Propensity Score-Based Analysis of Initial Endovascular and Open Revascularization for Peripheral Arterial Disease in a Long-Term Population-Based Study

Abstract

In a population-based study spanning two decades (1990-2009), we previously reported adverse limb-related outcomes following endovascular first (ER) compared with open first (OR) revascularization in patients with critical limb ischemia (CLI). The aim of this study was to assess long-term outcomes adjusting for propensity score (PS) to account for initial selection bias between OR and ER. Patients from a single county were categorized into OR or ER based on the index intervention, and CLI or claudication (CLAUD) at presentation. Data were assessed using multiple variable Cox proportional hazards regression including PS, and considering patient variables (age, diabetes mellitus, coronary artery disease, renal, and pulmonary) and urgency of procedure, which were significant on initial univariable analyses. Endpoints included major amputation (AMPU), major adverse limb events (MALE), amputation-free survival (AFS), reintervention-free survival (RFS), and mortality. There were 957 person-limbs (622 patients); mean age was 67.2 years (range, 26-97 years). The index intervention was for CLI in 415 limbs (OR, 270; ER, 145) and CLAUD in 540 (OR, 263; ER, 277). Median follow-up was 7.7 years. Major amputation was performed secondarily in 85 limbs. On PS-based multivariate analysis, ER was adversely associated with MALE (hazard ratio [HR], 1.5; P = .02), AMPU (HR, 2.0; P = .003), AFS (HR, 1.4; P = .004), and reintervention (HR, 1.7; P = .008) in patients with CLI and with increased reintervention (HR, 1.7; P = .01) in CLAUD. Mortality was not statistically significantly different between groups (HR, 1.3; P = .06). Considering mortality as competing risk, 5-year cumulative incidence of AMPU was 26.9% in ER and 13.0% in OR in patients with CLI. PS-based analysis corrected for selection bias in this population-based study showed endovascular first revascularization to be associated with significantly worse long-term limb-related outcomes in patients with CLI and greater reintervention in patients with CLAUD.