Summary
Network meta-analysis (NMA) has been an important tool to assess the relative efficacy and safety of the available treatments in chronic inflammatory diseases. However, concerns have been raised regarding the increasing number of redundant and discordant NMAs, and the influence of funding on the study results.
Objectives
The objective was to explore the association between industry funding and network meta-analyses’ conclusion, and the use in Clinical Practice Guidelines (CPGs) of NMAs.
Methods
This was an overview of NMAs and CPGs. We searched PubMed/MEDLINE, Epistemonikos, and several guideline databases up to February 18th 2023. We included CPGs from the last 5 years and NMAs of randomized controlled trials that evaluated targeted therapies in immune-mediated inflammatory diseases. Data extraction and outcome assessments were done in duplicate by independent authors.
Results
We included 216 NMAs and 99 CPGs. 31% (67/216) were industry-funded. The proportion of industry-funded NMAs that cited one treatment as being best was 44% (25/57) compared to 26% (30/116) for nonindustry-funded (OR = 2.24 [1.15-4.39]; aOR = 1.76 [0.81-3.81]). The abstract's conclusion of 39/67 (58%) industry-funded and 69/149 (46%) nonindustry-funded NMAs were considered unsupported by the results (OR = 1.61 [0.90-2.89]; aOR = 1.40 [0.71-2.78]). All industry-funded NMAs that cited one treatment as best cited their own sponsored drug. 59/99 (60%) CPGs included at least one NMA, with 23/59 (39%) of them citing industry-funded NMAs.
Conclusion
Industry funding were more likely to have unsupported conclusions and to cite only one treatment as being best, with substantial risk for sponsorship bias. We found important methological shortcomings, irrespective of the source of funding. Improvement is needed to increase transparency and accurate reporting of NMAs results and use in CPGs.
Funding and conclusions of network meta-analyses on targeted therapies in inflammatory diseases: an overview
Robin Guelimi, Sivem Afach, Thomas Bettuzzi, Antoine Meyer, Guillaume Padern, Zenas Yiu, Florian Naudet, Emilie Sbidian, Laurence Le-Cleach
PMID: 38852893 DOI: 10.1016/j.jclinepi.2024.111411
Published: 07 June 2024