Proposal for a New Predictive Scale for Recurrent Risk of Fall in a Cohort of Community-Dwelling Patients with Stroke

Abstract

Objectives: This study aimed to determine risk factors related to the occurrence of falls in stroke patients and to propose a new predictive scale for falls. Methods: Demographic and clinical data were collected and the following scales were applied: Barthel Index, Timed Up and Go Test (TUG), and National Institutes of Health Stroke Scale (NIHSS). Subjects were followed prospectively for 2 years for the occurrence of recurrent (≥2) falls. Kaplan–Meier curves were constructed and univariable associations were tested using log-rank test. Two separate multivariable models were then used: the first used Cox proportional hazards regression and the second used Poisson regression. In each model, significant associations were considered present with a P value less than .05. Results: We evaluated 150 individuals and the final analysis included 131 patients; the average age of the patients was 55.8 ± 13 years, 52% were women, and the median NIHSS score was 2 (interquartile range = 1-5). Falls occurred in 17% of patients, with a median of 23 months of follow-up (interquartile range = 16-26 months). In the multivariable Cox regression model, only TUG quartile, female gender, and posterior circulation territory involvement remained significant predictors of recurrent falls. We used the predictors from the Cox regression model to propose a new recurrent fall risk scale. The area under the receiver operating characteristic curve was 73%, 95% confidence interval = 62%-83%, P = .001, with 81.3% sensitivity and 41.8% specificity. Conclusions: The new predictive scale for recurrent risk (including TUG, posterior circulation territory involvement, and female gender) is presented as an instrument for monitoring the risk of recurrent falls.