Risk factors associated with falls in adult patients after stroke living in the community: baseline data from a stroke cohort in Brazil.

Background: Individuals with stroke have a high risk of falling, and their fall predictors may differ from those of other populations.
Purpose: To estimate fall frequency and identify factors related to fall occurrence in a sample of patients with stroke residing in the community.
Methods: Clinical data were collected from 150 consecutive stroke patients with independent gait, and the following scales were applied: modified Barthel Index (mBI), Timed Up & Go Test (TUG), and National Institutes of Health Stroke Scale (NIHSS). Univariate analysis was performed; variables with possible association (P < .1) were included in a logistic regression model. Receiver operating characteristic curves were used to identify the best cutoff point for TUG.
Results: Falls occurred in 37% of patients. In multivariate analysis, right hemisphere injury (odds ratio [OR], 2.621; 95% CI, 1.196-5.740; P = .016), time in TUG (OR, 1.035 for every increase in 1 second; 95% CI, 1.003-1.069; P = .034), and longer time since stroke onset (OR, 1.012 for every month increase; 95% CI, 1.002-1.021; P = .015) remained predictors. When we grouped individuals according to affected cerebral hemisphere, both hemispheres had similar accuracy, but TUG cutoff point was lower in individuals with right- versus left-hemisphere lesions.
Conclusions: Patients with poor TUG performance, longer times since stroke onset, and right-hemisphere injury have particularly high fall rates, and TUG cutoff points for fall prediction vary according to cerebral hemisphere.