RESUMO
PURPOSE: Routine health care data are a valuable source for the assessment of risks of drugs during pregnancy. Therefore, the aim of the current analysis was the evaluation of pregnancy outcome records in German health insurance data. METHODS: We used the German Pharmacoepidemiologic Research Database including data of more than 15 million insurants (about 18% of the German population). Among 10-49 year-old women, we identified six categories of pregnancy outcomes: term births, preterm births, stillbirths, induced abortions, spontaneous abortions and ectopic pregnancies. In order to assess the validity of these records, we have set our results in relation to representative data for the corresponding outcomes in Germany. We also investigated whether pregnancy markers (diagnoses, procedures or medical services which indicate an existing pregnancy) can be used to identify pregnancies. RESULTS: In total, we identified 94 261 pregnancy outcomes in 2005. The percentage of births outside hospital (1.2%) and of preterm births (11.6%), the rate of stillbirths (3 per 1000 live births) and the rate of ectopic pregnancies (20 per 1000 live births) agreed well with representative data for Germany. Compared to epidemiological data, the occurrence of spontaneous abortions was underestimated (5.4% of all pregnancies). There were 4.1 induced abortions per 100 live births, compared to 18.1 in national data. Positive predictive values and sensitivities of markers varied across marker categories and for different pregnancy outcomes. CONCLUSIONS: Completeness of pregnancy outcomes recorded in the database varied by pregnancy outcome. This should be taken into account in studies of drug safety in pregnancy.
