A Cross-Sectional Evaluation of Publication Bias in the Plastic Surgery Literature.

ABSTRACT

BACKGROUND: Publication bias (PB) is the preferential publishing of studies with statistically significant results. PB can skew findings of systematic reviews (SR) and meta-analyses (MA), with potential consequences for patient care and health policy. This study aims to determine the extent by which SRs and MAs in the plastic surgery literature evaluate and report PB. METHODS: This cross-sectional study assessed PB reporting and analysis from plastic surgery studies published between January 1, 2015 and June 19, 2020. Full-texts of SRs and MAs were assessed by two reviewers for PB assessment methodology and analysis. Post-hoc assessment of studies that did not originally analyze PB was performed using Egger's regression, Duval and Tweedie's trim-and-fill, and Copas selection models. RESULTS: There were 549 studies evaluated, of which 531 full-texts were included. PB was discussed by 183 (34.5%) studies, and formally assessed by 97 (18.3%) studies. Among SR and MAs that formally assessed PB, PB was present in 24 (10.7%), not present in 52 (23.1%), and inconclusive in 8 (3.6%) studies; 141 (62.7%) studies did not report the results of their PB assessment. Funnel plots were the most common assessment method (n=88, 39.1%), and 60 (68.2%) studies published funnel plots. The post-hoc assessment revealed PB in 17/20 (85.0%) studies. CONCLUSION: PB is inadequately reported and analyzed amongst studies in the plastic surgery literature. Most studies that assessed PB found PB, as did post-hoc analysis of non-reporting studies. Increased assessment and reporting of PB amongst SRs and MAs would improve quality of evidence in plastic surgery.