Predictors Of Retention And Attrition In A Study Of An Advanced Upper Limb Prosthesis: Implications For Adoption Of The Deka Arm

Purpose: The purpose was to identify factors associated with completion of the VA home study of the DEKA Arm. Design and methodological procedures used: Differences between groups were examined using chisquare and t-tests. A multivariable logistic regression model predicting completion was generated and odds ratios (OR) for significant variables calculated. Post-hoc analysis was performed to plot the receiver operating characteristics (ROC) curve.Results: Participants who completed were more likely to be prosthesis users at study onset (p 1/4 .03), and less likely to have a history of musculoskeletal problems (p 1/4 .047). There were no statistically significant differences between groups who completed and those who did not in gender, race, veteran status, age, body mass index (BMI), weight, height, musculoskeletal pain at baseline, satisfaction with current prosthesis, type of prosthesis, or months of prosthesis use. Two variables, prosthesis use and history of musculoskeletal problems were significant at p<. 10. The area under the curve (AUC) accuracy index was 0.78.Conclusions: We considered completion of the home use study a reasonable proxy for participant willingness to adopt the device; and believe that findings can be extrapolated to guide DEKA Arm prescription recommendations. Participants most likely to complete the study were already using a personal prosthesis, and without pre-existing musculoskeletal problems.IMPLICATIONS FOR REHABILITATIONData from the VA Study of the DEKA Arm were analysed to determine which factors were associated with likely successful adoption of the DEKA Arm.Participants most likely to complete the study were those who already using a personal prosthesis, and those without pre-existing chronic or re-occurring musculoskeletal problems.This information may be useful when attempting to identify and target the most appropriate candidates for DEKA Arm prescription.