AIMS: To study and compare both mean performance measures as well as intra-individual variability measures of stride length and reaction time in vulnerable recurrent and non-recurrent older fallers. METHODS: Stride length during walking and walking while dual-tasking (GAITRite®) and choice reaction time (CANTAB®) were assessed in geriatric outpatients and their informal caregivers (n=60, ≥ 60 yrs). Logistic regression and Receiver Operating Characteristic (ROC) analysis were used to generate models with mean performance measures and intra-individual variability measures (coefficients of variation; CV=[sd/mean]x100)), as risk factors for recurrent falls. RESULTS: Reaction-time CV was higher in recurrent fallers than in non-recurrent fallers: 21.3% [9.3-47.7] vs 15.8% [8.3-34.9] (p=0.04). Also, stride-length CV was higher in recurrent fallers during performance of the verbal fluency dual-task: 4.5% [1.2-31.4] vs 3.5% [0.9-9.7] (p=0.017). The model with CVs provided an explained variance of 23.7%, and an area under the curve (AUC) of 0.73, which was higher than that of the model including mean performance measures (8.6% and 0.65 respectively). CONCLUSIONS: Older recurrent fallers are characterized by increased within-task variability in reaction time and stride length while dual-tasking. In addition, variability in performance is a more sensitive measure in discrimination of recurrent falls than mean performance itself, suggesting deterioration in neurocognitive regulation mechanisms as part of the causal pathway for recurrent falls.
AIMS: To study and compare both mean performance measures as well as intra-individual variability measures of stride length and reaction time in vulnerable recurrent and non-recurrent older fallers. METHODS: Stride length during walking and walking while dual-tasking (GAITRite®) and choice reaction time (CANTAB®) were assessed in geriatric outpatients and their informal caregivers (n=60, ≥ 60 yrs). Logistic regression and Receiver Operating Characteristic (ROC) analysis were used to generate models with mean performance measures and intra-individual variability measures (coefficients of variation; CV=[sd/mean]x100)), as risk factors for recurrent falls. RESULTS: Reaction-time CV was higher in recurrent fallers than in non-recurrent fallers: 21.3% [9.3-47.7] vs 15.8% [8.3-34.9] (p=0.04). Also, stride-length CV was higher in recurrent fallers during performance of the verbal fluency dual-task: 4.5% [1.2-31.4] vs 3.5% [0.9-9.7] (p=0.017). The model with CVs provided an explained variance of 23.7%, and an area under the curve (AUC) of 0.73, which was higher than that of the model including mean performance measures (8.6% and 0.65 respectively). CONCLUSIONS: Older recurrent fallers are characterized by increased within-task variability in reaction time and stride length while dual-tasking. In addition, variability in performance is a more sensitive measure in discrimination of recurrent falls than mean performance itself, suggesting deterioration in neurocognitive regulation mechanisms as part of the causal pathway for recurrent falls.