Twenty-two-year experience with aortic valve replacement. Starr-Edwards ball valves versus disc valves.

From 1965 through 1986, 817 patients underwent aortic valve replacement at our institution. Six hundred forty-five patients received Starr-Edwards ball valves, including 286 Silastic ball valves (Models 1200/1260), 165 cloth-covered caged-ball prostheses (Models 2300/2310/2320), and 194 track-valve prostheses (Model 2400). In contrast, 172 patients received disc-valve prostheses, including 126 St. Jude Medical aortic bi-leaflet disc valves, 32 Lillehei-Kaster pivoting disc valves, and 14 Björk-Shiley valves (6 convexoconcave and 8 monostrut). With respect to preoperative data, the 2 groups were comparable, with the following differences. The Starr-Edwards group included 1) more men (77% versus 51%; p < 0.0001); 2) a significantly older patient population (59 +/- 10 years versus 56 +/- 15 years; p < 0.0001); 3) more patients in New York Heart Association functional class III or IV (72% versus 65%; p < 0.01); 4) fewer patients with angina pectoris as a limiting symptom (20% versus 36%; p < 0.0001); and 5) patients who tended to receive larger prostheses (26 +/- 2 mm versus 23 +/- 3 mm, p < 0.0001). The overall 10-year survival rate +/- standard error was 59% +/- 2% for patients receiving Starr-Edwards valves and 63% +/- 6% for those with disc valves. The linearized complication rates (expressed as percentage per patient-year +/- standard error) for the Starr-Edwards and disc-valve groups, respectively, were 2.0% +/- 0.2% and 1.4% +/- 0.5% for thromboembolism, 2.1% +/- 0.2% and 3.9% +/- 0.8% for Coumadin-related hemorrhage, 0.5% +/- 0.1% and 0.3% +/- 0.2% for endocarditis, 0.3% +/- 0.1% and 0.7% +/- 0.3% for other prosthesis-related complications, and 4.8% +/- 0.1% and 6.4% +/- 1.0% for all complications together. There were no instances of thrombotic occlusion or mechanical failure. After the 6th postoperative year, no thromboembolic events were encountered in patients with a Silastic ball valve; the 15-year freedom from thromboembolic events was 89%. Cox regression analysis showed that 1) a prosthetic orifice diameter of 15 mm or less was associated with an increased mortality; 2) disc valves entailed an increased rate of hemorrhage and prosthesis-related complications considered as a whole; 3) and Lillehei-Kaster valves led to an increased rate of prosthesis-related complications other than thromboembolism, hemorrhage, and endocarditis. Neither the type of prosthesis nor the size influenced the rate of thromboembolism, endocarditis, or prosthesis replacement. Because of their proven durability and relatively low price, we advocate the continued use of Starr-Edwards Model 1260 Silastic ball valves that have an orifice diameter of 16 mm or more.