Analysis of the Impact of Early Surgery on In-Hospital Mortality of Native Valve Endocarditis Use of Propensity Score and Instrumental Variable Methods to Adjust for Treatment-Selection Bias

Background-The impact of early surgery on mortality in patients with native valve endocarditis (NVE) is unresolved. This study sought to evaluate valve surgery compared with medical therapy for NVE and to identify characteristics of patients who are most likely to benefit from early surgery. Methods and Results-Using a prospective, multinational cohort of patients with definite NVE, the effect of early surgery on in-hospital mortality was assessed by propensity-based matching adjustment for survivor bias and by instrumental variable analysis. Patients were stratified by propensity quintile, paravalvular complications, valve perforation, systemic embolization, stroke, Staphylococcus aureus infection, and congestive heart failure. Of the 1552 patients with NVE, 720 (46%) underwent early surgery and 832 (54%) were treated with medical therapy. Compared with medical therapy, early surgery was associated with a significant reduction in mortality in the overall cohort (12.1% [87/720] versus 20.7% [172/832]) and after propensity-based matching and adjustment for survivor bias (absolute risk reduction [ARR]-5.9%, P < 0.001). With a combined instrument, the instrumental-variable-adjusted ARR in mortality associated with early surgery was -11.2% (P < 0.001). In subgroup analysis, surgery was found to confer a survival benefit compared with medical therapy among patients with a higher propensity for surgery (ARR -10.9% for quintiles 4 and 5, P = 0.002) and those with paravalvular complications (ARR -17.3%, P < 0.001), systemic embolization (ARR -12.9%, P = 0.002), S aureus NVE (ARR -20.1%, P < 0.001), and stroke (ARR -13%, P = 0.02) but not those with valve perforation or congestive heart failure. Conclusions-Early surgery for NVE is associated with an in-hospital mortality benefit compared with medical therapy alone. (Circulation. 2010; 121: 1005-1013.)