Current Practices for Cohort Reporting and Statistical Adjustment in Studies From The Society of Thoracic Surgeons Adult Cardiac Surgery Database.

Abstract

Current reporting and statistical adjustment practices of studies based on The Society of Thoracic Surgeons Adult Cardiac Surgery Database (STS ACSD) have not been described. A review identified all published studies based on STS ACSD data from January 2016 to May 2024. Data were extracted by 2 authors and independently checked by the senior author. A total of 134 studies were included. There were 46 (34.3%) studies on coronary artery bypass grafting, 22 (16.4%) on mitral valve surgery, 16 (11.9%) on aortic surgery, 14 (10.4%) on surgical aortic valve replacement, 3 (2.2%) on tricuspid valve surgery, 18 (13.4%) on combined procedures, and 15 (11.2%) on other interventions; 59 (44.0%) studies used the Participant User File (PUF) program, with a significant increase over time (P_trend = .005). Of the 1239 variables available in the STS ACSD (version 4.20.2), 136 (11.0%; median, 21 variables per study; interquartile range [IQR], 16-25) were used to describe the baseline characteristics of the patients and 191 (15.4%; median, 22 variables per study; IQR, 17-33) for statistical adjustment; 121 studies (90.3%) performed statistical adjustment, with multivariable regression used in most (80 [66.1%]). PUF manuscripts had significantly more junior first authors (20.3% vs 1.3%; P = .001) and were published in higher impact journals (median impact factor, 4.90 [IQR, 3.70-9.15] vs 3.90 [IQR, 3.60-4.70]; P = .002) compared with non-PUF manuscripts. Our analysis provides data that may inform efforts to standardize reporting and analytic practices in studies based on the STS ACSD.