P2.F110: Characterization of Funding Sources and Reported Significance in Neonatal Randomized Trials

Background: Industry funding has been associated with potential bias in biomedical studies; however, this has not been adequately addressed in the context of the neonatal randomized controlled trial (RCT) literature. Our objective is to analyze trends in funding sources for neonatal randomized controlled trials (RCTs) and correlate these sources with data reporting and measures of quality.

Methods: We analyzed RCTs between 2003 and 2017 from NeoCanon, a validated, hand-curated, and comprehensive database of all neonatal RCTs that includes information on study design, sample size, interventions tested, and funding, as reported in the abstracts and text of the publications. Cluster RCTs were excluded from the final analysis. We tested for trends in the rates of funding over time using the Mann-Kendall test. We then correlated funding source with whether any statistically significant results were reported in the publication abstract, using a chi-square analysis.

Results: We included 2,665 RCTs in our analysis. Of these, 1686 (63%) RCTs reported a funding source. Among trials that included a funding source, 27% reported exclusively federal funding, 15% exclusively industry funding, 14% exclusively foundation funding, 13% other (such as institutional), 19% any combination of these, and 13% no funding. There was no change over time in the proportion of trials reporting industry (p=.24), foundation (p=.14), or combined (p=.92) sources. However, there was a positive change over time in the proportion of trials reporting a funding source (p <.01), the proportion reporting no funding (p <.01), the proportion funded by other (p=.04), and a negative change over time in the proportion funded by federal (p=.01). A large proportion of RCTs (70%) reported a statistically significant result in the abstract; this was more common in studies that did not report a funding source than those that did (73% unreported and 68% reported, p=.02) with industry-funded studies being the least likely to report statistically significant results. The median sample size of studies that reported on funding was significantly greater than the median sample size of studies that did not report (Table 1).

Conclusion: More than one third of neonatal RCTs did not report funding status, and these studies were smaller and more likely to report statistically significant results in their abstracts. However, the rate of reporting funding sources in neonatal RCTs has increased over time. Our findings of differences in inclusion of positive results in abstracts of studies not reporting source may be attributable to publication bias, selection bias, or preferential placement of secondary outcomes in the abstract. Contrary to prior studies, our analysis did not identify an association between industry funding and the reporting of positive results, suggesting that these more-monitored studies may be at lower risk of reporting bias compared to those that do not disclose funding.