Are blood and valve cultures predictive for long-term outcome following surgery for infective endocarditis?

Objective: To evaluate whether perioperative bacteria identification in blood and/or in valve cultures can predict early and late outcome of surgery for infective endocarditis, a retrospective study was performed, Methods: Between January 1978 and December 1998, 232 patients, 79 (34.1%) female and 153 (65.9%) male with mean age of 44.95 +/- 1.03 years (range 8-79) underwent surgery for infective endocarditis on a native (162 cases) or prosthetic (70 cases) valve. Patients were divided into three groups according to the perioperative x of microbiological tests: Group A: patients with preoperative positive blood cultures (83 cases); Group B: patients with positive valve cultures (35 cases); Group C: patients with negative blood and valve cultures (114 cases). Categorical values were compared by chi(2) analysis, whereas continuous data were compared by ANOVA and Bonferroni correction for post hoc comparisons. Analysis of late survival and complications was performed with Kaplan-Meier and Log Rank test. Late mortality, reoperation, perivalvular leak, recurrence of infection were considered as treatment failure. All data were presented as mean +/- standard error. Results: Hospital mortality was 10.8% (9/83) in Group A, 8.6% (3/35) in Group B, and 14.9% (17/114) in Group C (P = 0.52; not significant (NS)). Ten-year survival was 62.7 +/- 8% in Group A, 43.9 +/- 19% in Group B and 62.7 +/- 7% in Group C (P = 0.38; NS). Ten-year freedom from reoperation was 85.2 +/- 6% in Group A, 37.9 +/- 20% in Group B and 80 +/- 6% in Group C (P = 0.0034). Ten-year freedom from treatment failure was 56.3 +/- 8% in Group A, 31.6 +/- 16% in Group B and 55.3 +/- 7% in Group C (P = 0.46, NS). Conclusions: Positive blood and tissue cultures are not predictive for hospital mortality and late treatment failure in patients with infective endocarditis. Positive valve cultures, a common finding in patients with staphylococcal endocarditis, are predictive for a higher risk of reoperation. (C) 2000 Elsevier Science B.V. All rights reserved.