Funding bias in shoulder arthroplasty research.

ABSTRACT

BACKGROUND: Prior research has shown that industry funding can impact the outcomes reported in medical literature. Limited data exist on the degree of bias that industry funding may have on shoulder arthroplasty literature outside of the Journal of Shoulder and Elbow Surgery. The purpose of this study is to characterize the type and frequency of funding for recently published shoulder arthroplasty studies and the impact of industry funding on reported outcomes. We hypothesized that studies with industry funding are more likely to report positive outcomes than those without. MATERIALS AND METHODS: We performed a retrospective study searching all articles with the term "shoulder arthroplasty," "reverse shoulder arthroplasty," "anatomic total shoulder arthroplasty," or "total shoulder arthroplasty" on PubMed from the years January 2020 to December 2022. The primary outcome of studies was coded as either positive, negative, or neutral. A positive result was defined as one in which the null hypothesis was rejected. A negative result was defined as one in which the result did not favor the group in which the industry-funded implant was used. A neutral result was defined as one in which the null hypothesis was confirmed. Article funding type, subcategorized as National Institutes of Health funding or industry funding was recorded. Author disclosures were recorded to determine conflicts of interest. Statistical analysis was conducted using the χ2 test and Fisher exact test. RESULTS: A total of 750 articles reported on either conflict of interest or funding source and were included in the study. Of the total number of industry-funded studies, the majority were found to have a positive primary endpoint (58.1%, 104 of 179), as compared to a negative (7.8%, 14 of 179) or neutral endpoint (33.5%, 60 of 179) (P = .004). Overall, 363 articles reported an author conflict of interest, and the majority of these studies had positive primary endpoint (55.6%, 202 of 363) as compared to negative (9.1%, 33 of 363) or neutral endpoints (34.4%, 125 of 363) (P = .002). CONCLUSION: The results of this study suggest that there is a significant relationship between conflicts of interest and the primary outcome of shoulder arthroplasty studies, beyond the overall positive publication bias. Studies with industry funding and author conflicts of interest both report positive outcomes more frequently than negative outcomes. Shoulder surgeons should be aware of this potential bias when choosing to base clinical practice on published data.