Good agreement between physician and self-reported hormone therapy data in a case–control study

ObjectiveIn a population-based case–control study examining the effects of postmenopausal hormone therapy (HT) on breast cancer risk, the authors conducted a validation study comparing prescription data from gynecologists with self-reports.Study Design and SettingThe study was conducted in the Rhein-Neckar and Hamburg regions of Germany from 2002 to 2005. A total of 224 cases and 225 controls, stratified by region, age, and hormone use were randomly selected for the validation study.ResultsFor ever/never use 88.2% agreement was seen, and agreement for ever/never use by type of HT was 80.6%, 80.3%, and 90.5% for mono-estrogen, cyclical combined, and continuous combined therapy, respectively. The intraclass correlation coefficient (ICC) for duration of use was high, 0.82 (95% confidence interval [CI]: 0.77, 0.85), as were the ICCs for age at first and last use, 0.88 (95% CI: 0.85, 0.91) and 0.98 (95% CI: 0.97, 0.98). Despite the exceptionally high number of different HT prescriptions available in Germany, comparison of exact brand name resulted in perfect agreement for 50.2% of participants, partial agreement for 29.3%, and no agreement for 20.7%. In general, agreement was not differential by disease status.ConclusionOverall, the self-reported HT of the study participants corresponded well with physicians' reports.