Lessons for cluster randomized trials in the twenty-first century: a systematic review of trials in primary care

Comparison of intraclass correlation coefficient estimates and standard errors between using cross-sectional and repeated measurement data: the Safety Check cluster randomized trial

Designing cluster randomized trials in clinical studies often requires accurate estimates of intraclass correlation, which quantifies the strength of correlation between units, such as participants, within a cluster, such as a practice. Published ICC estimates, even when available, often suffer from the problem of wide confidence intervals. Using data from a national, randomized, controlled study concerning violence prevention for children--the Safety Check--we compare the ICC values derived from two approaches only baseline data and using both baseline and follow-up data. Using a variance component decomposition approach, the latter method allows flexibility in handling complex data sets. For example, it allows for shifts in the outcome variable over time and for an unbalanced cluster design. Furthermore, we evaluate the large-sample formula for ICC estimates and standard errors using the bootstrap method. Our findings suggest that ICC estimates range from 0.012 to 0.11 for providers within practice and range from 0.018 to 0.11 for families within provider. The estimates derived from the baseline-only and repeated-measurements approaches agree quite well except in cases in which variation over repeated measurements is large. The reductions in the widths of ICC confidence limits from using repeated measurement over baseline only are, respectively, 62% and 42% at the practice and provider levels. The contribution of this paper therefore includes two elements, which are a methodology for improving the accuracy of ICC, and the reporting of such quantities for pediatric and other researchers who are interested in designing clustered randomized trials similar to the current study.

Hierarchical Bayesian modeling of random and residual variance-covariance matrices in bivariate mixed effects models

Bivariate mixed effects models are often used to jointly infer upon covariance matrices for both random effects (u) and residuals (e) between two different phenotypes in order to investigate the architecture of their relationship. However, these (co)variances themselves may additionally depend upon covariates as well as additional sets of exchangeable random effects that facilitate borrowing of strength across a large number of clusters. We propose a hierarchical Bayesian extension of the classical bivariate mixed effects model by embedding additional levels of mixed effects modeling of reparameterizations of u-level and e-level (co)variances between two traits. These parameters are based upon a recently popularized square-root-free Cholesky decomposition and are readily interpretable, each conveniently facilitating a generalized linear model characterization. Using Markov Chain Monte Carlo methods, we validate our model based on a simulation study and apply it to a joint analysis of milk yield and calving interval phenotypes in Michigan dairy cows. This analysis indicates that the e-level relationship between the two traits is highly heterogeneous across herds and depends upon systematic herd management factors.

The gain in quality-adjusted life months by switching to esomeprazole in those with continued reflux symptoms in primary care: EncomPASS--a cluster-randomized trial

OBJECTIVES

Proton pump inhibitors (PPIs) are effective in gastroesophageal reflux disease (GERD), but their cost effectiveness is unknown. This is usually determined by cost/quality-adjusted life year (QALY) gained, but whether PPI therapy improves QALYs has not been assessed in a randomized trial. The PPI acid suppression symptom (PASS) test is a five-item questionnaire that identifies patients with persistent acid-related symptoms. We evaluated whether a PASS test-based management strategy of changing GERD therapy to esomeprazole in those with continued symptoms on another PPI or H(2) receptor antagonist therapy would be cost effective. We expressed the data in terms of cost per quality-adjusted life months (QALM), as this was a 4-week trial.

METHODS

This is a multicenter, cluster-randomized, open-label study in primary care physician centers across Canada. Primary care physician centers were randomized to intervention or control arms. Patients on acid-suppressing medication were identified from primary care records and asked to complete the PASS test. PASS test failures at baseline assessment continued current therapy in control practices or switched to esomeprazole 20 or 40 mg daily (the dose was at the clinician's discretion) for 4 weeks in intervention practices. A planned secondary end point was QALM gain, measured using the validated Euroqol (EQ-5D) completed at baseline and 4 weeks. Medication use was also assessed by questionnaire. Canadian unit generic costs were applied to all GERD drugs, except to esomeprazole and lansoprazole, wherein proprietary costs were used (all costs in Canadian $). Data were analyzed using bootstrap sampling.

RESULTS

A total of 1,564 patients were recruited from 134 intervention sites and 92 control sites. Data were evaluable for 808 intervention and 445 control patients. The mean (±standard deviation) QALM at 4 weeks in the intervention group was 0.885±0.164 compared with 0.814±0.179 in the control group, resulting in a mean 0.071 (95% CI=0.091-0.051) QALM gain (P<0.0001). Esomeprazole was cost effective for PASS test failures, with a mean cost of $763 (95% CI=456-1,414) per QALM gain.

CONCLUSIONS

Esomeprazole was associated with a statistically significant gain in QALMs and was cost effective in primary care patients with persistent acid-related symptoms identified by the PASS test.

Community based lifestyle intervention for blood pressure reduction in children and young adults in developing country: cluster randomised controlled trial

OBJECTIVE

To assess the effectiveness of a community based lifestyle intervention on blood pressure in children and young adults in a developing country setting.

DESIGN

Cluster randomised controlled trial.

SETTING

12 randomly selected geographical census based clusters in Karachi, Pakistan.

PARTICIPANTS

4023 people aged 5-39 years.

INTERVENTION

Three monthly family based home health education delivered by lay health workers.

MAIN OUTCOME MEASURE

Change in blood pressure from randomisation to end of follow-up at 2 years.

RESULTS

Analysed using the intention to treat principle, the change in systolic blood pressure (adjusted for age, sex, and baseline blood pressure) was significant; it increased by 1.5 (95% confidence interval 1.1 to 1.9) mm Hg in the control group and by 0.1 (-0.3 to 0.5) mm Hg in the home health education group (P for difference between groups=0.02). Findings for diastolic blood pressure were similar; the change was 1.5 mm Hg greater in the control group than in the intervention group (P=0.002).

CONCLUSIONS

Simple, family based home health education delivered by trained lay health workers significantly ameliorated the usual increase in blood pressure with age in children and young adults in the general population of Pakistan, a low income developing country. This strategy is potentially feasible for up-scaling within the existing healthcare systems of Indo-Asia.

TRIAL REGISTRATION

Clinical trials NCT00327574.

Spatial effects should be allowed for in primary care and other community-based cluster RCTS

BACKGROUND

Typical advice on the design and analysis of cluster randomized trials (C-RCTs) focuses on allowance for the clustering at the level of the unit of allocation. However often C-RCTs are also organised spatially as may occur in the fields of Public Health and Primary Care where populations may even overlap.

METHODS

We allowed for spatial effects on the error variance by a multiple membership model. These are a form of hierarchical model in which each lower level unit is a member of more than one higher level unit. Membership may be determined through adjacency or through Euclidean distance of centroids or in other ways such as the proportion of overlapping population. Such models may be estimated for Normal, binary and Poisson responses in Stata (v10 or above) as well as in WinBUGS or MLWin. We used this to analyse a dummy trial and two real, previously published cluster-allocated studies (one allocating general practices within one City and the other allocating general practices within one County) to investigate the extent to which ignoring spatial effects affected the estimate of treatment effect, using different methods for defining membership with Akaike's Information Criterion to determine the "best" model.

RESULTS

The best fitting model included both a fixed North-South gradient and a random cluster effect for the dummy RCT. For one of the real RCTs the best fitting model included both a random practice effect plus a multiple membership spatial term, while for the other RCT the best fitting model ignored the clustering but included a fixed North-South gradient. Alternative models which fitted only slightly less well all included spatial effects in one form or another, with some variation in parameter estimates (greater when less well fitting models were included).

CONCLUSIONS

These particular results are only illustrative. However, we believe when designing C-RCTs in a primary care setting the possibility of spatial effects should be considered in relation to the intervention and response, as well as any explanatory effect of fixed covariates, together with any implications for sample size and methods for planned analyses.

Trials in primary care: statistical issues in the design, conduct and evaluation of complex interventions

Trials carried out in primary care typically involve complex interventions that require considerable planning if they are to be implemented successfully. The role of the statistician in promoting both robust study design and appropriate statistical analysis is an important contribution to a multi-disciplinary primary care research group. Issues in the design of complex interventions have been addressed in the Medical Research Council's new guidance document and over the past 7 years by the Royal Statistical Society's Primary Health Care Study Group. With the aim of raising the profile of statistics and building research capability in this area, particularly with respect to methodological issues, the study group meetings have covered a wide range of topics that have been of interest to statisticians and non-statisticians alike. The aim of this article is to provide an overview of the statistical issues that have arisen over the years related to the design and evaluation of trials in primary care, to provide useful examples and references for further study and ultimately to promote good practice in the conduct of complex interventions carried out in primary care and other health care settings. Throughout we have given particular emphasis to statistical issues related to the design of cluster randomised trials.

Electronic search strategies to identify reports of cluster randomized trials in MEDLINE: low precision will improve with adherence to reporting standards

Background

Cluster randomized trials (CRTs) present unique methodological and ethical challenges. Researchers conducting systematic reviews of CRTs (e.g., addressing methodological or ethical issues) require efficient electronic search strategies (filters or hedges) to identify trials in electronic databases such as MEDLINE. According to the CONSORT statement extension to CRTs, the clustered design should be clearly identified in titles or abstracts; however, variability in terminology may make electronic identification challenging. Our objectives were to (a) evaluate sensitivity ("recall") and precision of a well-known electronic search strategy ("randomized controlled trial" as publication type) with respect to identifying CRTs, (b) evaluate the feasibility of new search strategies targeted specifically at CRTs, and (c) determine whether CRTs are appropriately identified in titles or abstracts of reports and whether there has been improvement over time.

Methods

We manually examined a wide range of health journals to identify a gold standard set of CRTs. Search strategies were evaluated against the gold standard set, as well as an independent set of CRTs included in previous systematic reviews.

Results

The existing strategy (randomized controlled trial.pt) is sensitive (93.8%) for identifying CRTs, but has relatively low precision (9%, number needed to read 11); the number needed to read can be halved to 5 (precision 18.4%) by combining with cluster design-related terms using the Boolean operator AND; combining with the Boolean operator OR maximizes sensitivity (99.4%) but would require 28.6 citations read to identify one CRT. Only about 50% of CRTs are clearly identified as cluster randomized in titles or abstracts; approximately 25% can be identified based on the reported units of randomization but are not amenable to electronic searching; the remaining 25% cannot be identified except through manual inspection of the full-text article. The proportion of trials clearly identified has increased from 28% between the years 2000-2003, to 60% between 2004-2007 (absolute increase 32%, 95% CI 17 to 47%).

Conclusions

CRTs should include the phrase "cluster randomized trial" in titles or abstracts; this will facilitate more accurate indexing of the publication type by reviewers at the National Library of Medicine, and efficient textword retrieval of the subset employing cluster randomization.

Performance of analytical methods for overdispersed counts in cluster randomized trials: sample size, degree of clustering and imbalance

Many different methods have been proposed for the analysis of cluster randomized trials (CRTs) over the last 30 years. However, the evaluation of methods on overdispersed count data has been based mostly on the comparison of results using empiric data; i.e. when the true model parameters are not known. In this study, we assess via simulation the performance of five methods for the analysis of counts in situations similar to real community-intervention trials. We used the negative binomial distribution to simulate overdispersed counts of CRTs with two study arms, allowing the period of time under observation to vary among individuals. We assessed different sample sizes, degrees of clustering and degrees of cluster-size imbalance. The compared methods are: (i) the two-sample t-test of cluster-level rates, (ii) generalized estimating equations (GEE) with empirical covariance estimators, (iii) GEE with model-based covariance estimators, (iv) generalized linear mixed models (GLMM) and (v) Bayesian hierarchical models (Bayes-HM). Variation in sample size and clustering led to differences between the methods in terms of coverage, significance, power and random-effects estimation. GLMM and Bayes-HM performed better in general with Bayes-HM producing less dispersed results for random-effects estimates although upward biased when clustering was low. GEE showed higher power but anticonservative coverage and elevated type I error rates. Imbalance affected the overall performance of the cluster-level t-test and the GEE's coverage in small samples. Important effects arising from accounting for overdispersion are illustrated through the analysis of a community-intervention trial on Solar Water Disinfection in rural Bolivia.

Measures of between-cluster variability in cluster randomized trials with binary outcomes

Cluster randomized trials (CRTs) are increasingly used to evaluate the effectiveness of health-care interventions. A key feature of CRTs is that the observations on individuals within clusters are correlated as a result of between-cluster variability. Sample size formulae exist which account for such correlations, but they make different assumptions regarding the between-cluster variability in the intervention arm of a trial, resulting in different sample size estimates. We explore the relationship for binary outcome data between two common measures of between-cluster variability: k, the coefficient of variation and rho, the intracluster correlation coefficient. We then assess how the assumptions of constant k or rho across treatment arms correspond to different assumptions about intervention effects. We assess implications for sample size estimation and present a simple solution to the problems outlined.

Evaluation of cluster-randomized trials on maternal and child health research in developing countries

OBJECTIVE

To summarize and evaluate all publications including cluster-randomized trials used for maternal and child health research in developing countries during the last 10 years.

METHODS

All cluster-randomized trials published between 1998 and 2008 were reviewed, and those that met our criteria for inclusion were evaluated further. The criteria for inclusion were that the trial should have been conducted in maternal and child health care in a developing country and that the conclusions should have been made on an individual level. Methods of accounting for clustering in design and analysis were evaluated in the eligible trials.

RESULTS

Thirty-five eligible trials were identified. The majority of them were conducted in Asia, used community as randomization unit, and had less than 10,000 participants. To minimize confounding, 23 of the 35 trials had stratified, blocked, or paired the clusters before they were randomized, while 17 had adjusted for confounding in the analysis. Ten of the 35 trials did not account for clustering in sample size calculations, and seven did not account for the cluster-randomized design in the analysis. The number of cluster-randomized trials increased over time, and the trials generally improved in quality.

CONCLUSIONS

Shortcomings exist in the sample-size calculations and in the analysis of cluster-randomized trials conducted during maternal and child health research in developing countries. Even though there has been improvement over time, further progress in the way that researchers utilize and analyse cluster-randomized trials in this field is needed.

Evaluation of a standard provision versus an autonomy promotive exercise referral programme: rationale and study design

Background

The National Institute of Clinical Excellence in the UK has recommended that the effectiveness of ongoing exercise referral schemes to promote physical activity should be examined in research trials. Recent empirical evidence in health care and physical activity promotion contexts provides a foundation for testing the utility of a Self Determination Theory (SDT)-based exercise referral consultation.

Methods/Design

Design: An exploratory cluster randomised controlled trial comparing standard provision exercise on prescription with a Self Determination Theory-based (SDT) exercise on prescription intervention.

Participants: 347 people referred to the Birmingham Exercise on Prescription scheme between November 2007 and July 2008. The 13 exercise on prescription sites in Birmingham were randomised to current practice (n = 7) or to the SDT-based intervention (n = 6).

Outcomes measured at 3 and 6-months: Minutes of moderate or vigorous physical activity per week assessed using the 7-day Physical Activity Recall; physical health: blood pressure and weight; health status measured using the Dartmouth CO-OP charts; anxiety and depression measured by the Hospital Anxiety and Depression Scale and vitality measured by the subjective vitality score; motivation and processes of change: perceptions of autonomy support from the advisor, satisfaction of the needs for competence, autonomy, and relatedness via physical activity, and motivational regulations for exercise.

Discussion

This trial will determine whether an exercise referral programme based on Self Determination Theory increases physical activity and other health outcomes compared to a standard programme and will test the underlying SDT-based process model (perceived autonomy support, need satisfaction, motivation regulations, outcomes) via structural equation modelling.

Trial registration

The trial is registered as Current Controlled trials ISRCTN07682833.

The methodological quality of cluster randomised controlled trials for managing tropical parasitic disease: a review of trials published from 1998 to 2007

The aim of this review was to assess the methodological quality of cluster randomised controlled trials (CRCT) for the management of tropical parasitic disease published between 1998 and 2007. A literature survey was conducted using Medline for CRCTs of interventions aimed at managing any one of the six major tropical parasitic diseases: malaria, leishmaniasis, lymphatic filariasis, onchocerciasis, schistosomiasis and trypanosomiasis (Chagas disease). Information was extracted from the published articles in order that, for each trial, categorical responses could be made to a pre-specified list of 12 questions concerning issues relating to the methodological quality of the trial, including choice of design, generalisability, baseline assessment, blinding, use or non-use of a matched design, and accounting for the intraclass correlation in both design and analysis. The literature survey found 38 CRCTs. Of the 35 CRCTs that reported at least one human outcome, 27 were for interventions in the management of malaria whilst the rest were for managing leishmaniasis (4 trials), lymphatic filariasis (2 trials) and schistosomiasis (2 trials). For every one of the pre-specified questions that concerned an issue associated with methodological quality, the responses were consistent with the practice of trialists in relation to the given issue being generally poor.

Comparison of the risk difference, risk ratio and odds ratio scales for quantifying the unadjusted intervention effect in cluster randomized trials

This paper evaluates methods for unadjusted analyses of binary outcomes in cluster randomized trials (CRTs). Under the generalized estimating equations (GEE) method the identity, log and logit link functions may be specified to make inferences on the risk difference, risk ratio and odds ratio scales, respectively. An alternative, 'cluster-level', method applies the t-test to summary statistics calculated for each cluster, using proportions, log proportions and log odds, to make inferences on the respective scales. Simulation was used to estimate the bias of the unadjusted intervention effect estimates and confidence interval coverage, generating data sets with different combinations of number of clusters, number of participants per cluster, intra-cluster correlation coefficient rho and intervention effect. When the identity link was specified, GEE had little bias and good coverage, performing slightly better than the log and logit link functions. The cluster-level method provided unbiased point estimates when proportions were used to summarize the clusters. When the log proportion and log odds were used, however, the method often had markedly large bias for two reasons: (i) bias in the modified summary statistic used for cluster-level estimation when a cluster has zero cases with the outcome of interest (arising when the number of participants sampled per cluster is small and the outcome prevalence is low) and (ii) asymptotically, the method estimates the ratio of geometric means of the cluster proportions or odds, respectively, between the trial arms rather than the ratio of arithmetic means.

Consensus report of the coalition for clinical research-self-monitoring of blood glucose

The Coalition for Clinical Research-Self-Monitoring of Blood Glucose Scientific Board, a group of nine academic clinicians and scientists from the United States and Europe, convened in San Francisco, California, on June 11-12, 2008, to discuss the appropriate uses of self-monitoring of blood glucose (SMBG) and the measures necessary to accurately assess the potential benefit of this practice in noninsulin-treated type 2 diabetes mellitus (T2DM). Thirteen consultants from the United States, Europe, and Canada from academia, practice, and government also participated and contributed based on their fields of expertise. These experts represent a range of disciplines that include adult endocrinology, pediatric endocrinology, health education, mathematics, statistics, psychology, nutrition, exercise physiology, and nursing. This coalition was organized by Diabetes Technology Management, Inc. Among the participants, there was consensus that: protocols assessing the performance of SMBG in noninsulin treated T2DM must provide the SMBG intervention subjects with blood glucose (BG) goals and instructions on how to respond to BG data in randomized controlled trials (RCTs);intervention subjects in clinical trials of SMBG-driven interventions must aggressively titrate their therapeutic responses or lifestyle changes in response to hyperglycemia;control subjects in clinical trials of SMBG must be isolated from SMBG-driven interventions and not be contaminated by physician experience with study subjects receiving a SMBG intervention;the best endpoints to measure in a clinical trial of SMBG in T2DM include delta Hemoglobin A1c levels, hyperglycemic events, hypoglycemic events, time to titrate noninsulin therapy to a maximum necessary dosage, and quality of life indices;either individual randomization or cluster randomization may be appropriate methods for separating control subjects from SMBG intervention subjects, provided that precautions are taken to avoid bias and that the sample size is adequate;treatment algorithms for assessing SMBG in T2DM may include a dietary, exercise, and/or medication intervention, which are all titratable according to the SMBG values;the medical literature contains very little information about the performance of SMBG in T2DM from RCTs in which treatment algorithms were used for dysglycemic values; and research on the performance of SMBG in T2DM based on sound scientific principles and clinical practices is needed at this time.

Developing and evaluating complex interventions: the new Medical Research Council guidance

Evaluating complex interventions is complicated. The Medical Research Council's evaluation framework (2000) brought welcome clarity to the task. Now the council has updated its guidance

Internal and external validity of cluster randomised trials: systematic review of recent trials

OBJECTIVES

To assess aspects of the internal validity of recently published cluster randomised trials and explore the reporting of information useful in assessing the external validity of these trials.

DESIGN

Review of 34 cluster randomised trials in primary care published in 2004 and 2005 in seven journals (British Medical Journal, British Journal of General Practice, Family Practice, Preventive Medicine, Annals of Internal Medicine, Journal of General Internal Medicine, Pediatrics).

DATA SOURCES

National Library of Medicine (Medline) via PubMed.

DATA EXTRACTION

To assess aspects of internal validity we extracted data on appropriateness of sample size calculations and analyses, methods of identifying and recruiting individual participants, and blinding. To explore reporting of information useful in assessing external validity we extracted data on cluster eligibility, cluster inclusion and retention, cluster generalisability, and the feasibility and acceptability of the intervention to health providers in clusters.

RESULTS

21 (62%) trials accounted for clustering in sample size calculations and 30 (88%) in the analysis; about a quarter were potentially biased because of procedures surrounding recruitment and identification of patients; individual participants were blind to allocation status in 19 (56%) and outcome assessors were blind in 15 (44%). In almost half the reports, information relating to generalisability of clusters was poorly reported, and in two fifths there was no information about the feasibility and acceptability of the intervention.

CONCLUSIONS

Cluster randomised trials are essential for evaluating certain types of interventions. Issues affecting their internal validity, such as appropriate sample size calculations and analysis, have been widely disseminated and are now better addressed by researchers. Blinding of those identifying and recruiting patients to allocation status is recommended but is not always carried out. There may be fewer barriers to internal validity in trials in which individual participants are not recruited. External validity seems poorly addressed in many trials, yet is arguably as important as internal validity in judging quality as a basis for healthcare intervention.

Electronic medical record-assisted design of a cluster-randomized trial to improve diabetes care and outcomes

BACKGROUND

Electronic medical records (EMRs) have the potential to facilitate the design of large cluster-randomized trials (CRTs).

OBJECTIVE

To describe the design of a CRT of clinical decision support to improve diabetes care and outcomes.

METHODS

In the Diabetes Improvement Group-Intervention Trial (DIG-IT), we identified and balanced preassignment characteristics of 12,675 diabetic patients cared for by 147 physicians in 24 practices of 2 systems using the same vendor's EMR. EMR-facilitated disease management was system A's experimental intervention; system B interventions involved patient empowerment, with or without disease management. For our sample, we: (1) identified characteristics associated with response to interventions or outcomes; (2) summarized feasible partitions of 10 system A practices (2 groups) and 14 system B practices (3 groups) using intra-cluster correlation coefficients (ICCs) and standardized differences; (3) selected (blinded) partitions to effectively balance the characteristics; and (4) randomly assigned groups of practices to interventions.

RESULTS

In System A, 4,306 patients, were assigned to 2 groups of practices; 8,369 patients in system B were assigned to 3 groups of practices. Nearly all baseline outcome variables and covariates were well-balanced, including several not included in the initial design. DIG-IT's balance was superior to alternative partitions based on volume, geography or demographics alone.

CONCLUSIONS

EMRs facilitated rigorous CRT design by identifying large numbers of patients with diabetes and enabling fair comparisons through preassignment balancing of practice sites. Our methods can be replicated in other settings and for other conditions, enhancing the power of other translational investigations.

Intracluster correlation coefficients from the 2005 WHO Global Survey on Maternal and Perinatal Health: implications for implementation research

Cluster-based studies involving aggregate units such as hospitals or medical practices are increasingly being used in healthcare evaluation. An important characteristic of such studies is the presence of intracluster correlation, typically quantified by the intracluster correlation coefficient (ICC). Sample size calculations for cluster-based studies need to account for the ICC, or risk underestimating the sample size required to yield the desired levels of power and significance. In this article, we present values for ICCs that were obtained from data on 97,095 pregnancies and 98,072 births taking place in a representative sample of 120 hospitals in eight Latin American countries. We present ICCs for 86 variables measured on mothers and newborns from pregnancy to the time of hospital discharge, including 'process variables' representing actual medical care received for each mother and newborn. Process variables are of primary interest in the field of implementation research. We found that overall, ICCs ranged from a minimum of 0.0003 to a maximum of 0.563 (median 0.067). For maternal and newborn outcome variables, the median ICCs were 0.011 (interquartile range 0.007-0.037) and 0.054 (interquartile range 0.013-0.075) respectively; however, for process variables, the median was 0.161 (interquartile range 0.072-0.328). Thus, we confirm previous findings that process variables tend to have higher ICCs than outcome variables. We demonstrate that ICCs generally tend to increase with higher prevalences (close to 0.5). These results can help researchers calculate the required sample size for future research studies in maternal and perinatal health.

Training farmers in rational drug-use improves their management of cattle trypanosomosis: A cluster-randomised trial in south Mali

We carried out a stratified, cluster-randomised, controlled trial in south Mali in 2004 to evaluate the impact of providing information on the diagnosis and treatment of bovine trypanosomosis by farmers. We recruited cattle farmers (444) in 46 villages and used stratified, restricted-randomisation to assign villages to either the test or control group. Farmers in the test group received an information leaflet designed to address gaps in farmer knowledge likely to lead to inadequate treatment; their knowledge was assessed before the intervention, and at 2 weeks and 5 months after the intervention. We assessed the quality of farmer treatments by measuring clinical outcomes in cattle 2 weeks after selection and treatment. As an indicator of herd health, we assessed the mean hematocrit of the village herd before, and 5 months after, the intervention. To account for clustering, we analysed data using generalised estimating equations. Improvements in farmer knowledge of trypanosomosis diagnosis and treatment at 2 weeks and 5 months in the group receiving information were 23% and 14% greater at 2 weeks and 5 months, respectively. In the test group, 84% of farmer treatments were successful, compared to 73% in the control group. Giving rational drug-use information to farmers improved their knowledge and management of trypanosomosis as well as clinical outcomes in cattle they treated and had no discernible negative impacts.

Effect of general practitioner education on adherence to antihypertensive drugs: cluster randomised controlled trial

OBJECTIVE

To determine the impact of a simple educational package for general practitioners on adherence to antihypertensive drugs.

DESIGN

Cluster randomised controlled trial.

SETTING

Six randomly selected communities in Karachi, Pakistan.

PARTICIPANTS

200 patients with hypertension taking antihypertensive drugs; 78 general practitioners.

INTERVENTION

Care by general practitioners specially trained in management of hypertension compared with usual care.

MAIN OUTCOME MEASURE

Correct dosing, defined as percentage of prescribed doses taken, measured with electronic medication event monitoring system (MEMS) bottle.

RESULTS

200 patients were enrolled, and 178 (89%) successfully completed six weeks of follow-up. Adherence was significantly greater in the special care group than in the usual care group (unadjusted mean percentage days with correct dose 48.1%, 95% confidence interval 35.8% to 60.4%, versus 32.4%, 22.6% to 42.3%; P=0.048). Adherence was also higher among patients who had higher levels of education (P<0.001), were encouraged by family members (P<0.001), believed in the effect of drugs (P<0.001), and had the purpose of the drugs explained to them (P<0.001).

CONCLUSIONS

Special training of general practitioners in management of hypertension, emphasising good communication between doctors and patients, is more effective than usual care provided in the communities in Karachi. Such simple interventions should be adopted by other developing countries that are now facing an increasing burden of hypertension.

TRIAL REGISTRATION

Clinical trials NCT00330408 [ClinicalTrials.gov].

A simulation study of odds ratio estimation for binary outcomes from cluster randomized trials

We used simulation to compare accuracy of estimation and confidence interval coverage of several methods for analysing binary outcomes from cluster randomized trials. The following methods were used to estimate the population-averaged intervention effect on the log-odds scale: marginal logistic regression models using generalized estimating equations with information sandwich estimates of standard error (GEE); unweighted cluster-level mean difference (CL/U); weighted cluster-level mean difference (CL/W) and cluster-level random effects linear regression (CL/RE). Methods were compared across trials simulated with different numbers of clusters per trial arm, numbers of subjects per cluster, intraclass correlation coefficients (rho), and intervention versus control arm proportions. Two thousand data sets were generated for each combination of design parameter values. The results showed that the GEE method has generally acceptable properties, including close to nominal levels of confidence interval coverage, when a simple adjustment is made for data with relatively few clusters. CL/U and CL/W have good properties for trials where the number of subjects per cluster is sufficiently large and rho is sufficiently small. CL/RE also has good properties in this situation provided a t-distribution multiplier is used for confidence interval calculation in studies with small numbers of clusters. For studies where the number of subjects per cluster is small and rho is large all cluster-level methods may perform poorly for studies with between 10 and 50 clusters per trial arm.

Considering cluster analysis in sport medicine and injury prevention research

OBJECTIVE

In the field of sport medicine and injury prevention in sport, prospective study designs implementing cluster randomization or grouping of subjects by cluster (ie, team, clinic, school, community) are becoming increasingly common. However, there are very few published studies in the field that adequately account for clustering effects in the design and analysis, leading to potentially spurious conclusions. This paper will review the implications of using a cluster RCT or other intervention or observational design grouping individuals by cluster and to highlight the practical implications of appropriate analysis considering the effects of clustering.

DATA SOURCES/SYNTHESIS

Previously published papers have provided a foundation of expertise to discuss the often neglected impact of ignoring the effects of cluster in the design and analysis of cluster RCT and other study designs that group individuals by cluster in sport medicine.

RESULTS

The loss of statistical efficiency inherent when a study design implements randomization or grouping by cluster is reviewed. Specifically, the effect of cluster design on sample size considerations and analysis are discussed in the context of data from a recently published cluster RCT examining the effectiveness of a balance training prevention strategy in youth basketball.

CONCLUSIONS

Researchers in sport medicine are encouraged and challenged to consider appropriate research design and analytical techniques more consistently when study subjects function in the context of a cluster in order to avoid spurious results and misleading conclusions.

Developments in cluster randomized trials and Statistics in Medicine

The design and analysis of cluster randomized trials has been a recurrent theme in Statistics in Medicine since the early volumes. In celebration of 25 years of Statistics in Medicine, this paper reviews recent developments, particularly those that featured in the journal. Issues in design such as sample size calculations, matched paired designs, cohort versus cross-sectional designs, and practical design problems are covered. Developments in analysis include modification of robust methods to cope with small numbers of clusters, generalized estimation equations, population averaged and cluster specific models. Finally, issues on presenting data, some other clustering issues and the general problem of evaluating complex interventions are briefly mentioned.

Knowledge Translation: Closing the Evidence-to-Practice Gap

Knowledge translation describes any activity or process that facilitates the transfer of high-quality evidence from research into effective changes in health policy, clinical practice, or products. This increasingly important discipline attempts to conceptually combine elements of research, education, quality improvement, and electronic systems development to create a seamless linkage between interventions that improve patient care and their routine implementation in daily clinical practice. We outline the gap between research and practice and present a case study of an emergency medicine example of validated evidence that has failed to achieve widespread implementation. The authors describe a model of organization of evidence and its relationship with the process that links research from the scientific endeavor to changes in practice that affect patient outcomes. Obstacles to evidence uptake are explored, as well as the limitations of current educational strategies. Innovative strategies in realms such as computerized decision support systems designed to enhance evidence uptake are also described. The potential interface between knowledge translation and continuous quality improvement, as well as the role for bedside tools, is also presented. Research in knowledge translation includes studies that attempt to quantify and understand the discrepancies between what is known and what is done, as well as those that examine the impact and acceptability of interventions designed to narrow or close these gaps. Sentinel examples in this line of research conducted in the emergency department setting are described.

Enhanced relapse prevention for bipolar disorder--ERP trial. A cluster randomised controlled trial to assess the feasibility of training care coordinators to offer enhanced relapse prevention for bipolar disorder

BACKGROUND

Bipolar Disorder (BD) is a common and severe form of mental illness characterised by repeated relapses of mania or depression. Pharmacotherapy is the main treatment currently offered, but this has only limited effectiveness. A recent Cochrane review has reported that adding psycho-social interventions that train people to recognise and manage the early warning signs of their relapses is effective in increasing time to recurrence, improving social functioning and in reducing hospitalisations. However, the review also highlights the difficulties in offering these interventions within standard mental health services due to the need for highly trained therapists and extensive input of time. There is a need to explore the potential for developing Early Warning Sign (EWS) interventions in ways that will enhance dissemination.

METHODS AND DESIGN

This article describes a cluster-randomised trial to assess the feasibility of training care coordinators (CCs) in community mental health teams (CMHTs) to offer Enhanced Relapse Prevention (ERP) to people with Bipolar Disorder. CMHTs in the North West of England are randomised to either receive training in ERP and to offer this to their clients, or to continue to offer treatment as usual (TAU). The main aims of the study are (1) to determine the acceptability of the intervention, training and outcome measures (2) to assess the feasibility of the design as measured by rates of recruitment, retention, attendance and direct feedback from participants (3) to estimate the design effect of clustering for key outcome variables (4) to estimate the effect size of the impact of the intervention on outcome. In this paper we provide a rationale for the study design, briefly outline the ERP intervention, and describe in detail the study protocol.

DISCUSSION

This information will be useful to researchers attempting to carry out similar feasibility assessments of clinical effectiveness trials and in particular cluster randomised controlled trials.

Cluster randomized controlled trials in primary care: an introduction

BACKGROUND

Cluster randomized trials occur when groups or clusters of individuals, rather than the individuals themselves, are randomized to intervention and control groups and outcomes are measured on individuals within those clusters. Within primary care, between 1997 and 2000, there has been a virtual doubling in the number of published cluster randomized trials. A recent systematic review, specifically within primary care, found study quality to be both generally lower than that reported elsewhere and not to have shown any recent quality improvement.

OBJECTIVE

To discuss the design, conduct and analysis of cluster randomized trials within primary care in terms of the appropriate expertise required, potential bias, ethical considerations and expense.

DISCUSSION

Compared with trials that involve the randomization of individual participants, cluster randomized trials are more complex to design and analyse and, for a given sample size, have decreased power and a broadening of confidence intervals. Cluster randomized trials are specifically prone to potential bias at two levels-the cluster and individual. Regarding the former, it is recommended that cluster allocation be undertaken by a party independent to the research team and careful consideration be given to ensure minimal cluster attrition. Bias at the individual level can be overcome by identifying trial participants before randomization and at this time obtaining consent for intervention, data collection or both. A unique ethical aspect to cluster randomized trials is that cluster leaders may consent to the trial on behalf of potential cluster members. Additional costs of cluster randomized trials include the increased number of patients required, the complexity in their design and conduct and, usually, the need to recruit clusters de novo.

CONCLUSION

Cluster randomized trials are a powerful and increasingly popular research tool. They are uniquely placed for the conduct of research within primary-care clusters where intracluster contamination can occur. Associated methodological issues are straightforward and surmountable and just need careful consideration and management.

Differential recruitment in a cluster randomized trial in primary care: the experience of the UK back pain, exercise, active management and manipulation (UK BEAM) feasibility study

BACKGROUND

Cluster randomized trials, which randomize groups of patients rather than individuals, are commonly used to evaluate healthcare interventions such as training programmes targeted at health professionals. This article reports the dangers of randomizing entire primary care practices when participants cannot be identified before randomization, as shown by a UK national trial.

METHOD

The UK BEAM trial, a national cluster randomized 3 x 2 x 2 factorial trial, was designed to evaluate three treatments for back pain in primary care: "active management"; randomized by practice; and spinal manipulation and exercise classes, both randomized by individual.

RESULTS

Two hundred and thirty-one participants were recruited in the feasibility study, 165 (141% of expected recruitment) from active (management) practices but only 66 (54% of expected recruitment) from traditional (management) practices. The participants in active practices were significantly different from those in traditional practices, notably in suffering from milder back pain.

CONCLUSIONS

The feasibility study highlighted the dangers of randomizing clusters when individuals cannot be identified beforehand. Different numbers and types of participants were recruited in the two types of cluster. This differential recruitment led us to change the main trial design by abandoning practice level randomization. Instead all practices were trained in active management to maximize recruitment. Ideally cluster randomized trials should identify patients beforehand, to minimize the chance of selection bias. If this is not possible, patient recruitment should be independent in both intervention and control clusters. Pilot studies are especially important for cluster randomized trials, to identify unforeseen problems.

Informed patient consent to participation in cluster randomized trials: an empirical exploration of trials in primary care

Cluster randomized trials are increasingly common. Obtaining informed patient consent to participation in these trials raises practical challenges and ethical issues. The aims of this paper were to 1) develop a typology of interventions employed in cluster randomized trials in primary care; 2) assess whether the likelihood of seeking individual consent to participation varies by intervension type; 3) assess whether this likelihood has increased over time; 4) assess evidence for under reporting of consent procedures; 5) articulate reasons for not obtaining consent; and 6) make recommendations for future trial investigators. We collected data on trial interventions and consent procedures from reports of 152 recently published trials, and 47 unpublished trials. We develop a typology of interventions based on reasons for adopting a clustered design. We examine proportions seeking individual consent to participation among trials involving different types of intervention, in different periods, and among published and unpublished trials. Two-thirds of the trials had multifaceted interventions. Trials involving different types of intervention had different propensities to seek consent, largely because of practical obstacles to obtaining consent. Obtaining consent can compromise internal validity. More recent trials are no more likely to obtain consent than past trials. There was no evidence of under-reporting of consent procedures in publications. In conclusion, future trial investigators should consider both practical reasons and scientific arguments for not obtaining individual patient consent for all interventions in their trials. Where feasible, they should allow patients to opt out of the trial. Lay individuals should represent trial participants as part of the process of cluster consent to participation, and lay individuals could also be involved in considering ethical issues during trial planning. A more public debate may clarify the general acceptability of not obtaining consent in certain situations.

Sample size for cluster randomized trials: effect of coefficient of variation of cluster size and analysis method

BACKGROUND

Cluster randomized trials are increasingly popular. In many of these trials, cluster sizes are unequal. This can affect trial power, but standard sample size formulae for these trials ignore this. Previous studies addressing this issue have mostly focused on continuous outcomes or methods that are sometimes difficult to use in practice.

METHODS

We show how a simple formula can be used to judge the possible effect of unequal cluster sizes for various types of analyses and both continuous and binary outcomes. We explore the practical estimation of the coefficient of variation of cluster size required in this formula and demonstrate the formula's performance for a hypothetical but typical trial randomizing UK general practices.

RESULTS

The simple formula provides a good estimate of sample size requirements for trials analysed using cluster-level analyses weighting by cluster size and a conservative estimate for other types of analyses. For trials randomizing UK general practices the coefficient of variation of cluster size depends on variation in practice list size, variation in incidence or prevalence of the medical condition under examination, and practice and patient recruitment strategies, and for many trials is expected to be approximately 0.65. Individual-level analyses can be noticeably more efficient than some cluster-level analyses in this context.

CONCLUSIONS

When the coefficient of variation is <0.23, the effect of adjustment for variable cluster size on sample size is negligible. Most trials randomizing UK general practices and many other cluster randomized trials should account for variable cluster size in their sample size calculations.

Statistical Aspects of Design and Analysis of Clinical Trials for the Prevention of Caries

This paper considers the methods used in design and analysis of recent clinical trials of topical fluoride interventions designed to prevent the development of dental caries in children, with particular consideration given to issues related to cluster-randomized trials. Studies which met the inclusion criteria were recent clinical trials of topical fluoride interventions published since 1990, conducted in children under 16 years of age, with caries as the outcome variable. Papers not published in English were translated. Information was extracted from the published trial reports on the units of randomization and analysis. The papers were also studied to assess if reporting allowed the assessment of potential consent bias in cluster-randomized trials and the reproduction of sample size calculations. Fifteen trials published since 1990 were included, of which five were cluster randomized. Only 1 of the 5 accounted for the clustering in the analysis. For the other four trials, it was possible to calculate that values from 0.002 (for DMFS) and 0.08 (for being caries free) for the intracluster correlation coefficient within schools could result in statistically non-significant findings. 3 of the 5 cluster-randomized trials did not report the consenting procedure in enough detail to judge whether consent bias could be present. Only 1 of the total 15 trials reported a sample size calculation. In summary, researchers should be aware of the importance of correctly analyzing cluster-randomized data and thorough reporting of clinical trials according to the CONSORT guidelines.

Planning a cluster randomized trial with unequal cluster sizes: practical issues involving continuous outcomes

Background

Cluster randomization design is increasingly used for the evaluation of health-care, screeening or educational interventions. At the planning stage, sample size calculations usually consider an average cluster size without taking into account any potential imbalance in cluster size. However, there may exist high discrepancies in cluster sizes.

Methods

We performed simulations to study the impact of an imbalance in cluster size on power. We determined by simulations to which extent four methods proposed to adapt the sample size calculations to a pre-specified imbalance in cluster size could lead to adequately powered trials.

Results

We showed that an imbalance in cluster size can be of high influence on the power in the case of severe imbalance, particularly if the number of clusters is low and/or the intraclass correlation coefficient is high. In the case of a severe imbalance, our simulations confirmed that the minimum variance weights correction of the variation inflaction factor (VIF) used in the sample size calculations has the best properties.

Conclusion

Publication of cluster sizes is important to assess the real power of the trial which was conducted and to help designing future trials. We derived an adaptation of the VIF from the minimum variance weights correction to be used in case the imbalance can be a priori formulated such as "a proportion (γ) of clusters actually recruit a proportion (τ) of subjects to be included (γ ≤ τ)".

Cluster randomized controlled trials

Cluster randomized controlled trial (RCT), in which groups or clusters of individuals rather than individuals themselves are randomized, are increasingly common. Indeed, for the evaluation of certain types of intervention (such as those used in health promotion and educational interventions) a cluster randomized trial is virtually the only valid approach. However, cluster trials are generally more difficult to design and execute than individually randomized studies, and some design features of a cluster trial may make it particularly vulnerable to a range of threats that can introduce bias. In this paper we discuss the issues that can lead to bias in cluster randomized trials and conclude with some suggestions for avoiding these problems.

A priori postulated and real power in cluster randomized trials: mind the gap

BACKGROUND

Cluster randomization design is increasingly used for the evaluation of health-care, screening or educational interventions. The intraclass correlation coefficient (ICC) defines the clustering effect and be specified during planning. The aim of this work is to study the influence of the ICC on power in cluster randomized trials.

METHODS

Power contour graphs were drawn to illustrate the loss in power induced by an underestimation of the ICC when planning trials. We also derived the maximum achievable power given a specified ICC.

RESULTS

The magnitude of the ICC can have a major impact on power, and with low numbers of clusters, 80% power may not be achievable.

CONCLUSION

Underestimating the ICC during planning cluster randomized trials can lead to a seriously underpowered trial. Publication of a priori postulated and a posteriori estimated ICCs is necessary for a more objective reading: negative trial results may be the consequence of a loss of power due to a mis-specification of the ICC.

Performance of a mixed effects logistic regression model for binary outcomes with unequal cluster size

When a clustered randomized controlled trial is considered at a design stage of a clinical trial, it is useful to consider the consequences of unequal cluster size (i.e., sample size per cluster). Furthermore, the assumption of independence of observations within cluster does not hold, of course, because the subjects share the same cluster. Moreover, when the clustered outcomes are binary, a mixed effect logistic regression model is applicable. This article compares the performance of a maximum likelihood estimation of the mixed effects logistic regression model with equal and unequal cluster sizes. This was evaluated in terms of type I error rate, power, bias, and standard error through computer simulations that varied treatment effect, number of clusters, and intracluster correlation coefficients. The results show that the performance of the mixed effects logistic regression model is very similar, regardless of inequality in cluster size. This is illustrated using data from the Prevention Of Suicide in Primary care Elderly: Collaborative Trial (PROSPECT) study.

Intraclass correlation coefficients for cluster randomized trials in primary care: the cholesterol education and research trial (CEART)

Cluster randomization trials are increasingly being used in primary care research. The main feature of these trials is that patients are nested within large clusters such as physician practices or communities and the intervention is applied to the cluster. This study design necessitates calculation of intraclass correlation coefficients in order to determine the required sample size. The purpose of this study is to determine intraclass correlation coefficients for a number of outcome measures at the primary care practice level. The CEART study is a randomized trial testing the effectiveness of translating ATP III guidelines into clinical practice, with primary care physician practices as the unit of randomization and patients as the unit of data collection. The intraclass correlation coefficient (ICC) was<0.02 and the design effect ranged from 1.0 to 2.3, respectively, for weight, total cholesterol, LDL, non-HDL, glucose, creatinine, and % at non-HDL goal. For smoking status, body mass index, systolic blood pressure, HDL cholesterol triglycerides, total cholesterol/HDL ratio and % at LDL goal, the ICC was 0.02-0.047 and the design effect was 2.6-4.1. The largest ICCs (0.05-0.12) and design effects (4.4-9.4) were found for height and diastolic blood pressure. These findings suggest that cluster randomization may substantially increase the sample size necessary to maintain adequate statistical power for selected outcomes such as diastolic blood pressure studies compared with simple randomization for most outcomes evaluated in this study where the design effect is small to moderate. Overall, the ICCs presented will be useful in calculating sample sizes at the primary care level.

Testing the effectiveness of an innovative information package on practitioner reported behaviour and beliefs: the UK Chiropractors, Osteopaths and Musculoskeletal Physiotherapists Low back pain ManagemENT (COMPLeMENT) trial [ISRCTN77245761]

BACKGROUND

Low back pain (LBP) is a common and costly problem. Initiatives designed to assist practitioner and patient decisions about appropriate healthcare for LBP include printed evidence-based clinical guidelines. The three professional groups of chiropractic, osteopathy and musculoskeletal physiotherapy in the UK share common ground with their approaches to managing LBP and are amongst those targeted by LBP guidelines. Even so, many seem unaware that such guidelines exist. Furthermore, the behaviour of at least some of these practitioners differs from that recommended in these guidelines. Few randomised controlled trials evaluating printed information as an intervention to change practitioner behaviour have utilised a no-intervention control. All these trials have used a cluster design and most have methodological flaws. None specifically focus upon practitioner behaviour towards LBP patients. Studies that have investigated other strategies to change practitioner behaviour with LBP patients have produced conflicting results. Although numerous LBP guidelines have been developed worldwide, there is a paucity of data on whether their dissemination actually changes practitioner behaviour. Primarily because of its low unit cost, sending printed information to large numbers of practitioners is an attractive dissemination and implementation strategy. The effect size of such a strategy, at an individual practitioner level, is likely to be small. However, if large numbers of practitioners are targeted, this strategy might achieve meaningful changes at a population level.

METHODS

The primary aim of this prospective, pragmatic randomised controlled trial is to test the short-term effectiveness (six-months following intervention) of a directly-posted information package on the reported clinical behaviour (primary outcome), attitudes and beliefs of UK chiropractors, osteopaths and musculoskeletal physiotherapists. We sought to randomly allocate a combined sample of 1,800 consenting practitioners to receive either the information package (intervention arm) or no information above that gained during normal practice (control arm). We collected questionnaire data at baseline and six-months post-intervention. The analysis of the primary outcome will assess between-arm differences of proportions of responses to questions on recommendations about activity, work and bed-rest, that fall within categories previously defined by an expert consensus exercise as either 'guideline-consistent' and 'guideline-inconsistent'.

Methodological bias in cluster randomised trials

Background

Cluster randomised trials can be susceptible to a range of methodological problems. These problems are not commonly recognised by many researchers. In this paper we discuss the issues that can lead to bias in cluster trials.

Methods

We used a sample of cluster randomised trials from a recent review and from a systematic review of hip protectors. We compared the mean age of participants between intervention groups in a sample of 'good' cluster trials with a sample of potentially biased trials. We also compared the effect sizes, in a funnel plot, between hip protector trials that used individual randomisation compared with those that used cluster randomisation.

Results

There is a tendency for cluster trials, with evidence methodological biases, to also show an age imbalance between treatment groups. In a funnel plot we show that all cluster trials show a large positive effect of hip protectors whilst individually randomised trials show a range of positive and negative effects, suggesting that cluster trials may be producing a biased estimate of effect.

Conclusion

Methodological biases in the design and execution of cluster randomised trials is frequent. Some of these biases associated with the use of cluster designs can be avoided through careful attention to the design of cluster trials. Firstly, if possible, individual allocation should be used. Secondly, if cluster allocation is required, then ideally participants should be identified before random allocation of the clusters. Third, if prior identification is not possible, then an independent recruiter should be used to recruit participants.

Cluster randomised trials in the medical literature: two bibliometric surveys

BACKGROUND

Several reviews of published cluster randomised trials have reported that about half did not take clustering into account in the analysis, which was thus incorrect and potentially misleading. In this paper I ask whether cluster randomised trials are increasing in both number and quality of reporting.

METHODS

Computer search for papers on cluster randomised trials since 1980, hand search of trial reports published in selected volumes of the British Medical Journal over 20 years.

RESULTS

There has been a large increase in the numbers of methodological papers and of trial reports using the term 'cluster random' in recent years, with about equal numbers of each type of paper. The British Medical Journal contained more such reports than any other journal. In this journal there was a corresponding increase over time in the number of trials where subjects were randomised in clusters. In 2003 all reports showed awareness of the need to allow for clustering in the analysis. In 1993 and before clustering was ignored in most such trials.

CONCLUSION

Cluster trials are becoming more frequent and reporting is of higher quality. Perhaps statistician pressure works.

Intracluster correlation coefficients in cluster randomized trials: empirical insights into how should they be reported

Background

Increasingly, researchers are recognizing that there are many situations where the use of a cluster randomized trial may be more appropriate than an individually randomized trial. Similarly, the need for appropriate standards of reporting of cluster trials is more widely acknowledged.

Methods

In this paper, we describe the results of a survey to inform the appropriate reporting of the intracluster correlation coefficient (ICC) – the statistical measure of the clustering effect associated with a cluster randomized trial.

Results

We identified three dimensions that should be considered when reporting an ICC – a description of the dataset (including characteristics of the outcome and the intervention), information on how the ICC was calculated, and information on the precision of the ICC.

Conclusions

This paper demonstrates the development of a framework for the reporting of ICCs. If adopted into routine practice, it has the potential to facilitate the interpretation of the cluster trial being reported and should help the development of new trials in the area.