Background and Objective: We aimed to develop a clinical prediction rule for 30-day mortality in patients undergoing Björk-Shiley convexo-concave (BScc) valve replacement, based on primary valve implantation procedures. Methods: We studied 30-day mortality in 2,263 patients who received a BScc valve between 1979 and 1985 in The Netherlands. A logistic regression model was constructed and internally validated with bootstrapping techniques. Results: Predictors for mortality were emergency operation, coronary artery disease, mitral or double-valve replacement, BScc implant being a reoperation, endocarditis, poor left ventricular function, age, concomitant tricuspid valve plasty, and aortic root replacement. The area under the receiver operating characteristic curve (AUC) for the predicted probability of death was 0.76 (95% CI 0.72-0.80). The model calibrated well, and had adequate discriminative ability in 195 BScc valve patients who underwent explantation, including 53 prophylactic explantations (AUC 0.81, 95% CI 0.73-0.89). Conclusion: A prediction rule derived from BScc valve implantation data performed well in patients undergoing explantation. Because the model provides accurate individualized estimates of the risk of mortality after reoperation, it may be helpful for decision making in patients with BScc valves.

, , , , ,,
Journal of Clinical Epidemiology
Erasmus MC: University Medical Center Rotterdam

van Gorp, M., Steyerberg, E., Kallewaard, M., & van der Graaf, Y. (2003). Clinical prediction rule for 30-day mortality in Björk-Shiley convexo-concave valve replacement. Journal of Clinical Epidemiology, 56(10), 1006–1012. doi:10.1016/S0895-4356(03)00172-0