Performance-based balance tests, combined with the number of falls recalled in the past year, predicts the incidence of future falls in established unilateral transtibial prosthesis users.

BACKGROUND

Falls are common and consequential events for lower limb prosthesis (LLP) users. Currently, there are no models based on prospective falls data that clinicians can use to predict the incidence of future falls in LLP users. Assessing who is at risk for falls, and thus most likely to need and benefit from intervention, remains a challenge.

OBJECTIVE

To determine whether select performance-based balance tests predict future falls in established, unilateral transtibial prosthesis users (TTPU).

DESIGN

Multisite prospective observational study.

SETTING

Research laboratory and prosthetics clinic.

PARTICIPANTS

Forty-five established, unilateral TTPU.

INTERVENTION

Not applicable.

MAIN OUTCOME MEASURES

The number of falls reported over a prospective 6-month period. Timed Up-and-Go (TUG) and Four-Square Step Test (FSST) times, as well as Narrowing Beam Walking Test scores were recorded at baseline, along with the number of falls recalled over the past 12 months and additional potential fall-risk factors.

RESULTS

The final negative binomial regression model, which included TUG (P = .044) and FSST (P = .159) times, as well as the number of recalled falls (P = .009), was significantly better than a null model at predicting the number of falls over the next 6 months (X [3] = 11.6, P = .009) and fit the observed fall count data (X [41] = 36.12, P = .20). The final model provided a significant improvement in fit to the prospective fall count data over a model with fall recall alone X (1) = 4.342, P < .05.

CONCLUSION

No combination of performance-based balance tests alone predicted the incidence of future falls in our sample of established, unilateral TTPU. Rather, a combination of the number of falls recalled over the past 12 months, along with TUG and FSST times, but not NBWT scores, was required to predict the number of "all-cause" falls over the next 6 months. The resulting predictive model may serve as a suitable method for clinicians to predict the incidence of falls in established, unilateral TTPU.