gms | German Medical Science

Background/research question: The evidence of a relationship between hospital volumes and outcomes for several procedures has led to regulations aimed at centralisation in several health care systems. Study authors may have an interest to influence health policy decisions by preferentially publishing research results that favour the volume category of the hospital to which they are affiliated. This can be considered as a form of institutional conflicts of interest (COI). This study aimed to explore the existence of institutional COIs in volume-outcome studies.

Methods: We used a sample of studies included in a systematic review on the hospital volume-outcome relationship in total knee arthroplasty. We categorised study authors' conclusions (positive vs. non-positive), their affiliation (hospital vs. non-hospital), and their hospital volume (high, intermediate, low). We compared frequencies and odds ratios (OR) with corresponding 95% confidence intervals (CI). We performed two analyses comparing authors’ conclusions for 1) high vs. intermediate or low hospital volume categories and 2) hospital vs. non-hospital affiliated authors. We also asked study authors by e-mail for their opinion on institutional COIs.

Results: Fifty-three studies were eligible and 14 study authors provided their institutional volume at the time of abstract submission (17 responses out of 29 authors contacted, 59%), so that 18 studies were eligible for analysis 1. Preliminary results showed that studies by authors from high volume institutions tended to support a hospital volume-outcome relationship more often than studies by authors from intermediate- or low-volume intuitions, but the effect was statistically not significant (OR: 2.5, CI: 0.26–24.38). Affiliation of at least one author to a hospital seemed to have no effect on conclusions compared to no hospital affiliation (OR: 0.96, CI: 0.27–3.41). The survey results showed that 7/14 (50%) authors believed that institutional factors such as the case volume were (very) likely to influence the study design, analysis or conclusions of research in the field of volume-outcome studies; 3/14 (21%) were neutral and 4/14 (29%) believed this was (very) unlikely.

Conclusion: Overall, we found no robust evidence for institutional COIs in the field of volume-outcome studies. However, half of the surveyed authors believe that institutional factors influence the study design, analysis or conclusions. Institutional COIs may depend on the reimbursement system.

Competing interests: The authors declare no conflicts of interest.