The patient, a 34-year-old interior designer in Manchester, had assumed the diarrhoea and nausea were the only surprises semaglutide had in store. She had been on the combined pill for nine years. Four months in, with roughly twelve kilograms gone, she was sitting in her GP's office with a positive pregnancy test and a half-empty Wegovy pen. The vignette is a composite, drawn from the cluster of similar stories now showing up in clinic waiting rooms, in TikTok comments, and, more rigorously, in a 1.6-million-woman Australian primary-care cohort published in September [1][2].
The phrase doing the rounds, on social media and in the journals, is the "Ozempic baby". It is catchy, slightly frivolous, and slightly misleading. GLP-1 receptor agonists do not increase fertility in any direct sense. They cause weight loss, and weight loss restores ovulation in women whose cycles had quietly stopped working. For a substantial slice of users, that is a welcome reversal. For another slice, it is an unplanned pregnancy in a body that was not yet ready to host one.
The Signal in the Data
The most rigorous snapshot so far comes from Thapaliya and colleagues at Flinders University, working with the 1.6 million women aged 18 to 49 who attended Australian general practices between 2011 and 2022 [2]. Of 18,010 women first prescribed a GLP-1 receptor agonist during the study window, only 21 percent had any contraception recorded in their notes. Within six months of that first prescription, 2.2 percent of the cohort had conceived. Women with polycystic ovary syndrome were about twice as likely to fall pregnant as women without it, a ratio consistent with weight loss reversing chronic anovulation rather than with the drug itself doing anything new to the ovary.
A 2025 feature in The BMJ framed the same signal in plainer language: case reports and social media are filling with stories of women who conceived on semaglutide or tirzepatide, and the leading explanations are weight-loss-driven ovulation, a pharmacokinetic interaction with the oral contraceptive pill, and a quiet rise in sexual activity once people feel more comfortable in their bodies [1]. The Royal Australian College of General Practitioners has circulated clinical guidance in the wake of the MJA paper, advising that effective contraception be discussed at the point of prescribing [3][4].
Weight Loss, PCOS, and the Return of Ovulation
Polycystic ovary syndrome is the most common metabolic disorder of reproductive-age women, with pooled global prevalence in the ten to thirteen percent range under the Rotterdam criteria [5]. The syndrome ties together insulin resistance, hyperandrogenism, and often, but not always, a body mass index above the population norm. Even a five to ten percent drop in weight is enough to restore ovulation in a substantial fraction of those affected, and the GLP-1 class produces weight losses of that magnitude routinely.
A 2023 meta-analysis of randomised trials in PCOS, by Wu and colleagues, found that GLP-1 receptor agonists significantly increased natural pregnancy rates compared with comparator treatments, with a pooled risk ratio of 1.72 [6]. The same meta-analysis documented reductions in BMI, waist circumference, and androgen levels. The mechanism is not exotic. Less adipose tissue, less aromatised androgen, less suppression of the hypothalamic-pituitary-ovarian axis, more ovulatory cycles. The drug is not a fertility treatment. The weight loss it produces is.
For women with PCOS who have spent years wondering why they cannot conceive, this is a piece of luck rather than a scandal. The harder clinical question sits on the other side of the same statistic: women who did not want a pregnancy, who were on a contraceptive that was working for them in a heavier body, and whose contraception may or may not still be working now.
The Oral Contraceptive Question
The contraceptive mechanism worth worrying about is pharmacokinetic, not behavioural. Tirzepatide, the dual GIP/GLP-1 agonist sold as Mounjaro and Zepbound, slows gastric emptying substantially. A single 5 mg dose has been shown to reduce the peak concentration of co-administered ethinylestradiol and levonorgestrel by around twenty percent, and the FDA-approved US label advises women on oral hormonal contraceptives to switch to a non-oral method or add a barrier method for four weeks after starting tirzepatide and for four weeks after each dose escalation [7].
The data for semaglutide are more reassuring, at least for the subcutaneous formulation. Kapitza and colleagues, in a 2015 study in postmenopausal women with type 2 diabetes, found that steady-state semaglutide did not reduce the bioavailability of a combined oral contraceptive pill containing ethinylestradiol and levonorgestrel [8]. The Reproductive Health Access Project, summarising the field, recommends that women on tirzepatide prefer non-oral contraception (an IUD, an implant, or a depot injection) and that those on subcutaneous semaglutide, liraglutide, or dulaglutide can usually rely on a standard combined pill provided they take it consistently [9]. The older short-acting GLP-1 agonists, exenatide and lixisenatide, sit somewhere in between and are generally advised to be taken at least an hour before the pill.
The Australian cohort data make the practical point sharply. The 2.2 percent six-month pregnancy rate fell substantially among the women who did have contraception recorded [2]. A pill that is being vomited back up by a patient in the first weeks of dose escalation is, in practice, no pill at all. A non-oral method, set and forget, sidesteps the question.
What the Safety Data Actually Say
Reproductive toxicology in animals is the part of the story that most justifies the regulatory caution. A 2023 systematic review in Frontiers in Endocrinology, by Müller and colleagues, catalogues the animal evidence: reduced foetal weight and growth, delayed ossification, skeletal variants, and visceral abnormalities, usually alongside reduced maternal food intake [10]. A 2024 review in the American Journal of Obstetrics & Gynecology and a 2025 piece in Obstetrics, Gynaecology & Reproductive Medicine reach similar conclusions and recommend discontinuing GLP-1 receptor agonists at least two months before a planned pregnancy, in line with FDA labelling on the long half-life of semaglutide [11][12].
The human data are thinner, and the most rigorous piece, a 2024 prospective cohort from six Teratology Information Services led by Dao and colleagues, is genuinely encouraging in a narrow sense. Among 168 first-trimester GLP-1-exposed pregnancies, the major-birth-defect rate was 2.6 percent, statistically indistinguishable from both a diabetes reference group (2.3 percent) and an overweight or obese reference group (3.9 percent) [13]. The authors describe their finding as "reassurance" for inadvertent first-trimester exposure, while being careful to note the small sample. The cumulative incidence of live birth in that cohort was 59 percent, with pregnancy loss at 23 percent and termination at 18 percent, figures broadly consistent with what would be expected in the underlying populations.
Put the two bodies of evidence side by side and the picture is consistent. Animal reproductive toxicology is real and worrying. The human data, limited as they are, do not yet replicate that signal in women whose exposure was inadvertent and brief. The FDA's two-month washout recommendation before planned pregnancy is a precaution, not a verdict [11][12]. It is also, importantly, the kind of precaution that only works if a pregnancy is planned in the first place.
What Prescribers and Patients Are Doing Now
The clinical consensus, as far as one exists, is fairly straightforward. Women of reproductive age starting a GLP-1 receptor agonist for weight loss should be asked, at the point of prescribing, whether they could become pregnant and, if so, what they are using for contraception [1][3][9][14]. The default contraceptive of choice for someone on tirzepatide is a long-acting reversible method that does not depend on daily adherence or on a normally emptying stomach. Subcutaneous semaglutide, liraglutide, and dulaglutide are less likely to compromise a combined pill, but the pill still has to be inside the patient long enough to be absorbed.
The two-month washout before a planned pregnancy remains the standard advice, more because the human safety data are too thin to justify a tighter window than because there is a known teratogen waiting at the door [11][12][15]. For an unplanned pregnancy that has already happened, the early evidence is cautiously reassuring, but the right next step is a conversation with an obstetrician who knows the relevant Teratology Information Service contact, not a search of the Reddit thread where the pregnancy was first announced.
The deeper story is that GLP-1 agonists have arrived in a population that includes millions of women for whom pregnancy is not a hypothetical. Most of those women will use the drugs without incident, and a substantial minority will discover that a chronic condition they had stopped expecting to reverse, anovulation, insulin resistance, the metabolic syndrome of PCOS, has quietly lifted under their skin. Some of those discoveries will be wanted. Some will not. The clinical task, more than the regulatory one, is to make sure the second group has a chance to choose.