Search

Background: Infectious disease interventions are increasingly tested using cluster-randomized trials (CRTs). These trial settings tend to involve a set of sampling units, such as villages, whose geographic arrangement may present a contamination risk in treatment exposure. The most widely used approach for reducing contamination in these settings is the so-called fried-egg design, which excludes the outer portion of all available clusters from the primary trial analysis. However, the fried-egg design ignores potential intra-cluster spatial heterogeneity and makes the outcome measure inherently less precise. Whereas the fried-egg design may be appropriate in specific settings, alternative methods to optimize the design of CRTs in other settings are lacking. Methods: We present a novel approach for CRT design that either fully includes or fully excludes available clusters in a defined study region, recognizing the potential for intra-cluster spatial heterogeneity. The approach includes an algorithm that allows investigators to identify the maximum number of clusters that could be included for a defined study region and maintain randomness in both the selection of included clusters and the allocation of clusters to either the treatment group or control group. The approach was applied to the design of a CRT testing the effectiveness of malaria vector-control interventions in southern Malawi. Conclusions: Those planning CRTs to evaluate interventions should consider the approach presented here during trial design. The approach provides a novel framework for reducing the risk of contamination among the CRT randomization units in settings where investigators determine the reduction of contamination risk as a high priority and where intra-cluster spatial heterogeneity is likely. By maintaining randomness in the allocation of clusters to either the treatment group or control group, the approach also permits a randomization-valid test of the primary trial hypothesis.

Summary Data from multiple prevalence surveys can provide information on common parameters of interest, which can therefore be estimated more precisely in a joint analysis than by separate analyses of the data from each survey. However, fitting a single model to the combined data from multiple surveys is inadvisable without testing the implicit assumption that all of the surveys are directed at the same inferential target. We propose a multivariate generalized linear geostatistical model that accommodates two sources of heterogeneity across surveys to correct for spatially structured bias in non-randomized surveys and to allow for temporal variation in the underlying prevalence surface between consecutive survey periods. We describe a Monte Carlo maximum likelihood procedure for parameter estimation and show through simulation experiments how accounting for the different sources of heterogeneity among surveys in a joint model leads to more precise inferences. We describe an application to multiple surveys of the prevalence of malaria conducted in Chikhwawa District, Southern Malawi, and discuss how this approach could inform hybrid sampling strategies that combine data from randomized and non-randomized surveys to make the most efficient use of all available data.