OBJECTIVE: To test prospectively the unsubstantiated claim that patient-specific predictions of time-related outcome after coronary artery bypassgrafting (CABG) from multivariable parametric equations are reliable formedical decision making and for intra- and interdepartmental qualitycontrol in surgical training and practice. METHODS: 3720 survival curveswere generated prospectively for all primary, isolated CABG patientsoperated upon at the Katholieke Universiteit (KU) Leuven between July, 1987and January, 1992 using the published AHA/ACC guidelines multivariableequation derived from prior KU Leuven experience. The average of thesecurves (risk-adjusted predicted survival) was compared to the Kaplan-Meier(actual) estimates, overall and for patient subsets. Variables associatedwith systematic deviation of actual from predicted number of deaths weresought by multivariable residual risk analysis. RESULTS: Actual overallsurvival was less good than predicted (P = 0.03) and the excess risk wasdistributed uniformly across time. The excess risk was not attributable tosubstantial changes in prevalence of known risk factors. It wasattributable largely to a small subset of patients (n = 292) withlow-prevalence, but important risk factors not accounted for by theequation (P = 0.7, for difference in survival among the remaining 3428patients). CONCLUSIONS: Within the confines of a single institution,patient-specific predictions of outcome after CABG can be made reliably inmost patients using multivariable equations developed from a heterogeneousexperience, despite changes in prevalence of risk factors. New subsets ofhigh-risk patients, failure or inability to account for important rare riskfactors or for institutional changes, may lead to systematic errors ofprediction. Under these limitations it is an excellent tool for medicaldecision making and audit of surgical training and practice.