Network meta-analysis combining survival and count outcome data: A simple frequentist approach

Network meta-analysis for survival outcome data often involves several studies only reported dichotomized outcomes (i.e., the numbers of events and sample sizes of individual arms). To avoid the reporting biases via eliminating these studies in the syntesis analyses, Woods et al. (2010; BMC Med Res Methodol 10:54) proposed a Bayesian approach to combine the survival and dichotomized outcome data using hierarchical models. However, the Bayesian methods require complicated computations involving the Markov Chain Monte Carlo and the technical aspects are generally difficult to handle by non-statisticians (e.g., the convergence diagnostics). Besides, frequentist approaches have been alternative standard methods for statistical analyses of network meta-analysis. The methodology has been well established, and several effective software packages have been developed (e.g., netmeta package in R). In this article, we propose a simple method to synthesis the survival and dichotomized outcome data within the frequentist framework using the contrast-based models. We also provide a R package survNMA (https://doi.org/10.32614/CRAN.package.survNMA) to calculate hazard ratio statistics from the dichotomized outcome data that can be directly used for implementing general analyses of network meta-analysis using netmeta package of R.