Thus, they concluded that there is a high chance that the findings of Cipriani et al are false positives and that there might be no actual differences between the drugs. Their simulations showed that 72% of the simulated datasets found at least one statistically significant treatment comparison between the treatments, even though in truth there was none. They performed simulations where they replicated the network of antidepressants by Cipriani et al, assuming, however, no treatment effects between the drugs. Del Re et al claimed that such findings might be due to multiple testing. 8 This was a NMA on the efficacy of antidepressant drugs, which found several important differences between the drugs. For example, Del Re et al 7 claimed that considerations related to multiple testing shed doubts on the validity of results of a previously published NMA by Cipriani et al.
5, 6ĭespite the popularity of NMA, there have been some concerns regarding the validity of NMA findings. NMA has been increasingly popular, with hundreds of application being published every year.
1, 2, 3, 4 NMA offers several distinct advantages over a series of standard (pairwise) meta‐analyses, such as an increase in precision and power, the opportunity to compare interventions that have not been compared directly in any studies, and the capacity to provide a ranking of all competing treatments. Network meta‐analysis (NMA) is a statistical tool for synthesizing evidence from multiple studies comparing a range of alternative treatment options for the same disease.
0 kommentar(er)
