CGRP Migraine Injections and Miscarriage Risk: What a New JAMA Neurology Study Found

Why This Study Matters

Migraine is most common during the years when many people are trying to get pregnant, and the CGRP monoclonal antibodies are now among the most widely used preventive treatments. That overlap has left a lot of people asking the same question: what happens if I conceive while I am still on my injection?

Until now, the answer came almost entirely from drug safety reports and small case series. Those sources had not picked up a clear problem, which is what our recent review of CGRP drugs and women's health described. But safety reports cannot compare exposed pregnancies with a matched group of unexposed ones, so they can miss a moderate increase in risk.

A study published in JAMA Neurology in October 2026 took a different approach. Researchers from CHU Sainte-Justine in Montreal and the Montefiore Headache Center in New York used US insurance records to follow thousands of pregnancies in people with migraine and compare miscarriage rates by medication exposure.

How the Study Was Done

The team used the AM-PREGNANT cohort, which is built from the Merative MarketScan Commercial Database, a large collection of US employer-based insurance claims. Out of nearly 8 million pregnancies in that cohort, they kept those that met strict criteria:

• Age 15 to 45 and continuously insured
• Last menstrual period on or after November 1, 2018
• Migraine before pregnancy, defined as a migraine diagnosis or a triptan or ergot prescription in the 6 months before the last menstrual period
• A single baby (no twins or multiples)

That left 7,579 pregnancies in 7,119 people. Each pregnancy was followed up to 20 weeks or until a miscarriage occurred.

Exposure meant at least one filled prescription or injection claim for a CGRP antibody, or for propranolol (Inderal), during follow-up. Propranolol was included because it is a long-standing migraine preventive often used around pregnancy, so it gives a useful comparison. Exposure was treated as time-varying, meaning a pregnancy only counted as exposed from the point the drug was filled or given.

The researchers adjusted their models for other health conditions and other medications, then ran a separate bias analysis to test how much unmeasured factors could shift the result.

What It Found

Compared with time when no study drug was used, exposure to a CGRP antibody in early pregnancy was linked to about a 42% higher rate of miscarriage. The bias-adjusted estimate was slightly smaller, at about 33%, and still pointed in the same direction.

Propranolol showed no link. Its estimate sat almost exactly at 1.0, meaning no difference from unexposed time.

Reading the Results

This is an observational study. It shows an association between the antibodies and miscarriage, and the authors framed it that way. It cannot prove the drugs caused the losses. A few points help put the numbers in context.

Why the finding is taken seriously

The bias analysis landed close to the main result, which suggests the link is not easily explained away by the kinds of bias the researchers tested for. The propranolol comparison also helps. Both groups had migraine and both were on a preventive, so if migraine itself or needing a preventive explained the signal, propranolol users would be expected to show it too. They did not.

There is also a biological reason to look closely. CGRP plays a part in how the placenta develops, which is why researchers flagged the antibodies as a pregnancy question from the start.

What claims data cannot show

• Who gets prescribed what. People on CGRP antibodies often have more frequent or harder-to-treat migraine than people on propranolol. Severity is hard to capture fully in insurance records.
• Filled is not the same as taken. A pharmacy claim shows a prescription was picked up, not exactly when it was injected.
• Missing details. Claims usually lack information such as smoking, alcohol, body weight, and over-the-counter medicines, all of which can affect pregnancy outcomes.
• Early losses. About 1 in 4 pregnancies in this sample ended in miscarriage across all groups. Insurance data capture losses that reach medical care, and how completely they do so can vary.

The study looked at miscarriage only. It does not tell us about birth defects, growth, preterm birth, or child development, and it does not cover gepants, which are a different type of drug.

How It Fits With Earlier Data

Earlier evidence on CGRP antibodies in pregnancy came from voluntary safety reports and drug company case collections. Those sources did not flag a miscarriage problem, but they were not built to measure risk against a comparison group. The new study adds a comparison group, which the earlier sources lacked.

The two kinds of evidence answer different questions. Safety reports are good at spotting unusual or severe patterns, such as a specific birth defect. A cohort with a comparison group is better at detecting a moderate rise in a common outcome like miscarriage. The new result is one study, and it will need to be repeated in other datasets, ideally in pregnancy registries where exposure timing is recorded directly.

If You Use a CGRP Antibody

If you are planning a pregnancy

These antibodies stay in the body for weeks after each dose. Galcanezumab (Emgality), for example, has a half-life of 27 days, and drug levels can still be present into the second trimester when the last dose was given around conception. That is why timing is worth discussing well before you start trying. Your doctor can help you plan when to stop and what to use in the meantime. Our guide to migraine treatment in pregnancy and breastfeeding covers the options that have more safety data behind them.

This study adds some reassurance about propranolol (Inderal) and miscarriage specifically, though it did not look at other pregnancy outcomes.

If you find out you are pregnant while on one

Contact your prescriber and your obstetric provider soon so they can review your treatment and follow your pregnancy. A higher relative rate does not mean a loss is likely in your case. Most exposed pregnancies in this study and in earlier reports did not end in miscarriage. Do not take another dose until you have spoken with your doctor.

If pregnancy is possible but not planned

Reliable contraception matters while using a long-acting antibody. Migraine with aura affects which methods are a good fit, since estrogen-containing options are generally avoided in that group. The AAN and AHS prevention guideline lays out the full range of preventives, including older options your doctor may consider if a pregnancy is on the horizon.

Questions for Your Doctor

Bringing a short written summary of your attacks and treatments makes these conversations easier. Our free migraine appointment summary sheet is built for that.

• If I might want to get pregnant in the next year, when should I stop my CGRP antibody, and what would I use instead?
• How does this new study change your advice compared with a year ago?
• If I conceive while still on treatment, what monitoring would you recommend, and is there a pregnancy registry I can join?
• Is propranolol or another preventive a reasonable option for me while trying to conceive?
• Given whether I have aura, which contraception is right for me while I stay on my current treatment?

Key Takeaways

Earlier safety reports had not flagged a miscarriage problem with CGRP antibodies. This study, which used a comparison group, found one, and the finding fits with what is known about CGRP's role in the placenta. Registries that track exposure timing directly will be the next test of whether the link holds.