Approved Does
Not Mean Studied

Three Words People Use Interchangeably

Almost every argument about vaccines in pregnancy runs aground on three words that mean different things.

• ●Approved, or licensed, means the FDA reviewed an application and granted the product an indication. The indication is printed on the label and it states who the product is for.
• ●Recommended means a committee, usually ACIP, decided who should get it and when. A recommendation can sit entirely outside the approved indication, and for most of what's offered in pregnancy, it does.
• ●Studied means somebody ran the trial and the question was answered. This is the one nobody checks, and it's the only one that tells you anything about your own risk.

You'll hear that no vaccine is approved for use in pregnancy. That was true for a long time and it isn't quite true now. Abrysvo, Pfizer's RSV vaccine, was approved on August 21, 2023 specifically for pregnant individuals at 32 through 36 weeks gestation, to protect infants from birth through six months. It is the first and only US approval of its kind.

So the honest claim isn't that nothing is approved. It's that approval and study are not the same event, and this one approval is the clearest proof of that you will ever find, because the FDA wrote the whole thing down.

What the FDA's Own Reviewers Wrote

This comes from the Summary Basis for Regulatory Action, dated September 12, 2023, the document the review committee produces to explain its decision. It is public, and almost nobody reads it.

Read what that says. The trial enrolled women at 24 to 36 weeks. Across the window actually studied, the reviewers concluded that the risks outweighed the benefit.

So they approved a narrower slice of it. The document is explicit about why:

The product was licensed for a sub-window of its own trial, drawn to avoid the harm the trial had signalled. That is not a conspiracy theory, it's the stated reasoning, and once you know it you can't look at the 32-week floor the same way again.

The Vote You Were Never Told About

The Vaccines and Related Biological Products Advisory Committee met on May 18, 2023. Fourteen members voted on two separate questions.

Unanimous that it works. Nearly a third dissenting that it's safe. The summary records that the committee "discussed the balance between the convincing VE, including against severe LRTD, and AEs, particularly premature delivery/birth," and that members "expressed concern about preterm birth and stressed the importance of robust, timely postmarketing evaluation of preterm births."

Hold on to the word timely.

What "Timely" Turned Out to Mean

The FDA required four postmarketing studies to answer the preterm birth and hypertensive disorder question. Here are the timetables Pfizer submitted on August 10, 2023, which the FDA acknowledged in the approval letter.

Approved August 2023. The answers arrive between 2029 and 2031. That's six to eight pregnancy seasons of administration before anyone reports on the question the advisory committee split over.

And look at the registry, because it's the telling one

Of the four, only one is prospective, meaning it follows women forward rather than searching records afterward. That's the design that answers a safety question properly. It plans to enroll 927 vaccinated women against 927 unvaccinated.

Now recall the difference it's looking for. In the trial, preterm birth was 5.7% in the vaccine group against 4.7% in the placebo group, a gap of one percentage point.

Run that through 927 women per arm and here is what the study is being asked to do. At 5.7% you would expect about 53 early births in the vaccinated group. At 4.7% you would expect about 44 in the unvaccinated group. The entire question comes down to whether 53 is meaningfully different from 44.

It isn't, statistically. Counts that size swing by that much from chance alone, and the confidence interval around a nine-event difference would be wide enough to include no effect in almost any plausible result. The study is designed so that it will almost certainly report no significant difference whether or not one exists.

The one study designed to answer the question is the one least able to. That isn't a hidden flaw, it's arithmetic anyone can check, and it was in the approval package.

What the Label Says Was Never Looked At

Section 13 of a prescribing document is called Nonclinical Toxicology. Section 13.1 covers carcinogenesis, mutagenesis and impairment of fertility. On this label it reads:

Not "no risk was found." Not "studies showed no effect." Not evaluated. The question was not asked.

The only fertility data that exists is a rabbit developmental toxicity study, reporting no evidence of impaired female fertility at a dose with twice the antigen content of one human dose. Rabbits.

This language is standard across vaccine labels rather than unique to this one, which is the broader point. If you've been told a vaccine was tested for everything, the label is where you find out what "everything" covered. Ask for Section 13.1 by name and read it yourself, for any product, in pregnancy or out of it.

The Competitor That Was Stopped

This is the part that makes the preterm birth concern more than a statistical quibble, and it happened eighteen months before the approval.

GSK was developing its own maternal RSV vaccine, RSVPreF3-Mat, a different product aimed at the same target in the same population. In mid-February 2022 the independent committee monitoring its trial notified GSK of a preterm birth imbalance. On February 25, 2022, GSK stopped enrollment and vaccination in that trial and in every other ongoing trial of the product in pregnant women. The candidate has since been discontinued entirely.

The published conclusion was that the vaccine both reduced infant RSV disease and raised preterm birth risk. Both things at once.

And then the part that belongs in a conversation about consent rather than about safety. A 2023 BMJ investigation reported a dispute over whether Pfizer should have told the women enrolled in its own maternal trial that a trial of a similar vaccine had been halted over preterm birth. Same target, same mechanism class, same trimester, stopped for the same endpoint. The question is what the women still being vaccinated were told about it.

The Number the Label Doesn't Show You

The figure that gets quoted from Pfizer's trial is the pooled one, and pooled, it doesn't reach significance. Split by setting, it looks different.

In wealthy countries the two arms are identical. Everywhere else they separate by a margin the confidence interval supports.

In babies rather than percentages: the pooled one-point gap is about 10 extra early births per 1,000 women vaccinated, and that one does not reach significance. The three-point gap outside high-income countries is about 30 extra early births per 1,000, or roughly one in every 33 women vaccinated, and that one does.

The honest caveats, because this needs them to hold up. That split was a post-hoc analysis, meaning it was looked for after the fact rather than planned, and the number of events is low. Both of those weaken it. But a 20% difference that disappears in one half of a trial and reaches 73% in the other is not a finding you leave out of the summary, and it was left out of the summary.

Published in Obstetrics & Gynecology, February 2025, eighteen months after approval.

Now Compare the Clock on Everything Else

Here's where the three words at the top start to matter. Two products given to the same women in the same trimester, with their gestational windows set by completely different questions.

ACIP voted in October 2016 to prefer the early part of the Tdap window, and the stated reasons are all about the baby's antibody level: the mother needs about two weeks to respond, and cohort data from Israel and Switzerland found higher cord antibody at 27 to 30 weeks than at 31 to 36. Canada's advisory committee goes further and recommends 27 to 32 weeks ideally.

So one product has a regulator-imposed floor at 32 weeks because of a possible harm to the pregnancy. The other is preferentially given four or five weeks earlier than that floor, and the reason it moved earlier had nothing to do with the pregnancy.

The limit of this comparison, stated plainly, because the argument needs it. These are different products. A prefusion F protein vaccine and a tetanus-diphtheria-pertussis toxoid are not interchangeable, and a preterm signal in one does not transfer to the other by analogy. Anyone defending the schedule will say exactly that, and they're right to.

But that was never the claim. The claim is about which question got asked. For one product, "does this cause preterm birth at this gestational age" was asked, answered imprecisely, and the window drawn around the uncertainty. For the other, at 27 weeks, the question was how much antibody the baby gets. Both answers may be fine. Only one of the questions was about you.

What to Do With This

None of this tells you to decline anything. But "the vaccine works" is a claim with a number attached, and this page would be a hypocrite if it demanded numbers for the harms and accepted a word for the benefit. So here is the benefit, from the same trial, in the same units.

So the honest statement of benefit is this. For roughly every 139 to 167 women vaccinated, one infant hospitalization was prevented in the first six months. After that the measured benefit is no longer distinguishable from nothing.

For context on the background risk, the FDA's own review states that RSV-associated hospitalization runs at 1 to 3% of infants under one year, peaking in the first three months, with 1 to 3% mortality among those hospitalized. So the baseline chance of an infant dying of RSV works out somewhere around 1 in 1,000 to 1 in 10,000. Real, and rare.

A woman who reads all of that and goes ahead has made a real decision. So has one who doesn't. That is the entire purpose of writing it down.

What it does give you is a set of questions nobody minds you asking, and a way to check the answers yourself.

• ●"Is this approved for use in pregnancy, or recommended for it?" They are different and the answer is on the label.
• ●"Can I see Section 13.1?" Carcinogenesis, mutagenesis, impairment of fertility. Ask what it says rather than whether it was tested.
• ●"What were the postmarketing studies required at approval, and have any reported yet?" If the answer is that they're due in 2029, you've learned what's known today.
• ●"What gestational age window is this approved for, and why that one?" Sometimes the window is the safety finding.
• ●"What was the comparison group in the trial?" A worthwhile question, and for this trial the registry record says only "Placebo" without stating its composition.

Every document behind this page is public and free. The Summary Basis for Regulatory Action, the approval letter with its timetables, the label with its Section 13. They're on the FDA's own website and they're written in plain enough English to read in an afternoon.

Nobody hid any of it from you. It was just never going to arrive on its own.