A systematic survey of 200 systematic reviews with network meta-analysis (published 2020-2021) reveals that few reviews report structured evidence summaries

Objectives To map whether and how systematic reviews (SRs) with network meta-analysis (NMA) use presentation formats to report (a) structured evidence summaries – here defined as reporting of effects estimates in absolute effects with certainty ratings and with a method to rate interventions across one or more outcome(s) – and (b) NMA results in general. Study Design and Setting We conducted a systematic survey, searching MEDLINE (Ovid) for SRs with NMA published between January 1, 2020, and December 31, 2021. We planned to include a random sample of publications, with predefined mechanisms in place for saturation, and included SRs that met prespecified quality criteria and extracted data on presentation formats that reported: (a) estimates of effects, (b) certainty of the evidence, or (c) rating of interventions. Results The 200 eligible SRs, from 158 unique Journals, utilized 1133 presentation formats. We found structured evidence summaries in 10 publications (5.0%), with 3 (1.5%) reporting structured evidence summaries across all outcomes, including benefits and harms. Sixteen of the 133 SRs (11.7%) reporting dichotomous outcomes included estimates of absolute effects. Seventy-six SRs (38.0%) reported both benefits and harms and 26 SRs (13.0%) reported certainty ratings in presentation formats, 20 (76.9%) used Grading of Recommendations Assessment, Development and Evaluation and 6 (23.1%) used Confidence In Network Meta-analysis. Surface Under the Cumulative Ranking Curve was the most common method to rate interventions (69 SRs, 34.5%). NMA results were most often reported using forest plots (108 SRs, 54.0%) and league tables (93 SRs, 46.5%). Conclusion Most SRs with NMA do not report structured evidence summaries and only rarely do such summaries include reporting of both benefits and harms; those that do offer effective user-friendly communication and provide models for optimal NMA presentation practice.