On June 8, 2026, Annals of Internal Medicine published one of the most directly useful pregnancy-safety studies the GLP-1 field has produced to date.[1] Researchers at the Harvard T.H. Chan School of Public Health used insurance-claims data to ask a question that matters to a growing number of women: if you were taking a GLP-1 medication like semaglutide, tirzepatide, or liraglutide and became pregnant, does continuing it into the first trimester appear to harm the pregnancy?
The short answer from this analysis is reassuring — but the honest answer comes with real caveats, and it does not mean GLP-1 drugs are recommended in pregnancy. Here is what the study found, and what it carefully does not claim.
- What it studied: 3,572 pregnancies (2011–2024); women who continued a GLP-1 into the first trimester vs. those who stopped at conception
- Main finding: risks of nonlive birth, abnormal fetal growth, small-for-gestational-age, and major malformation were similar between groups — no definitive increase
- The catch: estimates were imprecise for the two rarest outcomes (major malformation and small-for-gestational-age) — this is reassurance, not a clean "all clear"
- What it does NOT say: it is not an endorsement of using GLP-1s in pregnancy. Label guidance still says to stop these drugs in pregnancy
Quick facts
- Journal
- Annals of Internal Medicine
- Published
- June 8, 2026
- Institution
- Harvard T.H. Chan School of Public Health
- Design
- Target-trial emulation (claims data)
- Pregnancies
- 3,572 (2011–2024)
- With type 2 diabetes
- 41.1% (n=1,467)
- Comparison
- Continued vs. stopped at conception
Why this study was designed the way it was
GLP-1 receptor agonists are not tested in pregnant women. Pregnant people are excluded from the trials that bring these drugs to market, so the labels for semaglutide, tirzepatide, and liraglutide all advise discontinuing them in pregnancy — a precaution, not a finding of harm. That leaves a real-world gap: with so many women of reproductive age now taking GLP-1s, accidental first-trimester exposure happens, and clinicians have had little hard data to counsel from.
You cannot ethically run a randomized trial that assigns pregnant women to keep taking a drug their label tells them to stop. So the Harvard team used a method called target-trial emulation — a way of analyzing existing data to mimic the trial you would have run if you could.[1] Using U.S. insurance-claims records, they identified pregnancies in women who had been on a GLP-1, then compared two groups:
- Women who continued the drug (at least one further dispensing after conception)
- Women who stopped it at conception
Of the 3,572 pregnancies analyzed (2011–2024), 41.1% (1,467) were in women with type 2 diabetes — an important detail we will return to.[1]
What the study found
Across the outcomes that worry families most, the two groups looked alike.[1][2]
For nonlive birth (the study's most clearly measured outcome), the risk was 29.7% among women who continued the GLP-1 versus 27.1% among those who stopped — an adjusted risk ratio of 1.09 (95% CI 0.98 to 1.23). The confidence interval crosses 1.0, meaning the data are consistent with no real difference.
For abnormal fetal growth, small-for-gestational-age, and major congenital malformation, the researchers similarly found no definitive increase in risk with continued use.
Read the conclusion carefully. The authors' framing is precise: as GLP-1 use among women of reproductive age becomes more common, these findings "can provide some reassurance for pregnant women with unintentional first-trimester exposure." That is a statement about accidental exposure — not a recommendation to keep taking the drug on purpose.[2]
The honest caveat: imprecision on the rarest outcomes
This is the part that gets lost in headlines, and it is the part that matters most for an honest read.
Major congenital malformations and small-for-gestational-age births are, thankfully, rare. Rare outcomes are hard to study, because even a few thousand pregnancies may include only a handful of cases. The authors say so directly: their estimates were imprecise, particularly for the rarer outcomes of major congenital malformation and small size for gestational age — and they conclude that further research would be valuable.[2]
What that means in plain terms: this study can rule out a large increase in malformation risk, but it is not powered to rule out a small one. A clean negative result — "GLP-1s are proven safe in early pregnancy" — is not what this study delivers, and the researchers do not claim it. The right takeaway is narrower and more useful: if you were exposed before you knew you were pregnant, the available evidence does not point to major harm.
What it does not change: the guidance to stop
None of this overturns standard advice. FDA-aligned labeling and manufacturer guidance for semaglutide, tirzepatide, and liraglutide still direct patients to discontinue these medications during pregnancy.[1] Pregnancy remains a reason to stop a GLP-1, not a reason to continue one. This study lowers the alarm for women who had accidental early exposure; it does not create a case for intentional use.
If you are pregnant, planning a pregnancy, or could become pregnant while taking a GLP-1, this is a conversation to have with your clinician — not a decision to make from a news article.
How it fits with earlier evidence
Two recent studies add important context. Both are older than this Annals analysis and answer different questions — we include them so the picture is complete, not because they are fresh news.
Stopping a GLP-1 came with its own risks (JAMA, Nov 2025)
A retrospective cohort from Mass General Brigham (1,792 matched pregnancies, 2016–2025, mostly women with obesity) looked at what happened after women discontinued a GLP-1 before or early in pregnancy.[3] Compared with matched pregnancies, the GLP-1 group had more excess gestational weight gain (65% vs. 49%; RR 1.32, 95% CI 1.19–1.47) and modestly higher risks of preterm delivery (RR 1.34, 1.06–1.69), gestational diabetes (RR 1.30, 1.01–1.68), and hypertensive disorders of pregnancy (RR 1.29, 1.12–1.49).
This is a useful counterweight: the conversation is not simply "exposure bad, no exposure safe." Stopping the medication — and the metabolic rebound that can follow — carries its own considerations, especially for women who were taking a GLP-1 for obesity or diabetes in the first place.
When you separate diabetes from the drug (Danish cohort, 2026)
A Danish nationwide registry study of 756,636 singleton pregnancies offered a sharp insight into why some GLP-1 pregnancy studies find a preterm-birth signal.[4] Among women exposed around conception, the elevated preterm-birth risk was confined to those using a GLP-1 for diabetes (liraglutide adjusted OR 1.70, 95% CI 1.17–2.48; semaglutide aOR 1.84, 1.24–2.71). Among women using the same drugs for weight management, there was no such increase.
The implication is that underlying diabetes — a long-recognized risk factor for preterm birth — appears to be driving the association in some datasets, rather than the medication itself. That is exactly the kind of confounding the Annals target-trial emulation was designed to account for.
Why these studies can look like they disagree. They are measuring different things — in-pregnancy continuation vs. discontinuation vs. periconceptional exposure — in different populations (diabetes vs. obesity), using different designs. Read together, they point in a consistent direction: the strongest pregnancy risks track with the underlying metabolic condition, and accidental early GLP-1 exposure does not show a major independent signal of harm. None of them tested intentional use in pregnancy, and none should be read as doing so.
What this means for you
If you are a woman taking a GLP-1 and you discover you are pregnant, this study is genuinely reassuring news: the best available evidence does not point to major harm from the exposure that occurred before you knew. That can ease a real and common source of fear.
It does not mean GLP-1 drugs are safe to use throughout pregnancy, and it does not change the guidance to stop them once pregnancy is confirmed. It also cannot fully rule out small effects on the rarest outcomes — the data simply are not precise enough yet, and the researchers are upfront about that.
The right next step is always the same: talk to your clinician. Do not start, stop, or change any medication in pregnancy based on a study summary — including this one.
For the full evidence profiles on the drugs involved, see our pages on semaglutide, tirzepatide, and liraglutide.
References
- [1]Wang Z, Hernández-Díaz S, Bateman BT, et al.. “Continuing Glucagon-Like Peptide-1 Receptor Agonists Into the First Trimester of Pregnancy and Pregnancy Outcomes: A Target Trial Emulation Study Using Claims Information.” Ann Intern Med. 2026. DOIcohort
Target-trial emulation on U.S. insurance-claims data. 3,572 pregnancies (2011–2024); 41.1% (n=1,467) with type 2 diabetes. Compared continuing a GLP-1 RA into the first trimester vs. stopping at conception. Nonlive birth risk 29.7% vs. 27.1% (adjusted risk ratio 1.09; 95% CI 0.98–1.23). Estimates imprecise for the rarest outcomes (major congenital malformation, small-for-gestational-age). Observational; not a randomized trial; does not establish safety of intentional use.
- [2]American College of Physicians. “Available evidence limited but does not suggest major adverse outcomes with early GLP-1 use in pregnancy.” 2026. Linkother
Press summary accompanying the Annals of Internal Medicine target-trial emulation. Frames the findings as reassurance for unintentional first-trimester exposure, and notes estimates were imprecise particularly for rarer outcomes.
- [3]Maya J, Pant D, Fu Y, et al.. “Gestational Weight Gain and Pregnancy Outcomes After GLP-1 Receptor Agonist Discontinuation.” JAMA. 2025. DOI PubMedcohort
Retrospective cohort, Mass General Brigham, 1,792 matched pregnancies (2016–2025), primarily women with obesity. After discontinuing a GLP-1 before or early in pregnancy: excess gestational weight gain RR 1.32 (95% CI 1.19–1.47), preterm delivery RR 1.34 (1.06–1.69), gestational diabetes RR 1.30 (1.01–1.68), hypertensive disorders RR 1.29 (1.12–1.49). Older context; describes the discontinuation period, not in-pregnancy exposure.
- [4]Damkier P, Cesta CE, et al.. “Periconceptional GLP-1 receptor agonist exposure and obstetric outcomes: a Danish nationwide cohort study.” Hum Reprod Open. 2026. Linkcohort
Danish nationwide registry cohort, 756,636 singleton pregnancies (2009–2023). Elevated preterm-birth risk seen only among women using GLP-1 RAs for diabetes (liraglutide aOR 1.70, 95% CI 1.17–2.48; semaglutide aOR 1.84, 1.24–2.71), not among weight-management users — implicating underlying diabetes rather than the drug. Older context.
Stay informed
Get evidence-based peptide updates. No spam, no vendor pitches — just the research and regulatory changes that matter.
Related content
Semaglutide
The GLP-1 agonist behind Ozempic and Wegovy. Full research profile covering mechanism, trials, and safety.
PeptideTirzepatide
The dual GIP/GLP-1 agonist (Mounjaro/Zepbound). Mechanism, clinical evidence, and safety data.
PeptideLiraglutide
The first-generation daily GLP-1 agonist (Victoza/Saxenda). Long-track-record profile and evidence.
GuideWhat Is 'Ozempic Face'?
The science behind facial volume loss on GLP-1 drugs — and what it does and doesn't mean.
GuideWhat Are Peptides?
A foundational guide to understanding peptides and their therapeutic potential.
Disclaimer
This article is for educational purposes only and is not medical advice. The study described is an observational target-trial emulation, not a randomized trial, and cannot establish that GLP-1 medications are safe to use in pregnancy. FDA-aligned labeling still advises discontinuing GLP-1 drugs during pregnancy. Do not start, stop, or change any medication before or during pregnancy without first consulting a licensed healthcare provider who knows your history.