Does stopping semaglutide before pregnancy make it safe?

This is what the new data complicates. Both studies found that women who stopped GLP-1 receptor agonists before becoming pregnant still had significantly elevated risks for gestational complications compared to non-users — including excessive gestational weight gain, gestational diabetes, large-for-gestational-age babies, and caesarean delivery. The elevated risk in people who stopped before conception was similar in magnitude to those who continued into pregnancy. Researchers suggested this could reflect 'rebound effects' after treatment discontinuation, or it may reflect the underlying metabolic conditions that led to GLP-1 RA use in the first place.

What are GLP-1 receptor agonists and who uses them?

GLP-1 receptor agonists (GLP-1 RAs) include semaglutide (Ozempic, Wegovy), liraglutide (Saxenda, Victoza), dulaglutide (Trulicity), and others. They are TGA-approved for type 2 diabetes management and, for some formulations, weight management in people with obesity. Their use has grown very rapidly — the Annals of Internal Medicine study found GLP-1 RA dispensations in the 90 days before a woman's last menstrual period rose from 2 per 1,000 pregnancies in 2020 to 15 per 1,000 pregnancies in 2024 in the US, a sevenfold increase in four years.

What should someone using a GLP-1 RA do if they are thinking about pregnancy?

Discuss it with a GP or treating specialist before making any medication changes. The current picture is one of emerging, association-based evidence — not established guideline change. Current Australian guidance from the TGA recommends avoiding semaglutide in pregnancy and discontinuing at least two months before a planned conception. These new studies add complexity to the 'just stop before conception' framing, but they do not yet change the formal recommendation. The decision about what to do — and when to stop — depends on what the GLP-1 RA was prescribed for, the underlying metabolic health, and the pregnancy timeline.

Stopping semaglutide before pregnancy may not be enough — two new studies

What just happened

Two new US studies have landed data that complicates one of the assumed safe harbours around GLP-1 receptor agonist use and pregnancy: the idea that stopping the medication before conception adequately removes the risk.

Both studies — one published in Obstetrics & Gynaecology in August 2026 using data from the Truveta network (covering more than 20% of US clinical care), and one in the Annals of Internal Medicine in June 2026 tracking GLP-1 RA dispensations from 2011 to 2024 — found elevated gestational complication rates in women with prior GLP-1 RA exposure, whether or not they stopped before becoming pregnant.

The Obstetrics & Gynaecology study followed 1,230 women who had received semaglutide before pregnancy, comparing them to 2,203 pregnancies in a no-GLP-1 group. Among the women who stopped semaglutide before conception, the findings were: excessive gestational weight gain (odds ratio 1.98), gestational diabetes (odds ratio 1.43), excessive foetal growth (odds ratio 1.54), and caesarean delivery (odds ratio 3.92) — all significantly elevated compared to non-users.

Crucially, whether semaglutide was ceased before pregnancy or continued into the first trimester made no statistically significant difference to these outcomes.

The Annals study found the background context has shifted rapidly: GLP-1 RA dispensations in the 90 days before a woman’s last menstrual period rose from 2 per 1,000 pregnancies in 2020 to 15 per 1,000 in 2024 — a sevenfold increase in four years.

Both-and

What the data says — and where the uncertainty lives

The findings are striking in their direction but the interpretation requires care. The studies were observational. The women using GLP-1 RAs before pregnancy were — by definition — women with obesity, type 2 diabetes, or both. These conditions themselves independently increase the risk of every adverse outcome the studies measured: gestational diabetes, excessive gestational weight gain, large-for-gestational-age babies, caesarean delivery.

This is the confounding problem that runs through all GLP-1 RA pregnancy research: you cannot easily separate the drug’s effect from the underlying metabolic health of the people prescribed it.

The Obstetrics & Gynaecology researchers acknowledged this, suggesting the elevated risks in women who stopped before pregnancy could indicate “the elevated risks were related to factors such as rebound effects after treatment discontinuation” — but they could not confirm that against a comparator of women with equivalent metabolic risk who never used GLP-1 RAs.

The “rebound” hypothesis is biologically plausible. GLP-1 RAs suppress appetite and improve insulin sensitivity. Stopping them removes those effects — and in people with underlying metabolic vulnerability, the rebound in weight and glucose regulation before pregnancy could itself drive the adverse outcomes, independent of any direct medication effect.

What was not elevated

Both studies found no significant difference in spontaneous abortion rates between GLP-1 RA users and non-users. The Annals study found no significant association between GLP-1 RA continuation and small-for-gestational-age babies, large-for-gestational-age babies, or major congenital malformations in live births. Elective terminations were more common in the continuation group — which the researchers attributed to anxiety about drug exposure, not to known teratogenicity.

This nuance matters. The signal is in metabolic and obstetric complications, not in fetal structural harm or pregnancy loss. That does not make the findings less important, but it shapes the clinical conversation appropriately.

The scale of the question

The sevenfold rise in GLP-1 RA pre-pregnancy exposure in the US over four years is not a niche phenomenon. Australia is tracking a similar trajectory in GLP-1 RA use overall. The Annals study mean age at last menstrual period was 34 years — well within the reproductive years for many women on these medications. For general practice, this is no longer a rare clinical question.

My two cents

The clinical instinct so far has been to tell women to stop GLP-1 RAs at least two months before a planned conception — which is still the current TGA recommendation for semaglutide — and assume the risk largely resolves with the drug’s clearance. This new data suggests that framing may be too simple.

What it might mean is that the period of pre-pregnancy care for women on GLP-1 RAs needs to be more active than a medication stop and a waiting period. If stopping the drug triggers metabolic rebound — weight regain, glucose dysregulation, rising cardiovascular risk markers — then the two months before conception might require its own clinical attention: monitoring, lifestyle support, and in some cases an alternative medication review.

None of this means GLP-1 RAs should not be prescribed, or that the clinical benefit in type 2 diabetes and obesity is outweighed by pregnancy risk. These are medications that work, that carry real metabolic benefit, and that the Australian health system has now made more accessible. The pregnancy question is a specific subset of the overall clinical picture.

What it does mean is that women of reproductive age using GLP-1 RAs — especially those who may become pregnant in the next one to two years — deserve a conversation about what the emerging evidence shows, and a plan for what happens when they want to conceive. That conversation starts in general practice.

Verdict: maybe — the data is US-sourced, observational, and confounded by underlying metabolic risk. It does not yet change Australian clinical guidelines. But the pattern is consistent across two studies, and the scale of GLP-1 RA use in reproductive-age women makes this a conversation that belongs in general practice now, not when guidelines eventually catch up.

Sources cited

Historical GLP-1 use could still affect pregnancy outcomes. The Medical Republic, 3 August 2026. https://www.medicalrepublic.com.au/historical-glp-1-use-could-still-affect-pregnancy-outcomes/127867