Scholarlink

Background <p>Frailty increases the risk of medication-related harm in older adults, emphasising the importance of early detection of drug-related problems during medication reviews. A new frailty screening tool, using electronic pharmacy data on age, sex, preferential reimbursement of medical expenses, number of chronic medications and medication use, offers a practical solution for community pharmacists to identify this risk group.</p> Objective <p>In this study, we aimed to externally validate this tool.</p> Methods <p>For this external validation, data of community-dwelling participants with available frailty status from the HISLink project were used. This database contains data from the national Health Interview Survey 2018, a cross-sectional population survey conducted by Sciensano, linked to health insurance data (i.e. claims data from the Belgian Compulsory Health Insurance, which includes records on reimbursed healthcare utilisation and medication, and level of reimbursement of medical expenses). The Health Interview Survey 2018 database also contains frailty status assessed using SHARE-FI, which was used to evaluate the frailty screening tool’s model performance and predictive accuracy. Logistic recalibration was applied to account for differences in baseline risk and predictor effects, between the development and validation cohorts. A subgroup sensitivity analysis was conducted.</p> Results <p>Of the 2191 (mean age 74.4 ± 7.2 years, 53.5% female) participants, 20.7% were frail according to SHARE-FI. The screening tool detected 31.6% as frail. Discrimination was good with an area under the receiver operating characteristic curve of 0.77 (95% confidence interval 0.74–0.79), sensitivity of 63.9% and specificity of 76.8% (positive predictive value 41.8% and negative predictive value 89.1%). Calibration was initially modest (Brier score of 0.17) but improved substantially after logistic recalibration (Brier score of 0.13). In the subgroup of individuals with polypharmacy, an area under the receiver operating characteristic curve of 0.72 (95% confidence interval 0.68–0.76) was reported with sensitivity of 76.6% and specificity of 55.6% (positive predictive value 49.3%; negative predictive value 80.8%).</p> Conclusions <p>The frailty screening tool provides a valid and reliable method for identifying older adults at risk for being frail in the community pharmacy. Its previously observed good performance is confirmed through external validation using data from the HISLink project.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

External Validation of a Community Pharmacy-Based Screening Tool for Frailty That Uses Routinely Collected Data

  • 期刊论文

  • 发表日期：

    2026年6月15日

查看全文
  • Marie Carrein,
  • Johan Van der Heyden,
  • Els Mehuys,
  • Mirko Petrovic,
  • Koen Boussery

摘要

Background

Frailty increases the risk of medication-related harm in older adults, emphasising the importance of early detection of drug-related problems during medication reviews. A new frailty screening tool, using electronic pharmacy data on age, sex, preferential reimbursement of medical expenses, number of chronic medications and medication use, offers a practical solution for community pharmacists to identify this risk group.

Objective

In this study, we aimed to externally validate this tool.

Methods

For this external validation, data of community-dwelling participants with available frailty status from the HISLink project were used. This database contains data from the national Health Interview Survey 2018, a cross-sectional population survey conducted by Sciensano, linked to health insurance data (i.e. claims data from the Belgian Compulsory Health Insurance, which includes records on reimbursed healthcare utilisation and medication, and level of reimbursement of medical expenses). The Health Interview Survey 2018 database also contains frailty status assessed using SHARE-FI, which was used to evaluate the frailty screening tool’s model performance and predictive accuracy. Logistic recalibration was applied to account for differences in baseline risk and predictor effects, between the development and validation cohorts. A subgroup sensitivity analysis was conducted.

Results

Of the 2191 (mean age 74.4 ± 7.2 years, 53.5% female) participants, 20.7% were frail according to SHARE-FI. The screening tool detected 31.6% as frail. Discrimination was good with an area under the receiver operating characteristic curve of 0.77 (95% confidence interval 0.74–0.79), sensitivity of 63.9% and specificity of 76.8% (positive predictive value 41.8% and negative predictive value 89.1%). Calibration was initially modest (Brier score of 0.17) but improved substantially after logistic recalibration (Brier score of 0.13). In the subgroup of individuals with polypharmacy, an area under the receiver operating characteristic curve of 0.72 (95% confidence interval 0.68–0.76) was reported with sensitivity of 76.6% and specificity of 55.6% (positive predictive value 49.3%; negative predictive value 80.8%).

Conclusions

The frailty screening tool provides a valid and reliable method for identifying older adults at risk for being frail in the community pharmacy. Its previously observed good performance is confirmed through external validation using data from the HISLink project.