Application of the E-Value to Assess Bias in Observational Research in Plastic Surgery

Abstract

The E-value is a statistical measure that is used to quantify the degree of unmeasured confounding that is necessary to undermine the treatment-outcome associations established in observational studies. Despite the substantial amount of observational research that informs evidence-based practice in plastic surgery, sensitivity analyses based on the E-value have not been conducted in the field. The authors performed a systematic search of the literature to identify meta-analyses of observational studies in plastic surgery. The authors calculated E-values for various treatment-outcome associations based on the risk, odds, or hazard ratios in each study, to assess unmeasured confounding effects that may influence the validity of the conclusions. The authors then analyzed the distribution of E-values from pooled versus individual studies. The authors identified 45 meta-analyses that met the inclusion criteria, with each containing an average of three pooled assessments of observational data. The E-value of the pooled effect estimates ranged from 1.11 to 19.49, with an average value of 3.82. As for the individual effect estimates from each primary study within the meta-analyses, the E-values ranged from 1.00 to 321.50, with an average value of 8.74. The authors determined that E-values vary substantially across the literature and that unmeasured confounding may be present in a high number of observational studies. Although extant statistical techniques will continue to be necessary to control for measured confounding, the E-value is a novel concept that can facilitate more robust sensitivity analyses in plastic surgery research.