Schirm Eric, Monster Taco B M, de Vries Robin, van den Berg Paul B, de Jong-van den Berg Lolkje T W, Tobi Hilde
Department of Social Pharmacy, Pharmacoepidemiology and Pharmacotherapy, Groningen University Institute for Drug Exploration, University of Groningen, Groningen, The Netherlands.
Pharmacoepidemiol Drug Saf. 2004 Mar;13(3):173-9. doi: 10.1002/pds.882.
Pharmacy-dispensing data are valuable sources of drug information, but the population that is covered by the pharmacies is often difficult to determine. We evaluated two methods using drug utilisation information to estimate the population size: a drug-use-based extrapolation of a known part of the population and a capture-recapture estimation without any prior knowledge of the population.
Using pharmacy-dispensing data of three towns with known populations in the Netherlands, we estimated age-and-sex specific population sizes by extrapolating the proportion of drug-using inhabitants. In addition, we applied two-source and three-source capture-recapture models with all combinations of the following drug groups as different sources: anti-asthmatics, analgesics, antibiotics and anti-histamines.
Drug-use-based extrapolation resulted in the best estimates with the least variability. All capture-recapture models provided underestimations of the true population. Three-source capture-recapture resulted in better average estimates than two-source capture-recapture, but also had more variability.
If a part of the population is known, and if there is reason to assume that drug utilisation patterns do not vary within the region, it is best to use drug-use-based extrapolation. In all other situations capture-recapture may be considered, with as main limitation that we found all models to underestimate the population considerably.
药房配药数据是药物信息的宝贵来源，但药房所覆盖的人群往往难以确定。我们评估了两种利用药物使用信息来估计人口规模的方法：一种是基于药物使用情况对已知部分人群进行外推，另一种是在对人群没有任何先验知识的情况下进行捕获再捕获估计。
利用荷兰三个已知人口城镇的药房配药数据，通过外推使用药物居民的比例来估计特定年龄和性别的人口规模。此外，我们应用了两源和三源捕获再捕获模型，将以下药物组的所有组合作为不同来源：抗哮喘药、镇痛药、抗生素和抗组胺药。
基于药物使用情况的外推得出的估计值最佳，变异性最小。所有捕获再捕获模型都低估了真实人口。三源捕获再捕获得出的平均估计值比两源捕获再捕获更好，但变异性也更大。
如果已知部分人群，并且有理由假设该地区内药物使用模式不变，那么最好使用基于药物使用情况的外推法。在所有其他情况下，可以考虑使用捕获再捕获法，主要限制是我们发现所有模型都大大低估了人口数量。
