CLINICAL USEFULNESS OF RISK-STRATIFIED OUTCOME ANALYSIS IN CARDIAC-SURGERY IN NEW-JERSEY

Background. The results of aortocoronary bypass grafting are under inc reasing scrutiny by the Health Care Financing Agency, health maintenan ce organizations, and the news media. Surgeons and hospital administra tors are concerned that erroneous conclusions may be drawn from raw ou tcome data, which do not reflect the patient's preoperative condition. It is our contention that any realistic comparison of results among s urgeons or institutions must take that condition into account through a process of risk management. Methods. We have developed a statistical model for risk stratification based on data compiled systematically a t the Newark Beth Israel Medical Center since 1980. Univariate analysi s and stepwise logistic regression are used to identify the most signi ficant risk factors and determine the appropriate weight for each. Our original risk stratification system has now been updated by eliminati ng the optional fields and reweighting the variables. This has reduced the subjective input and improved the accuracy. Results. Use of the m odified system shows good correlation between expected and observed ou tcomes at our institution and in other cases reported to the New Jerse y Department of Health. It has improved the results especially in high -risk cases: in total, a group of 5,336 patients have been assessed by the modified system: the expected mortality overall was 7.2% and the observed mortality was 5.4%. In 1,280 high-risk patients, ie, those wi th an expected mortality of greater than 11%, the expected mortality w as 16.2% and the observed mortality was 12.3%. Conclusions. Our result s suggest a decline in length of hospital stay and beneficial changes in operative procedures. They also indicate that exclusion of high-ris k cases will result in only minimal financial savings, perhaps less th an 2%.