A simple tool predicted probability of falling after aged care inpatient rehabilitation

Objective: To develop and internally validate a falls prediction tool for people being discharged from inpatient aged care rehabilitation. Study Design and Setting: Prospective cohort study. Possible predictors of falls were collected for 442 aged care rehabilitation inpatients at two hospitals. Results: One hundred fifty participants fell in the 3 months after discharge from rehabilitation (34% of 438 with follow-up data). Predictors of falls were male gender (odds ratio [OR] 2.32, 95% confidence interval [CI] = 1.00-4.03), central nervous system medication prescription (OR 2.04, 95% CI = 1.00-3.30), and increased postural sway (OR 1.93, 95% CI = 1.00-3.26). This three-variable model was adapted for clinical use by unit weighting (i.e., a score of 1 for each predictor present). The area under the receiver operating characteristic curve (AUC) for this tool was 0.69 (95% CI = 0.64-0.74, bootstrap-corrected AUC = 0.69). There was no evidence of lack of fit between prediction and observation (Hosmer-Lemeshow P = 0.158). Conclusion: After external validation, this simple tool could be used to quantify the probability with which an individual will fall in the 3 months after an aged care rehabilitation stay. It may assist in the discharge process by identifying high-risk individuals who may benefit from ongoing assistance or intervention. (C) 2011 Elsevier Inc. All rights reserved.