Proactive rehabilitation screening (PReS) – development and validation of a modified PReS tool to screen admitted patients needing in-hospital rehabilitation programs

To analyse the predictive performance of Proactive Rehabilitation Screening (PReS) tool in identifying hospitalised patients who require formal inpatient multidisciplinary rehabilitation program. A prospective observational cohort study was conducted in two Australian public hospitals. The aims were to (1) assess the original PReS tool’s ability to predict rehabilitation needs and (2) develop and validate a modified PReS tool. Regression analysis was used to re-estimate predictor weights. In the development cohort (<i>n</i> = 1600), 272 patients (17%) received rehabilitation. Using multivariate regression analysis, we identified five key variables (age, number of allied health sessions received in 5 days, assistance required to mobilise, new personal care impairment and discharge barriers) to modify the original PReS tool. The modified PReS scores were stratified into likelihood of rehabilitation: low (0–4), medium (5–13) and high (14–20). Performance was improved for medium likelihood when a rehabilitation clinician was included in decision making. This was validated in an independent cohort of 800 patients demonstrating good predictive performance: sensitivity of 81.1%, specificity of 88.0%, total accuracy 86.9%. The modified PReS tool can reliably identify patients at risk of needing rehabilitation using five easily obtained clinical variables, enabling timely evidence-based rehabilitation interventions to minimise disability. Identifying patients who require inpatient rehabilitation programs early is important to minimise disability.A simple screening tool was developed with only five easily obtained clinical variables to predict who needs rehabilitation.In a large sample this screening tool showed good predictive performance and is quick to administer. Identifying patients who require inpatient rehabilitation programs early is important to minimise disability. A simple screening tool was developed with only five easily obtained clinical variables to predict who needs rehabilitation. In a large sample this screening tool showed good predictive performance and is quick to administer.