Association of Volume and Outcomes in 234 556 Patients Undergoing Surgical Aortic Valve Replacement.

Background: The relationship between institutional volume and operative mortality after surgical aortic valve replacement (SAVR) remains unclear.

Methods: From January 2013 to June 2018, 234 556 patients underwent isolated SAVR (n = 144 177) or SAVR with coronary artery bypass grafting (CABG) (n = 90 379) within the Society of Thoracic Surgeons Adult Cardiac Surgery Database. The association between annualized SAVR volume (group 1 [1-25 SAVRs], group 2 [26-50 SAVRs], group 3 [51-100 SAVRs], and group 4 [>100 SAVRs]) and operative mortality and composite major morbidity or mortality was assessed. Random effects models were used to evaluate whether historical (2013-2015) SAVR volume or risk-adjusted outcomes explained future (2016-2018) risk-adjusted outcomes.

Results: The annualized median number of SAVRs per site was 35 (interquartile range, 22-59; isolated aortic valve replacement [AVR], 20; AVR with CABG, 13). Among isolated SAVR cases, the mean operative mortality and composite morbidity or mortality were 1.5% and 9.7%, respectively, at the highest-volume sites (group 4), with significantly higher rates among progressively lower-volume groups (P trend < .001). After adjustment, lower-volume centers had increased odds of operative mortality (group 1 vs group 4 [reference]: adjusted odds ratio [AOR] for SAVR, 2.24 [95% CI, 1.91-2.64]; AOR for SAVR with CABG, 1.96 [95% CI, 1.67-2.30]) and major morbidity or mortality (AOR for SAVR, 1.53 [95% CI, 1.39-1.69]; AOR for SAVR with CABG, 1.46 [95% CI, 1.32-1.61]) compared with the highest-volume institutions. Substantial variation in outcomes was observed across hospitals within each volume category, and prior outcomes explained a greater proportion of hospital operative outcomes than did prior volume.

Conclusions: Operative outcomes after SAVR with or without CABG is inversely associated with institutional procedure volumes; however, prior outcomes are more predictive of future outcomes than is prior volume. Given the excellent outcomes observed at many lower-volume hospitals, procedural outcomes may be preferable to procedural volumes as a quality metric.