Background: In Sweden, self-reported maternal medication use collected during antenatal visits has been systematically stored, providing a unique opportunity to study both over-the-counter medications and circumvent the (classic) assumption that prescribed or dispensed drugs are consumed.
Objectives: To determine the temporal, geographic, and sociodemographic patterns of self-reported intake of harmful or potentially harmful medications during pregnancy in Sweden.
Methods: We used data from the Swedish Medical Birth Register (>98% coverage of all deliveries) to capture nearly all pregnancies in Sweden between 1995 and 2019 (N=2,479,665). At the first antenatal visit (typically at 8–12 weeks of gestation), gravidas report current medication use, with updates recorded at subsequent visits (median: 8 visits). This information, curated by the National Board of Health and Welfare, was combined with the Catalogue of Registered Pharmaceutical Specialties, which includes a clinically used risk classification system for pregnancy (covering 946 unique ATC codes in our data). This allowed us to estimate the prevalence of intake of medications classified as ‘clearly harmful’ (human evidence indicating teratogenicity or strong animal evidence) or ‘potentially harmful’ (suspected teratogens based on animal data).
Results: In 1995, 0.5% of gravidas reported using clearly harmful medications, while 2.1% reported using potentially harmful medications. By 2019, these proportions had increased to 1.5% and 21.8%, respectively. Across the entire study period, the three most reported clearly harmful medications were ondansetron, zopiclone, and azathioprine, while the most frequently reported potentially harmful medications were promethazine, sertraline, and acetylsalicylic acid. After adjusting for delivery year, there was a large variation across Sweden; the adjusted prevalence of clearly harmful medication use ranged from 0.4% (95% CI: 0.3–0.6) to 4.2% (95% CI: 3.9–4.4) across hospitals in different regions. Similar variation was seen for potentially harmful medications, though hospitals with high rates of clearly harmful medication use did not necessarily have high rates of potentially harmful medication use. Sociodemographic differences were apparent after adjusting for delivery hospital, maternal age, and delivery year. For example, gravidas with primary education reported higher use compared to those with university education, with an odds ratio of 1.25 (95% CI: 1.18–1.32) for clearly harmful and 1.14 (95% CI: 1.11–1.16) for potentially harmful medications.
Conclusions: The use of harmful and potentially harmful medications during pregnancy has increased in Sweden since the mid-1990s, with clear geographical and sociodemographic variation.
