Metformin is increasingly used during pregnancy, but what happens years later to children exposed before birth?
Study: Metformin exposure in pregnancy and childhood developmental outcomes. Image Credit: Svitlana Hulko / Shutterstock.com
In a recent study published in JAMA Network Open, researchers examined whether metformin exposure during pregnancy was associated with developmental vulnerabilities in children during the first year of full-time school.
Why prenatal metformin exposure is under scrutiny
Metformin crosses the placenta, exposing the fetus to concentrations similar to those in maternal blood. While perinatal safety findings are reassuring, long-term outcomes remain unclear.
An animal study suggested that in utero metformin exposure could affect cortical development, while human studies have not shown an association with neurodevelopmental delay in younger children. Research has also raised questions about childhood adiposity after prenatal metformin exposure, but findings remain conflicting due to small samples, varied populations, and different reasons for metformin treatment.
Larger studies of school-aged children are needed, as developmental differences may become more apparent after preschool.
Tracking development from pregnancy to school entry
Researchers conducted a cohort study of singleton births in Victoria, Australia, from January 1, 2009, through December 31, 2020. Researchers linked pregnancy, birth, medication, and developmental data from the Victorian Perinatal Data Collection (VPDC), National Diabetes Service Scheme (NDSS), Pharmaceutical Benefits Scheme (PBS), Australian Early Development Census (AEDC), and Victorian Admitted Episodes Dataset. The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.
Metformin exposure was defined as dispensing at least 1 prescription between the last normal menstrual period and birth.
AEDC is a teacher-completed assessment administered during the first year of school at ages 4 to 6 years. Developmental vulnerability was defined as scoring below the 10th percentile in at least 2 of 5 domains: physical health, social competence, emotional maturity, language and cognitive skills, and communication and general knowledge.
Children with congenital anomalies and pregnancies with unclear or inconsistent diabetes or insulin records were excluded.
Associations were estimated using inverse probability-weighted regression adjustment (IPWRA), which accounted for differences between exposed and unexposed pregnancies, including maternal age, body mass index (BMI), education, Socio-Economic Indexes for Areas (SEIFA), diabetes, and assisted reproductive technology (ART), along with factors related to pregnancy, birth, and developmental assessment.
Metformin exposure shows no developmental disadvantage
Among 871,627 singleton births, 177,409 children had linked developmental outcome data. Of these, 1,095 children (0.6%) had antenatal metformin exposure. Most exposed children were assessed between 5 years 1 month and 6 years of age.
Of the exposed pregnancies, 64.2% first received a metformin prescription in the first trimester, 12.9% in the second trimester, and 22.9% in the third trimester.
As compared to women who were not exposed to metformin, those exposed had a higher mean BMI, were more likely to have conceived using ART, and more commonly had polycystic ovary syndrome, type 2 diabetes, and gestational diabetes. Their infants were more likely to be born preterm and by cesarean delivery, with a larger proportion having birth weights above the 97th percentile. Birth weights below the 10th percentile were similar between groups.
Developmental vulnerability was identified in 18.2% of exposed children compared with 13.9% of children without exposure. However, this difference disappeared after adjusting for confounding factors. Adjusted analyses similarly found no association between metformin exposure and any of the five individual AEDC domains.
Sensitivity analyses produced similar findings. Among women with gestational diabetes or type 2 diabetes, developmental vulnerability occurred in 20.7% of exposed children and 15.9% of unexposed children. Ultimately, this difference was no longer evident after adjustment, with no increased risk observed in any of the five developmental domains.
When children with special educational needs and incomplete outcome data were excluded, 11.6% of exposed children and 9.2% of unexposed children were developmentally vulnerable, with no association after adjustment. Analyses restricted to first-trimester exposure and those using alternative statistical models also produced similar findings.
Because the study was observational, it could not rule out residual confounding. Another limitation was the inability to assess whether women actually took the dispensed medication, as well as the lack of information on metformin dose and the specific indication for treatment. Analyses of multiple prescriptions and treatment beginning in the second or third trimester also lacked sufficient statistical power for fully adjusted models.
Findings offer reassurance, but long-term questions remain
In this cohort of children assessed during the first year of school, antenatal metformin exposure was not associated with developmental vulnerability after adjustment. These findings were consistent across all five individual AEDC domains.
Similar results were observed when analyses were restricted to pregnancies complicated by gestational diabetes or type 2 diabetes, when imputed outcomes and children with special educational needs were excluded, and when first-trimester exposure was examined. Clinicians and women considering metformin during pregnancy may find the results reassuring, though longer-term research is still needed on children’s growth and cardiometabolic health.
Intentional substance exposures at school rise sharply among younger children