Nobody warned me fertility could return quickly. What are the contraception considerations?

Anovulatory for most of my adult life with PCOS, told repeatedly that conception would be difficult and would need assistance. Five months into treatment, having lost 13 kg, I have had four consecutive apparently normal cycles and a positive pregnancy test. The pregnancy is wanted and I am pleased, and I am also angry, because at no point in the prescribing consultation was the possibility raised that my fertility might return, let alone that it might return quickly, let alone that I should think about whether I wanted to be pregnant while taking a drug that is not recommended in pregnancy.

I would like to understand, partly for my own sake and partly so it can be said clearly to whoever reads this:

• How quickly can ovulation resume, and does it require substantial weight loss first?
• What is the position on these drugs in pregnancy, and what washout is advised before trying to conceive?
• Are there interactions with contraception that people should know about? I have seen something about oral contraceptives and one of these drugs specifically.
• What else should someone planning a pregnancy after significant weight loss know?

I am under proper obstetric care now, so this is not a request for advice about my own situation. It is a request for the information that should have been in the consultation.

Worth stating why this gets omitted, because understanding the failure mode is how you defend against it in your own care.

Four structural reasons, none of which is a good excuse:

• The prescribing pathway is metabolic, not reproductive. These drugs are prescribed for weight and glycaemia, frequently by clinicians whose framing is cardiometabolic risk. Fertility is not on the mental checklist because it is not what the drug is for, even though it is a predictable consequence of what the drug does.
• The patient's own history argues against it. A woman who has been told for a decade that conception will be difficult is unlikely to raise contraception, and a clinician taking a history hears "PCOS, anovulatory, subfertile" and files it as a reason contraception is not a topic. The two parties reinforce each other's omission.
• Telehealth pathways are short. A consultation optimised for throughput covers the labelled contraindications and the common side effects. A conditional, patient-group-specific warning that requires a conversation about pregnancy intention is exactly the thing a fifteen-minute asynchronous intake drops.
• The people using these agents outside a prescribing relationship get no consultation at all. Anyone obtaining research-grade material has no consultation, no medication review, no pregnancy discussion and no follow-up. Research-use-only compounds are not approved for human use and this is one of the concrete consequences of that framing that is rarely spelled out: nobody tells you the things a prescriber would be obliged to tell you.

The defence, if you are the patient, is to raise it yourself with three specific questions:

1. Could this restore ovulation, and how quickly?
2. What contraception do you recommend while I am taking it, and does this specific agent interact with it?
3. If I decided to try to conceive, how long before that would I need to stop?

Those three questions take two minutes and they extract the entire content of the accepted answer from any clinician who knows the material. They are also the questions to put to a prescriber who did not raise the topic, since the answer will tell you something useful about the quality of the service.

One more asymmetry worth noting. The same restoration of fertility that is an unwelcome surprise for one woman is the outcome another woman has been pursuing for years. Framing it purely as a risk is as unbalanced as omitting it entirely. The correct framing is that it is a powerful and reasonably fast effect, in a direction the patient may or may not want, and which therefore requires asking the patient what she wants before it happens rather than after.

Adding a note on what the trial evidence says about pregnancy in these programmes, since the answer is "almost nothing, by design", and that itself is worth understanding.

Every large trial in this class excluded pregnancy, required contraception in participants of childbearing potential, and withdrew participants who became pregnant. That is standard and appropriate for an investigational agent with animal reproductive toxicity findings. The consequence is that:

• There is no randomised evidence on pregnancy outcomes. There will not be, and there should not be, from this design.
• The inadvertent exposures are small in number and unsystematically reported. Trial pregnancies were withdrawals, not endpoints, and follow-up varies.
• Everything we will eventually know will come from registries and from population database studies, both of which take years and both of which have the confounding problems that any observational study of pregnancy exposure has — the underlying obesity, diabetes and PCOS all independently affect pregnancy outcomes, so separating the drug's contribution from the indication's is hard.

Which means the current advice is precautionary rather than evidence-based in the strong sense, and precautionary advice is the correct response to genuine uncertainty about a first-trimester exposure. It also means that anyone claiming these agents are known to be safe in pregnancy, or known to be harmful, is over-reading the same absent dataset in opposite directions.

Two smaller points that follow.

First, the reason the trials required contraception is precisely the reason the prescribing consultation should discuss it. The trial protocols got this right; the routine care pathway did not inherit it. That is a reasonably damning observation about how trial protocols translate into practice.

Second, if you have had an inadvertent exposure, reporting it — to the manufacturer's pregnancy registry, to your national pharmacovigilance scheme, or through your obstetric team — is the mechanism by which the dataset improves. It is unglamorous and it is the only route by which the next person gets a better answer than the current precautionary one.