Contamination effects in cluster randomised trials of TB interventions.

Abstract

Cluster randomised trials (CRTs) of TB interventions have achieved mixed results, with many lacking significant reductions in outcomes. Contamination in CRTs, resulting from short and long-term movement between clusters and the general population, may dilute the impact of measured intervention.We systematically reviewed the literature to identify CRTs that aimed to capture the population-level effects of the intervention on TB. Details of trial designs, interventions, outcomes, populations, cluster configurations, and geographic data were extracted to produce text summaries, descriptive statistics, and spatial analyses.We screened 1,039 abstracts and included 20 reports from seven CRTs. The median number of clusters was 32 (IQR 23-61), with populations ranging from 400-50,000 individuals per cluster. Four trials reported spatial data, from which the mean distance between clusters was 12.3 km (range 3.71-35.9). Several trials acknowledged design limitations, such as small cluster sizes and population mobility, which could have led to underestimations of intervention impact. Trials used various geographic, social, and pre-existing TB measures to select and allocate study clusters. Data on the potential for contamination are inconsistent.Gaps remain in the reporting of methods and results, suggesting necessary improvements to standardised reporting tools. These insights can inform recommendations for improved CRT design and reporting practices..