Development of a physical mobility prediction model to guide prosthetic rehabilitation.

The data came from a private prosthetics practice and only included records with PLUS-M scores, so people without prosthetic care or without PLUS-M scores are not represented. Records were excluded if they did not contain 2 or more PLUS-M time points or missing covariates, which may remove patients with less complete follow-up. Some subgroups were limited, including partial foot amputation, bilateral amputation, certain etiologies and people early after amputation, so predictions may underestimate extremely poor outcomes. The model was developed and tested only on old records, not on new patients, and it was not compared with predictions from experienced prosthetists. The outcome is self-reported PLUS-M mobility, which is prone to reporter bias and may not match performance-based function. PLUS-M observations were treated as independent, and timing of collection depended on clinic patterns, which may limit generalizability. No alternative prediction methods were available for comparison, so the improvement over other prediction methods is not established.

No funding or conflict-of-interest declaration is supplied. The data were provided by Hanger Clinics, a private prosthetics practice.

Do not treat this model as a validated clinical decision tool. It predicted self-reported PLUS-M mobility in old records, but it was not compared with experienced prosthetist estimates and was not tested prospectively in patients.