COMPARATIVE-ANALYSIS OF CORONARY SURGERY RISK STRATIFICATION MODELS

Background. Preoperative risk assessment models for coronary bypass su rgery (CABG) have been proposed, but comparison of them using independ ent databases needs to be done. Methods. Models of CABG hospital morta lity were tested on a set of 3443 patients who underwent CABG includin g a subset of 3237 patients who had isolated CABG (no valve procedures ), in our database since 1991. Four models previously described were d esignated as Parsonnet (PS), Cleveland (CL), and Society of Thoracic S urgeons version 1 (ST1) and version 2 (ST2). We developed our own Baye sian (BA) and logistic regression (LR) models and calibrated the PS an d CL models on 2842 patients operated on prior to 1991. Models were co mpared with respect to 1) mean predicted mortality, 2) correlation of predicted to observed mortality, 3) Brier mean probability score, 4) d escriptive statistics, 4) the C-Index (area beneath the receiver opera ting characteristic curve), and 5) predictive efficiency. Since the ST 1 and ST2 models were developed for use only with isolated CABG patien ts, these models were compared with the others using an isolated CABG subset. Results. Observed mortality for all 3443 CABG patients was 4.0 %. For this group, the mean mortality predicted by PS, CL, BA, LR, was 9.0 +/- 8.0, 6.0 +/- 6.0, 7.6 +/- 15.6, and 5.1 +/- 7.7 (mean +/- sta ndard deviation) respectively. C-Indexes were .80 +/- .02, .80 +/- .02 , .83 +/- .02, and .80 +/- .02 (C-Index +/- standard error) respective ly. Observed mortality for 3237 isolated CABG patients was 3.7%. For t his subgroup, the mean mortality predicted by PS, CL, BA, LR, ST1, and ST2 was 8.4 +/- 7.4, 5.7 +/- 5.9, 6.5 +/- 13.9, 4.5 +/- 6.5, 9.6 +/- 9.1, and 3.0 +/- 3.3 respectively. C-Indexes were .80 +/- .03, .80 +/- .03, .83 +/- .02, .79 +/- .03, .77 +/- .03, and .81 +/- .02 respectiv ely. Conclusions. Existing CABG models can accurately discriminate out come about 80 percent of the time. Models developed on a national data base and those from non-local databases appear to have validity for ou r local data set. Predictions can vary widely between models and exist ing methods for comparing models appear to be inadequate. The methodol ogy presented here is applicable for use with patients undergoing inte rventions in the cardiac catheterization laboratory.