Did Fauci's Texts Prove He Knew Vaccines Caused Miscarriages?

The claim and why it splinters

On August 10, 2026, Senators Ron Johnson and Rand Paul released screenshots of 11 text messages from Dr. Anthony Fauci's former government iPhone, a device they say contains more than 34,000 messages and 522 voicemails.^[1] The centrepiece is a January 25, 2021, text in which Fauci raised a theoretical concern about first-trimester miscarriage risk tied to second-dose mRNA vaccine reactions. Within hours, the framing settled into a single viral assertion: Fauci "KNEW" vaccines caused miscarriages and hid it.

That assertion earns a False verdict. But the release has also revived two older, separately flawed claims (one about a 2021 NEJM study's denominator and one about an 82% miscarriage figure that circulated after a retracted paper) that predate the August 2026 texts by five years. Mixing them into a single "bombshell" inflates each beyond what its underlying evidence supports. All three threads need individual treatment.

What the texts actually say

The January 25-26, 2021 exchange involved Fauci, then-CDC Director Rochelle Walensky, and then-Surgeon General nominee Vivek Murthy. Murthy opened by asking whether there was "any data or theoretical reason" to prefer COVID-19 vaccination earlier or later in pregnancy.

Fauci initially replied that he knew of no data or theoretical reason to prefer one timing over another. That evening, at 6:45 PM, he returned with a caveat:

"Since many people have significant cytokines storm and fever after the 2nd dose, this theoretically could be associated with miscarriage in the 1st trimester." Dr. Anthony Fauci, text message, January 25, 2021^[2]

Walensky responded two minutes later: "Definitely a good point, esp after dose two." Murthy described it as "a really good point" roughly 40 minutes after that.^[3]

Walensky also described vaccine safety data in pregnancy as "data free" at the time. This phrase is critical context: pregnant women had largely been excluded from initial vaccine trials, so no observational data existed to confirm or refute the theoretical mechanism Fauci raised. Her use of the phrase was an acknowledgment that the field was operating without evidence in either direction — not a confirmation that vaccines were dangerous.

The public-private gap: a real but narrower question

Nine days after the January 25 text, Fauci told the Journal of the American Medical Association: "The FDA … have found, thus far … no red flags about that, about pregnant women."^[4] On February 10, at a White House briefing, he noted approximately 20,000 pregnant women had been vaccinated with "no red flags."^[5]

The gap between a private theoretical concern on January 25 and public "no red flags" language on February 3 and 10 is real. It has attracted scrutiny for a reason.

Fauci's public statements accurately described the observed data: no safety signals had emerged in vaccinated pregnant women. His January text raised a question about a biological mechanism, not observed outcomes. "No red flags" reports what monitoring found; it does not claim every theoretical question has been foreclosed. The CDC's V-safe Pregnancy Registry was already collecting the missing data, having enrolled more than 15,000 pregnant vaccinees by late January 2021.

That said, a fuller communication to the public could have explicitly noted that pregnancy safety data was still being gathered and that some questions remained under active scientific discussion. The "no red flags" framing was accurate within a specific scope. Whether that scope was adequately narrow is a governance question the released texts alone cannot resolve. The evidence does not support the viral claim's strong version — deliberate concealment of known causation — but it does leave open a weaker version: public reassurances that did not fully reflect the state of internal scientific uncertainty.

The 2021 NEJM denominator error

This thread is five years older than the August 2026 text release. In June 2021, the CDC published a study in the New England Journal of Medicine tracking 3,958 pregnant people enrolled in the V-safe Pregnancy Registry who had received mRNA COVID-19 vaccines.^[6]

Of 827 completed pregnancies, 104 ended in miscarriage: a rate of 12.6%. The paper noted its own limitation: 700 of those 827 had been vaccinated in the third trimester, a period when miscarriage risk falls sharply, and the study had not yet followed the full first-trimester cohort to the endpoint where miscarriage could occur.

NEJM issued a correction in October 2021, removing the 12.6% figure from a summary table. The correction noted that no valid denominator existed for a miscarriage risk estimate because follow-up through 20 weeks was unavailable for 905 of the 1,224 participants vaccinated before or during the first trimester.^[7] This was a methodological acknowledgment, not evidence of elevated risk.

A subsequent analysis of 2,456 participants found a miscarriage risk of 14.1%, within the documented background population rate of roughly 10-26%.^[8] The correction tightened the study's statistical precision; it did not reverse its conclusion.

Where did "80%" come from?

In late 2021, a paper in Science, Public Health Policy, and the Law by two New Zealand researchers, Simon Thornley and Aleisha Brock, claimed that over 80% of vaccinated pregnant women had miscarried.^[9] This figure circulates today, often alongside the August 2026 text release, as though it corroborates the Fauci story. It does not: it derived from a specific arithmetic error, and the paper was retracted.

The error is traceable. The NEJM study recorded 104 miscarriages from a total of 827 completed pregnancies (12.6%). The SPHPL authors removed the 700 third-trimester vaccinees from the denominator, leaving 127, and reported 104/127 = 82%. The problem is that removing every pregnancy that did not miscarry — and the third-trimester vaccinees were disproportionately those who had healthy births — guarantees an inflated miscarriage share in what remains.

A co-author of the SPHPL paper initiated retraction rather than publish a correction.^[10] Full Fact, PolitiFact, and others documented the flaw in November 2021, within days of publication.^[11][12] The 82% figure has recirculated in each subsequent wave of COVID vaccine controversy, most recently alongside the August 2026 text release.

What the research shows

Subsequent studies with larger and better-designed samples are consistent. A 2023 systematic review and meta-analysis in Human Reproduction pooled data from 21 studies including 149,685 women and found a risk ratio of 1.07 (95% CI 0.89–1.28) for miscarriage in vaccinated versus unvaccinated women: a difference that is not statistically significant and could easily reflect chance variation or confounding.^[13] The pooled miscarriage rate among vaccinated women was 9%, in line with pre-pandemic population estimates.

The authors noted that the certainty of pooled evidence remains low due to high heterogeneity and predominantly observational data, and called for larger, higher-quality studies. That caveat is appropriate scientific humility. It does not open the door to the viral claim: a pooled RR of 1.07 with a CI crossing 1.0 does not establish elevated risk.

The theoretical mechanism Fauci raised — that fever and cytokine response after second-dose mRNA vaccination could stress a first-trimester pregnancy — was a plausible hypothesis. It was not borne out by subsequent observational data. The American College of Obstetricians and Gynecologists continued recommending annual COVID-19 vaccination for pregnant people as of spring 2026.^[3]

Actors and stakes

The strongest case for concern

The most defensible version of this critique is about communication calibration, not causation.

In January 2021, Walensky acknowledged that vaccine safety data for pregnant women was "data free." Fauci flagged a plausible theoretical mechanism on January 25. Nine days later, he was publicly describing official findings as showing "no red flags." Sixteen days later, he said the same about 20,000 vaccinated pregnant women.

Those two registers — private scientific uncertainty and public "no red flags" — are not strictly contradictory. "No adverse signals detected" and "we don't yet have full data" can coexist when a safety monitoring system is operating and nothing alarming has emerged. But for a pregnant person deciding whether to vaccinate, the two framings do different work. "No red flags" reads as reassurance grounded in established evidence. "Data free" describes a situation where the absence of bad news may partly reflect a short observation window.

Informed consent, at its fullest, would include both signals: that no safety signals had appeared in vaccinated pregnant women and that the pregnancy data was still being collected and would take months to mature. Whether that fuller framing was possible amid simultaneous vaccine hesitancy pressure, a genuine early-pandemic data lag, and a public communication environment hostile to nuance is a fair question. The texts establish that officials were wrestling with real uncertainty in private. They do not establish that officials chose concealment over caution.

What to watch

Senators Johnson and Paul have indicated they will continue releasing material from the 34,000-message archive. The first release covered 11 texts and one specific three-person exchange. Additional releases may reveal whether the theoretical miscarriage concern was escalated within CDC or NIH, formally documented in V-safe monitoring protocols, or addressed in subsequent internal communications as the safety data matured.

Congressional hearings on the text archive appear likely. The relevant inquiry is whether any formal action was taken in response to the January 25 concern, and what documentation exists of how officials weighed the theoretical signal against the absence of observed red flags in the subsequent months of monitoring.

The timing of the release coincides with fall 2026 COVID vaccination recommendations. How the resulting controversy affects vaccine uptake among pregnant people in the coming months will be measurable through V-safe or equivalent registry data. That effect on actual vaccination decisions is arguably more consequential than the historical question of who knew what in January 2021.