Propensity Scores Used for Analysis of Cluster Randomized Trials with Selection Bias: a Simulation Study C. Leyrat a,b , A. Caille a,b,c,d , A. Donner e , B. Giraudeau a,b,c,d December 14, 2013 a INSERM UMR-S 738, Paris, France b INSERM CIC 202, Tours, France c Universit´ e Franc ¸ois-Rabelais, Tours, France d CHRU de Tours, Tours, France e Department of Epidemiology and Biostatistics, University of Western Ontario, Lon- don, Canada Corresponding author: Cl´ emence Leyrat, INSERM CIC 202, 2 bd Tonnell´ e, 37044 Tours cedex 9, France. E-mail: clemence.leyrat@univ-tours.fr This research was supported by a grant from the French Agence Nationale de la Recherche (ANR-2010-PRSP-010-01) Abstract Cluster randomized trials (CRTs) are often prone to selection bias despite ran- domization. Using a simulation study, we investigated the use of propensity score (PS)-based methods in estimating treatment effects in CRTs with selection bias when the outcome is quantitative. Of four PS-based methods (adjustment on PS, inverse weighting, stratification, optimal full matching method), three successfully corrected the bias, as did an approach using classical multivariable regression. However, they showed poorer statistical efficiency than classical methods, with higher standard error for the treatment effect, and type I error much smaller than the 5% nominal level. Keyword: cluster randomized trial; Monte-Carlo simulations; selection bias; propen- sity score. 1