Screening for risk of unplanned readmission in older patients admitted to hospital: predictive accuracy of three instruments

Background and aims: Hospital readmission after discharge is an important clinical and health policy issue. We compared the predictive accuracy of the Identification of Seniors at Risk (ISAR), the Flemish version of the Triage Risk Screening Tool (TRST) and Variable Indicative of Placement risk (VIP) in assessing unplanned readmissions. Methods: We included 213 patients (>= 65 years), hospitalized following admission to the emergency department. The ISAR, TRST and VIP were administered at admission. Unplanned readmissions were registered by telephone follow-up 14, 30 and 90 days post-discharge. Results: Unplanned readmission rates were 6.8%, 14.7% and 23.5% after 14, 30, and 90 days, respectively. The ISAR showed low to moderate sensitivity (54%-69%) and a high negative predictive value (>= 78%) at all measurement points. Specificity and positive predictive value were low (<= 33% and <= 24%, respectively). The TRST had low to moderate sensitivity (42%-67%) and a high negative predictive value (>= 82%). Specificity and positive predictive value were low (<= 45% and <= 27%, respectively). The VIP had very low sensitivity (<= 26%) and high specificity (>= 80%). Its negative predictive value was high (>= 79%) and its positive predictive value was low (<= 22%). Conclusions: Due to their moderate to low sensitivity, and low specificity and positive predictive value, none of the instruments was capable of accurately predicting unplanned readmission in older, hospitalized patients. Overall, reducing or increasing the original cut-off value by one point did not result in improved performance. Our findings suggest that these instruments lack the necessary sophistication to capture the complexity of (unplanned) readmissions. (Aging Clin Exp Res 2010; 22: 345-351) (C) 2010, Editrice Kurtis