You can be on GLP-1 and want a baby. You can have a thyroid condition and want to lose weight. Both situations have an answer. Here is the 2026 update: refreshed FDA washout windows by drug, the new Cleveland Clinic preconception data presented at SMFM February 2026, the "Ozempic baby" studies, MTC and REWIND thyroid data, Hashimoto's, and exactly how Levothyroxine fits in.
🔑 Key Takeaways
- The pre-conception washout is the most important rule. Semaglutide 2 months, tirzepatide 1 month per UK FSRH (US label says stop at recognition only), liraglutide 3 days, dulaglutide ~4 weeks, Bydureon ER 12 weeks. Stop earlier than you need to, not later.
- The 2026 Cleveland Clinic data presented at SMFM (Society for Maternal-Fetal Medicine, February 2026) found preconception GLP-1 use in 208 women with obesity did NOT increase gestational diabetes, severe maternal morbidity, or other adverse outcomes versus 8,000+ matched controls. First major reassurance signal in human data.
- Accidental pregnancy on GLP-1 is not a medical emergency. The 168-pregnancy BMJ Open cohort showed birth defect rates of 2.6% (GLP-1) vs 2.3% (diabetes reference) vs 3.9% (obese reference). No safety signal yet.
- If you find out you are pregnant on GLP-1: stop immediately, switch to insulin if you have diabetes, schedule an early anatomy scan around weeks 8 to 12, enroll in the registry (Wegovy: 1-877-390-2760; Zepbound: 1-800-LillyRx), do not assume termination is indicated.
- Tirzepatide is the only GLP-1 with a clinically significant interaction with oral birth control pills. Use a backup barrier method or non-oral method for 4 weeks after starting and after each dose increase. The other GLP-1s do not require added barrier contraception for absorption reasons (FSRH 2024 clinical statement).
- The "Ozempic baby" effect is real but mostly mediated by PCOS recovery, weight loss restoring ovulation, and (for tirzepatide users) reduced contraception efficacy. A 2025 PLOS ONE review found public sentiment outpaces the actual fertility evidence outside PCOS.
- The FDA boxed warning about thyroid cancer applies to medullary thyroid cancer (MTC) and MEN2 syndrome only (3 to 4% of thyroid cancers). The REWIND trial of dulaglutide (n=9,901) found zero new MTC cases in either arm over 5.4 years, the strongest human evidence against the rodent signal.
- Hypothyroidism and Hashimoto's do not contraindicate GLP-1, but they need management. Optimize TSH first. Re-test every 6 to 8 weeks while titrating. Expect Levothyroxine dose to drop 10 to 25% over the first year as you lose weight.
- If you do IVF, tell your fertility clinic and anesthesia team you are on a GLP-1. Delayed gastric emptying raises aspiration risk under sedation for egg retrieval. Hold timing varies by protocol.
This page covers the three safety questions GLP-1 users ask most: pregnancy planning (washouts, accidental exposure, postpartum re-start), fertility (Ozempic babies, PCOS, IVF, male partners), and thyroid disease (the boxed warning, MTC vs other thyroid cancers, Hashimoto's, Levothyroxine timing).
GLP-1 and Pregnancy: The Washout Rule
Stop the drug before you try, not after.
Every GLP-1 receptor agonist has a half-life that determines how long it stays in your system. The pregnancy warnings on every label are based on animal teratogenicity at supratherapeutic doses plus the absence of conclusive human safety data. The washout is dose-independent and indication-independent.
| Drug (brands) | Half-life | 5 half-lives (PK clearance) | Recommended pre-conception washout |
|---|---|---|---|
| Semaglutide injectable (Ozempic, Wegovy) | ~7 days | ~5 weeks | At least 2 months (FDA) |
| Semaglutide oral (Rybelsus) | ~7 days | ~5 weeks | At least 2 months (FDA) |
| Tirzepatide (Mounjaro, Zepbound) | ~5 days | ~3.5 weeks | US label: discontinue at pregnancy recognition (no numeric preconception window). UK FSRH/MHRA: ~1 month |
| Liraglutide (Saxenda, Victoza) | ~13 hours | ~3 days | Discontinue if planning pregnancy; ~3 days clears the drug |
| Dulaglutide (Trulicity) | ~5 days | ~3.5 weeks | ~4 weeks (no numeric US preconception window stated) |
| Exenatide IR (Byetta) | 2.4 hours | ~12 hours | ~10 days (FSRH) |
| Exenatide ER (Bydureon) | ~6 days release matrix | ~4 weeks | 12 weeks (FSRH, due to extended release matrix) |
| Retatrutide (R-30, trial-stage triple agonist) | ~6 days | ~4 weeks | 1 to 2 months (estimated, not yet on a marketed label) |
Sources: FDA prescribing information for each drug (2025-2026 labels), UK Faculty of Sexual and Reproductive Healthcare (FSRH) Clinical Statement on GLP-1 receptor agonists and contraception, MotherToBaby fact sheets.
Why the washout matters
Five half-lives clears about 97% of any drug from circulation. The recommended washouts above all hit at least 5 half-lives, with a buffer for the slow tissue release that GLP-1 medications can have. The animal data showing skeletal and visceral malformations occurred at doses many multiples above human therapeutic ranges, but the data exists, and the conservative path is to be off the drug well before conception.
US vs UK guidance: why they differ
If you have read different things in different places, this is why. The US FDA labels for tirzepatide (Zepbound, Mounjaro) and dulaglutide (Trulicity) say only "discontinue when pregnancy is recognized" without giving a numeric preconception window. The UK MHRA and FSRH chose to issue specific numeric guidance instead. The UK position uses ~1 month for tirzepatide and 12 weeks for Bydureon ER. Most US OB-GYNs informally follow the UK numeric guidance because patients want a real number, not "at recognition."
The fertility paradox
GLP-1 improves fertility (especially in PCOS, where liraglutide plus metformin doubled IVF pregnancy rates in one trial) and you have to stop the drug to get pregnant. The right framing is to use the time on GLP-1 to bank metabolic health, then plan the washout, then conceive. The fertility improvement persists past the drug.
The 2026 Cleveland Clinic / SMFM Preconception Data
The first large signal of reassurance.
At the Society for Maternal-Fetal Medicine (SMFM) pregnancy meeting in February 2026, Cara Dolin, MD, and Cleveland Clinic colleagues presented the largest preconception GLP-1 cohort to date. The team analyzed electronic medical records for more than 8,000 women with a pre-pregnancy BMI above 30 who delivered after 20 weeks gestation. They compared 208 women prescribed GLP-1 receptor agonists before pregnancy with the rest of the cohort.
What the Dolin / Cleveland Clinic 2026 cohort found
- The GLP-1 group was older, had higher BMI, higher rates of bariatric surgery, and higher rates of chronic hypertension at baseline, meaning more risk factors going in.
- Despite that, rates of gestational diabetes were similar between groups.
- Rates of severe maternal morbidity were similar between groups.
- Other adverse maternal outcomes were similar.
- The implication: preconception GLP-1 use may have helped mitigate the elevated baseline risk by optimizing metabolic health before conception.
This does not change the rule (stop before conception, do not use during pregnancy). What it does change is the conversation about whether to use GLP-1 in the months before a planned pregnancy in women with obesity or cardiometabolic risk. Dr. Dolin's recommendation is to consider a maternal-fetal medicine (MFM) referral for pre-pregnancy counseling rather than just stopping the drug abruptly.
The other relevant 2025-2026 signal: a separate 2025 study suggested that rebound weight gain after stopping GLP-1 may worsen some pregnancy outcomes. This is one reason why having a glycemic and weight-maintenance plan in place during the washout matters.
What to Do If You Get Pregnant on GLP-1
Step-by-step in the first 72 hours.
- Stop the GLP-1 immediately. Do not take another dose. Do not finish the pen.
- Contact your obstetrician or primary care. Same-day if possible.
- If you have type 2 diabetes, switch to insulin under clinician guidance. Do not abruptly stop a GLP-1 used for diabetes without a glycemic replacement plan. Insulin is the standard of care in pregnancy and has the longest safety record.
- If you do not have diabetes, no replacement medication is needed. Manage glucose and weight with diet.
- Schedule an early anatomy scan at 8 to 12 weeks. This is standard for any high-risk pregnancy, and it is reassuring early imaging.
- Enroll in the manufacturer's pregnancy exposure registry. Wegovy / semaglutide: Novo Nordisk at 1-877-390-2760 or wegovypregnancyregistry.com. Zepbound / tirzepatide: Eli Lilly at 1-800-LillyRx (1-800-545-5979). Your data helps future patients.
- Do not assume termination is indicated. The 168-pregnancy BMJ Open cohort and the 2026 Cleveland Clinic preconception data both show outcomes within the normal range. The data does not support routine termination based on GLP-1 exposure alone.
- Manage gestational weight gain. Maintenance is harder than expected after stopping a GLP-1; build a plan with your OB.
GLP-1 and Birth Control
The contraception gap is bigger than people realize.
The Flinders MJA cohort study of 18,010 women on GLP-1 RA found that only 21% were using contraception, and 2.2% became pregnant within 6 months of starting. Women with PCOS had twice the conception rate. This is the gap behind most "Ozempic babies."
Drug-specific contraception interactions
Per the FSRH 2024 Clinical Statement, the key distinction is that tirzepatide is the only GLP-1/GIP agonist with a clinically significant effect on oral contraceptive absorption. Semaglutide, liraglutide, dulaglutide, exenatide, and lixisenatide do not require added barrier contraception for pharmacokinetic absorption reasons.
| Drug | Effect on oral contraceptives | Recommendation |
|---|---|---|
| Tirzepatide (Mounjaro, Zepbound) | Reduces OC absorption during titration and after dose increases | Backup barrier method for 4 weeks after starting AND for 4 weeks after each dose escalation, OR switch to a non-oral method (IUD, implant, ring, injection) |
| Semaglutide (Ozempic, Wegovy, Rybelsus) | No clinically significant absorption effect | OCs work normally; vomiting from dose escalation can still mean a "missed pill" |
| Liraglutide (Saxenda, Victoza) | Minor delay in absorption | OCs generally fine |
| Dulaglutide (Trulicity) | No clinically significant interaction | OCs work normally |
| Exenatide IR (Byetta) | Timing matters | Take OC at least 1 hour before exenatide |
| Lixisenatide (Adlyxin) | Timing matters | Take OC at least 1 hour before or 11 hours after the dose |
The safest approach across all GLP-1s if you absolutely do not want to conceive: use a non-oral contraception method (IUD, implant, ring, injection), which is unaffected by any of the absorption issues above. The copper IUD is also the most effective emergency contraception option and is unaffected by any GLP-1.
The "Ozempic Baby" Phenomenon: What the Studies Actually Show
Three real mechanisms, one popular myth.
"Ozempic babies" is a term that emerged on social media around 2024 to describe unplanned pregnancies in women on GLP-1 medications, sometimes after years of trying without success. The phenomenon is real. The mechanisms behind it are not what the headlines suggest.
Mechanism 1: Weight loss restoring ovulation
In women with obesity-related anovulation or PCOS, modest weight loss (5 to 10% of body weight) is enough to restore regular ovulation. ASRM, ACOG, and the Endocrine Society all recognize this. GLP-1 medications produce 10 to 20% weight loss in 6 to 12 months in the average user. That is more than enough to restart cycles in many women whose only fertility barrier was metabolic.
Mechanism 2: PCOS-specific effects
GLP-1 medications do appear to have a PCOS-specific fertility benefit beyond pure weight loss:
- Salamun et al, European Journal of Endocrinology, 2018: liraglutide plus metformin produced superior IVF pregnancy rates in PCOS patients compared with metformin alone.
- Li et al, Archives of Gynecology and Obstetrics, 2023: significantly more spontaneous pregnancies in an exenatide group versus a metformin group at 24 weeks.
- Mechanism: GLP-1 reduces androgen levels, improves insulin sensitivity, and restores ovulation in a way that goes beyond weight effects alone.
Mechanism 3: Birth control failure on tirzepatide
The largest single contributor to truly accidental "Ozempic baby" cases is reduced oral contraceptive absorption on tirzepatide during dose escalation. Women using the pill plus tirzepatide who do not switch methods or add a barrier are the highest-risk group for unplanned pregnancy.
What is NOT happening
GLP-1 medications are not fertility drugs. They do not stimulate ovaries. They do not directly cause ovulation. The 2025 PLOS ONE review by Vu et al found that public sentiment about GLP-1s and fertility outpaces the clinical evidence, particularly outside PCOS. ASRM has acknowledged GLP-1 as an option for weight management in PCOS while warning that pregnancy safety data remain insufficient and effective contraception must be in place.
GLP-1 and Male Fertility
Limited data, no current safety signal.
Animal studies with semaglutide showed effects on sperm quality and fertility markers, but only at doses far exceeding the maximum human dose. Human data on male fertility effects from GLP-1s are limited, and current evidence does not show a pregnancy-risk signal from paternal exposure. MotherToBaby states that paternal GLP-1 exposure is unlikely to increase pregnancy risk.
Practical takeaway: men do not need to stop GLP-1 while their partner is trying to conceive. If a male user has unexplained subfertility on GLP-1 and wants to maximize their odds, a 3 month washout (the typical sperm production cycle) is the conservative option, but it is not currently a guideline recommendation.
GLP-1 and IVF / Egg Retrieval
The aspiration risk is real and underdiscussed.
If you are doing IVF and on a GLP-1, two things matter:
- Anesthesia aspiration risk during egg retrieval. GLP-1 medications delay gastric emptying, sometimes for many hours. Under sedation, that raises the risk of aspirating stomach contents into the lungs. Tell your fertility clinic and anesthesia team that you are on a GLP-1/GIP-GLP-1 medication. Hold-time protocols vary by clinic and procedure. Some clinics ask patients to hold the GLP-1 for several days before retrieval; others extend the fasting window. Coordinate directly with your reproductive endocrinologist.
- Embryo transfer timing. If you are doing fresh transfer, you need to be in the washout window already. If you are doing frozen transfer, you can stay on GLP-1 through the retrieval cycle (with the anesthesia hold above) and only stop before the transfer cycle.
For PCOS patients specifically: some fertility clinics now use GLP-1 in the months before stimulation to optimize metabolic health, then stop in time for transfer. This is the same logic as the 2026 Cleveland Clinic preconception data above.
Postpartum Return to GLP-1
Slow re-titration, not your previous dose.
Most women considering GLP-1 postpartum should plan around two milestones:
- Breastfeeding decision. GLP-1 transfer to breast milk is poorly studied. Semaglutide injectable was not detected in the milk of mothers using the injection, per the LactMed database. Oral semaglutide (Rybelsus) did show traces, which is why the FDA does not recommend Rybelsus during breastfeeding. Tirzepatide doses below 5 mg did not usually leave detectable traces. The conservative recommendation is still to be off GLP-1 while nursing.
- Postpartum recovery. Wait until bleeding, contractions, and significant hormonal shifts have settled, typically 6 to 12 weeks postpartum.
When restarting, expect to retitrate from the lowest dose (0.25 mg semaglutide, 2.5 mg tirzepatide), even if you were on a higher maintenance dose pre-pregnancy. The receptors have downregulated. Going straight to a higher dose causes severe nausea.
GLP-1 and Thyroid Cancer: The MTC Boxed Warning
The boxed warning is more specific than most articles tell you.
Every GLP-1 receptor agonist carries a boxed warning for medullary thyroid cancer (MTC) and Multiple Endocrine Neoplasia type 2 (MEN2) syndrome. The warning is based on rodent data: rats and mice given high doses of GLP-1 developed C-cell hyperplasia and thyroid C-cell tumors. The translation to humans is poor for two reasons.
- Species differences: Rodent C-cells express much higher GLP-1 receptor density than human C-cells. The "high-receptor tissue" effect that produces tumors in rodents is not replicable in human thyroid biology.
- MTC is a small fraction of thyroid cancer: Only about 3 to 4% of thyroid cancers are medullary. The remaining 95 to 97% are papillary, follicular, or Hürthle cell, and they are not the target of the boxed warning.
The REWIND data: the strongest human evidence
REWIND was the cardiovascular outcomes trial of dulaglutide. It enrolled 9,901 patients with type 2 diabetes and followed them for a median of 5.4 years. It is the longest, largest randomized GLP-1 dataset in existence with adjudicated thyroid outcomes.
| REWIND finding | Number |
|---|---|
| Patients on dulaglutide | 4,949 |
| Patients on placebo | 4,952 |
| Median follow-up | 5.4 years |
| New medullary thyroid cancer cases (dulaglutide arm) | 0 |
| New medullary thyroid cancer cases (placebo arm) | 0 |
| Calcitonin elevations > reference | No statistically significant difference between arms |
If GLP-1 caused MTC in humans the way it does in rats, REWIND should have shown it. It did not.
The detection bias problem with observational thyroid signals
Observational studies showing increased thyroid cancer in GLP-1 users (HR 1.85 in year 1, dropping to 0.78 after 2 years) follow a textbook detection bias pattern. New users get more imaging and more medical attention. That finds existing thyroid nodules, not new ones. The signal disappears with longer follow-up because there are no new cases to find.
| Source | Finding | Interpretation |
|---|---|---|
| REWIND RCT (n=9,901, 5.4 yr) | Zero new MTC cases in either arm | No human MTC signal |
| Diabetes Care retrospective (n=351,913) | Year 1 HR 1.85, year 2 HR 1.27, after year 2 HR 0.78 | Pattern consistent with detection bias |
| JAMA Otolaryngology meta-analysis | HR confidence interval crossing 1.0 | No statistically significant signal |
| Calcitonin levels in users | No consistent meaningful elevation | Routine calcitonin screening NOT recommended |
Who should not take GLP-1 because of thyroid risk
- Personal history of medullary thyroid cancer
- Family history of MTC (one or more first-degree relatives)
- MEN2 syndrome (genetic testing positive)
- Calcitonin elevated above baseline reference range
Who can take GLP-1 with thyroid history
- Personal history of papillary, follicular, or Hürthle cell thyroid cancer (post-treatment, not contraindicated by the boxed warning)
- Hashimoto's thyroiditis (not a cancer; not contraindicated)
- Graves' disease (managed; not contraindicated)
- Subclinical or treated hypothyroidism
- Simple thyroid nodules (followed appropriately)
- Post-thyroidectomy
GLP-1 and Hypothyroidism: The Levothyroxine Question
You can do both. The dosing changes.
If you take Levothyroxine and start GLP-1, three things happen over the first 6 months:
- GLP-1 delays gastric emptying. Levothyroxine absorption depends on stomach acid and timing. Delayed emptying can reduce absorption by 10 to 20%.
- Weight loss reduces Levothyroxine requirement. Less body mass means less thyroid hormone needed. The dose typically drops over 3 to 12 months on GLP-1.
- The two effects partially cancel. Reduced absorption pushes you toward needing more, but lower body weight pushes you toward needing less. Net effect is variable.
The protocol
- Optimize thyroid function before starting GLP-1 (target TSH 0.5 to 2.5 mIU/L)
- Take Levothyroxine on an empty stomach, 30 to 60 minutes before any food, fluid other than water, or other medications
- Do not take GLP-1 and Levothyroxine in the same window. Separate by at least 4 hours.
- Re-test TSH every 6 to 8 weeks during GLP-1 titration
- Expect to drop the Levothyroxine dose by 10 to 25% over the first year of GLP-1 use, with weight loss
- Watch for over-replacement symptoms (palpitations, heat intolerance, jitteriness) which can suggest the dose is now too high
GLP-1 and Hashimoto's Thyroiditis
Not contraindicated. Often beneficial.
Hashimoto's is autoimmune destruction of thyroid tissue, not cancer. It is not the target of the GLP-1 boxed warning. Mechanistically, GLP-1 may even help with autoimmune thyroid disease through reductions in metabolic inflammation, leptin levels, and gut barrier dysfunction (all of which are implicated in autoimmune thyroid pathology). No randomized trials yet, but the direction of the effect is plausibly positive.
Practical guidance: same as hypothyroidism above. Optimize first, monitor TSH every 6 to 8 weeks during titration, expect Levothyroxine dose to drop with weight loss. If you want broader context on the GLP-1 family and how the hormone itself works, see our explainer on what GLP-1 is, and for cost planning during a long preconception window, our guide to GLP-1 without insurance.
Frequently Asked Questions
Medical disclaimer. This article is informational only and does not replace individualized medical advice. Pregnancy planning, accidental exposure management, fertility decisions, and decisions about GLP-1 use in the context of thyroid disease must be made with your prescribing clinician, your obstetrician, and (where relevant) an endocrinologist or maternal-fetal medicine specialist. The recommendations summarize current FDA labeling, UK FSRH and MHRA guidance, the 2026 Cleveland Clinic / SMFM preconception data, the REWIND trial, and the BMJ Open accidental-exposure cohort, but do not substitute for case-by-case clinical judgment. Pregnancy and thyroid decisions are high-stakes and time-sensitive; if you are unsure, do not delay calling your clinician.