Applying mobility prediction models to real-world patients with major amputations

Area of Science:

• Rehabilitation Medicine
• Prosthetics and Orthotics
• Health Services Research

Background:

• Mobility prediction models (PMs) are widely used for patient counseling post-major amputation (MA).
• Existing PMs require validation in real-world, socioeconomically diverse populations.
• This study evaluates PM performance in a disadvantaged population with peripheral arterial disease-related MA.

Purpose of the Study:

• To assess the accuracy of established mobility prediction models in a socioeconomically disadvantaged patient cohort undergoing major amputation.
• To compare predicted versus actual mobility outcomes at one year post-amputation.

Main Methods:

• Retrospective review of 126 patients with MA due to peripheral arterial disease (2016-2022).
• Exclusion of nonambulatory patients pre-MA or with contralateral MA.
• Evaluation of three PMs: AmpPredict, Amputee Single Item Mobility Measure, and a Vascular Quality Initiative (VQI) model.

Main Results:

• The cohort (60% non-White, Area Deprivation Index 9/10) differed significantly from PM derivation cohorts.
• Actual one-year mobility was 43%; published models showed poor predictive accuracy.
• AmpPredict and VQI models overestimated mobility, with significant discrepancies between predicted and actual rates.

Conclusions:

• Existing mobility prediction models significantly overestimate functional mobility in socioeconomically disadvantaged populations.
• Demographic and comorbidity differences between patient cohorts impact model performance.
• Clinical application of PMs requires careful consideration of patient population characteristics relative to model derivation data.