Interhospital variability in failure to rescue rates following aortic valve surgery.

ABSTRACT

Objective: This study evaluated interhospital variability and determinants of failure-to-rescue for patients undergoing surgical aortic valve replacement. Methods: An observational study was conducted among 28,842 patients undergoing aortic valve replacement with or without coronary artery bypass grafting between July 2011 and June 2017 across 90 hospitals participating in the Society of Thoracic Surgeons Adult Cardiac Surgery Database. Postoperative complications were defined as major (stroke, renal failure, reoperation, prolonged ventilation, sternal infection) and overall (major plus 14 other morbidities). Hospital terciles of observed to expected (O/E) mortality were compared on crude rates of major and overall complications, operative mortality, and failure to rescue (among major and overall complications). The correlation between hospital observed and expected failure-to-rescue rates was assessed. Results: Median Society of Thoracic Surgeons Adult Cardiac Surgery Database predicted mortality risk was similar across hospital OE mortality terciles (P = .10). As expected, mortality rates significantly increased across terciles (low O/E tercile 1.6%, high O/E tercile 4.7%; P < .001). Failure-to-rescue rates increased substantially across hospital mortality terciles among patients with major (low tercile, 8.8% and high tercile, 20.8%) and overall (low tercile, 3.0% and high tercile, 8.9%) complications. Hospital-level expected failure to rescue had a higher correlation with observed complications for overall complications (R2 = 0.71) compared with Society of Thoracic Surgeons major complications (R2 = 0.24). Conclusions: Considerable interhospital variation exists in failure-to-rescue rates following aortic valve replacement. Hospitals in the low O/E mortality tercile experience failure to rescue nearly one-third less than those in the high O/E mortality tercile. Efforts to advance quality will benefit from identifying and disseminating optimal rescue strategies in this patient population.