A Practical Guide to Cluster Randomised Trials in Health Services Research
Published on 9. January 2012
320 pages
978-1-119-96672-2 (ISBN)
Description
Cluster randomised trials are trials in which groups (or clusters)of individuals are randomly allocated to different forms oftreatment. In health care, these trials often compare differentways of managing a disease or promoting healthy living, in contrastto conventional randomised trials which randomise individuals todifferent treatments, classically comparing new drugs with aplacebo. They are increasingly common in health services research.This book addresses the statistical, practical, and ethical issuesarising from allocating groups of individuals, or clusters, todifferent interventions.
Key features:
* Guides readers through the stages of conducting a trial, fromrecruitment to reporting.
* Presents a wide range of examples with particular emphasis ontrials in health services research and primary care, with bothprinciples and techniques explained.
* Topics are specifically presented in the order in whichinvestigators think about issues when they are designing atrial.
* Combines information on the latest developments in the fieldtogether with a practical guide to the design and implementation ofcluster randomised trials.
* Explains principles and techniques through numerous examplesincluding many from the authors own experience.
* Includes a wide range of references for those who wish to readfurther.
This book is intended as a practical guide, written forresearchers from the health professions including doctors,psychologists, and allied health professionals, as well asstatisticians involved in the design, execution, analysis andreporting of cluster randomised trials. Those with a more generalinterest will find the plentiful examples illuminating.
Reviews / Votes
"There are several unique strengths to this book. Inparticular the authors are very experienced statisticians havingworked for many years in the design and analysis of clusterrandomized trials and have written excellent methodologicalarticles many of which are cited in their book." (Journal of Biopharmaceutical Statistics, 2012)More details
Other editions
Additional editions
Sandra Eldridge | Obioha Ukoumunne
A Practical Guide to Cluster Randomised Trials in Health Services Research
Book
02/2012
Wiley
€94.50
Shipment within 10-20 days
Software
01/2012
Wiley
€88.19
Shipment within 15-20 days
Persons
Sandra Eldridge - Senior Lecturer, Medical Statistics,Queen Mary, University of London, UK
Dr Eldridge is well experienced in designing and analyzing highquality cluster randomized trials, having worked on ten such trialssince the mid 1990s. An award-winning speaker on randomized trials,Dr Eldridge has given numerous presentations in many settings andconferences over the last seven years.
Content
Preface xiii
Notation xv
Table of cases: Trials used as examples in more than onechapter in the book xviii
1 Introduction 1
1.1 Introduction to randomised trials 2
1.2 Explanatory or pragmatic trials 2
1.3 How does a cluster randomised trial differ from othertrials? 3
1.4 Between-cluster variability 9
1.5 Why carry out cluster randomised trials? 10
1.6 Quality of evidence from cluster randomised trials 13
1.7 Historical perspectives 16
1.8 Summary 18
References 19
2 Recruitment and ethics 22
2.1 Selecting clusters and participants to enhance externalvalidity 22
2.2 Ethics of cluster randomised trials 24
2.3 Selection and recruitment of participants to enhanceinternal validity 35
2.4 Retention of participants in the trial 41
2.5 Summary 41
References 41
3 Designing interventions 44
3.1 Lack of effectiveness of interventions evaluated in clusterrandomised trials 45
3.2 What is a complex intervention? 46
3.3 Phases in the development of a complex intervention 50
3.4 Identifying evidence for potential intervention effect(pre-clinical phase) 50
3.5 Understanding more about intervention components (modellingphase) 53
3.6 Developing the optimum intervention and study design(exploratory trial phase) 55
3.7 What is the intervention? 57
3.8 Summary 58
References 58
4 Pilot and feasibility studies 60
4.1 What is a pilot study? 60
4.2 Reasons for conducting pilot and feasibility studies 63
4.3 Designing a pilot or feasibility study 69
4.4 Reporting and interpreting pilot studies 71
4.5 Summary 72
References 73
5 Design 74
5.1 Parallel designs with only two arms 75
5.2 Cohort versus cross-sectional designs 85
5.3 Parallel designs with more than two arms 88
5.4 Crossover designs 92
5.5 Further design considerations 95
5.6 Summary 96
References 96
6 Analysis 99
6.1 Data collection and management 99
6.2 Analysis - an introduction 101
6.3 Analyses for two-arm, completely randomised, stratified orminimised designs 104
6.4 Analyses for other designs 124
6.5 Intention to treat and missing values 129
6.6 Analysis planning 131
6.7 Summary 132
References 133
7 Sample size calculations 137
7.1 Factors affecting sample size for cluster randomised designs138
7.2 Calculating sample size using the intra-cluster correlationcoeffi cient 142
7.3 Sample size calculations for rates 145
7.4 Restricted number of clusters 146
7.5 Trials with a small number of clusters 149
7.6 Variability in cluster size 150
7.7 Comparison of different measures of between-clustervariability 154
7.8 Matched and stratifi ed designs 160
7.9 Sample size for other designs 166
7.10 Summary 169
References 169
8 The intra-cluster correlation coeffi cient 172
8.1 What is the ICC? 173
8.2 Sources of ICC estimates 175
8.3 Choosing the ICC for use in sample size calculations 179
8.4 Calculating ICC values 185
8.5 Uncertainty in ICCs 192
8.6 Summary 193
References 193
9 Other topics 196
Richard Grieve
9.1 Systematic reviews 197
9.2 Cost effectiveness analyses 207
9.3 Process evaluation 212
9.4 Monitoring 213
9.5 Summary 215
References 215
10 Trial reporting 218
10.1 Trial quality and reporting quality 218
10.2 Steps to improve trial reporting in the early stages of thetrial 227
10.3 Reporting randomised trials in journal and conferenceabstracts 230
10.4 Application of CONSORT statement to cluster randomisedtrials 232
10.5 Summary 262
References 263
Index 267
Notation
Subscripts
i Represents clusters j Represents individuals 1 Represents intervention arm 2 Represents control armFrequently used notation
k Number of clusters in each arm N Sample size (total number of individuals analysed or number of person years) for individually randomised trial Nc Sample size (total number of individuals analysed or number of person years) for cluster randomised trial m Cluster size (numbers analysed) if cluster sizes are fixed mi Number of individuals in the ith cluster Mean cluster size ? Intra-cluster correlation coefficient (ICC) ? Mean of outcome ?b2 Between-cluster variance (?b between-cluster standard deviation) ?w2 Within-cluster variance (?w within-cluster standard deviation) ? Probability of ‘success’ ? RateNotation in Chapter 4
p Proportion of successes in sample s Observed total standard deviation of outcome between participantsNotation in Chapter 6
? Constant in generalised estimating equations or mixed effects models ? Effect size eij Residual for jth individual in the ith cluster xij Dummy variable indicating intervention arm. When there are only two arms, xij takes the value 1 when the ith cluster is in the intervention arm, and 0 when the ith cluster is in the control arm Yij Value of outcome for the jth individual in the ith cluster Yi Mean value of outcome in the ith cluster ?i Mean effect of being in the ith cluster for cluster-specific models ?ij Probability of ‘success’ for the ith individual in the jth clusterNotation in Chapter 7
?m Correlation between matched pairs cvc Coefficient of variation of cluster sizes, cvcs Coefficient of variation of cluster sizes within strata sc Standard deviation of cluster sizes cv Coefficient of variation for the outcome, or cvm Coefficient of variation for the outcome within a matched pair t Number of person years per cluster th Harmonic mean of the number of person years per cluster S Number of strata Deff Design effect ma ?b1 Between-cluster standard deviation in the intervention arm ?b2 Between-cluster standard deviation in the control arm ?bm Between-cluster standard deviation for clusters within a matched pair ?w Within-cluster standard deviation (consistent with Donner) ?1 Mean in the intervention arm (note the subscripts for ? and ? have a different meaning in Chapter 7 from their meaning in Chapters 6 and 8, although the usage in each chapter is consistent with the subscripts section at the top of this list ) ?2 Mean in the control arm (see note above) ?1 Proportion in the intervention arm (see note above) ?2 Proportion in the control arm (see note above) ?1 Rate in the intervention arm (see note above) ?2 Rate in the control arm (see note above)Notation in Chapter 8
ps Probability that any two subjects from the same cluster have the same outcome po Probability that any two subjects from different clusters have the same outcome MSB Mean square between clusters MSW Mean square within clusters L Total number of intervention arms Ns Total number of individuals in a single population Nl Total number of individuals in arm l of a trial Yli Mean outcome in cluster i in intervention arm l Yl Mean outcome in intervention arm l Yij Value of the outcome for the jth individual in the ith cluster Yli Mean outcome in cluster i in intervention arm l Yl Mean outcome in intervention arm l L Total number of intervention arms Ns Total number of individuals in a single population Nl Total number of individuals in arm l of a trial mli Number of individuals in ith cluster in intervention arm l ?2 Total variance ? Constant in generalised estimating equations or mixed effects...