Predictors of recurrence and reoperation for prosthetic valve endocarditis after valve replacement surgery for native valve endocarditis.

Objective Surgical treatment of native valve endocarditis remains challenging, especially in cases with paravalvular destruction. Basic principles include complete debridement and reconstruction. This study is designed to evaluate the outcomes of surgical reconstruction of complex annular endocarditis using standard techniques and materials, including autologous and bovine pericardium. Methods From 1975 to 2000, 358 cases (357 patients, mean age 49 ± 16 years, range 18–88 years) of native valve endocarditis were surgically managed. Bioprosthetic valves were implanted in 189 cases, and mechanical prostheses were implanted in 169 cases. A total of 78 cases of paravalvular destruction were identified: 62 annular abscesses, 8 fistulas, and 8 combined abscesses/fistulas. These were managed with 46 pericardial patches and 32 isolated suture reconstructions after radical debridement and prosthetic valve replacement. Results The overall early mortality was 8.4% (n = 30). The mortality with paravalvular destruction was 17.9%, and the mortality with simple leaflet infection was 5.7% ( P = .001). The unadjusted survival at 20 years was 26.4% ± 4.9% for bioprosthetic valves and 56.5% ± 8.1% for mechanical prostheses ( P = .007). The freedom from recurrent prosthetic valve endocarditis was 78.9% ± 4.4% at 15 years. The freedom from reoperation for recurrent prosthetic valve endocarditis was 85.8% ± 4.2% at 15 years. The freedom from reoperation after reconstruction for paravalvular destruction was 88.2% ± 6.9% at 15 years. The freedom from mortality for recurrent prosthetic valve endocarditis was 92.7% ± 3.4% at 15 years. The independent predictors of reoperation were age (hazard ratio 0.930, P = .005) and intravenous drug use/human immunodeficiency virus plus surgical technique (hazard ratio 12.8, P = .003 for patch reconstruction plus valve and hazard ratio 3.6, P = .038 for valve replacement only). Prosthesis type was not predictive when separated from intravenous drug use/human immunodeficiency virus (hazard ratio 3.268, P = .088). Conclusion Paravalvular destruction is associated with a higher operative mortality. Native valve endocarditis can be managed with reasonable long-term survival and low rates of reinfection with radical debridement and pericardial reconstruction with bioprostheses and mechanical prostheses. The type of prosthesis implanted does not influence long-term outcome. Patients with a history of intravenous drug use and human immunodeficiency virus are at increased risk for recurrent infection and reoperation.