Hospital characteristics associated with failure to rescue in cardiac surgery.

Objective:The study objective was to examine the association between hospital processes of care and failure to rescue in a diverse, multi-institutional cardiac surgery network. Methods:Failure to rescue was defined as an operative mortality after 1 or more of 4 complications: prolonged ventilation, stroke, renal failure, and unplanned reoperation. Society of Thoracic Surgeons data from 20,950 consecutive patients in the Columbia HeartSource network who underwent 1 of 7 cardiac operations-coronary artery bypass grafting, aortic valve replacement ± coronary artery bypass grafting, mitral valve repair or replacement ± coronary artery bypass grafting-were analyzed to calculate failure to rescue rates. Hospital-specific characteristics were ascertained by survey method. Multivariable mixed-effects logistic models assessed the association of these hospital characteristics with failure to rescue while adjusting for patient-related factors known to be associated with mortality. Results:Failure to rescue rates at affiliate hospitals ranged from 5.45% to 21.74% (median, 12.5%; interquartile range, 6.9%). When controlling for Society of Thoracic Surgeons-predicted risk of mortality with hospital as a random effect, 4 hospital characteristics were found to be associated with lower failure to rescue rates; the presence of cardiac-trained anesthesiologists (odds ratio, 0.41; CI, 0.31-0.55, P < .001), availability of extracorporeal membrane oxygenation mechanical circulatory support (odds ratio, 0.41; CI, 0.31-0.54, P < .001), ratio of intensive care unit beds to intensivists (odds ratio, 0.87; CI, 0.76-0.99, P = .039), and total number of intensive care unit beds (odds ratio, 0.97; CI, 0.96-0.99, P = .002). Conclusions:In a diverse multi-institutional cardiac surgical network, we were able to identify specific hospital processes of care associated with failure to rescue, even when adjusting for patient-related predictors of operative mortality.