Predictive Modeling for One-Year Lower Extremity Endovascular Revascularization Failure in Black Persons

Area of Science:

• Vascular Surgery
• Health Disparities
• Predictive Analytics

Background:

• Peripheral artery disease (PAD) disproportionately affects Black persons.
• Black individuals experience higher rates of endovascular revascularization failure (ERF) compared to non-Hispanic White individuals.

Purpose of the Study:

• To identify predictors of ERF in Black persons with PAD.
• To develop and evaluate predictive models for ERF in this population.

Main Methods:

• Retrospective study of 163 Black individuals undergoing endovascular revascularization for PAD (2011-2018).
• Development of three predictive models: logistic regression, penalized logistic regression (LASSO), and random forest (RF).
• Performance evaluation using repeated cross-validation.

Main Results:

• 63.1% of patients experienced ERF at one year.
• Significant differences in symptom status, lesion location, diabetes status, and surgeon procedural volume were observed in patients with ERF.
• Logistic regression and LASSO identified tissue loss, smoking, femoro-popliteal lesions, and diabetes control as risk factors.
• RF model highlighted procedural volume, age, PAD symptoms, comorbidities, and lesion location as key predictors.
• All models achieved ≥75% overall accuracy, with LASSO and RF showing higher sensitivity but lower specificity.

Conclusions:

• Black individuals undergoing endovascular revascularization for PAD are at high risk of ERF.
• Predictive models show promise for identifying high-risk patients, aiding targeted interventions.
• Further research into these models may enhance limb salvage rates for this demographic.