Identifying gait characteristics associated with freezing of gait in Parkinson's disease: An analysis of on and off medication states.

The study included only 36 participants, and the freezing severity models used only the 15 participants who had observed freezing, so the findings come from a small group. The study was primarily exploratory and cross-sectional, so it shows associations with freezing severity and presence, not that gait variability causes freezing or predicts future freezing. Freezing severity was assessed with a brief video-recorded observational test, and freezing can be transient, so the score may not capture all real-world freezing episodes. The gait analysis used straight-line walking at a self-selected speed and removed passes that included freezing, so it may miss gait problems that appear during turning, narrow spaces, or dual-task conditions. The classification model had overall good accuracy, but the individual gait variables in the logistic model were not all statistically significant.

The authors declared no conflicts of interest. The article is listed as supported by NIH extramural funding.

Do not use these gait measures as a ready-to-use clinical test for freezing of gait. The study was exploratory, included 36 people, and the logistic model's individual gait variables were not all statistically significant, so the overall accuracy needs confirmation in larger samples.