PGRx, LA-SER, Paris, France.
Pharmacoepidemiol Drug Saf. 2010 Jun;19(6):591-5. doi: 10.1002/pds.1952.
Patients' self-reported drug exposure is subjected to memory errors and different sources of bias. Utilization of prescription records is impaired with non-compliance and over-the-counter (OTC) drug utilization. This study compared patients' self-report (PS) to physician's prescriptions of cardiovascular drugs (CVDs).
The PGRx database is constituted by networks of specialized centers that recruited cases of 15 different diseases including myocardial infarction (MI) cases, and a network of general practitioners recruiting a pool of potential referents. For MI cases and referents, data on all drug utilization within the 2 years preceding the index date were obtained from PS and from physician's report of their prescriptions (PP). Patients' reports were obtained using a structured telephone interview complemented with an interview guide containing names of diseases and pictures of drug packages. Comparisons were made on exposure to each class of CVDs, for different time-windows, 2 months, 3-12 months and 13-24 months prior to the index date.
The concordance between physician and patient report was assessed on 2702 patient-physician pairs. Agreement was excellent overall (kappa = 0.83, 95% confidence interval (CI): 0.81-0.85). Prevalences of exposure were very close between PS and PP for all classes of prescription CVDs.
Using a standardized and systematic collection of information on drug exposure directly from patients appeared to provide similar information to using physician prescription records over a 2-year recall period.
患者自我报告的药物暴露情况容易受到记忆错误和各种来源的偏倚影响。处方记录的利用受到不遵医嘱和非处方（OTC）药物使用的影响。本研究比较了患者自我报告（PS）与医生开具的心血管药物（CVD）处方。
PGRx 数据库由专门中心的网络组成，这些中心招募了 15 种不同疾病的病例，包括心肌梗死（MI）病例，以及一个由普通医生招募潜在参照者组成的网络。对于 MI 病例和参照者，在索引日期前 2 年内从 PS 和医生报告的处方（PP）中获得所有药物使用的数据。使用结构化电话访谈获得患者报告，并辅以包含疾病名称和药物包装图片的访谈指南。比较了暴露于每种 CVD 类别的情况，时间窗口为 2 个月、3-12 个月和 13-24 个月。
在 2702 对医患对中评估了医生和患者报告之间的一致性。总体上一致性非常好（kappa=0.83，95%置信区间（CI）：0.81-0.85）。在 PS 和 PP 中，所有处方 CVD 类别的暴露率非常接近。
使用标准化和系统地直接从患者收集药物暴露信息的方法，在 2 年的回顾期内似乎提供了与使用医生处方记录相似的信息。
