The use and effect of surgical therapy for prosthetic valve infective endocarditis: a propensity analysis of a multicenter, international cohort.

BACKGROUND

Although surgical intervention is often used in the treatment of prosthetic valve infective endocarditis (PVIE), an understanding of its effect on survival has been limited by the biases of observational studies and lack of controlled trials.

METHODS

The International Collaboration on Endocarditis Merged Database is a large, multicenter, international registry of patients with definite endocarditis by Duke criteria, including 367 patients with PVIE. Clinical, microbiologic, and echocardiographic variables were analyzed to determine those factors associated with the use of surgery for PVIE. Logistic regression analysis was performed to create a propensity model of predictors of surgery use. Patients who underwent surgery during initial hospitalization were matched by propensity score with patients treated with medical therapy alone. Logistic regression analysis was performed to determine variables independently associated with inhospital mortality in this matched subset.

RESULTS

Surgical therapy for PVIE was performed in 148 (42%) of 367 patients. Inhospital mortality was similar for patients treated with surgery compared with those treated with medical therapy alone (25.0% vs 23.4%, P = .729). Surgical therapy was independently associated with patient age, microorganism, intracardiac abscess, and congestive heart failure. After adjustment for these determinants, inhospital mortality was predicted by brain embolization (OR 11.12, 95% CI 4.16-29.73) and Staphylococcus aureus infection (OR 3.67, 95% CI 1.29-9.74), with a trend toward benefit for surgery (OR 0.56, 95% CI 0.23-1.36).

CONCLUSIONS

Despite the frequent use of surgery for the treatment of PVIE, this condition continues to be associated with a high inhospital mortality rate in the contemporary era. After adjustment for factors related to surgical intervention, brain embolism and S aureus infection were independently associated with inhospital mortality and a trend toward a survival benefit of surgery was evident.