Prenatal Medication Profiles and Childhood Asthma in POMME

Key Takeaways

• In the French POMME birth cohort, some prenatal medication-dispensing profiles had above-average childhood asthma occurrence.
• Higher-occurrence profiles featured repeated dispensings throughout pregnancy or combinations of medications supplied late in pregnancy.
• Chronic maternal conditions and pregnancy-related complications were more frequent in the higher-occurrence profiles.

How do combinations of medications supplied during pregnancy, and the timing of those dispensings, relate to asthma identified later in childhood? Maternal dispensing records capture supplied medications rather than confirmed use, while childhood asthma identified through medication records reflects an algorithm-defined outcome. Prenatal dispensing patterns and later asthma ascertainment were examined together in a birth cohort.

Investigators used the population-based, observational PrescriptiOn Médicaments Mères Enfants (POMME) birth cohort, comprising 18,851 children born in Haute-Garonne, France, in 2010 or 2015 and followed until 2023. Maternal dispensing histories and children's medication records supplied the exposure and outcome data.

Maternal medication dispensings were classified by Anatomical Therapeutic Chemical (ATC) subclass and assessed separately for each trimester as none, one, or at least two dispensings. Childhood asthma from age 4 onward was identified using an algorithm based on dispensings of asthma medications. Bayesian profile regression simultaneously grouped children by prenatal dispensing patterns and asthma occurrence. Higher-occurrence groups were then characterized by medication classes, dispensing timing, and maternal and infant characteristics.

Investigators identified 17 prenatal medication-dispensing profiles; six had childhood asthma occurrence above the overall cohort occurrence of 15.2%. Higher-occurrence profiles featured repeated dispensings throughout pregnancy of medications for acid-related disorders, thyroid hormones, antibiotics, asthma medications, and antihistamines. Other patterns involved combinations of antibiotics, nasal medications, and cold medicines late in pregnancy. Mothers in higher-occurrence profiles more frequently had chronic conditions and pregnancy-related complications. The findings concern dispensing profiles, not independently estimated associations for each medication class.

A dispensed medication is not necessarily consumed, and asthma inferred from medication dispensings is not a confirmed clinical diagnosis. Maternal illnesses and pregnancy-related complications were also more frequent in the higher-occurrence groups, complicating attribution of the associations to medications themselves. These exploratory findings do not isolate individual-drug effects or establish that changes to indicated treatment during pregnancy would alter childhood asthma occurrence.

The investigators concluded that combinations of prenatal medication dispensings and possible exposure windows merit further investigation. They also suggested that Bayesian profile regression may aid signal detection in perinatal pharmacoepidemiology.