Background: Gait disturbance is an important risk factor for falls in Parkinson's disease (PD). Using wearable sensors, we can obtain the spatiotemporal parameters of gait and calculate the gait variability. This prospective study aims to objectively evaluate the gait characteristics of PD fallers, and further explore the relationship between spatiotemporal parameters of gait, gait variability and falls in PD patients followed for six months.
Methods: Fifty-one PD patients were enrolled in this study. A seven-meter timed up and go test was performed. Gait characteristics were determined by a gait analysis system. Patients were followed monthly by telephone until the occurrence of falls or till the end of six months. The patients were categorized into fallers and non-fallers based on whether fell during the follow-up period. Gait parameters were compared between two groups, and binary logistic regression was used to establish the falls prediction model. In the receiver-operating characteristic curve, area under the curve (AUC) was utilized to evaluate the prediction accuracy of each indicator.
Results: All subjects completed the follow-up, and 14 (27.5%) patients reported falls. PD fallers had greater gait variability. The range of motion of the trunk in sagittal plane variability was an independent risk factor for falls and achieved moderate prediction accuracy (AUC = 0.751), and the logistic regression model achieved a good accuracy of falls prediction (AUC = 0.838).
Conclusions: Increased gait variability is a significant feature of PD fallers and is more sensitive to detect PD patients at high risk of falls than spatiotemporal parameters.
Keywords: Parkinson’s disease; fall; gait variability; prediction; wearable sensors.