Crossroads of methodological choices in research synthesis: insights from two network meta-analyses on preventing relapse in schizophrenia

Abstract

In recent years, network meta-analyses have been increasingly carried out to inform clinical guidelines and policy. This approach is under constant development, and a broad consensus on how to carry out several of its methodological and statistical steps is still lacking. Therefore, different working groups might often make different methodological choices based on their clinical and research experience, with possible advantages and shortcomings. In this contribution, we will critically assess two network meta-analyses on the topic of pharmacological prevention of relapse in schizophrenia, carried out by two different research groups. We will highlight the implications of different methodological choices on the analysis results and their clinical–epidemiological interpretation. Moreover, we will discuss some of the most relevant technical issues of network meta-analyses for which there is not a broad methodological agreement, including the assessment of transitivity.