Development of a screening tool to flag older adults at a high risk of preventable medication‐related readmissions

Abstract Background Preventable medication‐related readmissions (pMRRs) are a significant contributor to older adult morbidity, mortality, and health care costs. There are currently no specific screening tools for identifying patients at a high risk of pMRRs. This study aimed to develop and internally validate a screening tool to identify general internal medicine patients aged 65 or older at risk of 30‐day pMRRs. Methods Data on 500 patients discharged between January 1, 2022 and October 18, 2023, and readmitted within 30 days were reviewed, and 116 pMRRs were identified. An additional 500 control patients without pMRRs were selected. A multivariable logistic regression model was developed using the four pre‐specified predictors of each patient's number of medications, hospitalizations in the past year, high‐risk medications at discharge (anticoagulants, opioids, and sedatives), and the number of the following diagnoses: heart failure, diabetes, chronic kidney disease, and chronic obstructive pulmonary disease. Results The screening tool's performance was evaluated and internally validated using bootstrapping. The screening tool demonstrated moderate discriminatory power, with a C‐statistic of 0.714 (95% confidence interval [CI]: 0.663–0.764). Overall performance was assessed as acceptable, with a Brier score of 0.177. Bootstrapping yielded an optimism‐corrected C‐statistic of 0.704 (95% CI: 0.654–0.755), indicating the tool's robustness. Conclusions The screening tool's discriminatory power suggested its potential utility for flagging patients at risk of pMRRs. External validation and a search for additional predictors will be necessary to enhance the tool's performance and generalizability. If validated, a cost‐effectiveness analysis would be needed to ensure its feasibility for clinical implementation.