Comparison of Medical and Surgical Treatment for Prosthetic Valve Endocarditis: A Propensity Score-Adjusted Study

Background: Patients with prosthetic valve endocarditis (PVE) who are treated medically have poor clinical outcomes. This could possibly be from medically-treated patients being too sick for surgery, but the possibility that medical treatment may be inferior to surgical treatment cannot be ignored. This study was undertaken to compare outcomes of medically-treated PVE with surgically-treated PVE adjusting for differences between them.

Methods: A retrospective cohort study was undertaken. Patients admitted to Cleveland Clinic with PVE from April 1, 2008 to August 31, 2010 were identified from our institution’s infective endocarditis (IE) registry. Treatment groups (medical or surgical treatment) were defined by decision to treat surgically or medically, and the date of this decision was defined as time zero. Propensity to be treated surgically was calculated in a logistic regression model including pre-selected variables felt to affect a decision on surgery, and all other baseline variables with univariable associations with treatment group assignment. Hazard for death was compared between surgically and medically-treated patients using a reduced Cox proportional hazards model that included propensity to be treated surgically, and all baselines variables with univariable associations with survival. In-hospital mortality, one-year mortality, subsequent surgery for endocarditis, and endocarditis relapse, were examined using logistic regression models adjusted for propensity to be treated surgically.

Results: Of 163 identified patients with PVE, 36 were treated medically and 127 surgically. Mean age was 61 yrs, 71% were male, and 25% had Staphylococcus aureusinfection. All but 2 had left sided involvement, and 60% had invasive disease. Medically-treated patients had a higher hazard of death (HR 4.50, 95% CI 2.30-8.81, p-value <0.0001) compared to surgically-treated patients. Medical treatment was associated with higher odds of death within one year (OR 6.32, 95% CI 2.12-20.16, p-value 0.001), subsequent surgery for IE (OR 10.5, 95% CI 2.12– 51.81, p-value 0.003), and IE relapse (OR 16.7, 95% CI 0.98-450.87, p-value 0.049), compared to surgical treatment.

Conclusion: Medical treatment is significantly less effective than surgical treatment in patients with PVE.