Scholarlink

Background <p>Medication adherence includes initiation, implementation, and persistence. Population-level estimates are commonly based on administrative claims, yet these may not fully capture patient behavior. This study developed a framework to estimate adherence to antihypertensive treatment using linked self-reported data and medication refill records from a representative Belgian sample.</p> Methods <p>We linked participants of the 2018 Belgian Health Interview Survey (BHIS) with Belgian Compulsory Health Insurance (BCHI) records via pseudonymized identifiers. Chronic hypertension patients were defined as those with at least one antihypertensive prescription in the year prior to the study and ≥ 30 consecutive days of treatment during the study year. Adherence was assessed using three approaches: (1) regimen-level Proportion of Days Covered in 2018 (PDCY) capturing implementation and persistence; (2) Proportion of Days Covered in a flexible period (PDCF) capturing implementation during persistence; and (3) BHIS-based self-reported medication use at the interview date cross-validated with prescriptions. We also calculated Medication Possession Ratios (MPR) for drug class–specific adherence. Agreement between methods was tested with Cohen’s Kappa and McNemar’s test; predictors of adherence were analyzed using logistic regression.</p> Results <p>Of 2,475 chronic hypertension patients (22.6% of BHIS sample; mean age 67.1 years, 52.8% female), adherence rates were similar across indicators (PDCY: 58.3%; PDCF: 57.1%; BHIS: 61.4%). Kappa values indicated fair, consistent agreement, while McNemar’s test showed no systematic differences between methods. Logistic models identified age, multimorbidity, and polypharmacy as significant predictors of adherence, with effect sizes varying by measurement approach.</p> Conclusions <p>Findings suggest that short-term adherence measured via self-report provides a reasonable proxy for long-term adherence. Linking survey and administrative data offers a robust and practical alternative to claims-only approaches.</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

Estimating hypertension medication adherence among adults using linked Belgian health interview survey and administrative pharmacy refill data

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年2月18日

查看全文
  • Arno Maetens,
  • Rana Charafeddine,
  • Stefaan Demarest,
  • Bernard Vrijens,
  • Johan Van der Heyden

摘要

Background

Medication adherence includes initiation, implementation, and persistence. Population-level estimates are commonly based on administrative claims, yet these may not fully capture patient behavior. This study developed a framework to estimate adherence to antihypertensive treatment using linked self-reported data and medication refill records from a representative Belgian sample.

Methods

We linked participants of the 2018 Belgian Health Interview Survey (BHIS) with Belgian Compulsory Health Insurance (BCHI) records via pseudonymized identifiers. Chronic hypertension patients were defined as those with at least one antihypertensive prescription in the year prior to the study and ≥ 30 consecutive days of treatment during the study year. Adherence was assessed using three approaches: (1) regimen-level Proportion of Days Covered in 2018 (PDCY) capturing implementation and persistence; (2) Proportion of Days Covered in a flexible period (PDCF) capturing implementation during persistence; and (3) BHIS-based self-reported medication use at the interview date cross-validated with prescriptions. We also calculated Medication Possession Ratios (MPR) for drug class–specific adherence. Agreement between methods was tested with Cohen’s Kappa and McNemar’s test; predictors of adherence were analyzed using logistic regression.

Results

Of 2,475 chronic hypertension patients (22.6% of BHIS sample; mean age 67.1 years, 52.8% female), adherence rates were similar across indicators (PDCY: 58.3%; PDCF: 57.1%; BHIS: 61.4%). Kappa values indicated fair, consistent agreement, while McNemar’s test showed no systematic differences between methods. Logistic models identified age, multimorbidity, and polypharmacy as significant predictors of adherence, with effect sizes varying by measurement approach.

Conclusions

Findings suggest that short-term adherence measured via self-report provides a reasonable proxy for long-term adherence. Linking survey and administrative data offers a robust and practical alternative to claims-only approaches.