Combining Data from Healthcare Databases in Europe: A Study of Dopamine Agonists

Background: With rare exposures and outcomes, a single automated database may not provide sufficient subjects and heterogeneity of exposure needed for comparative pharmacoepidemiologic drug safety research. Objectives: Data from 4 general practitioner (GP) databases and record linkage systems in 3 European countries were combined to assess the association between individual dopamine agonists (DA) and cardiopulmonary disorders. The objective of this abstract is to demonstrate the value of combining data from multiple databases to address potential safety issues. Methods: Four cohorts of 1) new users of DAs for Parkinson’s disease (PD) or hyperprolactinemia, 2) levodopa, 3) DA-naive persons with hyperprolactinemia, and 4) database, gender and age-matched controls were identified from THIN-UK, Italian HSD-THALES, Dutch-IPCI GP databases, and the Dutch-PHARMO record linkage system between 1996–2007. Outcomes were identified by automated and medical record abstraction after mapping of codes, and validated by an expert panel blinded to exposure. After database-specific extractions, a common output dataset was created. Results: During the study period, cohorts identified were 1) DA users for PD or hyperprolactinemia (>35,000), 2) levodopa (18,000), 3) DA-naive persons with hyperprolactinemia (16,000) and 4) controls (>35,000). Overall, cabergoline was the most frequently used DA (42%), with highest use in Italy (48%) and lowest use in UK (16%). Non-ergot DAs had the highest use in UK (35%) and NL (25%) and lowest use in Italy (18%). Apomorphine was the most frequently used non-ergot DA in UK (25%). Conclusions: Combining data from multiple databases increases study size, generalizability, and heterogeneity of exposure, which can facilitate drug comparisons.