Conflict of interest and risk of bias in systematic reviews on methylphenidate for attention-deficit hyperactivity disorder: a cross-sectional study

Abstract

Background: Systematic reviews (SRs) are pivotal to evidence-based medicine, yet there is limited research on conflicts of interest in SRs. Our aim was to investigate financial conflicts of interest and risk of bias (RoB) in SRs of a well-defined clinical topic. Methods: A librarian searched Medline, Cochrane Library, Embase, and PsycINFO for SRs investigating the effect of methylphenidate on ADHD in December 2020. The selection process adhered to the PRISMA guidelines. Two blinded reviewers independently searched open websites, including other publications, for information on financial conflicts of interest of all authors of the included SRs. A time limit of 3 years before or after the index SR was adopted. Declarations on conflict of interest were extracted from the included SRs for comparison. ROBIS was used for RoB assessment. Results: Out of 44 SRs included, 15 (34%) declared conflict of interest, 27 (61%) did not, and a declaration of conflict of interest was missing for 2 (5%). On open websites, conflict of interest was found for at least one author of 23 (52%) SRs: disclosed in 15 (34%) and not disclosed in 8 (18%) SRs. Seven (16%) SRs had low, 36 (82%) had high, and 1 (2%) had unclear RoB. Among SRs with financial conflict of interest found in open sources, 6/22 (27%) had low RoB compared to 1/21 (5%) if no such conflict of interest was identified. Among SRs with financial conflict of interest identified, 1/6 (17%) at low RoB did not disclose their conflict of interest, whereas the corresponding proportion among SRs at high RoB was 7/16 (44%). Eight (18%) SRs presented conflict of interest disclosed in the included primary studies. Four of them (50%) had low RoB, compared to 3/36 (8%) for SRs not reporting on this aspect. Conclusion: Financial conflict of interest was underreported in 18% of the SRs using our reference standard, and overall it was present for every second SR. This group embraced both SRs at low RoB disclosing conflict of interest and SRs at high RoB not disclosing their conflict of interest. Further studies to explore this heterogeneity are warranted.