Continuing a GLP-1 medicine into early pregnancy: a US study

Why this study matters

Most pregnancy studies of GLP-1 medicines compare people who used them with people who did not. This study asked a different and more practical question: for someone already using a GLP-1 medicine who becomes pregnant, what happens if they continue filling it into the first trimester compared with stopping? It used a method called target-trial emulation, which tries to make observational data mimic a randomized trial. It is one of the first studies of its kind.

Who was in the study

The study used US insurance claims data (Merative MarketScan) from 2011 to 2024. It included 3,572 pregnancies in people aged 16 to 55 who had a GLP-1 receptor agonist dispensed in the 90 days before their last menstrual period. Of these, 1,467 (41.1%) had type 2 diabetes. The summary we reviewed does not say why the others were using the medicine. Among 2,529 pregnancies that ended in a live birth, 1,499 could be linked to an infant's records, including 829 in the continuation group.

What the researchers did

This was an observational cohort study. Nobody was assigned to continue or stop; the researchers looked at what people actually did. They compared two strategies: continuation, defined as at least one further dispensing of a GLP-1 medicine during the first trimester, and noncontinuation.

To reduce bias they used a technique called clone-censor-weight target-trial emulation. In simple terms, each person's record was used to represent both strategies and then adjusted so that the two groups looked as similar as possible on measured factors. Risk of a nonlive birth was estimated with a weighted survival method.

The outcomes were nonlive birth (miscarriage, stillbirth or termination), abnormal fetal growth (babies small or large for gestational age), and major congenital malformations, meaning serious birth defects.

What they found

The weighted risk of a nonlive birth was 29.7% with continuation and 27.1% with noncontinuation (R32-C130). The adjusted risk ratio was 1.09, with a 95% confidence interval of 0.98 to 1.23. A ratio of 1.0 means no difference. A confidence interval is the range in which the true value most likely sits. This range just includes 1.0, so a small increase is possible but not established.

Among the 2,529 live-birth pregnancies, 1,443 (57.1%) had at least one GLP-1 dispensing after the last menstrual period.

The authors conclude that risks were "not definitively higher" with continuation, but they stress that the estimates for birth defects and small babies were imprecise. Those estimates are compatible with no increased risk and also with clinically important increases in risk (R32-C130).

Side effects and people who stopped

This study measured pregnancy and infant outcomes, not side effects in the pregnant person. "Stopping" here was one of the two strategies being compared rather than an adverse event, and the extracted data do not report discontinuation in the usual sense.

What this study does not tell you

• The authors name residual confounding by prior blood sugar control as a limitation. People who continued may have had different diabetes control from those who stopped, and claims data cannot fully capture that.
• Exposure was based on prescription dispensing, not on confirmed use.
• The confidence interval for major birth defects runs from 0.83 to 1.82, and for small babies from 0.82 to 2.06. Neither a real increase nor no effect can be ruled out.
• The data are from US insured people between 2011 and 2024. Which GLP-1 medicines were used, and at what doses, is not described in the summary we reviewed.
• The study cannot say anything about exposure later in pregnancy.
• This is a single observational study and does not settle whether continuing or stopping is safer.

What it means in Canada

This study used US data and does not tell anyone in Canada what to do about their own medicine if they become pregnant. It shows how uncertain the evidence still is. For information about specific medicines, see semaglutide in Canada and tirzepatide in Canada. For what provincial plans pay for, see public drug coverage by province.

Related reading

Other studies in this library that bear on the same question.

Source