Bayesian indirect and mixed treatment comparisons across longitudinal time points.

Meta-analysis has become an acceptable and powerful tool for pooling quantitative results from multiple studies addressing the same question. It estimates the effect difference between two treatments when they have been compared head-to-head. However, limitations occur when there are more than two treatments of interest, and some of them have not been compared in the same study. Indirect and mixed treatment modeling extends meta-analysis methods to enable data from different treatments and trials to be synthesized, without requiring head-to-head comparisons among all treatments; thus, allowing different treatments to be compared. Traditional indirect and mixed treatment comparison methods consider a single endpoint for each trial. We extend the current methods and propose a Bayesian indirect and mixed treatment comparison longitudinal model that incorporates multiple time points and allows indirect comparisons of treatment effects across different longitudinal studies. The proposed model only uses summary level longitudinal data. This model is particularly useful when a meta-analysis is performed on studies with different durations. It enables the borrowing of information from shorter studies even in the situation where the primary interest is in a time point beyond the duration of some of these shorter studies. We performed simulation studies, which demonstrate that the proposed method performs well and yields better estimations compared with other single time point meta-analysis methods. We apply our method to a set of studies from patients with type 2 diabetes.