Disclosures: Shane R. Wurdeman, PhD, CP: Nothing to disclose

Objective: The purpose of this study was to develop a model which provides the probability of an amputee functioning as either a limited community/household ambulator (i.e. K1/K2) or an unlimited community ambulator (i.e. K3/K4).

Design: Retrospective cross-sectional.

Setting: Multi-clinic collaboration of prosthetic limb providers.

Participants: 2,770 lower limb prosthesis users were included. 20% of users were randomly selected to build the predictive model (n=554, age: 57.0±14.7y, ht: 174.4±11.4cm, wt: 89.2±23.5kg), with the remaining patient records used to test the model (n=2216, age: 57.2±14.5y, ht: 174.6±11.2cm, wt: 89.9±23.1kg)

Interventions: Not applicable.

Main Outcome Measures: Prosthetic Limb Users Survey of Mobility (PLUS-M), Functional Comorbidity Index, Classification Tree.

Results: The resultant classification tree had classification accuracy of 87.4% (SEM 1.4%) for the training sample and 81.6% (SEM 0.82%) for the testing sample. Twelve of the nodes were found to have class assignment significantly different from a random assignment (P<.05).

Conclusions: The classification tree is a useful tool for aiding K-level determination as it provides a probability of functional potential based on results of nearly 2,800 other lower limb prosthesis users with similar characteristics. For example, a clinician is now aware that a diabetic amputee, age 62, PLUS-M T-Score of 53, and body weight 80.0kg, has a 75% probability of being an unlimited community ambulator (i.e. K3/K4).

Level of Evidence: Level III

« Back to AAPM&R Annual Assembly 2019

PM&R Meeting Abstracts - https://pmrjabstracts.org/abstract/classification-tree-to-determine-probability-of-amputee-functional-potential/