Research Review: Conflicts of Interest (COIs) in autism early intervention research - a meta-analysis of COI influences on intervention effects

BACKGROUND
The presence, types, disclosure rates, and effects of conflicts of interest (COIs) on autism early intervention research have not previously been studied. The purpose of this study was to examine these issues.

METHODS
This study is a secondary analysis of a comprehensive meta-analysis of all group-design, nonpharmacological early intervention autism research conducted between 1970 and 2018. We coded reports for the presence/absence of COI statements, the types of COIs that were disclosed, and for 8 types of COIs, including (a) the author developed the intervention, (b) the author is affiliated with a clinical provider, (c) the author is employed by a clinical provider, (d) the author is affiliated with an institution that trains others to use the intervention, (e) the author receives payment or royalties related to the intervention, (f) the study was funded by an intervention provider, (g) the study used a commercially available measure developed by the author, and (h) proceeds of the intervention fund the author's research. Frequencies and proportions were calculated to determine prevalence of COIs and COI disclosures. Meta-analysis was used to estimate summary effects by COI type and to determine if they were larger than for reports with no coded COIs.

RESULTS
Seventy percent of reports were coded for ≥ 1 COI, but only ~ 6% of reports contained COI statements fully accounting for all coded COIs. Metaregressions did not detect significant influences of any COI type on summary effects; however, point estimates for each COI type were larger than for reports with no coded COIs.

CONCLUSIONS
Conflicts of interest are prevalent but under-reported in autism early intervention research. Improved reporting practices are necessary for researcher transparency and would enable more robust examination of the effects of COIs on research outcomes.