Prediction of falls in rehabilitation and acute care geriatric setting

AbstractPurposeTo examine the feasibility to develop an efficient and applicable geriatric hospital-specific fall risk-prediction tool.Patients and methodsThis is a prospective cohort study in five rehabilitation units and one acute care geriatric unit at a geriatric hospital. In total, 1013 patients aged over 65 years were admitted during a 6-month period. Fourteen patient characteristics found in previous studies to be risk factors for falls were tested for predictive validity. The characteristics included: age, gender, history of falls, dementia, delirium, use of psychoactive drugs, using a wheelchair, acute illness, state of post joint arthroplasty, Parkinson's disease, stroke, heart disease, vision disturbance, and hospitalization for deconditioning. Each of the characteristics was assessed individually for their predictive power based on logistic regression models.ResultsDementia (p < 0.001) and delirium (p = 0.005) predicted falls in patients hospitalized for rehabilitation. In the multivariate model, only dementia was a significant predictor in these patients, (p = 0.014), while delirium only approached significance. Being hospitalized for rehabilitation after arthroplasty was a negative predictive factor of falls (p = 0.022). Among acute care patients, only being operated on in the past for joint arthroplasty (p = 0.035) predicted falls in the multivariate model, while using a wheelchair was a negative predictive factor (p = 0.023).ConclusionThe current study reveals a poor predictive value for falls for most patient characteristics (except delirium and dementia) in elderly hospitalized patients, illustrating the incertitude of developing and using predictive falls tools based on such characteristics in hospitalized elderly patients