2157. Design, Implementation, and Analysis Considerations for Cluster Randomized Trials in Infection Control and Hospital Epidemiology: A Systematic Review

Abstract

BackgroundA cluster randomized trial (CRT) is a comparative study in which clusters or groups rather than individuals are randomized to interventions or treatments. CRT are being utilized with increasing frequency in the study of interventions in infection control and hospital epidemiology. The aims of this study were (1) to identify critical design, implementation, and analysis principles to consider when planning a CRT of interventions in the healthcare setting and (2) to review published CRT in infection control and hospital epidemiology and synthesize key characteristics of these published studies using the principles identified above.MethodsAuthors reviewed articles and book chapters to identify key methodological principles relative to the design, implementation, and analysis of CRT. We undertook a systematic review of studies conducted between 1997 and 2017 in infection control and hospital epidemiology that used a CRT design, and evaluated each study on those key principles.ResultsSeven epidemiological principles were identified as most critical (Figure 1). Among the 44 studies included in the review, the most commonly used design was a CRT with cross-over (n = 15, 34%), followed by a parallel CRT (n = 11, 25%), and a stratified CRT design (n = 7, 16%). Twenty-two (50%) offered justification for their use of a CRT. Twenty (45%) accounted for clustering at the design phase when estimating sample size. Only 15 (34%) reported the intraclass correlation coefficient, coefficient of variation, or design effect. Fifteen studies (34%) obtained waived consent, 14 (32%) did not report how they dealt with consent, 8 (18%) studies obtained consent from individuals, and 7 (16%) sought consent at the cluster level. Seventeen studies (39%) matched or stratified at time of randomization, while 27 (61%) did not employ either of these techniques. Notably, 10 (23%) studies did not report any efforts to reduce the potential for bias and/or contamination. Twenty-seven (61%) accounted for clustering in their analyses.ConclusionCRT in infection control and hospital epidemiology are common but are still lacking in methodological rigor. It is crucial to continue improving the design and reporting of these studies to better evaluate the effectiveness of interventions. Disclosures All authors: No reported disclosures.