Prediction of late results following valve replacement in aortic valve stenosis. Seventeen years of follow-up examined with the Cox regression analysis.

: A total of 122 patients (mean age: 55 years, range 20-75 years) were alive 30 days after isolated aortic valve replacement (AVR) performed in 1965-73 for aortic valve stenosis (AS). Cold chemical Bretschneider cardioplegia was used in all operations. The maximum follow-up was 17 years, mean 9 years. The 10-year cumulative survival +/- SE was 63 +/- 4%. Twenty-one preoperative clinical, invasive and valve-related variables were entered into a Cox regression analysis. Variables with independent predictive value regarding long-term survival were: Left ventricular failure (regression coefficient: b = 1.078, p less than 0.0002), age (b = 0.749, p less than 0.009), pulse pressure (b = -0.663, p less than 0.02) and cardio-thoracic index (CTI) (b = 0.603, p less than 0.04). Based on these variables, a prognostic index with 16 different risk groups was made. In the "best" group (n = 8), the observed 10-year survival +/- SE was 88 +/- 11%, compared to 13 +/- 12% for the "worst" group (n = 8). Multivariate analysis of complication-free survival showed that the total valve-related complication-rate (VRC) (4.2/100 pat.-yrs) was influenced by valve type (Starr-Edwards ball valves, n = 118, versus Lillehei-Kaster and Smeelof-Cutter valves, n = 4) and CTI. The valve- and heart-related events (VRC together with late myocardial infarctions and pacemaker implantations, 6.5/100 pat.-yrs) depended on CTI and previous myocardial infarction. No predictors of thromboembolism (1.6/100 pat.-yrs) or coumarin-related hemorrhage (1.8/100 pat.-yrs) could be identified.(ABSTRACT TRUNCATED AT 250 WORDS)