Backward walking sensitively detects fallers in persons with multiple sclerosis

Individuals with multiple sclerosis experience deficits in mobility resulting in injurious falls. Fall detection has proved challenging; the majority of clinical measures rely on forward walking and balance measures, yet these measures have poor sensitivity and predictive value for differentiating between fallers and non-fallers. Backward walking better differentiates fallers from non-fallers in the elderly and other neurodegenerative diseases; therefore, the objective of this study was to examine both forward and backward walking to determine the strongest, unique contributor that differentiates fallers from non-fallers in persons with multiple sclerosis.

In a single session, spatiotemporal measures of forward and backward walking and fall history were collected. For the subsequent six months, individuals recorded falls in a fall diary. Discriminant function analysis was used to determine what variables most strongly and uniquely differentiate multiple sclerosis fallers from non-fallers.

Thirty-eight individuals with multiple sclerosis participated. Forward and backward velocity, stride length, and double support time as well as age, disease severity, and symptom duration were included in the models. Together, the variables differentiated between fallers and non-fallers (Wilk's lambda χ2 (8, N = 36) = 0.497, p<0.001) and in rank order, backward walking velocity was the strongest unique predictor. Repeating the analysis with a stepwise approach yielded that backward walking velocity in the first step (χ2 (1, 34) = 0.68, F = 15.96, p<0.001) and symptom duration in the second step (χ2 = 0.59, F (2, 33) = 11.46; p<0.001) most strongly differentiated retrospective fallers and non-fallers. This stepwise model with backward walking velocity and symptom duration accurately classified 76.3% of cases. Addition of forward walking measures did not significantly improve the models, and indeed the accuracy of classification was reduced to 71.1%. Exploratory analysis showed that backward walking velocity was the best predictor of prospectively reported fallers and non-fallers (χ2 (1, 7) = 0.43, F = 9.20, p = 0.02).

Backward walking velocity exhibits the highest effect magnitude and specificity in differentiating fallers from non-fallers in individuals with MS and demonstrates potential as clinically feasible and efficient fall detection tool.