Interpretation of network meta-analyses of the efficacy of rheumatoid arthritis drugs.

WHAT IS KNOWN AND OBJECTIVE: Both biological agents and Janus kinase (JAK) inhibitors have been shown to be effective in rheumatoid arthritis (RA) patients with inadequate responses to tumour necrosis factor (TNF) inhibitors. Network meta-analyses are useful for comparing drug effectiveness in the absence of direct head-to-head comparisons. We comment on two such meta-analyses reported in this Journal to highlight aspects of their interpretation.

COMMENT: In network meta-analyses, it is important to critically examine whether the direct and indirect estimates are divergent, keeping in mind that the surface under the cumulative ranking probability curve (SUCRA) is a relative value in such analyses. In two studies, examining the effect of therapeutic agents for RA by using NMA, the SUCRA method, produced a distinct ranking for JAK inhibitors in monotherapy, and non-TNF inhibitors and JAK inhibitors, in cases of TNF failure. Nonetheless, the differences in treatment effects were small and their clinical significance requires validation.

WHAT IS NEW AND CONCLUSION: NMA is a useful method for studying the comparative effectiveness of multiple treatments. However, increased attention should be paid when using this method to distinguish statistically significant differences from clinically meaningful differences.