Menu Expand
BMJ Research Methods and Reporting: General topics & statistics volume 1

BMJ Research Methods and Reporting: General topics & statistics volume 1

Professor Adrian Hunnisett

(2016)

Additional Information

Book Details

Abstract

Sample size calculations: should the emperor's clothes be off the peg or made to measure? Implementation research: what it is and how to do it How to obtain the confidence interval from a P value The Cochrane Collaboration's tool for assessing risk of bias in randomised trials Clinical prediction rules Prognosis research strategy (PROGRESS) 1: A framework for researching clinical outcomes Out of sight but not out of mind: how to search for unpublished clinical trial evidence Interpreting diagnostic accuracy studies for patient care Demystifying trial networks and network meta-analysis Interpreting and reporting clinical trials with results of borderline significance Assessing equity in systematic reviews: realising the recommendations of the Commission on Social Determinants of Health Assessing the value of diagnostic tests: a framework for designing and evaluating trials Three techniques for integrating data in mixed methods studies Dangers of non-specific composite outcome measures in clinical trials Cascade diagrams for depicting complex interventions in randomised trials Recommendations for examining and interpreting funnel plot asymmetry in meta-analyses of randomised controlled trials Brackets (parentheses) in formulas Comparisons within randomised groups can be very misleading Uncertainties in baseline risk estimates and confidence in treatment effects How to obtain the P value from a confidence interval Diagnostic accuracy studies: how to report and analyse inconclusive test results Making inferences on treatment effects from real world data: propensity scores, confounding by indication, and other perils for the unwary in observational research The impact of outcome reporting bias in randomised controlled trials on a cohort of systematic reviews Prognosis research strategy (PROGRESS) 4: Stratified medicine research Value of composite reference standards in diagnostic research Statistics Notes: Missing outcomes in randomised trials Multiple imputation for missing data in epidemiological and clinical research: potential and pitfalls

Table of Contents

Section Title Page Action Price
Book Cover C
Title i
Copyright ii
About the publisher iii
About The BMJ iii
Contents iv
About the editors vii
Introduction to Research Methods and Reporting series viii
Chapter 1 Sample size calculations: should the emperor’s clothes be off the peg or made to measure? 1
An example 1
What is the distribution of blood pressure in the population you intend to study? 1
How much do you think your treatment will affect systolic blood pressure? 1
What α and β levels do you want? 1
Other approaches to sample size 2
Sample size norms for different designs 2
Differences between groups 2
Measured outcome variable 2
Binary outcome variable— proportions 2
Relations between continuous variables 3
Conclusions 3
Chapter 2 The tyranny of power: is there a better way to calculate sample size? 4
Power calculations 4
Confidence intervals 5
Base sample sizes on estimation 6
Chapter 3 How to obtain the confidence interval from a P value 7
(a) Calculating the confidence interval for a difference 7
(b) Calculating the confidence interval for a ratio (log transformation needed) 7
Limitations of the method 7
P values presented as inequalities 7
Chapter 4 How to obtain the P value from a confidence interval 8
(a) P from CI for a difference (no transformation needed) 8
(b) CI for a ratio (log transformation needed) 8
Limitations of the method 8
Chapter 5 Clinicians are right not to like Cohen’s κ 9
Introduction 9
Calculation of Cohen’s κ 9
Absolute and relative measures to quantify observer variation 10
Cohen’s κ is a reliability measure 10
Do clinical questions concern reliability or agreement? 10
Proportion of specific agreement as preferred measure of agreement 11
Conclusion 11
Chapter 6 The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials 13
Development of risk assessment tool 13
Evaluation phase 13
The risk of bias tool 14
Evaluation of initial implementation 15
Presentation of assessments 15
Summary assessment of risk of bias 16
Assessments of risk of bias and synthesis of results 16
Discussion 16
Chapter 7 Taking healthcare interventions from trial to practice 18
Study protocol 18
Study fidelity: planned versus actual treatment 18
Publication of single studies 19
Synthesis of evidence and systematic reviews 20
Mapping the components of an intervention 20
Taxonomies 20
Using the study 20
Chapter 8 Clinical prediction rules 22
Establishing a clinical prediction rule 22
Advantages and disadvantages of prediction rules 23
Types of prediction model 23
Scoring systems derived from univariate analysis 23
Prediction models based on multivariate analysis 23
Nomograms 23
Prediction using artificial neural networks 24
Decision trees (CART analysis) 24
Conclusion 24
Chapter 9 When is a further clinical trial justified? 26
Two problems, and solutions 26
The smallest worthwhile effect of exercise for chronic back pain 27
Existing evidence of the effect of exercise on chronic back pain 27
Does exercise produce worthwhile reductions in chronic back pain? 27
What influence would the findings of a new trial have? 27
Conclusions 28
Chapter 10 Out of sight but not out of mind: how to search for unpublished clinical trial evidence 30
Trial registries and results databases 30
Regulatory agencies 31
Contacting trialists and sponsors 32
Other sources of information 32
Conclusions 32
Chapter 11 Interpreting diagnostic accuracy studies for patient care 34
Reporting test accuracy at different thresholds 34
Presenting results at a single threshold 34
Presenting results at multiple thresholds 34
Presenting a performance measure combined across thresholds 35
Are false positive and false negative diagnoses equally important? 35
Presenting diagnostic accuracy for patients 35
Comparing the performance of two diagnostic tests 36
Paired measures at specific thresholds 36
Summary measure at specific thresholds: net benefit methods 36
Single measure averaged across multiple thresholds 36
Problems with ROC AUC for diagnostic performance 36
AUC or partial AUC? 36
Extrapolation beyond available data 36
Incorporating relative misclassification costs 37
Incorporating disease prevalence 38
Summary 38
Chapter 12 Demystifying trial networks and network meta-analysis 40
Introduction 40
Part 1: network geometry 40
Part 2: heterogeneity and incoherence 41
Part 3: data synthesis 42
Summary 43
Chapter 13 Interpreting and reporting clinical trials with results of borderline significance 44
What is the problem? 44
Using confidence intervals 44
Inconsistency in language in clinical trial reports 45
Possible solutions 45
Recommendations 47
Chapter 14 Statistics Notes: Missing outcomes in randomised trials 48
Chapter 15 Strategy for intention to treat analysis in randomised trials with missing outcome data 50
Attempt to follow up all randomised participants 50
Perform a plausible main analysis 50
Perform sensitivity analyses 50
Account for all randomised participants in the sensitivity analyses 50
Example of strategy in action 51
Discussion 51
Chapter 16 Multiple imputation for missing data in epidemiological and clinical research: potential and pitfalls 53
Consequences of missing data 53
Statistical methods to handle missing data 53
What is multiple imputation? 54
Pitfalls in multiple imputation analyses 54
Omitting the outcome variable from the imputation procedure 54
Dealing with non-normally distributed variables 54
Plausibility of missing at random assumption 55
Data that are missing not at random 55
Computational problems 55
Practical implications 55
Reporting in recent literature 55
Suggested reporting guidelines 56
Summary 56
Chapter 17 Assessing the value of diagnostic tests: a framework for designing and evaluating trials 58
Effect of tests on patient health 58
Direct test effects 58
Test procedure 58
Altering clinical decisions and actions 58
Feasibility and interpretability 58
Test accuracy, diagnostic yield, therapeutic yield, and treatment efficacy 59
Diagnostic and therapeutic confidence 59
Changing timeframes of decisions and actions 59
Influencing patient and clinician perceptions 60
Patients 60
Adherence to treatment 60
Doctors 60
Systemic approach to evaluating tests 60
Conclusion 62
Chapter 18 Three techniques for integrating data in mixed methods studies 64
Triangulation protocol 64
Following a thread 65
Mixed methods matrix 65
Conclusion 66
Chapter 19 Dangers of non-specific composite outcome measures in clinical trials 68
Motivating example: planning a clinical trial 68
Explaining treatment effects with a causal mechanism model 68
Composite outcomes 69
Misclassification of outcome 69
Examples from real trials 70
Discussion 70
Chapter 20 Cascade diagrams for depicting complex interventions in randomised trials 72
Hierarchical interventions 72
Trials 72
More complex interventions 73
Discussion 73
Chapter 21 Recommendations for examining and interpreting funnel plot asymmetry in meta-analyses of randomised controlled trials 76
What is a funnel plot? 76
Implications of heterogeneity, reporting bias, and chance 76
Heterogeneity 76
Reporting bias 77
Chance 78
Interpreting funnel plot asymmetry 78
Statistical tests for funnel plot asymmetry 78
Funnel plots and meta-analysis models 79
Fixed and random effects models 79
Extrapolation of a funnel plot regression line 79
Discussion 80
Chapter 22 Brackets (parentheses) in formulas 82
Chapter 23 Comparisons within randomised groups can be very misleading 83
Chapter 24 Uncertainties in baseline risk estimates and confidence in treatment effects 85
Risk of bias 85
Publication bias 86
Imprecision 86
Inconsistency 86
Indirectness 86
Discussion 87
Chapter 25 Value of composite reference standards in diagnostic research 89
What is a composite reference standard? 89
Developing a composite reference standard 90
Defining rules to combine component tests 90
Selection of component tests 90
Extensions to the basic composite reference standard 91
Missing values on component tests 91
Reporting guidelines 91
Conclusions and recommendations 91
Chapter 26 Diagnostic accuracy studies: how to report and analyse inconclusive test results 93
Inconsistent reporting of inconclusive test results in research on diagnostic accuracy 93
Defining inconclusive results 94
Types of inconclusive test results 94
Invalid inconclusive results 94
Uninterpretable and missing index test results 94
Valid inconclusive results 94
Continuous inconclusive index test results 94
How to report inconclusive results 95
Valid inconclusive results 95
Continuous inconclusive test results 95
Categorical/ordinal inconclusive test results 95
Invalid inconclusive results 95
Reporting uninterpretable and missing inconclusive test results 95
How to analyse inconclusive test results 96
Scenario 1: Exclude valid inconclusive results completely 96
Scenario 2: Exclude valid inconclusive results from binary statistics but report an additional summary statistic that accounts for them 96
Scenario 3: Group valid inconclusive results with positive or negative results 96
Analysis of continuous inconclusive test results 97
Conclusion 97
Making inferences on treatment effects from real world data: propensity scores, confounding by indication, and other perils for the unwary in observational research 99
Propensity score analyses 99
When things go awry 100
Unknown bias 100
Confounding by indication 101
When is it helpful to use a propensity score analysis? 101
More titles in The BMJ Series 103
More titles in The BMJ Research Methods and Reporting Series 104
More titles in The BMJ Research Methods and Reporting Series 105
More titles in The BMJ Easily Missed? Series 106
More titles in The BMJ Easily Missed? Series 107
More titles in The BMJ Easily Missed? Series 108
More titles in The BMJ Clinical Review Series 109
More titles from BPP School of Health 110
More titles in The Progressing your Medical Career Series 111
More titles in The Progressing your Medical Career Series 112
More Titles in The Progressing Your Medical Career Series 113
More titles in The Essential Clinical Handbook Series 114
More titles in The Essential Clinical Handbook Series 115