Selecting Treatments for Peripheral Artery Disease: Differences Between Registry and a Randomized Controlled Trial Population

Abstract

Introduction

Patients enrolled in randomized trials are carefully selected and may have different comorbidities than patients treated in everyday practice.

Methods

We compared characteristics of 1815 patients enrolled in the Best Endovascular or Surgical Treatment for Critical Limb Ischemia (BEST-CLI, NCT02060630) with 104,877 patients receiving endovascular treatment and 32,120 patients undergoing bypass in the Vascular Quality Initiative's (VQI) registry from 2014 to 2020 using descriptive statistics. We studied mortality by treatment type among patients in both the trial and registry using Cox regression. We adjusted for differences in patient characteristics using inverse probability weighting with propensity scores.

Results

Compared to the BEST-CLI participants, patients in the VQI registry were commonly older, female, and of non-Hispanic ethnicity. Chronic obstructive pulmonary disease and congestive heart failure were more prevalent in VQI, while coronary artery disease and diabetes rates were higher in BEST-CLI. The unadjusted 1-y mortality in VQI was 12.5% following endovascular treatment and 10.2% following bypass. After weighting VQI patients to represent the BEST-CLI sample, the cumulative 5-y mortality was higher in those undergoing endovascular treatment versus bypass (26.3% versus 23.7%, P < 0.001). Bypass was associated with an 8% lower mortality than endovascular treatment (hazard ratio = 0.92, 95% CI:0.87-0.98, P = 0.005). This effect remained across all weighting schemes, even when limiting to patients treated at a BEST-CLI site.

Conclusions

Patients enrolled in BEST-CLI differ from patients treated in VQI. However, reweighting VQI data to represent BEST-CLI yields similar estimates of treatment effects in VQI data, supporting a role for registry-based analytic models in answering comparative, real-world clinical questions.