Longitudinal Assessment of Safety of Femoropopliteal Endovascular Treatment With Paclitaxel-Coated Devices Among Medicare Beneficiaries: The SAFE-PAD Study

Abstract

Importance: Paclitaxel-coated peripheral devices have been associated with increased mortality, yet this harm signal has not been replicated outside of meta-analyses of small trials.

Objective: To provide a longitudinal assessment of the safety of femoropopliteal endovascular treatment with peripheral drug-coated devices (DCDs) among Medicare beneficiaries.

Design, setting, and participants: SAFE-PAD (Safety Assessment of Femoropopliteal Endovascular Treatment With Paclitaxel-Coated Devices) was a retrospective cohort study designed with the US Food and Drug Administration to evaluate the noninferiority of mortality between DCDs and non-drug-coated devices (NDCDs) for femoropopliteal revascularization performed in 2978 inpatient and outpatient facilities in the US from April 1, 2015, through December 31, 2018. Evaluation of the primary outcome was assessed through May 31, 2020. Participants were Medicare fee-for-service beneficiaries 66 years and older with 1 or more years of enrollment prior to femoropopliteal revascularization. Prespecified subgroups included low-risk cohorts, procedure location, disease severity, and device type. Inverse probability weighting was used to account for imbalances of observed characteristics. Sensitivity analyses were used to evaluate the potential influence of unmeasured confounding.

Exposures: Treatment with DCDs vs NDCDs as determined by claims codes during the index procedure.

Main outcomes and measures: The primary outcome was all-cause mortality. Secondary outcomes included repeated hospitalization, repeated lower extremity revascularization, and lower extremity amputation. Falsification end points were acute myocardial infarction, congestive heart failure, and pneumonia.

Results: Of 168 553 patients, 70 584 (41.9%) were treated with a DCD. The mean (SD) age was 77.0 (7.6) years, 75 744 (44.9%) were female, 136 916 of 167 197 (81.9%) were White individuals, 85 880 of 168 553 (51.0%) had diabetes, 82 554 of 168 553 (49.0%) used tobacco, 78 665 of 168 553 (45.7%) had critical limb ischemia (CLI), and 13 296 of 168 553 (7.9%) had a prior amputation. Median follow-up was 2.72 years (interquartile range, 0.87-3.77; longest, 5.16 years). After weighting, the cumulative incidence of all-cause mortality was 53.8% with DCDs and 55.1% with NDCDs (hazard ratio [HR], 0.95; 95% CI, 0.94-0.97; noninferiority P < .001). Cox regression and instrumental variable analyses were consistent with the primary findings. No harm associated with DCDs was observed among subgroups, including those treated with stents (HR, 0.97; 95% CI, 0.95-1.00) or balloons (HR, 0.94; 95% CI, 0.92-0.96), with or without CLI (CLI: HR, 0.95; 95% CI, 0.93-0.97; non-CLI: HR, 0.97; 95% CI, 0.95-0.99), and those within the lowest quartile of total comorbidities (HR, 0.95; 95% CI, 0.92-0.99).

Conclusions and relevance: In this initial report from the SAFE-PAD cohort study, DCDs were found to be noninferior to NCDCs in respect to mortality through a median follow-up of 2.72 years. This finding remained robust in sensitivity analyses and was consistent across prespecified subgroups.

Figure 1.. Unweighted and Weighted Cumulative Incidence of Mortality Curves, Stratified by Treatment With or Without Drug-Coated Devices

Displayed are the unweighted (A) and weighted (B) cumulative incidence curves of mortality, stratified by treatment with a drug-coated or non–drug-coated device. Of the 168 553 Medicare beneficiaries studied, the median (interquartile range) follow-up was 993 (319-1377) days, and longest follow-up was 1883 days. At 1883 days, the crude cumulative incidence of mortality was 56.7% in the non–drug-coated device group and 51.6% in the drug-coated device group, with an unadjusted hazard ratio of 0.84 (95% CI, 0.82-0.85). After weighting, the cumulative incidence of mortality at 1883 days was 55.1% among those treated with non–drug-coated devices and 53.8% among those treated with drug-coated devices, with an adjusted hazard ratio of 0.95 (95% CI, 0.94-0.97). The 1-sided P value for the noninferiority test was <.001.

Figure 2.. Simulation of a Hypothetical Uncontrolled Confounder and Association With Measured Covariates in SAFE-PAD

Displayed is the simulation of a hypothetical uncontrolled confounder or group of confounders with a prevalence of 50% (orange curves), 20% (blue curves), and 10% (gray curves). The black dots represent a scatterplot of the odds ratio of treatment vs hazard ratio of mortality for the covariates adjusted for in the primary analysis of SAFE-PAD. In the SAFE-PAD cohort, no single measured covariate would have met the criterion to overturn the noninferiority margin. SAFE-PAD indicates Safety Assessment of Femoropopliteal Endovascular Treatment With Paclitaxel-Coated Devices.

Figure 3.. Adjusted Risks of All-Cause Mortality Between Drug-Coated and Non–Drug-Coated Devices Among Prespecified Subgroups

Displayed are the adjusted risks of mortality associated with drug-coated vs non–drug-coated devices across the different prespecified subgroups. Each subgroup underwent separate inverse probability weighting. No evidence of harm associated with drug-coated devices was observed across any of the subgroups. HR indicates hazard ratio.