Motor dual-tasking deficits predict falls in Parkinson's disease: A prospective study

Introduction: Falls severely affect lives of Parkinson's disease (PD) patients. Cognitive impairment including dual-tasking deficits contribute to fall risk in PD. However, types of dual-tasking deficits preceding falls in PD are still unclear. Methods: Walking velocities during box-checking and subtracting serial 7s were assessed twice a year in 40 PD patients over 2.8 +/- 1.0 years. Fourteen patients reported a fall within this period (4 excluded fallers already reported falls at baseline). Their dual-task costs (DTC; mean +/- standard deviation) 4.2 +/- 2.2 months before the first fall were compared with 22 patients never reporting falls. ROC analyses and logistic regressions accounting for DTC, UPDRS-III and disease duration were used for faller classification and prediction. Results: Only walking/box-checking predicted fallers. Fallers showed higher DTC for walking while box checking, p = 0.029, but not for box-checking while walking, p = 0.178 (combined motor DTC, p = 0.022), than non-fallers. Combined motor DTC classified fallers and non-fallers (area under curve: 0.75; 95% confidence interval, CI: 0.60-0.91) with 71.4% sensitivity (95%CI: 41.9%-91.6%) and 77.3% specificity (54.6%-92.2%), and significantly predicted future fallers (p = 0.023). Here, 20.4%-points higher combined motor DTC (i.e. the mean difference between falters and non-fallers) was associated with a 2.6 (1.1-6.0) times higher odds to be a future faller. Conclusion: Motor dual-tasking is a potentially valuable predictor of falls in PD, suggesting that avoiding dual task situations as well as specific motor dual-task training might help to prevent falls in PD. These findings and their therapeutic relevance need to be further validated in PD patients without fall history, in early PD stages, and with various motor-motor dual-task challenges. (C) 2016 Elsevier Ltd. All rights reserved.