BACKGROUND: Key pieces of information that are needed to fully understand studies on drug utilization and safety in pregnancy are sometimes omitted from publications—e.g., the decision to include multiple pregnancies in the study population, which can impact the prevalence of some outcomes.
OBJECTIVES: To identify key methodological data elements necessary for understanding observational pharmacoepidemiological research in pregnancy and quantify the proportion of studies that report these key data elements in a sample of published studies.
METHODS: Key methodological data elements were identified from the pregnancy pharmacoepidemiology draft guidelines from the Food and Drug Administration (2019) and European Medicines Agency (2020), relevant literature, and subject matter knowledge: pregnancy start (date, clinical entity, source of information); mother-infant, birth certificate, and other linkages (process, success rate); unit of analysis; and whether non-live births and fetuses with various anomalies were included in the study population. We searched PubMed for observational studies on drug utilization or safety during pregnancy published in 2015-2018. After screening of titles and abstracts, fulltext screening was conducted. For quality control, extracted data were confirmed against the publication. We estimated the prevalence of reporting of key data elements across studies.
RESULTS: The PubMed search retrieved 1,981 entries; data were extracted from a convenience sample of 50 eligible papers. Of these, 6% were published in epidemiology or pharmacoepidemiology journals; 16% were drug utilization studies, and 84% were safety studies. The mean study size was 109,060 subjects. Thirty-three percent of studies reported the method for determining pregnancy start; 59% reported whether the study population included multiple pregnancies; 44%, more than 1 pregnancy per woman; 60%, fetuses with major congenital malformations; 15%, fetuses with chromosomal abnormalities; and 85%, non-live births. Of the 5 studies that sought mother-infant linkage, 40% described the process, reported the linkage success rate, and specified which outcomes had been ascertained from maternal or infant files. Among the studies with more than one pregnancy/offspring per woman, 27% reported methods to address sibling correlation.
CONCLUSIONS: In this sample of pregnancy pharmacoepidemiology studies, completeness of methods reporting can be improved. A pregnancy-specific checklist would help to increase transparency in the dissemination of study results.
