Does type of funding affect reporting in network meta-analysis?
Areti Angeliki Veroniki, Unity Health Toronto
2Institute for Health Policy, Management, and Evaluation, University of Toronto, Toronto, ON, Canada, canada
3Department of Pediatrics, University of Antioquia, Colombia
4Department of Pediatrics, University of Antioquia; Department of Health Research Methods, Evidence and Impact; McMaster University; Paediatric Intensive Care Unit, Clinica Las Américas, Colombia
5Epidemiology Division & Institute of Health Policy, Management, and Evaluation, Dalla Lana School of Public Health, University of Toronto, canada
6Department of Geriatric Medicine, University of Toronto, Canada
Background: Industry-sponsored randomized clinical trials (RCTs) are more likely to report intervention-favourable results. However, the impact of different funding types on the reporting of network meta-analysis (NMA) of RCTs has not been explored before.
Objectives: To assess whether the funding type influenced: (a) completeness of Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting in NMAs of pharmacological interventions according to their funding type, (b) journal impact factor, and (b) the recommendation to use their own or another company’s intervention.
Methods: We collected NMAs comparing pharmacological interventions published between 2013 and 2018. We assessed completeness of reporting in NMAs using the 32-item PRISMA-NMA checklist and categorized NMAs according to their funding type and overall conclusion. We explored whether NMAs recommended their own or another company’s intervention and classified NMAs according to their primary outcome findings and the direction of the overall reported conclusion. We compared the underlying distributions of the journal impact factors by funding type. We also assessed reporting of conflicts of interest (CoI) for the first and senior author in the NMA using barplots.
Results: We included 658 NMAs with median PRISMA score 72% (IQR 66%, 81%). NMAs were categorized as 136 industry/mixed-sponsored, 314 publicly sponsored, and 208 non-sponsored NMAs with median PRISMA scores of 66% (59%, 75%), 77% (69%, 84%), and 72% (63%, 78%), respectively. Publicly funded NMAs were associated with more comprehensive reporting and published their findings in higher impact factor journals compared to industry-sponsored NMAs (median impact-factor: 4.39, [2.97, 7.09] vs. 3.00, [2.27, 4.16]). Most industry-sponsored NMAs suggested a statistically significant positive treatment effect for their underlying intervention in their primary outcome (112/136, 82%) and an overall positive conclusion (125/136, 92%). Of the 115 first and senior authors of publicly funded NMAs affiliated with the study sponsor (55%), none reported CoI and, of the 82 first and senior authors of industry-sponsored NMAs that were affiliated with the study sponsor, 78% reported CoI.
Conclusions: This is the largest empirical study used to evaluate reporting in NMAs with different funding sources. Differences in completeness of reporting were apparent between different types of funding in NMAs, and guideline developers using NMAs should be aware of this. Patient, public, and/or healthcare consumer involvement: None