Estimating the Effects of Friendship Networks on Health Behaviors of Adolescents

Statistical methods have been developed for cost-effectiveness analyses (CEA) of cluster randomised trials (CRTs) where baseline covariates are balanced. However, CRTs may have systematic differences in individual and cluster-level covariates between the treatment groups. This paper presents three methods to adjust for imbalances in observed covariates: seemingly unrelated regression (SUR) with a robust standard error, a ‘two-stage’ bootstrap (TSB) approach combined with SUR, and multilevel models (MLMs). We consider the methods in a CEA of a CRT with covariate imbalance, unequal cluster sizes and a prognostic relationship that varied by treatment group. The cost-effectiveness results differed according to the approach for covariate adjustment. Our simulation study assessed the relative performance of methods for addressing systematic imbalance in baseline covariates. The simulation extended the case study and considered scenarios with: different levels of confounding, cluster size variation and few clusters. Performance was reported as bias, root mean squared error and confidence interval (CI) coverage of the incremental net benefit. Even with low levels of confounding, unadjusted methods were biased, but all adjusted methods were unbiased. MLMs performed well across all settings, and unlike the other methods, reported CI coverage close to nominal levels even when with few clusters of unequal sizes.