Background: Frailty is associated with increased susceptibility to medication-related harm, highlighting the importance of medication review for frail older adults. Community pharmacists are increasingly involved in the initiation of medication reviews. Yet, current frailty measurement methods are impractical in this setting. Alternative approaches, leveraging routinely collected data, are needed.
Objective: To develop a frailty screening tool utilising routine electronic pharmacy records.
Methods: Community-dwelling older adults (≥ 70 years) using ≥ 5 chronic medications were recruited in 196 Belgian community pharmacies. Frailty was assessed using SHARE-FI75+ (based on Fried's frailty phenotype). Model development was on the basis of a two-stage approach using multivariable logistic regression with split-sample internal validation. Stage 1 considered only electronic pharmacy record variables, while stage 2 also included other variables that can easily be collected in the community pharmacy. Model performance was evaluated for discrimination, calibration and predictive accuracy.
Results: We recruited 875 participants [mean ± standard deviation (SD) age 79.3 ± 5.9 years], with 14.8% identified as frail. At stage 1, the frailty screening model included age, sex, reimbursement level of medical expenses, number of chronic medications and medication-derived comorbidities (anxiety, congestive heart failure, hypertension) [area under the receiver operating characteristic curve (AUC) 0.77, 95% confidence interval (CI) 0.69-0.85; sensitivity 78.0%; specificity 60.1%]. At stage 2, additional information on difficulties with basic activities of daily living or pharmacist's intuitive frailty assessment further improved the model (AUC 0.81, 95% CI 0.74-0.88 and AUC 0.82, 95% CI 0.75-0.89, respectively).
Conclusions: We developed a screening tool for frailty using data from electronic pharmacy records. This tool offers the opportunity for frailty screening in community pharmacy and to identify individuals that may benefit the most from medication review. External validation is warranted.
© 2024. The Author(s), under exclusive licence to Springer Nature Switzerland AG.