Clinical Trials

❘❙❚■ Chapter 15 | Cluster Randomized TrialsTable 4. Reporting cluster randomized trials.• Explain the baseline distribution of important characteristics of the population• Include sample size calculations and assumptions of correlation with a cluster• Provide values of the cluster effect, as calculated for the primary outcome variables• Explain how the reported statistical analyses account for the cluster effectThe CONSORT (Consolidated Standards of Reporting Trials) statement on trialconduct and reporting has now been extended to take into account the specialfeatures of CRTs, such as rationale for the cluster design, and implications of thecluster effect in design and analysis. These changes hope to increase the reportingquality of these trials [17].Reports of CRTs should include sample size calculations and statistical analysesthat take clustering into account. Information regarding other details (eg, estimatesof design effects, baseline distribution of important characteristics in theintervention group, number of clusters, and average cluster size for each group)can guide researchers to design better trials and avoid key errors when conductingand reporting CRTs (see Table 4) [18]. Indeed, the ideal scenario would be topublish and submit the design of a CRT for open peer review before the trial getsunder way; this would allow assumptions to be adequately challenged by expertsin the field.Ethical issues in cluster randomized trialsSeveral reports, guidelines, and codes provide an extensive overview of theimportance of ethical issues in individual patient randomized trials. However,the ethical issues raised by CRTs have also drawn the attention of experts.Individuals consenting in conventional randomized trials are likely to consent witha higher degree of freedom and independence compared to those who participatein CRTs. In these trials, participants are likely to impinge on each other’s choiceswhen informed consent for trial entry (that is, for randomization) is obtained. Thedecision to participate in the trial or intervention may depend not just on theindividual, but also on the guardian (eg, the hospital chief executive or managingpartner of the primary care trust) in the CRT. For example, a hospital managermay agree to have a counseling service for all patients with a stroke, and, since thisintervention will be accessible to all patients admitted to that hospital, it mightbecome protocol rather than a specific consent-requiring therapy – although thetrial design will, of course, have been approved by an ethics committee [19].148