most accurate and powerful for detecting procedures for comparing the means. We have already defined earlier that power is the ability to detect a difference where it really exists, and the Null Hypothesis is correctly rejected. Dunnett's Procedure Dunnett's procedure is applicable in studies wherein multiple treatment means are compared with a single control mean. Non-Parametric ANOVA The nonparametric ANOVA is based on the ranks of the observation rather than the original observations. For one-way ANOVA, the nonparametric procedure is the Kruskal-Wallis one-way ANOVA. Post-Hoc comparison between pairs of means may be made using the Wilcoxon Rank Sum test. COMPARISON OF FREQUENCIES OR PROPORTIONS IN MORE THAN TWO GROUPS If the comparison is to be made between the two groups, either the Chi-Square test or Z test can be applied, but if there are more than two rows and two columns in the frequency table, Z test is not suitable. However, the Chi-square test can be used irrespective of the numbers or categories within each variable. The Chisquare can be used in cases where the research question has been stated in terms of frequencies or in terms of proportions.

54 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

CONCLUSION 1. Probability: "Defined as a given outcome in terms of the number of items that occur divided by the number of trails (times). It can be : ● ●

Objective Probability Subjective Probability 2. Byes' theorem: A formula for calculation of the conditional Probability of one event from the conditional Probability of other events. 3. Methods of Random Sampling: Has the following types:

● ● ● ●

Simple Random sampling Systematic Random sampling Stratified Random sampling Cluster Random sampling 4. Steps of Hypothesis testing:

●

● ● ● ● ●

Making statements in terms of statistical Hypothesis. It involves stating Null Hypothesis (Ho) and Alternate Hypothesis(Ha). Making the decision on the suitable test statistics. Selecting the level of Significance. Determining the critical value. Choosing suitable test statistics. Making calculations.

Elements Of Clinical Study Design, Biostatistics & Research, 2023, 55-70

55

CHAPTER 4

Statistical Methods for Relationship Variables Abstract: In the language of statistics, research is a planned & systemic method of data collection, analysis, & drawing conclusions. In this chapter, a demonstration of the relationship between numerical, nominal & ordinal data & calculation of other statistical techniques applicable in critical research, is described in a nutshell.

Keywords: Anova, Ara, Bayes theorem, Correlation coefficient, Cer, Eer, Likely hood ratio, Multiple regression, Nnt, Nnt, Odds ratio (or), Probability, Regression, Roc curve, Rrr, Sensitivity, Specificity. INTRODUCTION Research questions that require the study of relationships between two numerical variables, Correlation, and Regression, are the statistical methods to be applied. Demonstrations of relationships amongst nominal and ordinal data require statistical treatment calculating Experimental Event Rate (EER), Control Event Rate (CER), Absolute Risk Reduction(ARR), Relative Risk(RR) & other appropriate calculations like number Needed to Treat (NNT), Absolute Risk Increase (ARI), Odds Ratio, ANOVA, Logistic Regression and Multivariant Analysis of Variance (MANOVA). The diagnostic procedure requires the calculation of Sensitivity, Specificity, Likelihood ratio & plotting of the Receiver Operating Characteristic ROC curve [17]. This chapter discusses the principles of choice of their statistics depending on the requirement of the research question. THE RELATIONSHIP BETWEEN TWO NUMERICAL OBSERVATIONS (CHARACTERISTICS) Correlation Correlation and Regression are statistical methods to examine the linear relationship between numerical variables. The relationship between two numerical S.S.Patel & Aditya Patel All rights reserved-© 2023 Bentham Science Publishers

56 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

observations is demonstrated by the correlation coefficient, which is also called “Pearson Correlation Coefficient.” The formula for the correlation coefficient is as under: ݎൌ

ᎂሺܺ െ ܺതሻሺܻ െ ܻതሻ ඥᎂሺܺ െ ܺതሻଶ ᎂሺܻ െ ܻതሻଶ

Where r is the correlation coefficient, X is the numerical value of one variable, and Y is the numerical value of other observations. The correlation coefficient ranges from -1 to +1. -1 denotes a perfect negative relationship, and +1 depicts a perfect positive relationship. A correlation of “0” (zero) means no linear relationship. Assumptions in Correlation The assumptions needed to draw valid conclusions about the correlation coefficient are: i. Random selection was made in selecting the sample. ii. Two variables are normally distributed, called a bivariate normal distribution. iii. The normal distribution of the two variables should be in joint distribution. It is important to remember that individually each variable is distributed normally separately. It does not guarantee that jointly they have a bivariate normal distribution. Pearson's product is unsuitable for use if one of the two variables is not normally distributed. In this situation, the Spearman Rank Correlation is suitable for statistics [5, 17]. To study the relationship, in addition to calculating the Coefficient of correlation, the scatter plots of the data should form an essential part of the analysis. A scatter plot, also known as a scatter graph or a scatter chart, is a two-dimensional data visualization that uses dots to represent the values obtained for two different variables - one plotted along the x-axis and the other plotted along the y-axis. The formula for the Coefficient of the relationship demonstrates the extent of the linear relationship between the variables (Fig. 4.1). Without the scatter, the researcher may miss a plot-important non-linear relationship if he uses only formulae.

Statistical Methods

Elements Of Clinical Study Design, Biostatistics & Research 57

To understand the said statement better, let us take an example. If we record the discharges from nerve fiber and change the temperature, we know that the discharges are more at the temperature of 250 C. The discharges decrease both at lower than 250 and also at a higher temperature. In this scatter plot, there is no linear- relationship- between- the temperature- and discharges- of the nerve fiber (r = 0.0). However, there is a relationship between the temperature and discharges from nerve fiber which remains obscured if the only value of r is used to calculate the relationship.

Fig. (4.1). Scatter Plots and Co- Relations.

The six hypothetical scatter plots demonstrate the usefulness of the graphical presentation of correlation. The six scatter plots, namely A, B, C, D, E, and F, demonstrate the shape of the scatter plot along with values of r. I. Scatter plot A: r = ±1 and Scatter plot shows a straight line, indicating a complete positive relationship between increasing temperature and nerve impulse discharges.

58 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

II. Scatter plot B: r= 0.7, which indicates a linear relationship; however, the scatter plot is oval, indicating a small relationship. III. Scatter plot C: r= -0.9, demonstrating a straight-liner relationship but in a negative direction. IV. Scatter plot D: r= -0.4, the shape of the scatter plot is oval, showing a small relationship. V. Scatter plot E: r= 0, scatter plot also demonstrates no relationship. VI. Scatter plot E: r= 0, scatter plot also demonstrates no relationship. VII. Scatter plot F: r= 0, scatter plot demonstrating no relationship. However, the shape of this scatter plot demonstrates a bimodal relationship between temperature and nerve impulse discharges. Thus we find that the r-value alone may not indicate the complete picture of the relationship between the two variables. Regression Linear Regression Linear Regression is used for making a statistical (mathematical)prediction of the value of one characteristic from the value of other known characteristics. The other term used is linear Regression, simple linear Regression, or least square Regression. The term linear regression refers to the fact that Regression measures the straight line between two variables. The term simple regression is used when only one independent variable is used for making predictions about an outcome. If more than one independent variable is used in making the prediction, it is known as multiple Regression. THE RELATIONSHIP BETWEEN TWO ORDINAL CHARACTERISTICS The relationship between two ordinal characteristics is described by Spearman Rank Correlation, also known as Spearman's rho. This test is also used to describe numerical observations skewed by extreme observations. The symbol for Spearman is rs'. The formula and calculations are commonly available in computer programs. THE RELATIONSHIP CHARACTERISTICS

BETWEEN

TWO

NOMINAL

The relationship between two nominal characteristics is described in the form of ratios. The two ratios, Relative risk and Odds ratios, are used to describe the

Statistical Methods

Elements Of Clinical Study Design, Biostatistics & Research 59

relationship. This Relative risk and Odds ratio is sometimes called the risk ratio. These ratios are used to describe the relationship between the risk factor and the occurrence of a disease or a given outcome. This ratio is also utilized to estimate the relationship between the number of patients and the side effect of a drug or beneficial effect. Some important concepts of statistics are required to be understood for application and understanding of the meaning of relative risk and odds ratio, which are as follows: Experimental Event Rate (EER) The Experimental Event Rate (EER) is a measure of how often a particular statistical event (such as a response to a drug, adverse event, or death) occurs within the experimental group (non-control group) of an experiment. It is calculated by dividing the number of subjects who have experienced the said event in the experimental group by the total number of subjects present in the experimental group [18, 19]. Let's say in an experimental group A is the number of subjects who have experienced the given event, and B is the number of subjects who have not experienced the said event. So the total number of subjects in the group will be A+B In mathematical terms, EER will be written as A/(A+B). i.e., EER = A/(A+B). The Control Event Rate (CER) The Control Event Rate (CER) is a measure of how often a particular statistical event (such as a response to a drug, adverse event, or death) occurs within the control group of an experiment. It is calculated by dividing the number of subjects who have experienced The said event in the Control group by the total number of subjects present in the Control group. Let's say in a Control group C is the number of subjects who have experienced the given event, and D is the number of subjects who have not experienced the said event. So the total number of subjects in the group will be C+D. In mathematical terms, CER will be written as: CER = C/(C+D).

60 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

Relative Risk (RR) The Relative Risk (RR) or Risk Ratio is the ratio of the Probability of an outcome in an exposed/experimental group to the Probability of an outcome in an unexposed/Control group. It is used to estimate the strength of the association between exposures (treatments or risk factors) and outcomes. The symbol for Relative Risk is RR. The RR ratio is the ratio of the incidence of a said experience in exposed subjects (of an experimental group or with risk factors) to the incidence in unexposed people (of a control group or without risk factors). Mathematically this can be calculated by EER (Experimental Event Rate)/CER (Control Event Rate), i.e., RR = EER/CER. Relative Risk Reduction (RRR) It is the relative decrease in the risk of an adverse event in the exposed/experimental group compared to an unexposed/control group. It can be calculated as follows: RRR = (CER − EER) / CER, Or simply, it can also be calculated by subtracting Relative risk from 1 i.e., RRR = 1- RR. Absolute Risk Reduction (ARR) Absolute Risk Reduction is the difference between the risk of an outcome in the exposed/experimental group and the unexposed/control group. It can be calculated by subtracting control event rate from experimental event rate [18]. i.e.,ARR = EER – CER For example, suppose a study is conducted to find out the effect of aspirin therapy in reducing the incidence of myocardial infarction. The sample consists of two groups of young male people, ten thousand people in each group. The collected data shows that out of ten thousand people taking a placebo, 217 suffered from myocardial infarction. Whereas in the group taking aspirin, 126 suffered from myocardial infarction. An Absolute Risk Reduction would be:

Statistical Methods

Elements Of Clinical Study Design, Biostatistics & Research 61

ARR = EER-CER = 126/10000-217/10000 = 0.0126-0.0217 = 0.0091. meaning thereby that the absolute reduction is 91 out of 10000. Absolute Risk Reduction(ARR) is a more valuable index as compared to Relative Risk Reduction (RRR). Sometimes in research publications, only relative risk reduction is mentioned. In those circumstances, the conversion of relative risk reduction to absolute risk reduction can be done by multiplying rrr by the Controlled Event Rate (CER). In this example: ARR = 0.4194 (RRR) x 0.0217 (CER) =0.0091 (ARR). Number Needed To Treat (NNT) The Number Needed to Treat (NNT) is an epidemiological measure used in communicating the effectiveness of an intervention or treatment. It's the reciprocal of absolute risk reduction, i.e., NNT = 1/ARR. The NNT is the average number of patients who need to be treated to prevent one additional bad outcome. For the above-mentioned experiment NTT = 1/ARR = 1/0.0091 = 19.9 It means 110 people are to be treated to avoid one myocardial infarction. This information helps in clinical decision-making by way of assessing the relative risk and benefits of a particular treatment or drug. Absolute Risk Increase (ARI) Some drugs or surgical procedures increase the risk of serious side effects or adverse outcomes. The absolute risk increase is essentially the negative impact of a treatment strategy in a given population. Absolute Risk Increase (ARI) can be calculated as the absolute difference between the Experimental Event Rate and Control Event Rate. Mathematically it can be written as: ARI = EER − CER Let us take an example: In a study, it was found that with the use of an “X” drug in 100 patients incidence of skin rashes was 35, whereas, in the control group

62 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

(unexposed group), it was 12 patients in every 100 patients. So, in this case: ARI = EER-CER = 0.35 - 0.12 = - 0.23 or 23% That suggests that the risk of developing skin rashes in the subjects exposed to the drug “X” is 23%. Number Needed to Harm (NNH) It is a measure that shows how many persons, on average, need to be exposed to a risk factor to cause harm in an average of one person who would not otherwise have been harmed. It is the reciprocal of absolute Risk Increase i.e., NNH = 1/ARI. so for the above-mentioned example, if ARI is 0.23, then NNH = 1/0.23 = 4.34. This suggests that 1 out of every 4.34 subjects exposed to drug X are likely to develop skin rashes. Odds Ratios The odds ratio is applied to estimate the risk in the case of controlled studies or if an individual with an adverse outcome was at risk divided by the odds that an individual without an adverse outcome was at risk [19]. The odds are the Probability that an event will occur divided by the Probability that the event will not occur, i.e. Odds= P divided by 1-P wherein P is the Probability. The odds ratio can be calculated by plotting the observation in to table. Multiple Regression Multiple Regression is used when more than two independent variables are used to predict the values of the outcome. The dependent variable is denoted by 'Y', and the independent variable is denoted by 'X.' Letters a and b are used to denote a sample estimate. Ready-made computer programs are available for various multiple Regression models to be applied. Computerized programs are also

Statistical Methods

Elements Of Clinical Study Design, Biostatistics & Research 63

available to control the problem of missing data in studies involving many variables. The sample size for multiple Regression is estimated by a computer program. However, the researcher must remember that: 1. The sample size should be 10 times the number of the independent variable. For example: If the independent variables are 10, then the minimum size of the sample should consist of 100 subjects. 2. The assumption about normality in multiple Regression is complicated depending on whether the independent variable is considered to be fixed or random. In such cases, an expert statistician's advice should be taken. ANALYSIS OF COVARIANCE(ANCOVA) It is the statistical technique used to control the effect of a confounding variable that occurs in research when the subject cannot be assigned to various groups by random allocation. For example: If we want to study whether smokers have more ventricular wall motion abnormality. Ventricular wall motion abnormalities are also related to the degree of coronary stenosis. Therefore any difference in the ventricular wall motion abnormalities between smokers and non-smokers may also occur because of a difference in the degree of coronary stenosis between these groups. Therefore in such studies, the researcher has to control the degree of coronary stenosis so that it does not confuse (confound) the relationship between smoking and wall motion abnormalities. In such a situation, ANCOVA is to be used in the said example. If ANCOVA is used, the coronary occlusion score is called the COVARIATE. LOGISTIC REGRESSION This technique is used in research studies wherein the independent variable includes numerical and nominal data both, and the outcome variable is also dichotomous. The overall results from a logistic regression can be tested with the Hosmer and LemeshowGoodness of Fit Test. The principle of this test is based on the Chi-Square Distribution. The P-value ≥ 0.05 means that the estimates fit the data at an acceptable level. The data can be further analyzed [20].

64 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

MULTIVARIATE ANALYSIS OF VARIANCE (MANOVA) This statistical technique is used in studies with multiple dependent and independent variables. The MANOVA is suitable wherein the independent variables are nominal or categorical, and the dependent variables (outcomes) are numerical. If the results of MANOVA are found statistically significant, further the data is analyzed by using a multivariate statistic called Wilks'lambda. Another statistic that can be used for the analysis of multiple independent and multiple dependent variables is called Canonical Correlation Analysis (CCA). The technique is suitable for the data when both the independent variables and the outcomes are numerical, and the research question requires to study of the relationship between the set of independent variables and the set of dependent variables. DIAGNOSTIC PROCEDURES The measurement of diagnostic tests or procedures requires an understanding of two aspects [21]. The first is Sensitivity, and the second is Specificity. Sensitivity As discussed earlier in Chapter 3, Sensitivity has been defined as the proportion of time the diagnostic test is positive in patients who have the disease or condition. If a diagnostic test has high Sensitivity, it has a low false-negative rate. In other words, a high-sensitivity test helps us to rule out the patient if the test is negative. Sensitivity has also been described as its positivity in disease or Sensitivity to disease. In other words, it demonstrates the true positive rate. Specificity Specificity has been defined as the proportion of time that a diagnostic test is negative in patients who do not have a disease or condition. It means a more specific test has a low false-positive rate. Some authors have described or defined Specificity as negative in health or specific to health. The Sensitivity and Specificity of a diagnostic procedure are determined by the administration of the test to two groups of patients who are known to have the disease or condition and another group of patients who are known not to have the disease or condition, and then the proportions are calculated. A sensitivity equals the percentage of patients known to have the disease and test positive. For example, If the number of patients suffering from a disease is 50 and

Statistical Methods

Elements Of Clinical Study Design, Biostatistics & Research 65

the diagnostic test is positive in 45 patients, the Sensitivity will be: Sensitivity = 45/50x100=90%. Similarly, Specificity is calculated by the proportion of subjects known to be free of the disease and for whom the test is negative. For example: If the healthy subject were 50 and the test was negative in 45 subjects. Specificity will be equal to: Specificity = 45/50x100=90%. However, sometimes gold standard may not be available, and even an autopsy is required to make the final diagnosis. Use of Sensitivity and Specificity for Revision of Probabilities: The values of Sensitivity and Specificity alone cannot determine the value of a diagnostic test in a patient presented to the clinician for the diagnosis of a disease. The clinician has to use its index of suspicion (provisional diagnosis). In statistical language, it is called Prior Probability that the patient has the disease index of suspicion not only based on Probability determined by experiment or observations, which has simply the best guess. A clinician begins with the baseline prevalence, then it is revised upwards or downwards. The revision is based on the clinical examination of the patient. Further, the Probability of the disease is revised in light of the diagnostic test's Sensitivity and Specificity. These new probabilities are the predictive value of a positive test and the predictive value of a negative test. The predictive value of a positive test gives the percentage of patients with a positive test who have the condition. LIKELIHOOD RATIO Another method of using the information obtained by the Sensitivity and Specificity of a test is the calculation of the likelihood ratio. Calculation of likelihood ratio The use of odds in place of Probability is increasing in medical literature, especially in the context of the Likelihood Ratio in a diagnostic test is equal to positive divided by false positive. Thus, a positive test has a one likelihood ratio. Only a negative test has another likelihood ratio for a positive test. A likelihood ratio is the Sensitivity divided by the False-positive rate. The likelihood ratio is multiplied by pre-test odds.

66 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

The post-test odds of a positive test are obtained as a formula. This can be written as post-test odds equal to pre-test odds multiplied by the likelihood ratio. For example: if the Sensitivity of a test for a disease is 78% and the Specificity is 67%, the false positive rate will be = 100% - 67%=33%. Therefore the likelihood ratio for a positive test will be calculated as under:

LR = 0.78 = 2.36 0.33

or LR = 0.78/0.33 = 2.36

value of the positive test and the predictive value of a negative test. The predictive value of a positive test gives the percentage of patients with a positive test who have the condition. BAYES' THEOREM Bayes' theorem is a formula for the calculation of the conditional Probability of one event, P (A | B), from the conditional Probability of other event P (B|A) [14]. In terms of Sensitivity and Specificity, the formula can be written as: ௌ௦௧௩௧௬ൈ௧௬ ௌ௦௧௩௧௬ൈ௧௬ାሾሺி௦௦௧௩௧௫ሺଵି௧௬ሻ

For example: if the prior Probability of a disease is 0.20, the Sensitivity of a test for the disease is 78%, and the Specificity of the test is 67%. In the numerator Sensitivity x Prior probability = 0.78 x 0.2= 0.156. In the denominator, again Sensitivity x Prior probability = 0.156, and the False positive rate = 1-specificity= 1-0.67 =0.33, 1- Prior probability will be = 0.80. Thus we get:

Statistical Methods

Elements Of Clinical Study Design, Biostatistics & Research 67

ௌ௦௧௩௧௬ൈ௧௬ ௌ௦௧௩௧௬ൈ௧௬ାሾሺୟ୪ୱୣ୮୭ୱ୧୲୧୴ୣ୰ୟ୲ୣ୶ሺଵȂ୰୧୭୰୰୭ୠୟୠ୧୪୧୲୷ሻሿ

ሺܦା ȁܶ ା ሻ ൌ

ൌ

ͲǤͺ ൈ ͲǤʹͲ ሺͲǤͺ ൈ ͲǤʹͲሻ ሺͲǤ͵͵ ൈ ͲǤͺͲሻ

ͲǤͳͷ ൌ ͲǤ͵ ͲǤͳͷ ͲǤʹͶ

Use of Sensitivity and Specificity in Making Clinical Diagnosis (Figs. 4.2 and 4.3) When the diagnostic tests are based on values on a numerical scale; in such cases, when the test values are measured on a continuum, the Sensitivity and Specificity depend on the cutoff values between the positive and negative. This situation is like the normal distribution of the laboratory test. One distribution is for the patient who has the disease and one is for the subject who does not have the disease.

T.N. = True negative, TP = true positive F.N. = false negative; FP = false positive Fig. (4.2). Two Hypothetical Distribution With Cut Off At 60.

68 Elements Of Clinical Study Design, Biostatistics & Research

People Without Disease (TN)

People With Disease (TP)

(FN) 45 Normal

Patel and Patel

55

(FP) 75 Abnormal

Specificity = 0.85; Sensitivity = 0.95 Two hypothetical distributions with cut off at 60 T.N. = True negative, TP = true positive F.N. = false negative; FP = false positive Fig. (4.3). Another situation with Cutoff value of 55.

A hypothetical situation can be created to understand ROC. Suppose the mean value of the test for the patient with the disease is 75, and the mean value of the test for the subject without the disease is 45, and if the cutoff point is placed at 60, about 10% of the people without the disease are diagnosed as False positive because their test values are greater than 60 and about 10% of the patients with the disease or diagnosed as normal, i.e., False negative because of their test values less than 60. Thus it can be concluded that this particular test has a sensitivity of 90% and a specificity of 90%. RECEIVER OPERATING CHARACTERISTIC (ROC) CURVE ROC curves are a better depiction of the relationship between Sensitivity and Specificity for the test, which has a continuum. The ROC curve is a plot of the Sensitivity or true positive rate against the false-positive rate (Fig. 4.4). The hypothetical ROC curve can be plotted as under:

Statistical Methods

Elements Of Clinical Study Design, Biostatistics & Research 69

Fig. (4.4). Receiver Operating Characteristic Curve.

In the said figure, in the graph diagonal line correspond to a test, i.e., positive or negative, just by chance. The upper left corner represents the accuracy of the test graphs. In these circumstances, the true positive rate is 1, and the false positive rate is 0. In other words, the closer is the ROC curve to the upper left-hand corner of the graph, the more accurate is the diagnostic test. It can be observed that if the criterion for a positive test becomes more strict. The point on the curve corresponding to Sensitivity and Specificity (point A) moves down and to the left, indicating lower Sensitivity and higher Specificity, and if less evidence is required for a positive test, the point on the curve corresponding to Sensitivity and Specificity that is point B moves up and to the right indicating higher Sensitivity and lower Specificity [22, 23]. ROC curves are very good methods for the comparison of two or more diagnostic tests. They can also be used for selecting a cutoff level for a test. CONCLUSION 1. The Correlation coefficient is a measure between two numerical measurements made on the set of the same subjects. It ranges from -1 to +1. O

70 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

denotes no relationship. 2. Experiment (in Probability) planned way of data collection. 3. Experiment Event Rate (EER): The number of research subjects in the interventional group who developed the outcome under the study. 4. Control Event Rate (CER): The number of research subjects in the control group who developed the outcome under the study. 5. Absolute Risk Reduction (ARR): The measure of the reduction of risk with new therapy compared with risk without the new therapy. It is equal to EER – CER. 6. The number needed to treat (NNT): The number of patients needed to be treated with the proposed therapy in order to prevent or cure one patient. It is the reciprocal of absolute risk reduction, i.e., 1/AER. 7. The Odds Ratio (OR): It is an estimate of reductive risk calculated in casecontrol studies. 8. Odds: It's the measurement of the Probability of an event that will occur divided by the Probability of an event that will not occur. ODDs=P/1-P. 9. Bayes' theorem; if formulae to calculate the conditional Probability of one event from the conditional Probability of another event. 10. Sensitivity means Sensitivity to disease, and the test is positive for the disease. 11. Specificity means specific to health, negative in health.

Elements Of Clinical Study Design, Biostatistics & Research, 2023, 71-88

71

CHAPTER 5

Clinical Research Abstract: The clinical trial could either be an Explanatory Trial/ Randomised Control Trial (RCT) or Correlational Trial/ Pragmatic Trial. For developing new molecules as a drug, RCTs require human studies conducted in 4 phases called Phase I to phase IV of clinical trials. Pragmatic Trials are Correlational Trials. As such RCT & Pragmatic Trials are not dichotomous, there is a continuum. A PRECIS has been developed to assess the said trials. The reverse pharmacology approach is recommended to generate scientific evidence to make herbal drugs more efficacious & safe. These aspects are elaborated on in this chapter.

Keywords: Clinical trial phases precis, CONSORT, Randomized control trial (RCT), Reverse pharmacology. INTRODUCTION Any research wherein study subjects are human beings, is denoted as clinical research. The commonly used study designs in clinical research are discussed in Chapter 1. This chapter deals with clinical trials (RCT) in little detail as the outcome of the clinical trials have become applicable to inpatient care. Presently, the clinical research health policy decision-makers and funding agencies prefer Real World Evidence (RWE), Comparative Evidence Research (CER), and Pragmatic Trials. Now even for demonstration of cause and effect relationship explanation, trials (RCT) remains the gold standard. In the case of generating authentic, more efficacious, and safer use of herbal medicines, the approach of reverse pharmacology is considered the technique of choice. CLINICAL RESEARCH Clinical Research: Clinical research is defined as “a branch of health care science that determines the safety and efficacy of medications, diagnostic products, and treatment regimens intended for human use”. Clinical research aims to find better methods to control, diagnose and treat the disease to increase overall human well-being. Clinical research generally is conducted through clinical trials. S.S.Patel & Aditya Patel All rights reserved-© 2023 Bentham Science Publishers

72 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

TYPES OF CLINICAL TRIALS Clinical trials are interventional studies conducted in humans. Clinical trials are conducted to evaluate the effectiveness and safety of drugs, devices, and therapies in human beings, healthy or suffering from the disease [24]. Clinical Trials are broadly classified into two groups: I. Randomized Clinical trials (RCT) II. Pragmatic Randomised Clinical Trials (pRCT) Randomized Clinical Trials are explanatory trials wherein the cause-and-effect relationship of the intervention is evaluated under strictly controlled conditions with the application of principles of de-confounding. In contrast, Pragmatic Trials focus on the correlation between treatments (interventions) and outcomes in a real-world health system practice rather than focusing on providing a causative explanation for the outcome. Thus this can be stated that RCTs are explanatory trials while pragmatic trials are correlation trials. PHASES OF CLINICAL TRIALS There are four phases of RCTs: Phase I The new drug, after preclinical studies, is administered for the first time to healthy volunteers. The purpose of Phase-I is to study the Pharmacokinetic and the pharmacodynamic effect of the new drug in human young adult healthy male volunteers. Phase II This Phase is subdivided into Phase II (a) and Phase II (b) Phase II(a) The drug is tested in a selected group of patients to evaluate its efficacy and safety in the patients suffering from the disease for which the drug is intended to be used. In this Phase, the appropriate doses of the drug are also determined.

Clinical Research

Elements Of Clinical Study Design, Biostatistics & Research 73

Phase II(b) (Pivotal Trial) In this Phase, the drug is administered to patients who are suffering from the disease for which the drug is to be used. The primary difference between Phase II, a, and b is that the Phase II (b) trials are more rigorously conducted to determine the new molecule's safety and efficacy in a selected group of diseased patients. Phase III(a) These trials are conducted after the successful completion of Phase-II trials. Only after successful completion of Phase III, (a) trials application to F.D.A. (Food and Drug Administration) can be submitted. These clinical trials are conducted in the group of patients in which the new drug is intended to be used. Phase-III clinical trial generates additional data on the efficacy and safety of the drug in a larger group of patients. Phase-III (a) trials may be required to be conducted in the following: ●

●

Specified group of patients, i.e., renal failure, cardiac function compromised patients, or other co-morbid conditions. Patients suffering from specific conditions and the nature of the drug Phase III trial provide information that is needed for the packaging insert and labeling of the medicine. Trials are conducted after the medicine's efficacy is demonstrated but before regulatory submission of a New Drug Application (NDA) or other dossiers. These clinical trials are conducted in patient populations for which the medicine is eventually intended. Phase III(a) clinical trials generate additional data on both safety and efficacy in relatively large numbers of patients in both controlled and uncontrolled trials.

Clinical trials are also conducted in special groups of patients (e.g., renal failure patients) or under special conditions dictated by the nature of medicine and disease. These trials often provide much of the information needed for the package insert and labeling of the medicine [25]. Phase III Clinical trials are further designed into three sub-types: i. Superiority Clinical Trials ii. Equivalence Clinical Trials iii. Non-inferiority Clinical Trials Phase-III (a) i) Superiority Clinical Trials When the research question is to show that one treatment is superior to another

74 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

treatment, the RCT design applicable is called a superiority clinical trial design. For this purpose, a statistical test is applied called a Superiority Clinical test. Remember that non-significant differences in a Superiority Trial should not be considered as any difference in the two treatments. For the purpose of demonstrating no difference in the two treatment groups, an equivalence clinical trial design is required to be applied. In the Superiority Clinical trial, the superiority margin (effect size) is to be chosen by the Clinician. It is recommended that the superiority margin should be decided in a meeting involving 2 to 3 experienced Clinicians of the same discipline [26]. The formula (F1) for the calculation of sample size for the Superiority Trial Nominal Variable (Dichotomous) is as under (Fig. 5.1): F1 F1:

N= 2 X Z1- α+ Z 1-β x p 1 - p Δ

Fig. (5.1). Superiority Trial Nominal Variable (Dichotomous).

Wherein: ● ● ● ● ●

N is the number of study subjects in each group. P is the response rate of the standard treatment group. PO: is the response rate of the new treatment group. Zx: is the standard normal deviation (SD) for one or two sided 'x'. Δ is the real difference between the two treatment groups which is the clinically acceptable margin (superiority margin).

Formula (F2) for Superiority Trial for Continuous Variable is as follows (Fig. 5.2): F2: F2:

N= 2 X Z1- α+ Z 1-β2x S2 Δ - Δ2 Fig. (5.2). Superiority Trial for Continuous Variable.

Clinical Research

Elements Of Clinical Study Design, Biostatistics & Research 75

Where in: ● ● ● ● ●

●

N is the number of study subjects in each group. P is the response rate of the standard treatment group. PO: is the response rate of the new treatment group. Zx: is the normal standard deviation (SD) for one or two-sided 'x' Δ is the real difference between the two treatment groups which is the clinically acceptable margin (superiority margin). S is the pooled standard deviation of both comparison groups.

Phase-III (a) ii)Equivalence Clinical Trial The trial aims to demonstrate that there is no difference between the two treatment groups, i.e., standard treatment and interventional treatment. In this trial design, the Clinician decides the equivalence margin with a difference ± degree would be acceptable. Usually, such trials are conducted when the new treatment gives some other benefits/advantages over the existing treatment. For example: ● ● ●

The side effects of the new treatment may be less The cost of the new treatment may be lower. For widening therapeutic options.

The formula for taking sample size for Equivalence Trial clinical design (nominal variable) is as follows (Fig. 5.3): F3:

N= 2 X

Z1- α+ Z 1-β 2 x p 1 - p

Δ2 Fig. (5.3). Equivalence Trial (nominal variable).

For equivalence clinical design, the formula for Equivalence Trial Continuous Variable is (F4) as follows (Fig. 5.4):

76 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

F4:

N= 2 X

2

Z1- α+ Z 1-β

x S2

Δ2 Fig. (5.4). Equivalence Trial (Continuous Variable).

Phase-III (a) iii)Non-inferiority Clinical Trial The design aims at providing information that the new drug/treatment is not inferior to the available treatment. But it does not demonstrate that the interventional drug/treatment is equivalency or maybe superior as the test measures only the non-inferiority margin. This trial design demonstrates that the said treatment is not inferior to the referred standard trial, more than the normal pre-specified margin. This margin is called the non-inferiority margin. The noninferior margin is to be decided by the Clinician. The following formula can calculate the sample size for the non-inferiority trial design. The formula for non-inferiority clinical design Non-Inferiority Trial Nominal Variable is as follows (Fig. 5.5): F5:

N= 2 X

Z1- α+ Z 1-β

x px (1- p)

d-Δ Fig. (5.5). Non-Inferiority Trial (Nominal Variable).

The formula for non-inferiority design Continuous variable is as follows (Fig. 5.6): F6:

N= 2 X

Z1- α+ Z 1-β

2

x S2

Δ Fig. (5.6). Non-Inferiority Trial (Continuous Variable).

Phase-III (b) Phase-III(b): Trials are conducted after a proposal to FDA is submitted but before

Clinical Research

Elements Of Clinical Study Design, Biostatistics & Research 77

NDA approval. The NDA approval is granted only after the results of Phase-III(b) are favorable. These trials may supplement the observations and inferences of Phase-III(a) trials, or the information of III(b) may direct for conduction of additional information to conduct some more trials for Quality of Life (QOL). Phase IV Clinical Trials Post-marketing surveillance and Phase IV trials are conducted after marketing the new drug. The new drug is kept further under observation regarding its efficacy and safety [27]. In phase-IV trials, the following aspects of the new drugs are evaluated: ● ● ● ● ● ● ●

Different formulations Different dosage forms Optimum duration of treatment Drug interactions New age groups Different races Any other relevant issue/s

If a new indication of the drugs is required to be evaluated, it is considered a Phase-II trial and has to be conducted per the guidelines of the Phase-II clinical trial. The post-marketing surveillance and Phase-IV trials are not synonymous. The post-marketing surveillance is an observational study, while there is no intervention. Whereas Phase IV Clinical trials are well-designed interventional studies to establish cause-and-effect relationships. RISK OF BIASES IN RANDOMIZED CONTROLLED TRIALS (RCTS) Cochrane Tool for Bias Risk Assessment 1. Random sequence generation (selection bias) 2. Allocation concealment (selection bias) 3. Blinding of participant personnel (performance bias) 4. Blinding of outcome assessment (detection bias) 5. Incomplete outcome data (attrition bias)

78 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

6. Selective reporting (reporting bias) 7. Other biases CONSORT GUIDELINES Adequate reporting of randomized, controlled trials (RCTs) is necessary to allow accurate, critical appraisal of the validity and applicability of the results. The CONSORT (Consolidated Standards of Reporting Trials) [27]. It consists of a twenty-two items checklist and a flow diagram (Fig. 5.7).

Fig. (5.7). CONSORT Guideline Flow Chart.

CONSORT Checklist The CONSORT Group has prepared a checklist for reporting RCT, and the

Clinical Research

Elements Of Clinical Study Design, Biostatistics & Research 79

checklist has been prepared in accordance with CONSORT guidelines. It can be divided into five sections: Title & Abstract, Introduction, Methods, Results, and Discussion. PRAGMATIC TRIALS (PRCT) (THERAPY EVALUATION) The pragmatic trials are conducted in real-world situations. There are no rigid inclusion or exclusion criteria. The pRCTs are also called co-relational trials, whereas the RCTs are called explanatory trials. Both trials (RCTs and pRCTs) are important study designs in clinical research. Pragmatic trials are most suited for the comparison of two therapies. The results of pRCTs are applicable in the real therapeutic world setting as they do not have strict inclusion or exclusion criteria. They observe the effectiveness of therapy in patients who consult a clinician for relief of their sufferings. As such, a pRCT may contain some elements of RCT. Main differences of (i) RCT and (ii) pRCT i. Pragmatic trials are co-relational trials, whereas RCTs are explanatory trials. ii. pRCT is conducted to know the effectiveness of the treatment in routine dayto-day life setup, not in Rigid Research Conditions (RRC). Whereas the RCTs are explanatory trials, they are conducted with the application of rigid inclusion and exclusion criteria for the purpose of de-confounding. Both explanatory and pragmatic trials have an important place in evaluating healthcare interventions, but they answer different research questions. Pragmatic trials are useful in answering questions about how effective a therapy is when compared to some standard or accepted treatment. They also overcome some specific difficulties that can be encountered with explanatory trials or complementary therapies. For example, when evaluating phytomedicines, pragmatic trial results can be generalized to wider clinical settings where they can provide evidence of how well therapies might perform as alternatives or adjuncts to conventional interventions. They can also help facilitate decision-making about whether therapies should be utilized more widely. RCT & pRCT are not dichotomous; there is a continual. The Pragmatic Explanatory Continuum Indicator Summary assesses the placement of a clinical trial [25, 26]. ELEMENTS OF PRAGMATIC TRIALS The pragmatic trials can be discussed in nine headings:

80 Elements Of Clinical Study Design, Biostatistics & Research

I. II. III. IV. V. VI. VII. VIII. IX.

Patel and Patel

Formulation of research questions Patients to be included A reference standard group Study protocol formulation Calculation of sample size Randomization Observation variables Statistical Analysis Publication

Formulation of Research Question As applicable to any research study design, formulation of the research question is very important. According to the research question study design is chosen. A pragmatic trial (pRCT) is most suited to answer the research question of which one of the two packages of therapy is better. A pragmatic trial generally compares the effectiveness of the two therapies. Pragmatic trials are more suited for the evaluation of 2 therapies. Usually, two systems of therapies, for example, Ayurvedic therapies and Modern Medicine. However, this approach may also be used to compare two regimes or evaluate adjunct therapy used in combination with the standard treatment. Pragmatic trials provide evidence that will help policymakers, practitioners or patients choose between two interventions [27]. Defining the patient group The aim of pRCTs is to generalize the outcomes of the two therapeutic modalities. Therefore, the patient needs to come from (representative of) a wider population. It is usually applied in those conditions where no treatment can be labeled as standard. For example, with Low backache, the selection criteria take into consideration: a. The condition itself b. Patterns of care and referrals To generalize results, the research subjects (study patients) should be representative of the population wherein the therapy is to be applied. Identifying comparison group The aim of pRCT is to generate evidence to facilitate decision-making for

Clinical Research

Elements Of Clinical Study Design, Biostatistics & Research 81

choosing which treatment option is better among available treatment options. In pRCTs, a placebo group/arm is not advisable. The pragmatic trial design needs both arms of the trial based on normal routine practice. The reference standard group should be one that is widely accepted. Defining the Treatment Protocol Pragmatic trials are primarily designed to model everyday clinical practice. In these trials, therapists (researchers) have the freedom to choose their patients without applying rigid inclusion and exclusion criteria. In pRCT a large group of patients is enrolled, and the study continues for longer. While analyzing the results, subgroup cohorts are required to be formed. As there are a large number of variables, the data needs appropriate statistical treatment to account for the consideration of confounding variables (de-confounding). Sample Size pRCTs require a much larger sample size as compared to explanatory clinical trials. This is important as the research subjects are recruited from a population wherein there is a mix of heterogeneous patients. Although there is variability between patients, which may dilute the effect of the treatment. However, this does not undermine the credibility of a pragmatic trial. A homogenous group of patients is selected in an explanatory clinical trial to control for confounding variables. The explanatory clinical trial demonstrates a direct relationship between the intervention and its outcome [5, 24]. Enrolment, Randomization, and Referral pRCT is aimed at exploring the potential role of study therapy in primary therapy. Therefore, a referral from general practitioners becomes an important source of study subjects for recruitment in pRCT. In other respect of recruitment and randomization, the pRCTs are similar to RCTs [18]. Outcome In the pragmatic trial, outcome measurements are primarily on long-term effects on the quality of life (QOL) rather than on a single outcome. This requires longterm follow-up to assess that the changes brought by the therapy under evaluation are sustainable. This is more important with respect to Ayurvedic and other Complementary Medicine, as it has been reported that some positive effect or benefit takes a longer time for their manifestation.

82 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

Analysis The analysis of the results of pragmatic trials is based on 'Intention To Treat’ (I.T.T.) basis. The groups are considered randomized. In the randomization process, it should be ensured that both groups are as near identical as possible. In comparing Ayurvedic care with standard care, the patient's choice of treatment is given. In the pragmatic trial, a patient may change his or her choice at any time during the course of treatment. The important point is to record the variations with the reasons thereof. The following points are important while analyzing the results of pragmatic trials: i. The statistical analysis should be done on the “intention to treat” basis. ii. The two therapy/treatment groups are considered randomized. iii. In view of (ii) the baseline data of the said two groups should be as near as possible identical. iv. If some study subjects (patients) opt for a change in their groups, they are allowed to do so, but the reasons thereof are to be reported. v. The researcher should consider the patients in the group in which they were randomized. vi. Above IV and V are required as this study is designed to reflect what is happening in a realistic world. Reporting and Publication Like RCTs, pragmatic trial reporting should also adhere to CONSORT guidelines. In addition, specific therapy guidelines and AYUSH guidelines for Ayurveda treatment should also be strictly adhered to. In reporting pragmatic trials, there is an additional challenge in the form of more complex treatment as compared to RCTs. Therefore, details regarding the various aspects should be reported so that the interventions can be repeated in practice. In disseminating the outcome, it should be clear that a pragmatic trial is appropriate to the research question [20, 21, 27]. REVERSE PHARMACOLOGY IN DRUG DEVELOPMENT The classical approach in new drug development commences from the synthesis of a new chemical moiety or alteration in an existing molecule to increase the safety and efficacy of the drug. It is carried out through the creation of the chemical library, finding out hits and leads; that is, it begins with the molecule. The new drug molecule is subjected to pre-evaluation of clinical efficacy and safety studies in animals, and then the clinical trial is conducted in 3 phases. This

Clinical Research

Elements Of Clinical Study Design, Biostatistics & Research 83

process is a very cost-intensive affair because of the enormous number of synthetic molecules, single-molecule fulfills the testing criteria. The classical approach can be called Molecule to Mice to Man [28]. In reverse pharmacology, the process takes place in the reverse direction that commences from Man to Mice to Molecule and then to its mechanism of action. Reverse Pharmacology Concept The concept of reverse pharmacology is inspired by traditional medicine. It is a systematic approach to the identification of a new drug candidate applying scientific principles and processes for its more effective and safe use (Fig. 5.8). While undertaking drug research through reverse pharmacology, it is essential to understand that the regulatory requirements of herbal medicine are completely different than those for the classical approach. Actually, the concept is not to develop a new drug but rather to increase the better utilization of herbal drugs. WHO guidelines state that “if the product has been traditionally used without demonstrating harm, no specific restrictive regulatory action should be undertaken unless new evidence demands a revised risk-benefit assessment.” WHO maintains that the plants or parts of the plants have been used in practice safely, and animal toxicity studies are not mandatory. Quite often, the plant or its products are used as food as well as medicine. Preclinical toxicity studies are required only for an herbal medicinal product that contains herbs that have no record of earlier safe use in humans [28]. THERE ARE FOUR STAGES OF REVERSE PHARMACOLOGY Stage-1: Selection of herbal remedy Stage-2: Dose escalating studies Stage-3: Randomized controlled trials Stage-4: Isolation and testing of active ingredient Stage-1:Selection of Herbal Remedy Stage-1:Selection of Herbal Remedy Literature Research It could be through published research articles or through standard ancient literature like Samhitas or other texts.

84 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

Fig. (5.8). Stages of Reverse Pharmacology.

Retrospective Treatment Outcome Study (RTO) The classical way of identifying medicinal plants for further research is through ethnobotanical studies. Many plants are “supposed” to be good for one disease or another but are not the preferred treatment used in everyday life. Graz et al. developed a method called a “Retrospective Treatment Outcome Study” (RTO) to circumvent these problems. This simply adds two essential elements to the ethnobotanical method: (a) clinical information and (b) statistical analysis. For the execution of the RTO study, clinical information is collected retrospectively on the presentation and progress of a defined disease episode. Further treatment and subsequent clinical outcomes are statistically analyzed to establish the statistical significance of associates between the two.

Clinical Research

Elements Of Clinical Study Design, Biostatistics & Research 85

There are 3 components of RTO: Component-1 Understanding local concepts and terms of diseases, the aim is to obtain maximum information about the disease of interest. Component-2: Choosing a representative sample A random sample of households in the study area should be done by cluster sampling method. Then enquire the family members of the family whether he or she had a disease of interest in the recent past. Component-3 If the answer to the second component is positive, then enquire about the treatment they had taken in what order and at what stage they have recovered. An extensive search of the literature is required to determine whether the safety of the remedy has been established or not in the past. A literature search is required with the aim of answering the following questions: ●

● ●

Whether human toxicity studies of the plant and remedies prepared by the same method are available? Are there laboratory studies conducted to assess toxicity? What is the probable pharmacological active principle present in the plant?

Interview With Traditional Healer The herbal remedy can also be selected through an interview with a traditional healer. Stage-2: Dose Escalating Studies The purpose of D.E.S. is to find out the best effective dose with minimum side effects. If the preparation was used by a patient, but no literature is available. In such a case, an observational clinical study in a small group of patients meeting the inclusion and exclusion criteria is required to be conducted (pilot study). The said study is to be conducted in the location (population) where patients are taking the remedy. It is a mandatory requirement (Fig. 5.9). In dose-escalating studies, it is better to start with consulting those patients who were taking the remedy and are familiar with the preparation. The research aims to find the minimum and maximum doses.

86 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

In the research protocol, 2 or 3 doses of the remedy, as proposed/used by the traditional healer, should be mentioned. It is important to apply the same inclusion and exclusion criteria to all research subjects to ensure that the two groups are as similar as possible. Based on the recommendation/practice of the use of a dose of the remedy, the study is commenced with the minimum dose following the below-depicted algorithm.

Fig. (5.9). DOSE ESCALATING FLOW CHART.

Dose Optimization The dose of the remedy is optimized based on the observation of the doseescalation studies. Preservation of the Study Plant (Herb) A specimen of the plant (with signature) should be deposited in the herbarium of the Institute for identification of herbal medicine. Further Phyto-chemical analysis

Clinical Research

Elements Of Clinical Study Design, Biostatistics & Research 87

should be done if no literature is available. Although according to W.H.O., it is not mandatory to conduct animal toxicity studies. It is better to determine ED50 and LD 50 in animals [27]. Stage-3 Randomized Controlled Trial (RCT) Suppose results from all previous stages are encouraging. In that case, the clinical study will be conducted to compare Phytomedicine to the standard first-line treatment to establish safety and efficacy with reference to the suitability of modalities with pragmatic (practical) inclusion criteria and outcome [25, 26]. Stage-4 Isolation And Testing Of Active Compounds Isolation And Testing Of Active Compounds For Standardizationquality Control, And Pharmaceutical Development is the last step of “Reverse pharmacology.” Although this is not mandatory, Ayurvedic principles state the plant remedy is more efficacious in combination as the ingredient has synergistic beneficial effects and moderates the adverse effects. However, isolation of the active ingredient is useful for quality control and standardization of plant medicine and to determine whether the new drug can be developed in the laboratory. CONCLUSION 1. 2. 3. 4.

In general, a study conducted in humans is called clinical research. There are mainly two types of clinical trials RCTs and pRCTs. The RCTs are conducted to establish cause-and-effect relationships. They are also called explanatory trials conducted under strictly restricted conditions with rigid inclusion and exclusion criteria. 5. The pragmatic trials are conducted in real-world situations without the application of rigid inclusion or exclusion criteria. 6. There are four phases of RCTs. ● ● ● ●

Phase I – studies in healthy volunteers. Phase II- Studies in a selected group of patients. Phase III- In a larger group of patients, usually multicentric studies. Phase IV- If post-marketing surveillance warrants further studies.

7. Pragmatic trials provide real-world evidence. It is the best approach for therapy evaluation. 8. As such, no clinical trial could be purely RCT pragmatic.

88 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

9. Pragmatic Explanatory Continuum Indicator Summary (PRESCIS) has been developed to demonstrate whether a clinical trial is Explanatory or Pragmatic. 10. Reverse pharmacology approach is indicated for the generation of scientific evidence regarding more efficacious & safe use of herbal medicine.

Elements Of Clinical Study Design, Biostatistics & Research, 2023, 89-98

89

CHAPTER 6

Survey Research Abstract: A survey is one of the most common methods in health sciences research. As in this case, choosing a suitable study design depends on the research question, so also survey research questions also decide methods. The decision of application of scale of measurement depends on the nature of the observations, which could be numerical, nominal, or ordinal. If research develops his/her new survey instrument (survey questionnaire), its validity & reliability requires other important treatments. This forms the subject of discussion in this chapter.

Keywords: Confidentiality, Likert scale, Reliability, Scale of measurement, Validity. INTRODUCTION Survey Research is one of the common forms of research. Calling it the most common form of research is not inappropriate. It involves generally asking questions to people about a topic. As is the case with any research, the research question is the most important part of the research, and survey research is no exception. Clear research questions with specific objectives help and guide in choosing a method of survey and designing an appropriate questionnaire [29]. There are three important components of survey research: I. Designing survey tool (questionnaire) II. Administering the questionnaire III. Interpreting the results

S.S.Patel & Aditya Patel All rights reserved-© 2023 Bentham Science Publishers

90 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

DESIGNING SURVEY TOOL The Research Question (Framing) As applicable to other research questions, a survey research question also requires a thorough review of the literature to know what is known on the topic and also to learn what other methods have been used earlier. Sometimes it becomes difficult to specify the issues precisely to be addressed in a survey, even after a search by important search engines. Under such situations, a 'focused group' discussion with a small number of concerned people, about 6-10, can provide better information for the determination of the survey research question. ADMINISTERING THE QUESTION Survey Methods Survey methods could be either: ● ●

●

●

Self-administered questionnaire or structured interview. Self-administered questionnaires can be done by e-mail/ mail or in person. Inperson administered questionnaires are considered to yield the best responses. Structured interviews could be conducted in person or through a Multi-Activity Device (mobile phone). Personal structured interviews provide a better result as compared to interviews taken on Multi-Activity Device (MAD) [30].

Developing Survey Questionnaire Format of question close-ended or open-ended. Open-ended questions permit the subject to respond in his/her own words. These types of questions are more useful when the topic has not been studied earlier. Close-ended questions are more difficult to write, but the main advantage is the comparatively easy analysis and reporting. Scales of Measurements In the case of closed-ended questions, the researcher is required to decide the level of detail required in the answer. The scale helps determine which method can be used to analyze the results. In some cases, only nominal data, i.e., Yes or No, is sufficient. If the information is required to be analyzed and to be presented in

Survey Research

Elements Of Clinical Study Design, Biostatistics & Research 91

grading, an ordinal scale is to be used. If the questionnaire is in the form of openended questions and the subject presents information in numerical form, the numerical scale, as described in Chapter 3, is to be used. It is advisable that the researcher should collect information in detail as much as possible at the same time; too many details are to be avoided. Too much information usually leads to unreliable data. Positive and negative categories If data is collected in the form of ordinal responses, there should be an equal number of positive and negative responses. If there is a neutral answer question, it should be placed in the middle position between negative and positive answers [31]. Vague adjectives: The use of qualifiers like sometimes, often, rarely, etc., should be avoided. For example: How many cold drinks did you drink last week? None A few Several Many Making less vague: How many cold drinks did you drink last week? Revised balanced probe (questionnaire): 00 1-5 6-10 11-20

92 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

Balancing the responses Mutually exclusive categories: One should not use questions that combine two categories of information in one question. It is advisable to make two separate questions (mutually exclusive). Very personal questions Some questions are considered very personal; use of such questions is to be avoided. For example, regarding questions related to income, sexual activity, personal habits, etc., people do not want to provide information on personal questions. Using don't know There is controversy in using the don't know option. Some researchers advocate that if this opportunity is not provided, the subject may commit to a wrong answer. Others say that opportunity for not committing to an answer should not be provided. If these categories are to be used, the probe (question) should be placed at the end of the questionnaire rather than placed in the middle. This increases the response rate by 9%, as has been reported. Balancing the Probes Problem: How much are you satisfied with the introduction of early clinical exposure? Unbalance probes (questionnaire): Completely satisfied Mostly satisfied Somewhat satisfied Neither satisfied nor dissatisfied Dissatisfied Revised balanced probe (questionnaire) How much are you satisfied with the introduction of Early Clinical Exposure?

Survey Research

Elements Of Clinical Study Design, Biostatistics & Research 93

Completely satisfied Mostly satisfied Somewhat satisfied Neither satisfied nor dissatisfied Dissatisfied Using the Likert Scale Likert scales are used in many surveys. This scale is suitable for probes with an ordinal scale (Table 6.1). The responses range from strongly agree to strongly disagree and vice-versa. The responses could also be from most important to not important. The options could be listed vertically if there is one –probe for responses. Suppose there is one response to each probe. If several probes use the same scale, the options are placed horizontally, i.e., in a row. For example: Likert scale for several probes: Probe – How important are the following factors in choosing a Medical College for your post-graduate course? Please circle '1' being not important and 5 being very important Table 6.1. Likert 5-Point Scale.

1

National reputation excellence

1

2

3

4

5

2

Availing good clinical material and hands-on training

1

2

3

4

5

3

Good hostels and campus ambiance

1

2

3

4

5

4

Senior Faculty

1

2

3

4

5

5

Academic Standards

1

2

3

4

5

6

Learns participation in academic decision making

1

2

3

4

5

94 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

Suggestions for Writing Probes ● ● ● ● ●

The use of user-friendly probes increases valid response rates Instructions should be clear, short, and specific Instructions should be stated in a neutral manner Continuing jargon and abbreviations to be avoided Probes with two details should be avoided

Layout Of Questionnaire A shorter questionnaire is preferable compared to a longer one. However, it is reported that questionnaires up to four pages in length have a similar response rate. Instruction should be placed wherever needed, even at the cost of repetition. The branching of questions is to be avoided. If not, possible directional arrows and other visual aid to be used to assist the subject. Easier questions to be placed; first, further questions to be placed in a logical order, and the difficult questions to be placed last. When scales are used, the scale direction is listed inconsistent manner. In this approach, the subjects are less confused. RELIABILITY AND VALIDITY OF SURVEY INSTRUMENT In a survey, it is better to use existing and pre-validated questionnaires. If a new survey instrument is developed, the survey instrument is to be pilot tested for its Reliability and Validity [32]. Reliability The researchers should use the existing pre-validated questionnaire or survey instruments as far as possible. This saves time and effort in developing the questionnaire and then establishing the reliability and validity of the instrument. The term reliability refers to the reproducibility of the finding of the instrument on repeated administration to the same subject. There are five types of reliability: Test Re-Test The responses are stable after repeated administration. This can be measured by the administration of the same questionnaire twice or more to the same subjects. Internal Consistency: Internal consistency refers to an agreement among items, i.e., the various items

Survey Research

Elements Of Clinical Study Design, Biostatistics & Research 95

measure the same thing. This can be measured by Cronbach's alpha test, i.e., average correlation. Alternative form: Alternative form means the measurement of the same topic by different items. In other words, the different items measure the same topic. This can be decided by the correlation of a square between the items. Intra-Observer Consistency: It is read by McNemar Statistics. Inter Observer Consistency: It is measured by the agreement between the different observers. The significance can be decided by kappa (k) statistics. Validity Validity is defined as the property of an instrument that indicates the instrument's capability of how well it measures the characteristic (variable). The measures measuring the validity are Face Validity, Content validity, Criterion validity, and Construct validity (FC3). Face Validity: Face validity is the degree to which a questionnaire or a test appears to be measuring what is supposed to be measured. Content Validity: Content validity indicates the degree to which the items on the instrument represent the knowledge of the characteristic being investigated. Criterion Validity: Criterion validity is the instrument's capacity to predict an associated characteristic. For example, an instrument is developed to measure the quality of life. The score thus obtained should be comparable with the physical examination and the patient's subjective feelings.

96 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

Construct Validity: It is very difficult to define construct validity. It demonstrates that the instrument is related to an instrument that assesses the same character. It is not related to the instruments which measure the other characteristic. It requires administering many tests or instruments on the same group of individuals and then evaluating the relationship pattern. ADMINISTRATION OF THE SURVEY INSTRUMENT Pilot testing As such, it is almost impossible to carry out a perfect survey. However, pilot testing of the survey instrument can detect many problems. Pilot testing is carried out after the questionnaire is designed. It should be done before the instrument is printed. Pilot testing can reveal that: ● ●

The reading level is not suitable for the intended subjects. Some questions are not clear or objectionable.

Under such observations, the probes are to be suitably modified. For the pilot study, a small sample of a small size is required. However, the sample should be representative in nature. Response Rate A response rate of more than 50% is considered satisfactory if the questionnaire is administered through the mail. A high response rate increases the researcher's confidence in the validity of the results [30, 31]. Methods to Increase Response Rates a. 3 to 4 follow-ups at the interval of two weeks. b. Advance notification to the subjects before administration. Pre-notification may be done by: ● ● ●

Letters Telephone E-mail c. Covering the letter along with return envelopes:

Survey Research

Elements Of Clinical Study Design, Biostatistics & Research 97

The covering letter should contain information on confidentiality and anonymity. d. Incentives: Giving incentives to research subjects is a controversial issue. But it has been reported that incentives had an increased response rate than any other single action. SELECTION OF REPRESENTATIVE SAMPLE AND DETERMINATION OF SAMPLE SIZE ● ● ● ●

Proper selection is very important. The sample should be selected by randomization. Selection bias to be eliminated. There are a good number of sampling methods. Some of them have been described in earlier chapters.

Determination of Sample Size Sample size determination requires consideration of the following: ● ● ● ●

●

Study outcome The difference in the type of respondents is two or more types. Relationship among probes (questions). The nature and number of comparisons, whether proportions or means and the number of two or more. Depending on the said variables, there are different methods of calculation of sample size. It is advisable to consult an expert to calculate the sample size before launching the study [32].

ANALYSIS OF SURVEY OBSERVATIONS Almost all procedures of statistics are used for analysis and drawing inferences from the observations recorded in survey research. Examples are as under: ● ● ●

●

Confidence limits for proportions Confidence limits to means Differences between means and proportions are important depending on the data, whether its numerical or ordinal. Chi-square test is used for nominal observation

98 Elements Of Clinical Study Design, Biostatistics & Research ● ● ●

Patel and Patel

Co-relation and Association Analysis of variance Regression and logistic regression

In addition, there are other advanced methods that are beyond the scope of nature of this book. It is advised to consult an expert in statistical methods. The information provided in this chapter is basic general guidelines about survey research. A good number of excellent information is available on the internet; a few sources are: ●

● ●

American statistical association section on survey research –stat pack survey software NCS Pearson National multiple sclerosis society – MUCS glossary asp.

CONCLUSION ● ● ● ●

● ● ● ● ● ●

● ● ●

A survey is one of the common methods in Health Science research. Survey methods are decided by the research question. An appropriate type of questions and response rates should be selected. The suitable scale of measurement should be chosen, whether Nominal, Ordinal or Numerical. Probes (questions) should be balanced in an equal number of. Positive and negative responses. Potentially objectionable questions to be avoided. Likert scales are popular and easy for respondents. Reliability indicates the reproductively of results of a survey instrument. Validity indicates the measurability of the survey instrument measures for which it is intended to measure. Pilot testing of the survey instrument is essential for its validation. A high response rate increases the confidence in findings. Confidentiality of the information is to be ensured, and the research subjects are assured of the same.

Elements Of Clinical Study Design, Biostatistics & Research, 2023, 99-107

99

CHAPTER 7

Planning and Writing Research Projects Abstract: Reading research papers/literature should be done through reviewing. It should not be not skimming. Biases in a study seriously affect the conclusion of the study adversely. While writing research reports, discrepancies in text, tables, and figures may reflect after-thought data treatment by the researchers: these and other aspects during the planning of research projects and report writing. Precautions required are the very purpose of this chapter.

Keywords: Biases, Fishing expedition, Power of study, Research gap, Study design, Sample size, Test of significance. INTRODUCTION Planning and developing research projects is the most important part of any research project. In medical research, the researcher must decide what he wants to do and why he wants to do it. The answer to Why requires a thorough Review of the Literature as per PRISMA Flow Chart mentioning the search strategy and search engine used. Depending on the research question, appropriate study design is chosen. All statistical methods are based on the assumption that the sample is free from bias and sample size is sufficient for accurately measuring desired effect size and reporting the research data in a concise, precise, and coherent manner in order to facilitate understanding by the readers. Depending on the type of study design, definite guidelines have been formed to bring global reporting uniformity. These are to be adhered to strictly. These issues are treated in this chapter. PLANNING OF A RESEARCH PROJECT Planning the research project is the most important part of any research work. At this juncture, it would be pertinent to quote Abraham Lincoln, the former president of the United States of America. He commented, “If I am given 8 hours to cut a tree, I would like to spend 6 hours sharpening my ax.” This statement is more true for research than for anything else [33, 34]. For the convenience of description and understanding, the research project can be divided into the following artificial sections, although overlapping is unavoidable. S.S.Patel & Aditya Patel All rights reserved-© 2023 Bentham Science Publishers

100 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

REVIEW OF LITERATURE RESEARCH PROBLEM

OF

AND

IDENTIFICATION

THE

The common practice, more applicable to the researchers in health sciences, is that while reviewing an article, the researcher generally skims the article without going through the details of the study design, sample size, Test of significance, and power of the study. It is important to remember that the sample size for comparing the mean in three groups would differ from that of two or more groups. The said practice of skimming the research article during the literature review is not justifiable unless it is evident that the conclusions drawn by the researcher are valid and justified to be incorporated into the process of decision-making [35]. To do so, the reader (reviewer) should focus attention on the following: Whether the study design of the concerned research project is appropriate to answer the research question/questions, it is important to remember that no statistical analysis by any expert or statistician, can compensate for shortcomings in the study design with respect to sample size, various types of biases, and poverty of observations. The research article should be reviewed and evaluated with a critical eye to detect the desired level of evidence. ●

● ●

Whether the statistical analysis is appropriate in accordance with the data, whether numerical, ordinal, or categorical (non-parametric, dichotomous). Whether the results are statistically significant. If the results are statistically significant, whether the magnitude of change is worth it for its clinical application (application inpatient care) and for making administrative decision-making. It has to be remembered that statistical significance does not necessarily signify clinical significance; sometimes, the reverse may be true.

RESEARCH GAP ANALYSIS & FRAMING OF RESEARCH QUESTION The clarity of the research project to be undertaken, i.e., the problem to be solved or the research gap to be filled, is the most important part of any research work. In other words, it can be stated that the research question/question to be addressed is the beginning of a research project, and for framing the research question/generating a hypothesis appropriate and thorough review of the literature is required. The researcher can test the appropriateness of his or her research question in interventional research studies on the PICO format wherein:

Planning and Writing

Elements Of Clinical Study Design, Biostatistics & Research 101

P - Stands for Population I - for Intervention C - for Comparison O - Outcome CHOOSING THE APPROPRIATE STUDY DESIGN TO ANSWER THE RESEARCH QUESTION Commonly use of study designs in health science research can be grouped broadly into two groups: Experimental/Interventional studies It includes clinical trials or observational studies. The observational studies are cohort studies, case-control studies, surveys, case series & cross-sectional studies, as discussed in earlier chapters of this book. Clinical trials Clinical trials with controls and randomization, along with proper planning and properly conducted, are the study design of choice. These studies are particularly free from all biases & problems which are associated with other study designs. Therefore, RCTs provide the strongest evidence. Cohort Studies These are longitudinal studies. The cohort study could be prospective, retrospective, or amphispective. The Prospective Cohort Study design provides stronger evidence as compared to retrospective studies, as the researcher can exercise his control over biases. The retrospective or historical cohort suffers from biases. Hence, the evidence is weaker as compared to Prospective Cohort. The cohort studies are most suited to investigate the causes of a disease or risk factors responsible for the disease or to study the course of a disease [2]. Case-Control Study Design It helps in the determination of whether exposure is associated with an outcome. In this study design, the cases are the individuals who suffer from the disease (Outcome), and the controls are those without the disease. Afterward, the previous clinical history of both the cases and controls is analyzed to find out the probable cause of the outcome (disease). This study design is good for the generalization of a hypothesis or for investigating a preliminary hypothesis. Case-control studies can be utilized to study rare diseases, and these studies suffer from biases.They

102 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

require very good health record keeping.The practice of keeping the health record is yet to come to the expected standards in developing countries. The selection of the appropriate control group is a major challenge to this type of study design. Surveys & Cross-Sectional Studies These studies are similar to case-control studies. These are generally used for the determination of the status of a disease at a particular point in time. These studies have a limitation in that they provide information at a particular point in time. This can lead to an erroneous conclusion for generalized applicability. Case Series Case series, in the true sense, is not a research study. The case series is only good for the generation of hypotheses or framing research questions for further research. They provide very weak evidence. Meta-Analysis Generally, meta-analysis focuses on clinical trials, randomized controlled trials, or observational studies. The difference between the traditional review of literature or systematic review of literature & meta-analysis is that in Meta-Analysis quality of research is evaluated. In this process, the summary data is also quantified. This study assists in drawing better conclusions or resolving conflicting conclusions. However, it is generally agreed, by most statisticians, that Meta-Analysis cannot take the place of Randomized Controlled Trials as far as the level of evidence is concerned. Fishing Expedition Sometimes the researchers do not have clear-cut research questions for their research work. They collect data, and then they analyze data for results that are significant. This type of study is known as a fishing expedition. The most important shortcoming of this type of study is that the conclusions drawn suffer from the chance of occurrences. Therefore, such type of conclusions may be inaccurate, and they may not come true if the studies are replicated with planning. LEVEL OF SIGNIFICANCE & SELECTION OF SAMPLE SIZE Determination of the sample size, which is needed to detect a difference or to detect the effect of a given magnitude, is the most important information in medical research. It has been discussed earlier. If the sample size is insufficient, a significant difference can be missed. This is called a type two error. While calculating the sample size following should be considered and mentioned.

Planning and Writing

Elements Of Clinical Study Design, Biostatistics & Research 103

Power of the Study The power of the study detects a difference at X level of significance with the assumed standard deviation of Y. In general language, power is the ability of a study to detect a difference or an actual effect at an assumed standard deviation. This standard deviation is assumed based on a pilot study or reported earlier studies. Variation in Standard Deviation The power of the study, in general language, is the ability of a study to detect a difference or an actual effect while carrying out the study. If the deviation is found to be more as compared to the assumed standard deviation, the calculated sample size is required to be increased in accordance. A hypothetical example will make the above statement clear. Suppose a study is designed to have a power of 80% to detect a difference of 20mm on a scale at a 0.05 level of significance, assuming that the expected standard deviation is ±5mm. Accordingly, 35 subjects will be needed in each cohort. On actual conduction of the study, the variance of actual response is found to be greater than anticipated, i.e. more than the standard deviation of more than ±5 mm. The sample size is to be increased, i.e. more subjects with the same inclusion and exclusion criteria are to be included. The quality of the sample should be representative and free from bias. BIASES IN THE STUDY Bias means prejudice; it is related to the errors related to the difference in the sample and targeted population. Bias is a threat to the validity of the study Procedure Bias The procedure bias occurs when the subjects in the two groups are not treated in the same manner. Recall Bias The recall bias occurs when the patients are asked to recall certain events of the past. In this situation, it is possible that subjects in one group may be more likely to remember the event. For example, people may take aspirin for many reasons. Patients diagnosed as suffering from acid peptic disease may remember the ingestion of aspirin with greater accuracy than those without suffering from peptic ulcers.

104 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

Insensitive Measurement bias This may occur when the instruments used for measuring the characteristic of interest may not be able to detect the difference properly. Detection bias is important in monitoring the survival under procedure or on a new drug. For example, with a sensitive procedure, the disease or condition may be diagnosed in a very early stage. This will falsely increase the survival rate. Compliance Bias This occurs when one drug or procedure is easier or more pleasant to comply with as compared to the other. Selection bias This is related to the selection of a subject or selection of procedure to measure a variable. Admission Rate Bias Admission rate bias can occur if the study admission rate differs, which causes a major distortion in the risk ratio. APPLICATION OF STATISTICAL PROCEDURE AND TEST OF SIGNIFICANCE As an oversimplification, Statistics in medicine is used for the following three main purposes: ● ● ●

To answer research questions concerning differences To answer research questions related to association To control confounding factors to make predictions

The observation of the research study could be (Chapter 2) ● ● ●

Nominal/Categorical Ordinal or Numerical

Multiple Tests of Significance Multiple tests in statistics result in increased chances of making a type one error or false positive significance. The best method to avoid this problem is the application of a global test with an analysis of variance prior to making individual

Planning and Writing

Elements Of Clinical Study Design, Biostatistics & Research 105

group comparisons. An appropriate method to analyze multiple variables should be applied. Multiple comparisons can occur in situations wherein the researcher perform many subgroup comparison. Either multivariate methods or clearly stated prior hypotheses are needed in this situation [36]. REPORT WRITING &DATA PRESENTATION As far as publication in journals is concerned, presently, practically every journal prescribes its own format, and the researcher has to abide by that which is binding for publication [36]. If writing a thesis is concerned, a scholar has to present research work per the guidelines of the University/evaluating institute. In general, the chapters are as under: ● ● ● ● ● ● ●

● ● ●

Introduction Review of literature Research gap analysis Framing of a research question Generation of hypothesis Aim& objectives Material and methods including study design sample size, statistical Test, and other material & methods (procedure) as applicable. Observation & discussion Conclusions Direction for further research.

SENDING RESEARCH WORK/ARTICLE FOR PUBLICATION To maintain uniformity in research publication, guidelines have been prescribed, and the researcher depending on his/her nature of study design, the study-specific guidelines required to be adhered to. Sources of guidelines for specific study design ●

Randomized control trials (RCTs) CONSORT guidelines Meta-analysis of observational studies in epidemiology MOOSE guidelines Observational studies in epidemiology STROBE guidelines Systematic reviews and Meta-analysis ❍

●

❍

●

❍

●

106 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

QUOROM guidelines Studies of diagnostic accuracy STARD guidelines ❍

●

❍

The observations and inferences are to be presented in figures and tables. The tables and figures should be clearly labeled so that they can be understood without referring to the text of the article. It is very important to see that there is no inconsistency in the information presented in tables or figures and the data discussed in the text. Though such inconsistency may arise because of typographical errors, such errors give the impression that the authors have reanalyzed and rewritten the results or that the research was carried out with carelessness in procedures. GUIDELINES FOR REPORTING STATISTICS ● ●

● ●

● ●

●

As far as possible, findings should be quantified. Indications of measurement errors or uncertainty, for example, Confidence limits, should be mentioned. Non-technical use of technical statistical terms must be avoided. Some examples are: ‘random,’ ‘normal,’ ‘significant,’ ‘co-relation,’ and ‘sample. The computer software used in statistical methods should be specified. For all ‘P’ values, the exact ‘P’ value must be mentioned as not less than 0.05 or 0.001. Mean the difference in continuous variables, proportions in categorical variables, and relative risks, including odds ratio and hazard ratio, must be accompanied by their confidence interval [37].

Presentation of Results ● ●

●

Results should be presented in a logical sequence giving important findings first. All data given in tables or graphs and figures need not be repeated as it interrupts the flow of text. If required, the data may be annexed. Only important observations should be summarized.

DISCUSSION AND CONCLUSION A researcher must remember that; ●

●

The most important point in writing a discussion is to ensure that there are no inconsistencies in the questions posed in the introduction and the data presented in the result. The second point to remember in a discussion is that there should be no extrapolation beyond the findings of the research work.

Planning and Writing ●

●

●

Elements Of Clinical Study Design, Biostatistics & Research 107

The researcher must ensure that there is no inconsistency in the data presented in the form of a table and graph with that mentioned in text form [38]. The researcher should also discuss the research question that arises from the study or remains unanswered. No other person than the researcher is in a better position to discuss these issues because the researchers are intimately involved in the study design, recording observations, and analyzing the data generated by their research work. The investigator should point out the strength and limitations of their study, particularly those which affect the conclusions.

CONCLUSION i. In Reading/Reviewing the medical literature, skimming should not be done. The researcher must thoroughly go through the details of the study design, sample size, a test of significance, and the power of a study. ii. Common Biases in a Study: • Procedural Bias • Recall Bias • Insensitive measure Bias • Compliance Bias • Selection Bias • Admission Rate Bias Report Writing: There should not be any discrepancy in data presentation in the form of Tables and figures and in text.

108

Elements Of Clinical Study Design, Biostatistics & Research, 2023, 108-114

APPENDIX-A ETHICAL GUIDELINES FOR CLINICAL RESEARCH INTRODUCTION “The Philosophy of Biomedical research is that the interest of the research subject be weighed heavier than the interest of science.” “Ethics is not definable, is not implementable, because it is not conscious; it involves not only our thinking but also our feeling” - Valdemar W. Setzer Ethics is the moral value of human behavior and the principles which govern these values. Every profession is bound by code of ethics (Greek word Ethos meaning Custom or Character) and which dates back to Hippocratic Oath, which is a guiding principle for the physician on professional ethics and mandates. The Hippocratic oath prescribes only beneficial treatments refraining from causing harm or hurt to their patients. This ultimately puts the interests of their patients above their own interests. When the clinician assumes himself the role of the researcher, the situation becomes complicated. It was realized that a code of ethics for clinical research was needed and to address the need Good Clinical Practice (GCP) guidelines for human research were framed. The role of the ethics committee has become supremely important. Thus, knowledge about the ethics committee and its functioning is not only the administration's advantage but also important from the researcher's viewpoint.

PRINCIPALS OF ETHICS (CORE VALUES) Ethical principles of research in the Biomedical Research are based on three core values: ● ● ●

Respecting the autonomy of research subjects Avoiding harm Privacy and data protection

Respecting the Autonomy Of Research Subjects Voluntary participation Participation in research should be voluntary and based on informed consent. An exception from the principle of voluntary consent can be made when research is conducted on published and public information and archived materials. Research concerning official registries and documents and carried out without the consent of research subjects is governed by legislation. S.S.Patel & Aditya Patel All rights reserved-© 2023 Bentham Science Publishers

Patel and Patel

Elements Of Clinical Study Design, Biostatistics & Research 109

Research subjects can give consent orally or in writing, or their behavior can otherwise be interpreted to mean that they have given consent to participate. For example, assenting to a polite request for an interview or responding to a questionnaire or request for a written response indicates that the subject has consented to be studied. ●

●

●

●

●

●

●

In institutional settings (prisons, child protection institutions, hospitals, homes for the elderly,etc) it is important to make sure that consent is given voluntarily by each and every subject. In evaluating the matter, attention must also be paid to the nature of the study, i.e .the degree to which personal matters are dealt with (need to protect privacy). If the research intervenes in personal integrity, it is particularly important to ensure the genuineness of consent. On the whole, researchers should always take into account the constitutional rights guaranteed to each individual. If research intervenes in the physical integrity of subjects, consent must always be given in writing or in some other certifiable way, unless this is contrary to the interests of subjects. For example, a person with AIDS may not want his or her name registered on written consent. Consent can be specific or general. General consent applies to research use in general. General consent can include conditions regarding the form in which data are recorded and archived and conditions set for the use of data in secondary research. If the information obtained from subjects is combined with information in official registers, subjects must be given detailed information on the registers that will be used. Specific consent concerns the use of information in a particular study. Specific consent with regard to the use of data may be justified on the grounds that data cannot be anonymized and that archiving the data with identifiers for secondary research would in all likelihood be harmful to subjects. Subjects have the right to withdraw from a study at any stage, but this does not mean, however, that their prior input (interviews,etc) cannot be used in the study.

Autonomy and research involving minors According to Indian Good Clinical Practice Guidelines and ICMR guidelines children must be treated equally and as individuals and must be allowed to influence matters pertaining to them to a degree corresponding to their level of development. In practice, it cannot be assumed that researchers should always request separate consent from a guardian when research involves minors. First, according to the above-mentioned principles children should be able to influence matters pertaining to themselves to a degree corresponding to their level of development. Second, there are situations where there may be differences in values and interests between a guardian and a minor, and requesting the guardian's consent may endanger the collection of comprehensive research data on the conditions and behavior of minors, thus restricting the freedom of science, which is guaranteed by the Constitution. Third, there are studies that do not include risks, and where requesting consent from the guardian would be difficult.

110 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

Autonomy and Research in Schools Many studies that are conducted in schools and institutions of early childhood education and care can be carried out as part of the normal work of the institution or school. It is not necessary to request a guardian's permission if the director of an institution of early childhood education and care or the headteacher of a school has evaluated that the study would produce useful information for the institution or school and can be carried out as part of the normal activities of the institution or school. For example, observations, broad questionnaires, and open interviews which do not collect directly identifying information(names, IDs, addresses) for research purposes can be carried out without the consent of parents or some other guardian. In other cases, they must be informed of the study.

Age When studying minors outside an institution of early childhood education and care or school, researchers must themselves evaluate when it is necessary to ask for a guardian's separate consent or inform a guardian of the study so that the guardian can forbid the child from participating in the study. A study involving children under the age of 15 can be conducted without a guardian's separate consent or informing a guardian if this is justified from the viewpoint of: ● ● ●

the age and development level of subjects the subject and research method or the need for information

If a study is to be conducted without a guardian's separate consent or informing a guardian, an ethical review must be requested for studies involving subjects under the age of 18. Researchers must always respect a minor's autonomy and the principle of voluntary participation, regardless of whether a guardian's consent has been obtained or not.

Information to subjects The information that must be provided to research subjects depends on the nature of data collection methods. In studies based on observation, interviews, or questionnaires, subjects must be told what the study is about and what participating in the study means in concrete terms and how long it will take. Information regarding a study should include at least the following:

1. 2. 3. 4.

The researcher's contact information. The research topic. The method of collecting data and the estimated time required. The purpose for which data will be collected, how it will be archived for secondary use, and 5). the voluntary nature of participation.

Patel and Patel

Elements Of Clinical Study Design, Biostatistics & Research 111

Exceptions from informed consent An exception from the principle of informed consent can be made if advance information would distort the results of the study. As a matter of principle, studies on the use of power should be allowed without the consent of those in power. There are also groups and subcultures that researchers cannot approach without using an assumed identity for the sake of their own safety. An ethical review must be requested from the ethics committee, if a study deviates from the principle of informed consent, an ethical review from the research ethics committee is always required.

Avoiding Harm Possible harm resulting from research can stem from the collection of data, the storage of data and consequences following the publication of studies. The harm could be: ● ● ●

Mental harm Financial Harm Social Harm

Privacy and Data Protection Research ethics principles concerning the protection of privacy fall into three categories: ● ● ●

Protecting research data and confidentiality Storing or disposing of research data and Research publications

Protecting research data and confidentiality The protection of data with identifiers must be carefully planned. The protection of subjects' privacy may not be jeopardized by the careless storage of data or unprotected electronic data transfers.[36,38]

Storing/ destroying research data Research is not always repeatable, but the scientific community should have the possibility, if necessary, to verify research findings from the data analyzed in a study. Openness is a key characteristic of science and also a precondition for testing the validity of scientific information, critically evaluating information, and advancing science. Data that are carefully archived for secondary research reduce the need to collect research

112 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

data containing identifiers. Archiving also reduces the research pressure on small population groups. It is particularly important to archive for secondary research data that have cultural, historical, and/or scientific value.[38] When necessary, the protection of privacy should be ensured through anonymization measures and through the regulation of access to data for secondary research.

Protecting privacy in research publications Unlike research data, research publications are in the public domain. The need to protect privacy in publications must be evaluated on a case-to-case basis. For most studies, there is no need to present subjects in an identifiable way in published findings. The results of quantitative research are reported statistically, which means that there is no risk of identification even when the publication is based on data containing identifiers. In the case of qualitative data, the risk of identification must always be evaluated before any samples/quotations from the data are published: what indirect identifiers (workplace, school, place of residence, age, profession, etc) will be left in the sample as such, what will be masked and what will be omitted altogether. If research concerns archived materials, the identifiability or non-identifiability of subjects in research publications depends on the conditions the distributing archive has set on the use of the data.

ETHICAL REVIEW Ethical Review Required The researcher is always responsible for the ethical and moral solutions in a study. Researchers must submit their research plan to ethical review if a study contains any of the following features: ● ●

●

●

●

●

The study involves an intervention in the physical integrity of subjects. If the study deviates from the principle of informed consent (ethical review is not required if the research is based on public documents, registries, or archived data). The subjects are children under the age of 15 and the study is not part of the normal activities of a school or an institution of early childhood education and care and the data are collected without parental consent and without providing the parents or guardians the opportunity to forbid the child from taking part in the study. The study exposes research subjects to exceptionally strong stimuli and evaluating possible harm requires special expertise (for example studies containing violence or pornography). The study may cause long-term mental harm (trauma, depression, sleeplessness) beyond the risks encountered in normal life. The study can signify a security risk to subjects (for example studies concerning domestic violence).[38]

Patel and Patel

Elements Of Clinical Study Design, Biostatistics & Research 113

Concerns of the Institutional Ethics Committee An ethical review examines the plan for collecting data, how the study will be carried out, the information that will be given to subjects, and the plan for processing and storing data from the perspective of avoiding risks and harm. The review weighs possible negative effects or harm to subjects resulting from participation in the study in relation to the intended scientific value of the study. If necessary, ethical guidelines in the particular research field should also be applied. In the Research, evaluating scientific value and risks is not a utilitarian cost-benefit analysis but rather a question of normative evaluation of values that are in themselves incommensurable. Evaluation ensures that a study does not contain unnecessary risks that could be avoided without reducing the scientific value of the study. Next, one must decide whether risks, on the whole, are morally acceptable. Research which entails higher risks may be morally acceptable if the scientific value of the study is very high, and the study does not cause harm to subjects (studies that are not based on informed consent. If a study does not have the features listed above (1-6) and does not present a risk of causing long-term mental harm beyond the risks encountered in normal life, this should be mentioned in the request for an ethical review. In this case, the committee will primarily evaluate the information supplied to subjects as well as matters concerning privacy and data protection. If a study entails any of the features listed above (1-6), the ethics committees must also evaluate the proposed research methods in relation to research questions and the value of the new information that the study is intending to provide.

Information to the research subjects The committee will check whether the informing of research subjects is planned appropriately. Information regarding a study includes at least the following: ● ● ● ●

●

The researcher's contact information, The research topic, The method of collecting data and the estimated time required, The purpose for which data will be collected, used in secondary research and archived, and The voluntary nature of participation.

In interventional studies, sufficient information must be provided concerning the design of the experiment. Experimental designs vary considerably from one research field to another. Ethics committees will determine whether the proposed level of information is adequate. If a study intervenes in the physical integrity of subjects, the information given to subjects must comply with the guidelines issued based on the Act on Medical Research, as far as these apply.

114 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

Privacy and data protection An ethical review examines a study's data management plan and ensures that technical data security solutions have been planned. The data management plan must describe: ● ●

●

How data containing identifiers will be protected or identifiers removed, Whether signing a pledge of confidentiality will be required from persons using or processing the personal data and The plan for archiving the data for secondary research or alternatively destroying personal data after the study has been completed.

Ethics committees do not review the protection of privacy in research publications. Researchers and editors are responsible for compliance with ethical principles in research publications.

Elements Of Clinical Study Design, Biostatistics & Research, 2023, 115-126

115

GLOSSARY (A). Absolute Risk Increase (ARI)- The increase inside treatment/procedure as compared with without new therapy.

effect

(risk)

with

new

Absolute Risk Reduction (ARR)- It is the difference between the event rate in the interventional group and the event rate in the control group it is also known as absolute risk difference. Absolute Value- It is the positive value assigned to a number iris respective of the original positive or negative value of the number. Addition Rule- It is the rule for calculation of probability of two mutually exclusive event the probability is calculated by adding. Alpha Error- It is wrong rejection of Null Hypothesis. Also called type I error. Alpha Value- The level of alpha selected in hypothesis testing. Alternative hypothesis- Alternative hypothesis it is the research hypothesis (alternative to) Null Hypothesis. Amphispective Study- Amphispective means combination of retrospective and prospective study. Analysis Covariance (ANCOVA)- Special type of analysis of variance used to controlled effect of confounding factor. Analysis of Variance-It is statistical procedure determine if there is any difference among two or more groups. (B). Bayes’sThorem – Mathematical formula to calculate conditional probability of one event from the given probability of the other event. Bell shaped distribution – It is a term used to describe shape of Gaussian distribution. Beta Error – It is the error related to wrong acceptance of Null Hypothesis. Also cold type II. Bias- Any influence that distorts the results of a research study. It is related to the ways the sample and target population or the control and the interventional group differs. Binary Observation – observation which has only two out comes also known as nominal observation best example is gender either male or female. S.S.Patel & Aditya Patel All rights reserved-© 2023 Bentham Science Publishers

116 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

Biometric- Methods of the study measures and statistical analysis in medicine and biological sciences. Biostatistics- Research study design and application of in statistical procedure in biological sciences. Blind Study – Study design in clinical trial to prevent bias that he researcher and the subject involved in the research presses or not aware of the treatment or interventional that a participant is receiving. (C). Case series – An observational study in a group of cases where in interesting features are observed. Cases – Object are entities whose behavior or characteristics we study. Categorical Variable- It is an observation nominal scale which false in category it is also called as qualitative observation. Central Limit Theorem – A mathematical derivation which proves that the derivation of the mean is approximately normal if the sample size is 30 or more, provided that the character stick is normally distributed in the population. Central Tendency – a major of centrality of a set of measurements. Three main measures of centre tendency are mean median and mode. Chi Square Test – Statistical applied to nominal observations or proportions or characteristics which or not dependent. Clinical Study- An experimental (interventional) study of a drug or procedure in which the study subjects are humans. Clinical Trial – An experiment designed to test the efficacy or effectiveness of a clinical treatment. Clinically Important Difference- A difference in a quantitative variable which will be clinically important to those in whom the variable is measure and in target population. Closed Question- A question to which individual or required to gift own of the several answers is specified in advanced by investigator. Cohort-A group of subjects, having common characteristics who remain together in the study over a period of time. Cohort Study – A quantitative observational is study in which a group of individual (The Cohort) or followed of over a period of time and measurement or taken at several times. Complementary Event – An event which is apposite to the event of the study.

References

Elements Of Clinical Study Design, Biostatistics & Research 117

Concurrent Control – Control group of subject being studied along with interventional group at the same period of time. Conditional probability – Probability of one event calculated by the given probability of other event. Confidence Interval – interval computed from the sample data that has a given probability that the unknown parameter, such as the mean or proportion, is contained within the interval. Commonly used confidence interval or 90%, 95% and 99%. Confidence Limit – The limits of confidence interval. This are computed from data of the sample of the study. Confounding Variable - A variable, other than the variables under investigation, which is not control and which may distort the results of experimental research. CONSORT Guideline - The full form is Consolidated Standards Of Reporting Trials. Controlled Event Rate – number of the subject in the controlled group who developed the outcome (variable) under study. Controlled Trial – A Trial in which interventional group is compeered with a similar non interventional group or with a reference standard control group. Correlation – The degree of association between two variables. A tendency for variation in one variable to we link to variation in a second variable. Correlation Coefficient - A measure of degree of relationship between two variables. It is also known as Pearson’s correlation coefficient. Covariate – A confounding variable which as potential to affect the outcome. Critical Appraisal – Interpreting the strength it weakness of the research process and applying the judgments to assts how used full the research is for practitioners. Critical Ratio – It is used in a statistical test for Z score. Critical Value – It is the value a test statistical must achieve for rejection of Null Hypothesis. Cross Over Study – It is a type of clinical trial in which the control group and the interventional group or crossed over in a time sequence. (D). Data Analyses – Processing, interpretation and analyses of findings. Deferential Bias – Bias that affects one group, in research study, different from another group. Degree of Freedom – A parameter used in t in t distribution and chi square distribution for

118 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

probability distribution. Dependent variable- In experimental research, the dependable variable is the variable presumed within the research hypothesis to dependent another variable (the independent variable). Descriptive Statist sic – Statistical method used to describe or surmised data collected from a specific sample. Discrete scale – It is type of numerical scale which is used to measure a character stick having integer (whole Number) value. Double Blind Study – A type of RCT were in neither the reaseather nor the subjects no which treatment he is receiving. Dunnat’s Procedure- It is a Statistical procedure to campair multiple interventional group with a control group i.e., –in dose response analyses dose is in depended variable and response is dependent variable. (E). Effect Size- It is a magnitude of a difference or relationship which is to be precisely measure in the interventional. It is used for calculation of sample size. Effectiveness – The extent to which an intervention, when used under ordinary circumstances, does what it is intended to do. Efficacy- The extent to which in interventional, when used under ideal circumstances, does what it is intended to do. Estimation- It is a Statistical process to draw conclusion from a sample information to it’s applicability in target population. Event – It is a outcome of in interventional study. Event should be single outcome or set of outcomes. Experiential Research – Also known as interventional research it is used to demonstrate cause and effect relationship. Experiment – A plant process of data collection. Experimental Event Rate– It is the number of subject in the interventional group to develop the outcome under study. (F). Face Validity- It is the correlation of the survey questionnaire with the purpose of study. Factor – In analysis of variance, factor is a characteristic of focus of enquiry.

References

Elements Of Clinical Study Design, Biostatistics & Research 119

False Negative – Term Describe when a subject received a negative test result but he does have the disease. False Positive – Term Describe when a subject has received a positive test result but actually he dose not have the disease. First Quartile – It Refers to 25th percentile frequency it is also called counts. It refers to no of times a given value of observation occurs. Fishing Expedition- Refers to research which is undertaken without a research question. Frequency polygons – In histograms it is a line connecting mead points of the columns. Frequency Table- A table which shows the percentage or number of observations a occurring at different values of a variable. (G). Gaussian Curve- It is a vell shaped crave of normal distribution. Geometric Mean – It is used for data recorded on log sale or skewed observation. Geometric mean is calculated by multiplies all observation the each other and then finding out the root value. Gold Standard – It refers to a test/treatment that may be considered to be best available (A Yard – Stick) against which the performances of other test/ treatment should be judged. (H). Hard Data – Precise data, like date of birth or blood group which is not open to interpretation by those providing the data. Histogram- histogram is a graph of frequency distribution of numerical observation. Historical Cohort – Previously collected observations, when used as controlled, to compare interventions in clinical studies are called historical cohort. Historical Cohort Study – It is a cohort studies which uses the previous data (case record) to determine effect of risk factors. Hypothesis Test – One of the approach to draft is statistical inference whether to reject Null Hypothesis or not to reject. (I). In depended Event – An outcome which is no effect on the probability of the other outcome. In depended Variable – It is a predictor variable in a study. It is also called as a factor in ANOVA.

120 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

Incidence- Number of new cases accounting per year. Inference– Process of Drawing Consultation. Intension TO Treat (ITT) – Statistical analysis of all subject in a group who were assigned the group in the beginning of the study. Inter Quartile Range – The Difference between 25th percentile and 75th percentile. Internal Constancy – Degree of Resemblance of the items to each other in a serve to measure a single character stick. Interval Estimate – When the variability is expressed in terms of range it is called as interval estimate. Intra ratter Reliability – The reliability of two measurement/ observations mead by two different persons. Intra ratter Reliability – The reliability of two measurement/ observation made by the same person at two different point of time. (K). Kappa – Kappa is a statistic which is use to measure interacterorintrarater agreement of nominal data. (L). Level of Significance – The probability of wrongly rejecting Null Hypothesis. It is also known as alpha or P value. Likely Hood Ratio – It is the ratio of true positive to true negative in a diagnostic test. Linear Regression – It is the process of determination of a regression or prediction of Y from X. Linear Relationship – A relationship which indicate that X and Y very together according to a constant increments. Logistic Regression – It is the regression technique used when the outcome is binary. Longitudinal Study – A study which is conducted for a long period of time. (M). Maching- A process by which individual within a quantitative research or found in to pairs. McNamara’s Test – The chi square test for comparing propositions from two paired groups. Mean – A descriptive used as a measure of tendency. All measurement in a set of

References

Elements Of Clinical Study Design, Biostatistics & Research 121

measurement or add together and divided by the number of subject. Measurement of Central Tendency – It is the index number which the notes the middle of distribution i.e., .mean, mode, median. Measurement of Dispersion – The index number which the notes the spread of observations above the central tendency (Usually Mean). Median – A descriptive static used has a measure of central tendency the median value in a list of quantitative measurement. Meta Analysis – A Statistical technique for combining and integrating the data collected from a number of experiential studies undertaken on a specific topic. Minimally Important Change (MIC) – The smallest difference in score in the domain of interest which patients fell as beneficial. Minimally Important Difference (MID) – The smallest difference between groups which may be considered to be of clinical importance. Mode – A descriptive statics used as a measure of central tendency. It is the value that occurrence most frequently in a distribution of measurement. Multiplication Rule – This rule a states that probabilities of two or more events accurse, is equal to the producer of the each event. Multivariate Analysis Of Variance (MANOVA) – It is advanced statistical methods applied in cases where there are multiple dependent variables and the independent variable are nominal. Mutually Exclusive Events – The events in which occurrence of one rules out the occurrence of other. (N). Nominal Scale – Nominal Scale is used for characteristic which have numerical values for example gender male or female birth and death. Non parametric Test – The statistical test which are used to analysis nominal or ordinal data. Non Probability Sample – A sample in which probability of selection of a subject is not known this sampling suffers from selection bias of the researchers. Normal Distribution – It is also known as Gaussian distribution it is symmetric distribution. Null Hypothesis – Statement that there is no difference in the means of interventional and control group. Number Needed to Harm (NNH) – It is the number of patient required to be treated to cause harms – side effect in one patient.

122 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

Number Needed To Treat (NNT)- It is the probable number of patient who are required to be treated to prevent or cure one patient. Numerical Scale – It is scale of measurement to measure a characteristic which has a numerical value. (O). Objective Probability – Probability calculated from observable events. Odds – In experimental research the number of times an events is observed divided by the number of times it not observed. Odds Ratio – It is the ratio of the to odds that is the odds that a patient was exposed to a given risk factor divided by the odds that a control was exposed to the risk facto. One Tailed Test – A test of hypothesis where in the Null Hypothesis is specifies direction only in one direction. One Way ANOVA – Analysis of variances in which means of three or more groups are compared. Open Ended Question – The type of question where it is left to the responded to answer the observational study a study in which investigators collect data but do not interning any process or treatment of the subject. Ordinal Scale – A scale of measurement were in the numbers or arbitrary. Outcome- The result of an experiment or trial. Outcome – It is the result of an experiment. (P). P Value – p the symbol for the probability that is associated with the outcome test. Parameter – This are character stick distribution in population like mean and standard deviation which are fixed as a custom statisticians used Greek letter for population parameter where as the mean and a standard deviation in a sample or called as estimate. The sample estimate are the noted by Roman letter. Percentage – It is the proportion multiplied by 100. Percentile – It is a number that the notes the percentage of distribution which is less than or equal to the number. Placebo – It is a sham doses from or prosier which externally a pairs like the real doses form are prosier used in clinical trials. Point Estimate – Mean standard deviation and proportions are called point estimate (A

References

Elements Of Clinical Study Design, Biostatistics & Research 123

common term used for the three). Poisson Distribution – It is the probability distribution of a rear event. Positive Correlation – A relationship between two variables where higher values of one variable are associated with higher values of the second variables. Power of the Study – The ability of a quantitative study to detected a difference or estimate to a given precision. Pragmatic Trial – A clinical trial conducted under Real World circumstances involving a wide range of participant which aims to asset the effectiveness of a treatment are intervention. Predictor – A variable that is assumed tiered to predict the outcome in a research study. Primary Outcome – The most important outcome that investigators bond to investigate. Probability Sampling – Any sampling scheme in which the probability of choosing each individual is the same (Or at list known, so it can be readjustedmathematically to be equal. Proportion – It is the observed number of the characteristic of interest divided by total number of observations. Procedure Bias – This bias across when the two groups or not treated in similarmanner. Prospective Cohorts – A longitudinal study in which study is design to collect data of cohorts after finalization of the study protocol. Prospective Study- The study which commences in a forward manners i.e., . data collection starts after the finalization of the study protocol. (Q). Qualitative Observation – Characteristic that are measured on a nominally scale. Quantitative Data – Quantitative observations or characteristic which are measure on a numerical scale. Quartile Test - The 25th percentile is called as first quartile 5th percentile is called as second quartile and 75th percentile is called as 75th percentile. (R). Random Assignment – The allocation of the subject to experimental and control group throw computer generated process. Random Variable – A variable in which values are hypothesis to ochre according to a given frequency distribution. Range - The difference between the a small as an the largest observation.

124 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

Recall Bias – The bias which ochre because the research subjects can not exactly recall events behaviour or dates. Receiver Operating Caractor (ROC) – Curve it is a in diagnosis test it is plot of false positive on X access against true positive on by access. References Standard – Reference standard is the drug or prosier which is censured as the best amongst the available it is used as control to compare with the interventional drug or prosier. Regression- Case Linear regression. Relative Risk – The ratio of the incidences of the diesis in exposed person to the incidence of the diesis in unexposed person. Relative Risk Reduction (RRR) – The reduction in risk with a new therapy relative to the risk without new therapy can be it is the difference by between experimental event rate and the control event rate divided by control event rate Formula is RRR is equal EER mines CER divided by CER. Reliability – Reliability is concern with the consistency and dependability of a mesmerising instrument that is it an indication of the degree to which it gives the same answer over time. Representative Sample – Is sample that is similar to the population in which the finding of study are to be applied. Research Question – A problem for which the researcher wants the answer by an experiment. (S). Sample– Subset of population which is the focus of the research. Sampling Distribution – It is the frequency distribution of the statistic for many sample. Scale of Measurement – There are three types of a scale namely numerical, ordinal and nominal. Selection Bias – Bias which occurs when the sample is not representative of the population that the researcher wishes to generalize. Sensitivity – It is the ability of the diagnostic test to detect True Positive. Significance Level – Establish at the out said by the researcher to test the hypothesis (e.g. 0.05 level or 0.01 significance level). Skewed Distribution – It is a distribution of observations in which a few outlying observationochre in one direction only. Specificity – It is the ability of the diagnostic test to detect True Negative (Negative in

References

Elements Of Clinical Study Design, Biostatistics & Research 125

Health). Standard Deviation – A descriptive statistic used to measure the degree of variability within a set of measurement. Standardised Mean Difference – It is also referred as the Effect Size (ES), this the mean difference divided by the standard deviation. Statistical Analysis – Statistical analysis is best on the principles of gathering data from a sample of individuals and using the same to make inference about the wider population. Statistical Significance – A term used to indicate whether the results of an analysis of data drown from a sample or unlikely to have been caused by chance at a specified level of probability (0.05 or 0.01). Statistical Test – A statistical procedure that allow a researcher to determine the probability that the result obtain from a sample have not arisen by chance rather they reflect genuine difference. Study Hypothesis – The hypothesis that has driven in investigator to undertake a quantitative study. Subject – A term most of an used in prospective research to describe those who participate in research. Subjective Probability – It is the best guest depending on the person’s previous experiences i.e., .provisional diagnosis in clinical setting. Their it the probability estimate reflecting the person’s opinion. Survey Research – A research design to collect systematically description of existing phenomena in order to describe what is going on. The data or obtain throw a question are. Systematic Review – A review of lecture best on a systematic method on searching for and identifying relevant literature and extracting data from the included literature. (T). T Distribution – A symmetrical distribution with mean 0 for a small sample size that is less then 30. As sample size increase t distribution text the shape of normal distribution. T Test – The statistical test for comparing a mean with a norm or for comparing to means with a small sampling size that is less than 30. Target population – The population to which the results of the sample are to be generalized. To way ANOVA – It is to away analysis of variance with to independent variables. True Positive – the term describe when a subject has received a positive test result and also has the disis true negative the term describe when a subject has received a negative test results and does not have the diesis.

126 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

Tukey’s (HSD) – It is a very popular method to make multiplepair wise compression. Type One Error – In error that ochre when a researcher reject the Null Hypothesis when it is true and concludes that a statistically significance difference/ relationship exits when it is not show. Type to Error - In error that ochre when a researcher aspect the Null Hypothesis when it is false and concludes that there is know significance difference/ relationship exits when it does. (U). Uncontrolled Study – An experimental study which is without control. (V). Validity– In term of research, validity research to the accuracy and truth of the data so that the feedings or reproduce bale. Variable – An attribute or capture stick of subject that text on different values for example age weight pulse rate. Variance – A measure of dispersions or variability. (W). Wilcoxan Rank Signed Rank Test – It is a nonparametric test used for todependent samples with ordinal data. Wilcoxan Rank Sum Test – It is a non parametric test to compare data of to independent samples it is also used for numerical data that are not normally distributed. (Z). Z distribution – The normal distribution with mean 0 and standard deviation 1. Z Ratio – It is calculated by subtractingthesypothezed mean from the observed mean and divided by the standard error of the mean. Z Ratio = Observed mean – Hypothesis.

Elements Of Clinical Study Design, Biostatistics & Research, 2023, 127-128

REFERENCES [1]

"Beth Dawson and Robert G. Trapp, Basic and clinical Biostatistics, 4th ed. Lange Basic Sciences :McGraw Hill : North America", 2004.

[2]

G. Bordage, and B. Dawson, "Experimental study design and grant writing in eight steps and 28 questions", Med. Educ., vol. 37, no. 4, pp. 376-385.

[3]

R.S. Greenberg, ”Prospective studies. In S. Kortz, NL.(editors),Encyclopedia of Statistical Sciences, Wiley ,vol.7, 315-319, 1986.

[4]

R.S. Greenberg, ”Retrospective studies. In S. Kortz, NL.(editors),Encyclopedia of Statistical Sciences, Wiley , vol.8, 120-124, 1988.

[5]

J. Lerman, "Study design in clinical research: sample size estimation and power analysis", Can. J. Anaesth., vol. 43, no. 2, pp. 184-191.1996,

[6]

J. LeLorier, G. Grégoire, A. Benhaddad, J. Lapierre, and F. Derderian, "Discrepancies between metaanalyses and subsequent large randomized, controlled trials", N. Engl. J. Med., vol. 337, no. 8, pp. 536-542.1997,

[7]

M.H. Briscoe, Preparing Scientific Illustrations:A Guide to Better Posters,Presentations & Publications.2nd edSpringer Verlag, .2nd ed1996,

[8]

G. Berry, J.N.S. Matthews, and P. Armitage, Statistical Methods in Medical Research.4th edBlackwell Scientific, .4th ed2001,

[9]

J.C. Bailar, and F. Mosteller, Medical use of statistics.2nd edMassachusetts Medical Society, .2nd ed1992,

[10]

J.J. Bartko, "Rationale for reporting standard deviations rather than standard errors of the mean", Am. J. Psychiatry, vol. 142, no. 9, p. 1060.1985,

[11]

S.N. Goodman, "Probability at the bedside: the knowing of chances or the chances of knowing?", Ann. Intern. Med., vol. 130, no. 7, pp. 604-606.1999,

[12]

L.D. Fisher, G.Van belle:Biostatistics:A Methodology for Health sciences.Willey, . 1996.

[13]

M.J. Gardner, and D.G. Altman, "Confidence intervals rather than P values: estimation rather than hypothesis testing", BMJ, vol. 292, no. 6522, pp. 746-750.1986,

[14]

TJ. Fagan, "Nomogram for Bayes Therom", N Eng J Med, 257-293, 1975.

[15]

M.J. Gardener, "Statistics :Study Design in clinical Research :sample size estimation and power analysis," CAN J ANAESTH, vol.43, no. 2, 184-91, Oct. 1995.

[16]

M.A. Gonzalo, C. Grant, I. Moreno, F.J. Garcia, A.I. Suárez, J.L. Herrera-Pombo, and A. Rovira, "Glucose tolerance, insulin secretion, insulin sensitivity and glucose effectiveness in normal and overweight hyperthyroid women", Clin. Endocrinol. (Oxf.), vol. 45, no. 6, pp. 689-697.1996,

[17]

T. Colton, Statistics in Medicine.Little, Brown, . 1974.

[18]

J. Nuovo, J. Melnikow, and D. Chang, "Reporting number needed to treat and absolute risk reduction in randomized controlled trials", JAMA, vol. 287, no. 21, pp. 2813-2814.2002.,

[19]

JL.Fleiss: for Statistical Methods Rate and Praportions, 2nd ed. Wiley, 1981.

[20]

D. Gillings, and G. Koch, "The application of the principle of intention to treat to the analysis of clinical trials", Drug Inf. J., vol. 25, no. 3, pp. 411-424.1991,

[21]

T. Greenhalgh, "Integrating qualitative research into evidence based practice", Endocrinol. Metab. Clin. North Am., vol. 31, no. 3, pp. 583-601, ix.2002, S.S.Patel & Aditya Patel All rights reserved-© 2023 Bentham Science Publishers

128 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

[22]

J. He, P.K. Whelton, B. Vu, and M.J. Klag, "Aspirin and risk of hemorrhagic stroke: a meta-analysis of randomized controlled trials", JAMA, vol. 280, no. 22, pp. 1930-1935.1998,

[23]

M. Müllner, H. Matthews, and D.G. Altman, "Reporting on statistical methods to adjust for confounding: a cross-sectional survey", Ann. Intern. Med., vol. 136, no. 2, pp. 122-126.2002,

[24]

T.V. Sakpal, "Sample size estimation in clinical trial", Perspect. Clin. Res., vol. 1, no. 2, pp. 6769.2010,

[25]

E. Christensen, "Methodology of superiority vs. equivalence trials and non-inferiority trials", J. Hepatol., vol. 46, no. 5, pp. 947-954.2007,

[26]

K. Loudon, F. Sullivan, P. Donnan, and E. Kevin, "Thorpe, “The PRECIS-2 tool: designing trials that are fit for purpose,”", BMJ, vol. 350, no. 2147, pp. 1-11.2015,

[27]

D.G. Altman, K.F. Schulz, D. Moher, M. Egger, F. Davidoff, D. Elbourne, P.C. Gøtzsche, and T. Lang, "The revised CONSORT statement for reporting randomized trials: explanation and elaboration", Ann. Intern. Med., vol. 134, no. 8, pp. 663-694.2001,

[28]

S. Patel, V. Mishra, and M. Deshmukh, ”Concept NoteOn Reverse Pharmacology, ”Copyright Office Government of India L-88924/2020,Jan. 01,2020,

[29]

J.L. Fleiss, Design and analysis of clinical experinments.Willey, .1999,

[30]

JH. Abramson, "”Survry Methods editionChurchill-Livingstone", 1999,

[31]

L.M. Rea, and R.A. Parker, Designing and conducting Survey Research; A comprehensive Guide.2nd edJossey-Bass, .2nd ed 1997.

[32]

J.J. Schlesselman, Case Control Studies:Design,Conduct.Analysis: Oxford, .1982,

[33]

A. Fink, and J. Kosecoff, How to conduct Survey:A step by step Guide.Sage Publication, .1998,

[34]

T. Greenhalgh, How to read a paper:Papers that reportdiagnostic or screening test. 1997.

in

Community

Medicine:Epidemiologic

studies,5th

[35]

T. Greenhalgh, How to read a paper: The basics of evidence based Medicine. 1997.

[36]

M.H. Katz, Multivariable Analysis: A Practcal guide for clinicians.Cambridge University Press, . 1999.

[37]

J.M. Kirkwood, J. Ibrahim, V.K. Sondak, M.S. Ernstoff, L. Flaherty, F.J. Haluska, M.B. Lens, and M. Dawes, "Use and abuse of statistics in evidence-based medicine", J. Clin. Oncol., vol. 20, no. 19, pp. 4122-4124.2002,

[38]

S. Saunders Dawson, "Azen,RS. Greenberg,AH.Reed,” The instruction of biostatisticsin medical school", Am. Stat., vol. 41, pp. 263-266.1987,

Elements Of Clinical Study Design, Biostatistics & Research, 2023, 129-133

129

SUBJECT INDEX A

C

Absolute risk reduction (ARR) 55, 60, 61, 70 Analysis 51, 52, 53, 55, 63, 64, 98, 104 of covariance 63 of variance (ANOVA) 51, 52, 53, 55, 64, 98, 104 ANOVA 51, 52, 53 nonparametric 53 one-factor 52 one-way 51, 52, 53 two-way 51, 52 Arithmetic 15, 16 average 16 operations 15 Aspirin 103 Ayurveda treatment 82 Ayurvedic therapies 80 AYUSH guidelines 82

Canonical correlation analysis (CCA) 64 Cardiac function 73 Cerebrospinal fluid 39 Chi-square 46, 47, 53, 63, 97 distribution 63 test 46, 47, 53, 97 Cluster sampling method 85 Cochrane tool for bias risk assessment 77 Co-efficient of variation (CV) 20 Comparing Ayurvedic care 82 Complementary medicine 81 Computer programs 49, 58, 62, 63 Consolidated standards of reporting trials 78 CONSORT 71, 78, 79, 82, 105 checklist 78 Group 78 guidelines 78, 79, 82, 105 Constitutional rights 109 Control event rate (CER) 55, 59, 60, 61, 70, 71 Conversion, peripheral 51 Coronary stenosis 63 Cronbach’s alpha test 5, 95

B Bayes’ theorem 34, 35, 36, 66, 70 Biases 10, 38, 77, 78, 99, 100, 101, 103, 104, 107 admission rate 104, 107 attrition 77 detection 77, 104 performance 77 recall 103, 107 reporting 78 Bias risk assessment 77 Blood pressure 16, 17, 19, 27, 42 systolic 16, 17, 19, 27, 42 Body 15, 39 fluid 39 mass Index 15 Breast cancer 15

D Data 108, 110, 111, 113, 114 collection methods 110 protection 108, 111, 113, 114 Designing survey tool 89, 90 Developing 90, 99 research projects 99 survey questionnaire 90 Deviation, standard normal 74 Diagnosis 36, 65 spinal malignancy 36 Diagnostic 35, 55, 64, 65, 67, 69, 71, 106 accuracy 106 procedure 55, 64 products 71

S.S.Patel & Aditya Patel All rights reserved-© 2023 Bentham Science Publishers

130 Elements Of Clinical Study Design, Biostatistics & Research

tests 35, 64, 65, 67, 69 Dichotomous 15, 63, 71, 74, 79, 100 Didactic teaching 44 Domestic violence 112 Drug 9, 12, 15, 27, 59, 61, 62, 71, 72, 73, 77, 82, 83 antihypertensive 27 herbal 71, 83 interactions 77 Dunnet’s test 31

E Early childhood education 110, 112 EBM 43, 44 data 44 teaching 43 training 44 Effect 72, 87 adverse 87 pharmacodynamic 72 Elements of pragmatic trials 79 Equivalence clinical trials 73, 75 Ethical principles 108, 114 of research 108 Ethics 108, 111, 113, 114 committees 108, 111, 113, 114 professional 108 Ethnobotanical studies 84 Event 32, 33, 34, 35, 39, 40, 54, 59, 60, 62, 66, 70, 103 adverse 59, 60 binary 40 complimentary 33 Evidence-based medicine (EBM) 43, 44 Experimental event rate (EER) 55, 59, 60, 61, 70 statistical treatment calculating 55

F Factors ANOVA 52 Fever 33 Fit test 63

Patel and Patel

Food and drug administration 73

H Hazard ratio 106 Health 39, 89, 98, 101, 102 record 102 sciences research 39, 89, 98, 101 Herbal 83, 85 medicinal product 83 remedy 83, 85 Hyperthyroidism 52 Hypothyroidism 51, 52

I Impaired glucose tolerance 52 Insulin secretion 52 Intake capacity 22 Interventional studies in medicine 9 Intra-ratter reliability 49

K Kruskal-Wallis one-way ANOVA 53

M Malignancy 36 Margin 74, 75, 76 acceptable 74, 75 non-inferior 76 non-inferiority 76 McNemar’s test 47 Measurement of dispersion 29 Measures 47, 61 epidemiological 61 nominal 47 Medicine 9, 14, 16, 17, 32, 37, 71, 73, 83, 86, 87, 88, 104 herbal 71, 83, 86, 88 plant 87 traditional 83

Subject Index

Elements Of Clinical Study Design, Biostatistics & Research 131

Mental harm 111, 112, 113 long-term 112, 113 Meta-analysis quality of research 102 Methods, ethnobotanical 84 MOOSE guidelines 105 Multi-activity device (MAD) 4, 47, 90 Multiple 53, 55, 58, 62, 63, 104 comparison procedure 53 regression 55, 58, 62, 63 tests of significance 104 Multivariant analysis of variance (MANOVA) 55, 64 Myocardial infarction 60, 61

N Nerve fiber 57 Non-parametric ANOVA 53 Nonparametric procedure 53 Null hypothesis 45, 46, 47, 48, 49, 53 Number needed to 55, 61, 62, 70 harm (NNH) 62 treat (NNT) 55, 61, 70 Numerical data 19, 22, 26 skewed 22 symmetric 22

O Observations 2, 16, 17, 18, 19, 20, 46, 47, 49, 50, 51, 52, 53, 55, 56, 58, 77, 97, 107, 110 curiosity-provoking 2 nominal 97 numerical 49, 55, 58 recording 107

P Pearson’s 49, 56 correlation coefficient 49 product 56 Peptic ulcers 103 Phyto-chemical analysis 86

Phytomedicine 87 Pilot testing 6 Plant remedy 87 Poisson distribution 39 Poliomyelitis 23 Pornography 112 Post-Hoc 53 comparison 53 Pragmatic randomised clinical trials 72 Probability of malignancy 36

Q Quality of life (QOL) 6, 77, 81, 95 QUOROM guidelines 106

R Random 10, 31, 38, 54, 63 allocation 10, 63 assignment 38 systematic sampling 31, 54 Randomised control trial (RCTs) 71, 72, 77, 78, 79, 81, 82, 87, 101, 105 Randomized 10, 71, 105 assignments 10 control trials 71, 105 double-blind clinical trial 10 Rare disease 3, 101 Real world evidence (RWE) 71 Receiver operating characteristic (ROC) 55, 68, 69 Regression measures 58 Relationship 51, 52, 55, 56, 57, 58, 59, 63, 64, 68, 70 bimodal 58 Relative risk (RR) 32, 47, 55, 58, 59, 60, 61, 106 reduction (RRR) 55, 60, 61 Renal failure 73 Research ethics 111 committee 111 principles 111 Respiratory 18

132 Elements Of Clinical Study Design, Biostatistics & Research

Patel and Patel

infections 18 tract infections 18 Retrospective 7, 84 cohort study 7 treatment outcome study 84 Reverse pharmacology 71, 82, 83, 84, 87, 88 approach 71, 88 concept 83 Revised balanced probe 91, 92 Rigid inclusion 79, 81, 87 application of 79, 87 applying 81 Rigid research conditions (RRC) 79 Risks, security 112

Superiority clinical test 73, 74 trials 73, 74 Superiority trial 74 for continuous variable 74 nominal variable 74 Surgical procedures 15, 61 Survey 4, 5, 6, 43, 89, 90, 94, 97, 98 observations 97 questionnaire 89 research questions 89, 90 Survey instrument 5, 94, 96, 98 measures 98

S

Temperature 57 linear-relationship-between-the 57 Therapies 15, 60, 70, 72, 79, 80, 81 adjunct 80 aspirin 60 complementary 79 Thyroid 51, 52 dysfunction 52 function 51, 52 gland 51 hormone replacement therapy (THRT) 51 Thyroiditis 52 Thyroperoxidase antibodies 52 TSH levels 51 Tukey’s HSD test 53

Scale 14, 15, 16, 19, 22, 24, 49, 67, 90, 91, 93, 94, 103 informative 16 logarithmic 19 metric 15, 16 numerical 14, 49, 67, 91 Search 99 engine 99 strategy 99 Selection of herbal remedy 83 Sensitivity 35, 36, 50, 55, 64, 65, 66, 68, 69, 70 diagnostic test’s 65 indicating lower 69 Sexual activity 92 Skin rashes 61, 62 developing 62 Sleeplessness 112 Software, computer 106 Spearman rank correlation 56, 58 STARD guidelines 106 Statistical 7, 20, 43, 47, 48, 49, 55, 63, 64, 74, 105 software 43 techniques 55, 63, 64 test 7, 20, 47, 48, 49, 74, 105 STROBE guidelines 105

T

V Variables 27, 39, 41, 49, 50, 51, 53, 55, 56, 58, 63, 95, 97, 104, 106 numerical 49, 55 Visual comparison 24 Vivax malaria 33

Subject Index

W WHO guidelines 83 Wilcoxon 47, 53 rank sum test 53 signed rank test 47

Elements Of Clinical Study Design, Biostatistics & Research 133