Gait variability predicts falls in older adults with multiple sclerosis

Abstract Background Falls and gait variability are prevalent in older adults with and without multiple sclerosis. Gait variability has been associated with an increased likelihood of reporting falls in older adults, yet its prediction of falls in older adults with multiple sclerosis remains unclear. Methods We examined whether gait variability measured under single- and dual-task walking conditions predicted falls during longitudinal follow-up in older adults with multiple sclerosis (OAMS) and healthy older adults (HOA). HOA (n = 106, mean age=69 years) and OAMS (n = 95, mean age=65 years) completed a single and dual-task walking paradigm and reported falls during a longitudinal follow-up. Gait variability was measured using an instrumented walkway. Results Cox-regression models indicated that larger coefficients of variation (CVs) of dual-task-walk stride length (HR = 1.04, p = 0.017), dual-task-walk swing time (HR = 1.03, p = 0.044), and single-task-walk swing time (HR = 1.05, p < 0.001) were significantly associated with increased hazards of incident falls, even after adjustment. Cohort-stratified cox-regression models with adjustment showed that larger single-task-walk swing time CV was associated with a higher hazard of incident falls in HOA (HR = 1.10, p = 0.028), while larger CVs in dual-task-walk stride velocity (HR = 1.04, p = 0.017), stride length (HR = 1.06, p = 0.016), and swing time (HR = 1.05, p = 0.018) were significantly associated with a higher hazard of incident falls in OAMS. Conclusions Findings suggest that greater gait variability predicts increased fall risk in OAMS and HOA participants, with a 1% increase in CV associated with a 4-10% fall hazard increase. However, walking condition influenced this association with single-task-walk variability being predictive in HOA, while dual-task-walk variability is more predictive in OAMS.