The Miscarriage Texts

What Fauci and Walensky discussed privately in January 2021, and what pregnant women were told

The Miscarriage Texts

What Fauci and Walensky discussed privately in January 2021, and what pregnant women were told

---

This essay is written for both a general audience and for expert scientists and physicians interested in this topic. I have been trained as a pathologist and taught pathology, as well as molecular biology, to medical students for many years, and so it should be no surprise that some of the language here is rather technical.

For this I apologize to and empathize with general-interest readers, and ask that you bear with me so I can communicate precisely with the experts and policy specialists who may also read the essay. Keep in mind that until he left the federal government, even Fauci or one of his minions used to open and read some of our essays, and on occasion our work reaches and affects policymakers at the highest levels of government.

---

What was found in the placenta

In March 2026 a group at the University Hospital of Wuerzburg published what they had found in 106 term placentas collected between November 2020 and October 2022. Thirty-one were positive for SARS-CoV-2 spike protein by immunohistochemistry. The staining sat predominantly in Hofbauer cells, the fetal macrophages inside the chorionic villi, and in syncytiotrophoblast, with the trophoblast layer and the endothelium of villous capillaries also involved. Eleven of those positive placentas came from women who had been vaccinated and who reported no infection of any kind during the pregnancy.

No viral RNA was detected in any sample tested. Vaccine RNA was. A Comirnaty probe lit up cells of the decidual surface in one placenta. A Spikevax probe lit up the endothelium of villous capillaries in another, from a woman whose last dose preceded her pregnancy (Bartmann et al. 2026).

That study carries severe limits. It is descriptive work on fixed tissue and cannot establish mechanism. The cohort was drawn from uncomplicated births at term, and the authors state plainly that women with miscarriages were not included, so the study cannot speak to pregnancy loss at all. The RNA findings rest on two samples out of nine tested. And the literature is split. Santos and colleagues reported no spike glycoprotein in placentas after mRNA vaccination, and Prahl and colleagues excluded vaccine traces in twenty placentas by Western blot and PCR.

The finding establishes that the question is real, answerable by experiment, and answerable in human tissue. The compartments where the staining appeared are the compartments that express ACE2, the receptor for the protein being stained. ACE2 is a carboxypeptidase. It degrades angiotensin II to angiotensin 1-7, and in the placenta that reaction governs vasodilation, oxidative balance and the vascular remodeling on which perfusion depends.

None of this was known when the federal government moved from telling pregnant women they could choose vaccination to telling them they should have it. Nor had anyone looked. The pivotal trials excluded pregnant women, the preclinical biodistribution work was done in nonpregnant rats, and the reproductive toxicity studies measured outcomes rather than tissue distribution.

To be blunt, as a physician and scientist with long experience in biological product development and clinical research, I am shocked by this absence of critical data given the policy decisions that followed. This is an example of public health malpractice.

What did exist, in January 2021, was a conversation

On January 25, 2021, five days after the inauguration, Dr. Vivek Murthy, Biden’s nominee for Surgeon General and a senior adviser to the incoming administration’s COVID response, texted Anthony Fauci and newly installed CDC Director Rochelle Walensky. He asked whether any data or theoretical reason favored vaccinating early or late in pregnancy, and when more robust data on risk might arrive. He added that he had been telling pregnant women no concerning outcomes had appeared in the trials, while noting that the number of pregnant subjects was probably limited (The Hill 2026).

Walensky answered first, describing a thin evidence base and pointing to the V-safe pregnancy registry, which by then had enrolled more than 15,000 pregnant women (Tampa Free Press 2026). Fauci replied that there were “no data or theoretical reason” to prefer early over late vaccination (The Hill 2026). He added that some people, including female health care professionals, were uneasy about a genetic vaccine early in pregnancy, and that the concern about effects on genes was a misperception.

Roughly two hours later he wrote again. He had asked around further, and another issue had come up that the others needed to know about. Because many people experience significant cytokine response and fever after the second dose, this “theoretically could be associated with miscarriage in the 1st trimester” (CNN 2026). Walensky replied that it was a good point, especially after dose two (Forbes 2026).

Israel had reached the same caution six days earlier and acted on it. Its Ministry of Health recommended vaccinating pregnant women on January 19, 2021, restricted to the second and third trimesters, and extended the recommendation to all stages of pregnancy on February 1 (Guetzkow et al. 2025). A national health authority held a first trimester restriction in force during the week of that text exchange, then dropped it.

The affirmative federal recommendation that pregnant women be vaccinated was issued on August 11, 2021. Nothing in that exchange, and no equivalent of it, reached the public in the intervening seven months.

That is what makes these texts consequential.

The scandal is not that scientists discussed uncertainty privately. Scientists are supposed to discuss uncertainty.

The scandal is that the uncertainty may have disappeared from public view not because the science had answered the question, but because someone judged that the public health message required certainty.

Pregnant women deserved to know the difference. And they deserved to know it before they rolled up their sleeves.

What the public heard in the same weeks

On January 26 Murthy flagged the World Health Organization’s recommendation against routinely administering the Moderna vaccine to pregnant women. His stated concern was not whether WHO was correct. It was what the recommendation might do to pregnant women’s confidence in vaccination.

Fauci’s answer set out the standard he was applying. Where data are lacking in pregnant women, one must weigh potential risks against benefits. COVID can be serious in pregnancy. More than 10,000 pregnant women had been vaccinated with no issues reported. FDA, the CDC advisory committee and the American College of Obstetricians and Gynecologists all left the decision to the individual (CBS News 2026).

That position was defensible and it matched CDC’s clinical guidance, which held that a pregnant woman in an eligible group may choose vaccination. It is also not what he said in public a week later. Speaking with Howard Bauchner, then editor of JAMA, Fauci reported that FDA had found no red flags among the pregnant women vaccinated to that point, and said he would rather take his chances with the vaccine than with infection while pregnant (The Hill 2026).

The private discussion included a plausible mechanism for harm, specific concern about the second dose, and an acknowledgment that the necessary data did not yet exist. The public discussion emphasized the absence of red flags and offered a personal endorsement.

A week earlier Fauci had privately identified a theoretical risk of miscarriage. When he spoke publicly, that part of the risk-benefit calculation was gone. The uncertainty had not disappeared. It had disappeared from the message.

One mechanism, two registers

Fever is the clinical expression of cytokine release. The mechanism Fauci described privately in January was therefore not a private discovery, and that makes the sequence worse.

CDC’s clinical guidance in December 2020 already advised that pregnant women who develop fever after vaccination may be counseled to take acetaminophen, on the stated ground that fever has been associated with adverse pregnancy outcomes (Newsweek 2020). The agency had accepted the causal premise in print before the first pregnant woman was vaccinated. Fauci’s January text restated the same pathway in mechanistic language and added two specifics: that the response concentrates after the second dose, and that the outcome at issue is loss in the first trimester.

So the agency was not unaware. It published the countermeasure in December, discussed the mechanism privately in January, and by August had dropped the subject. The August 11 statement contained no fever caution of any kind. A risk CDC had acknowledged in its own guidance eight months earlier stopped being mentioned precisely when its recommendation became universal.

Evidence ran the other way too. The April 2021 registry analysis found that pregnant participants reported fever, chills, headache and myalgia less frequently than nonpregnant women of the same ages. Of the whole inflammatory pathway, reactogenicity was the one element CDC measured, and the measurement was reassuring. That finding belonged in the public conversation as much as the caution did, and neither appeared in August.

Since these texts became public the mechanism has been misdescribed in both directions. Post-vaccination fever is a finding consistent with a cytokine-mediated systemic response. In most recipients that response is transient reactogenicity and nothing more, and anyone claiming these vaccines routinely produce cytokine storm is misusing a term that names a defined pathological syndrome.

A pathologist asks a different question, and that question does not appear to have been asked at all.

Any biological insult applied across a large population produces a distribution of host responses. Clinically important events often occur in the tails of that distribution rather than at its mean. A reassuring average cannot by itself exclude an uncommon susceptible subgroup.

The relevant question was not whether the average pregnant woman experienced a cytokine storm. She plainly did not. The question was whether the inflammatory response following vaccination could become sufficiently pronounced in some subset of pregnant women to alter pregnancy risk, particularly during the first trimester, and if so whether those women could be identified in advance.

A woman with an underlying inflammatory or thrombophilic tendency, a history of recurrent pregnancy loss, or some other susceptibility is not the modal recipient. Her risk cannot be inferred from the modal result.

That was the safety question hiding inside Fauci’s January text, and before the recommendation became universal it deserved an answer.

Why there were no data

Pregnant women were excluded from the pivotal trials of both mRNA vaccines. That exclusion followed a convention in vaccine and drug development long predating 2020. Its consequence was unavoidable. At the moment of authorization there was no randomized evidence for pregnancy outcomes, and once mass vaccination began that missing randomized evidence could not be recreated after the fact.

What largely replaced it was V-safe, CDC’s voluntary smartphone-based surveillance system, with a pregnancy registry attached. Participants enrolled and reported their own outcomes. A registry of that design can be useful for detecting a large safety signal quickly. It is not a randomized trial. It cannot by itself establish a population rate without an appropriate denominator, it has no contemporaneous unvaccinated control group, and it necessarily reflects the women who enroll and continue reporting.

Those limitations do not make the data worthless. They define what conclusions the data can support.

On August 11 CDC told pregnant women the vaccines were safe and recommended vaccination. Twelve days later FDA approved Pfizer’s vaccine while still requiring the company to conduct the pregnancy study necessary to characterize pregnancy and infant outcomes.

That distinction matters, because much of the reassurance offered about vaccination during pregnancy between December 2020 and September 2021 rested on this kind of observational surveillance. The government was not reassuring pregnant women on the basis of evidence generated in pregnant women before authorization. It was vaccinating them first and accumulating the pregnancy safety evidence afterward.

A distinction has to be drawn here, because it is the first one a critic will reach for. The Nuremberg Code, the Declaration of Helsinki and the Belmont Report govern research on human subjects. A public health recommendation to use an authorized product is not a clinical trial, and invoking those instruments as though it were invites the reply that the categories do not match.

That reply is fair as far as it goes. It does not reach the actual problem.

Both products were under emergency use authorization for the entire period in question. Pfizer’s biologics license did not issue until August 23, 2021, twelve days after the pregnancy recommendation. And the emergency use statute carries its own disclosure duty. Section 564 of the Food, Drug, and Cosmetic Act, codified at 21 U.S.C. 360bbb-3, requires as a condition of authorization that individuals receiving the product be informed of the significant known and potential benefits and risks of that use, and of the extent to which such benefits and risks are unknown.

FDA’s own guidance explains why the clause exists. Informed consent as generally required under its regulations does not apply to an authorized product, so section 564 supplies conditions to ensure recipients are informed instead. Congress wrote that provision because a population receiving a product before its evidence base exists occupies a position the ordinary approval pathway does not contemplate. The ethical instruments are the reason the clause is in the statute. They are not being asked to do the work themselves.

The recipient fact sheets did note that data in pregnancy were limited. That is not the same disclosure. A general statement that evidence is thin conveys something quite different from the specific concern these officials exchanged privately: a named inflammatory mechanism, concentrated after the second dose, bearing on loss in the first trimester. The statute asks that recipients be told the extent to which the risks are unknown. That extent was known to them with some precision, and it was set down in a text message rather than conveyed to the women being vaccinated.

Suspending the default

Pregnancy medicine has traditionally operated from a precautionary default. When adequate evidence of safety is absent, exposure to a new drug or biological product is approached cautiously. Thalidomide and diethylstilbestrol helped build that posture, and modern federal labeling rules preserve the underlying principle by requiring labels to state when human pregnancy data are unavailable. The reason is obvious. Pregnancy is precisely the circumstance in which controlled experimentation is difficult and the cost of guessing wrong may fall on someone who never consented.

Fauci identified the principle that could override that default. Where data are lacking in pregnant women, he wrote on January 26, potential risks must be weighed against potential benefits.

That is the correct standard. Performing such a calculation requires both sides of the equation.

The harm term was largely unquantified because the pivotal trials had excluded pregnant women. The benefit term rested principally on the risk COVID posed to the mother, and CDC had attempted to measure that.

Zambrano and colleagues, reporting in November 2020 on 409,462 symptomatic women aged 15 to 44, found that pregnant women were more likely than nonpregnant women to be admitted to intensive care, at 10.5 versus 3.9 per thousand; to receive extracorporeal membrane oxygenation, at 0.7 versus 0.3 per thousand; and to die, at 1.5 versus 1.2 per thousand. The adjusted risk ratios were 3.0, 2.4 and 1.7 respectively.

Their own summary captured an important distinction. Although the absolute risks of severe outcomes were low, pregnant women were at increased relative risk for severe illness.

Both statements were true. Only the second travelled.

Consider mortality. The observed difference was 0.3 deaths per thousand symptomatic infections, or three additional deaths per ten thousand. CDC’s earlier June 2020 analysis had found mortality among pregnant women statistically indistinguishable from nonpregnant women of the same ages, with an adjusted risk ratio of 0.9.

The November report contained another comparison that received considerably less attention. Citing the 2009 H1N1 pandemic, its authors noted that the absolute risks of severe outcomes among pregnant women during H1N1 exceeded those then being observed with COVID-19.

There was another curious feature in the COVID data. The relative risks diminished as the endpoints became harder. ICU admission carried an adjusted risk ratio of 3.0, ECMO 2.4, and death 1.7. In the earlier June analysis death had carried a ratio of 0.9 and had not reached statistical significance.

That pattern does not prove bias. It should have raised an obvious question about how much of the apparent elevation reflected biology and how much reflected clinical decision-making.

ICU admission is not purely a biological endpoint. It is also a clinical decision, and obstetrics has long maintained a lower threshold for escalating care because the physician is managing two patients and fetal monitoring may be easier at higher acuity.

ECMO introduces another selection problem. Candidacy favors younger patients without major underlying disease, characteristics disproportionately represented among pregnant women. A higher rate of receiving ECMO does not translate directly into an equivalent increase in underlying disease severity.

The clinical environment could reinforce these effects. Throughout 2020 and 2021 physicians were repeatedly told that pregnancy was a high-risk condition for COVID-19, with these same surveillance figures offered as evidence. That creates the possibility of a feedback loop. Clinicians expecting greater risk may admit or escalate pregnant patients sooner, and those admissions then enter surveillance datasets as evidence of greater risk.

CDC acknowledged a version of this problem one level down. Its data could not reliably distinguish hospitalization because of COVID deterioration from hospitalization for another reason, including delivery. The COVID-NET hospitalization surveillance report from September 2020 illustrates it. Among 598 hospitalized pregnant women with laboratory-confirmed SARS-CoV-2 infection, 54.5 percent were asymptomatic on admission. Many had entered the hospital for obstetric care and were discovered to be infected through testing.

Two further limitations appear in the November report’s own footnotes. Death information was missing for 22.0 percent of pregnant women and 17.2 percent of nonpregnant women, with those missing outcomes assumed to represent survival. ECMO information was missing for roughly 78 percent of both groups, with missing values assumed to mean ECMO had not occurred. The missingness on the mortality variable differed between the groups.

None of this establishes that pregnancy carried no additional COVID risk. There were sound biological reasons to suspect that it might. Pregnancy reduces functional residual capacity, elevates the diaphragm, increases oxygen consumption, produces a hypercoagulable state, and alters immune function. Influenza had already demonstrated that respiratory infections can behave differently during pregnancy. Zambrano also restricted the analysis to symptomatic women, reducing part of the ascertainment problem.

The subsequent literature was not uniform either. A prospective single-center comparison of obstetric and non-obstetric COVID cohorts in India found no difference in ICU admission, length of stay or mortality, and significantly less need for ventilatory support among pregnant patients, although differences between the cohorts limit what can be inferred from that comparison (Kumari et al. 2022).

The point is not that CDC had demonstrated there was no additional maternal risk. The point is that it had not measured that risk with anything approaching the precision implied by the public message. Its estimates rested on endpoints partly determined by clinical judgment, were generated within a clinical environment already primed to regard pregnancy as high risk, and carried important missing-data and ascertainment limitations. Yet those estimates supplied the benefit side of the calculation used to overcome the precautionary default.

Those qualifications persisted into the September 2021 advisory. Its 161 deaths among more than 125,000 confirmed cases corresponds to roughly 1.3 deaths per thousand, broadly consistent with the absolute mortality Zambrano had reported a year earlier. Its more striking figure, that approximately 97 percent of hospitalized pregnant patients were unvaccinated, included hospitalization for COVID illness or for labor and delivery, as the advisory itself acknowledged.

What reached the public were principally the counts and the relative risks. The absolute rates, the uncertainties, the endpoint problems and the confounders remained largely buried in the underlying literature.

Meanwhile the other side of Fauci’s equation remained poorly characterized. The mistake was not literally assigning vaccine risk a value of zero. It was allowing an absence of adequate evidence of harm to function, increasingly, as evidence of safety.

That distinction is especially important in pregnancy. A risk-benefit calculation can tolerate uncertainty when uncertainty exists on both sides, but it cannot manufacture a missing quantity. If the potential benefit is estimated from imperfect observational evidence while the potential harm cannot yet be reliably measured, the result is not a conventional risk-benefit calculation. It is a decision made under profound uncertainty.

Profound uncertainty is precisely why the precautionary default exists. Invoking risk-benefit analysis to override a precaution designed for circumstances in which one side of that analysis cannot yet be measured turns the purpose of the precaution on its head.

There is an uncomfortable parallel in Fauci’s own public assessment of COVID risk. On February 8, 2020 his diary recorded a denominator-adjusted case fatality estimate of 0.2 to 0.3 percent. On February 28, writing with H. Clifford Lane and Robert Redfield in the New England Journal of Medicine, he suggested that the overall clinical consequences of COVID-19 might ultimately resemble those of a severe seasonal influenza, with a case fatality rate perhaps near 0.1 percent. Less than two weeks later, on March 11, he told the House Oversight Committee that COVID-19 mortality was approximately one percent and that the virus was therefore roughly ten times more lethal than seasonal influenza.

Those numbers were being generated from a rapidly evolving epidemic with an uncertain denominator, so changing estimates were inevitable. That is precisely the point. Uncertainty was treated as grounds for precaution when estimating the danger of the virus. When estimating the danger of vaccination during pregnancy, uncertainty increasingly became a reason for reassurance.

The paper, the denominator, and the correction

On April 21, 2021, Shimabukuro and colleagues published preliminary V-safe pregnancy findings in the New England Journal of Medicine. Among 827 participants whose pregnancies had reached a recorded outcome, 104 had experienced spontaneous abortion before 20 weeks. The paper presented that as 12.6 percent and described the proportion as similar to the background rate in the general population. In the accompanying table the authors placed a published incidence range of 10 to 26 percent beside it.

There was a serious denominator problem.

A footnote to the same table disclosed that 700 of those 827 women had received their first eligible vaccine dose during the third trimester. By definition those women had already passed 20 weeks of pregnancy and were therefore no longer at risk of an event defined as spontaneous abortion before 20 weeks. Including them in the denominator necessarily drives the apparent miscarriage proportion downward.

Put simply, the calculation asked how often an event occurring before 20 weeks happened in a group dominated by women who were already beyond 20 weeks when they entered the analysis. The resulting 12.6 percent was not a valid estimate of miscarriage risk following vaccination in early pregnancy.

Hong Sun raised this objection in the American Journal of Obstetrics and Gynecology on August 3, 2021. On September 8 three of the CDC authors responded in the New England Journal of Medicine and agreed with the essential point. Eight hundred twenty-seven completed pregnancies was not an appropriate denominator for estimating the risk or rate of spontaneous abortion, and such an estimate needed to be based on women vaccinated before 20 weeks of gestation (Meaney-Delman, Ellington, and Shimabukuro 2021).

On October 14 the journal formally corrected the article. The 10 to 26 percent background comparator beside the spontaneous abortion figure was replaced with “Not applicable,” removing the comparison that had made the original 12.6 percent figure appear directly comparable with the expected population rate.

There is an important arithmetic trap here, because the opposite error has circulated ever since.

If the 700 third trimester pregnancies are simply removed from 827, that leaves 127 completed pregnancies. Dividing 104 miscarriages by 127 produces approximately 82 percent. That number is also invalid.

It fails for the mirror-image reason. The March 30 cutoff preferentially captured early pregnancies that had already ended, while many women vaccinated early in pregnancy were still pregnant and therefore had no completed outcome to enter the denominator. Miscarriages occur earlier than live births, so counting only pregnancies that had already finished at an early cutoff inevitably overrepresents losses.

The original calculation diluted the numerator with women who could no longer miscarry. The 82 percent calculation does the opposite, excluding many pregnancies that had not yet had time to reach a successful outcome. Neither 12.6 percent nor 82 percent is a valid miscarriage rate.

The scientifically defensible conclusion from the April dataset was simpler and much less reassuring than the headline number suggested. At that point the registry did not yet contain enough appropriately followed early pregnancy exposures to calculate a reliable miscarriage rate.

August 11

CDC announced its affirmative recommendation on August 11, 2021, under the headline “New CDC Data: COVID-19 Vaccination Safe for Pregnant People.” The agency cited an analysis of roughly 2,500 women who had received at least one vaccine dose before 20 weeks of pregnancy. It reported no increased risk of miscarriage, with a rate near 13 percent, within the expected background range.

This was in substance the corrected calculation. It restricted the analysis to women who had actually been vaccinated during the period in which spontaneous abortion could occur, precisely the denominator problem critics had identified in the April paper. Its central finding has held up. Published as Zauche and colleagues on September 8, 2021, the analysis estimated a cumulative spontaneous abortion risk of 14.1 percent, or 12.8 percent after age standardization, among 2,456 participants. Subsequent studies encompassing more than a million pregnancies have likewise found no overall increase in miscarriage following COVID-19 vaccination.

The picture is not uniform, and honesty requires saying where it is not. A group working with Maccabi Healthcare Services in Israel examined 226,395 singleton pregnancies and reported that first doses given during gestational weeks 8 to 13 were followed by roughly 13 fetal losses per hundred exposed pregnancies against about 9 expected, an excess of 3.9 with a confidence interval of 2.55 to 5.14 (Guetzkow et al. 2025). Their design addresses the exact problem raised here about the V-safe cohort. Women vaccinated for influenza during pregnancy showed consistently fewer losses than expected, which the authors read as healthy vaccinee bias made visible, and women vaccinated before pregnancy came in at or below expectation.

It also carries real weaknesses. It remains a preprint fourteen months after posting, unpublished and not peer reviewed. The expected values come from a regression model trained on 2016 to 2018 data rather than from a concurrent control group, so pandemic-era differences in behavior, care and reporting sit outside the model. The authors themselves note that the registry cannot separate elective from medically indicated terminations, and that women vaccinated early may have been more likely to report a loss. Two of the authors now hold federal positions, Retsef Levi on the immunization advisory committee and Tracy Beth Hoeg at FDA, which will be used to discount the work.

One internal control in that paper deserves separate attention. Exposure to SARS-CoV-2 infection during pregnancy produced no significant excess of fetal losses over expected in the same dataset, including among unvaccinated women.

That evidence matters, on both sides. So do the limits of what each side can establish.

The cohort underlying a universal recommendation is therefore worth examining. Of the 2,456 participants in the CDC analysis, 88.8 percent were health care personnel and 78.3 percent were non-Hispanic white. More than one quarter reported at least one previous spontaneous abortion. Thirty-five women who reported a loss before six completed weeks were excluded from the analysis. That is a methodologically defensible choice, given how frequently very early pregnancy losses go unrecognized in the general population, but it is still a choice that belongs in any description of what was measured.

Composition determines more than whom a study represents. It also affects what the study is capable of detecting. A cohort composed overwhelmingly of working health care personnel is not necessarily representative of the entire pregnant population to whom a universal recommendation will apply. More importantly, a study of approximately 2,500 pregnancies can provide useful reassurance about a substantial increase in overall miscarriage risk while remaining poorly equipped to identify a smaller increase confined to an uncommon susceptible subgroup. Those are different questions.

The timing presents another problem. CDC announced the conclusion four weeks before the analysis was published, meaning the public received the answer before outside scientists could examine the methods and the data underlying it. The announcement also arrived while the April New England Journal paper still stood uncorrected, with its inappropriate denominator and population comparator remaining in print until October 14.

Walensky’s August 11 statement urged all pregnant people, those considering pregnancy, and those breastfeeding to be vaccinated, describing COVID-19 vaccines as safe and effective during the urgency of the Delta wave (CDC 2021a).

Notice the distance between the evidence and the claim. The new analysis addressed one outcome, spontaneous abortion before 20 weeks, in one voluntary self-reported surveillance cohort. The public conclusion was that COVID-19 vaccination was safe in pregnancy. Those are not equivalent statements.

On September 29 CDC followed with a health advisory, CDCHAN-00453, calling for urgent action. Its summary stated that CDC “strongly recommends COVID-19 vaccination either before or during pregnancy because the benefits of vaccination outweigh known or potential risks” (CDC 2021b). The advisory reported more than 125,000 laboratory-confirmed cases in pregnant people, more than 22,000 hospitalizations and 161 deaths as of September 27, with 22 of those deaths in August alone, and noted that roughly 97 percent of pregnant people hospitalized with confirmed infection were unvaccinated. Those figures are serious and they are the strongest part of the case CDC made. The advisory now carries a retired designation in the agency’s own document archive.

The question that was not asked

Fauci’s concern ran from cytokine response and fever to first trimester loss. Fever in early pregnancy is a recognized risk, so the pathway is real, and it is the pathway an immunologist reaches for. It was also the only pathway anyone in that exchange raised.

The antigen these vaccines instruct cells to produce carries biological activity of its own. Spike is a receptor-binding protein, and its receptor had already been mapped to the tissue at issue. Li and colleagues, publishing in PLOS ONE in April 2020 from a preprint that February, reanalyzed roughly 65,000 cells from placentae of six to fourteen weeks and found ACE2 expressed in decidual stromal cells, decidual perivascular cells, villous cytotrophoblast and syncytiotrophoblast. Ashary and colleagues reported that July that syncytiotrophoblast carried the highest proportion of ACE2-expressing cells in the first trimester placenta, and noted that extravillous trophoblast, which acquires ACE2 expression later, is the population that invades the maternal decidua and remodels the spiral arteries.

That map was not new even then. Valdes and colleagues had localized ACE2 and angiotensin 1-7 to syncytiotrophoblast, cytotrophoblast, villous endothelium and vascular smooth muscle, and to invading trophoblast and decidual cells in the maternal stroma, in Placenta in 2006 (Valdes et al. 2006).

The two-proline modification does not change any of this. The substitutions sit at the junction of heptad repeat 1 and the central helix, in the fusion machinery. The receptor-binding domain is untouched. What 2P prevents is the transition to the postfusion state, and receptor engagement occurs upstream of that transition. Wrapp and colleagues established the point in the paper that defined the antigen. Their cryo-EM structure showed the predominant state of the stabilized trimer with one receptor-binding domain rotated up in a receptor-accessible conformation, and their surface plasmon resonance measured ACE2 binding at roughly 15 nanomolar, some 10 to 20 fold tighter than SARS-CoV spike (Wrapp et al. 2020). Both measurements were made on the 2P construct. Corbett and Graham of the NIAID Vaccine Research Center are authors on it. The canonical statement that this spike binds ACE2 with high affinity is a measurement of the vaccine antigen.

Spike binding drives ACE2 into the endocytic pathway and lysosomal degradation, an effect mapped to both the receptor-binding domain and S2, and one that tracked with clinical severity across variants. A physiology review of ACE2 in pregnancy states the consequence directly. Downregulation of placental ACE2, whether by inhibition or by spike protein binding, produces placental inflammation and oxidative stress, limits fetal growth, and generates the classical picture of preeclampsia.

Whether the antigen reaches that tissue at concentrations approaching its affinity is the operative question, and the arithmetic constrains the argument. Ogata and colleagues measured circulating vaccine antigen after mRNA-1273 and found S1 detectable from day one, peaking on average near 68 picograms per millilitre (Ogata et al. 2022). At roughly 76 kilodaltons that is about 0.9 picomolar, some four orders of magnitude below the 15 nanomolar dissociation constant. In bulk plasma receptor occupancy would be negligible. The question lives in local tissue concentration, not in blood.

The Wuerzburg placentas are not the only place investigators have found the material since. A team at NYU Langone reported, in a research letter funded by the Eunice Kennedy Shriver National Institute of Child Health and Human Development, vaccine mRNA in human placental tissue by in situ hybridization, with signal in the decidua and in the villi from recipients of both products (Lin et al. 2024). In pregnant mice given intramuscular mRNA-1273, the construct crossed the placenta within an hour, reached fetal circulation, accumulated in fetal tissue and was translated into spike protein there (Chen et al. 2025). Lipid nanoparticles resembling those in the vaccines transfect placental trophoblast, endothelium and immune cells, which is why an entire field now engineers them deliberately for placental delivery.

The decidua and the villi are where the mRNA was found. They are also where ACE2 sits. What the antigen does to decidual stroma, decidual perivascular cells, villous cytotrophoblast and syncytiotrophoblast has not been characterized, and it was not characterized before pregnant women were told the vaccines were safe for them. The receptor map dated to 2006, the affinity measurement to February 2020 from the antigen’s own designers, and the experiment joining them to the tissue has still not been done.

The same receptor distribution ties together presentations that were handled as separate topics. Decidual stromal and perivascular cells express ACE2. The decidua governs menstrual bleeding, sheds whole in decidual cast shedding, and fails in early pregnancy loss. Through the spring of 2021 women were reporting all three. Kate Clancy and Katharine Lee gathered roughly 130,000 accounts of menstrual disturbance in the same months the pregnancy registry was being assembled. The menstrual reports went to content moderation and to a survey run by two anthropologists. The pregnancy question went to a registry. No federal analysis treated them as observations on one organ.

Redfield

Robert Redfield left the CDC directorship on January 20, 2021, five days before that text chain, and is not a participant in it. He is the third author on the February 2020 New England Journal editorial with Fauci and Lane, and he ran the agency when ACIP voted in December 2020 and when the permissive guidance issued.

On August 11, 2026, responding to the released texts, he said he never thought pregnant women should be vaccinated, described it as a policy recommendation he had not favored, and called it an error in judgment (NewsNation 2026). He said he has seen no data establishing a link between the vaccine and miscarriage. His objection was to the risk-benefit reasoning and, more sharply, to the absence of the open discussion that would have let women decide for themselves.

Redfield’s current positions on mRNA vaccines contradict the account that academic public health and the obstetric professional societies still promote, and that CDC and FDA officials promoted through 2024. That makes it easy for critics to dismiss his 2026 comments as retrospective dissent.

But there is a contemporaneous record against which his claim can be tested. When Redfield left CDC on January 20, 2021, the agency he had been running did not recommend COVID vaccination for all pregnant women. It explicitly left the decision to them. Five days later Fauci privately raised the theoretical possibility of first trimester miscarriage. Seven months later CDC recommended vaccination for every pregnant woman.

What the record supports

No one concealed a known harm. The mechanism Fauci raised has not been borne out as a miscarriage signal, and the large studies that followed have not found an overall increase in miscarriage after COVID-19 vaccination.

It does not dispose of what the documentary record shows.

Three of the most influential officials shaping the federal COVID response privately acknowledged that pregnancy data were inadequate. Fauci raised a specific theoretical concern about first trimester miscarriage following the inflammatory response to vaccination, particularly after the second dose. Walensky immediately recognized the point. That specific concern was never communicated to pregnant women.

The reassurance offered in the months that followed rested in part on a miscarriage calculation that the agency’s own authors later acknowledged used an inappropriate denominator. The corrected analysis was considerably more reassuring and its central finding has held. But CDC announced its affirmative recommendation four weeks before that analysis was published, and therefore before independent scientists could examine it.

And the headline went considerably further than the evidence beneath it. “New CDC Data: COVID-19 Vaccination Safe for Pregnant People.” The new data addressed one outcome, miscarriage before 20 weeks, in a voluntary surveillance registry heavily composed of health care personnel. It did not establish the absence of every potential pregnancy risk, nor could a cohort of that size exclude a risk concentrated in a small susceptible subgroup. A limited and reassuring finding about one outcome became a categorical statement about safety in pregnancy.

The studies that were supposed to fill the evidentiary hole did not fill it. Pfizer announced a randomized pregnancy trial in February 2021 with a planned enrollment of approximately 4,000 women. Enrollment was terminated on October 25, 2021 with roughly 350 participants, and the label records 346 in the maternal safety population. Dosing occurred at 24 to 34 weeks of gestation, so the first trimester concern Fauci had raised privately was never tested at all. In the limited data that trial did produce, Pfizer reported congenital anomalies in 8 of 156 vaccinated participants against 2 of 159 in placebo, a numerical imbalance in a sample far too small to interpret in either direction.

Pfizer attributed the early termination to the difficulty of enrolling into a placebo-controlled study once universal vaccination of pregnant women had been recommended. The recommendation issued in the absence of randomized evidence became the reason the randomized evidence could no longer be generated.

It is also unclear whether ACIP ever voted to recommend COVID-19 vaccination specifically for pregnant women in 2021. No corresponding 2021 Morbidity and Mortality Weekly Report recommendation has been identified, and the pregnancy guidance instead appears in CDC’s Interim Clinical Considerations, a staff document that changed repeatedly over time. An inquiry by Malone.News is now with CDC and the agency has said it is looking into the question.

The answer matters. If ACIP did vote on the pregnancy recommendation, the record should be straightforward to produce: the meeting, the motion, the vote, the evidence presented, the conflicts disclosed, and the resulting recommendation. If it did not, then CDC appears to have moved from individual choice to a universal recommendation for a population excluded from the pivotal trials without a specific vote of the independent advisory committee normally expected to examine vaccine recommendations in public. That would mean the most consequential change in pregnancy vaccine policy during the pandemic occurred without the public deliberation, recorded vote, conflict disclosures and recusal procedures the advisory process exists to provide.

Redfield identified the deeper failure correctly. Government scientists will discuss uncertainties privately, and they should. Medicine advances because scientists argue about mechanisms, challenge assumptions, worry about signals, and ask questions for which they do not yet have answers.

The ethical failure is not that Fauci had doubts. It is that pregnant women were not permitted to share in those doubts while being asked to bear the consequences of the decision.

They were entitled to know that the pregnancy data were limited. They were entitled to know what was reassuring. They were entitled to know what remained unknown. And they were entitled to know that the government’s own senior scientists had discussed a theoretical risk that had not yet been resolved.

That is informed consent. It does not require government scientists to know every answer. It requires them to say so when they do not.

RWM/JGM

References

Ashary, Nancy, Anshul Bhide, Priyanka Chakraborty, et al. 2020. “Single-Cell RNA-seq Identifies Cell Subsets in Human Placenta That Highly Expresses Factors Driving Pathogenesis of SARS-CoV-2.” Frontiers in Cell and Developmental Biology 8: 783.

Bartmann, Catharina, Vanessa Schmidt, Michael Moerz, Michael Schwab, Monika Rehn, Bettina Blau-Schneider, Achim Woeckel, and Ulrike Kaemmerer. 2026. “Detection of Spike Protein in Term Placentas of COVID-19 Vaccinated and/or SARS-CoV-2 Infected Women.” PLOS ONE 21 (3): e0344185.

CBS News. 2026. “Fauci Discussed Potential Risks, Benefits of COVID Vaccine for Pregnant Women in 2021 Texts Released by GOP Senators.” August 11.

CDC. 2021a. “New CDC Data: COVID-19 Vaccination Safe for Pregnant People.” Media statement, August 11.

CDC. 2021b. “COVID-19 Vaccination for Pregnant People to Prevent Serious Illness, Deaths, and Adverse Pregnancy Outcomes from COVID-19.” CDC Health Alert Network, CDCHAN-00453, September 29.

Chen, J.-C., M.-H. Hsu, R.-L. Kuo, L.-T. Wang, M.-L. Kuo, L.-Y. Tseng, et al. 2025. “mRNA-1273 Is Placenta-Permeable and Immunogenic in the Fetus.” Molecular Therapy Nucleic Acids 36 (1): 102489.

CNN. 2026. “Republican Senators Release Fauci’s Text Messages Discussing Safety of Covid Vaccine and Pregnancy.” August 11.

Delahoy, Miranda J., Lindsay Whitaker, Alissa O’Halloran, et al. 2020. “Characteristics and Maternal and Birth Outcomes of Hospitalized Pregnant Women with Laboratory-Confirmed COVID-19, COVID-NET, 13 States, March 1 to August 22, 2020.” MMWR Morbidity and Mortality Weekly Report 69 (38): 1347-54.

Ellington, Sascha, Penelope Strid, Van T. Tong, et al. 2020. “Characteristics of Women of Reproductive Age with Laboratory-Confirmed SARS-CoV-2 Infection by Pregnancy Status, United States, January 22 to June 7, 2020.” MMWR Morbidity and Mortality Weekly Report 69 (25): 769-75.

Fauci, Anthony S., H. Clifford Lane, and Robert R. Redfield. 2020. “Covid-19: Navigating the Uncharted.” New England Journal of Medicine 382 (13): 1268-69.

Forbes. 2026. “Republican Senators Release Fauci Texts on Safety of COVID-19 Vaccine for Pregnant Women.” August 11.

Guetzkow, Josh, Tal Patalon, Sivan Gazit, Tracy Beth Hoeg, Joseph Fraiman, Yaakov Segal, and Retsef Levi. 2025. “Observed-to-Expected Fetal Losses Following mRNA COVID-19 Vaccination in Early Pregnancy.” medRxiv preprint 2025.06.18.25329352, posted June 20. Not peer reviewed.

Kumari, Anupama, Latika Chawla, Anissa Atif Mirza, et al. 2022. “Impact of Pregnancy on Susceptibility and Severity of COVID-19: A Hospital-Based Prospective Observational Study.” Cureus 14 (4).

Li, Mengmeng, Lin Chen, Jing Zhang, Chunlin Xiong, and Xi Li. 2020. “The SARS-CoV-2 Receptor ACE2 Expression of Maternal-Fetal Interface and Fetal Organs by Single-Cell Transcriptome Study.” PLOS ONE 15 (4): e0230295.

Lin, X., B. Botros, M. Hanna, E. Gurzenda, C. M. De Mejia, M. Chavez, et al. 2024. “Transplacental Transmission of the COVID-19 Vaccine Messenger RNA: Evidence from Placental, Maternal, and Cord Blood Analyses Postvaccination.” American Journal of Obstetrics and Gynecology 230 (6): e113-16.

Meaney-Delman, Dana M., Sascha R. Ellington, and Tom T. Shimabukuro. 2021. Reply, “On Preliminary Findings of mRNA Covid-19 Vaccine Safety in Pregnant Persons.” New England Journal of Medicine, September 8.

NewsNation. 2026. “It Was a Mistake to Vaccinate Pregnant Women: Dr. Robert Redfield.” Katie Pavlich Tonight, August 11.

Newsweek. 2020. “COVID Vaccine and Pregnancy: What You Need to Know.” December 17, quoting CDC clinical guidance then in force.

Ogata, Alana F., Chi-An Cheng, Michael Desjardins, et al. 2022. “Circulating SARS-CoV-2 Vaccine Antigen Detected in the Plasma of mRNA-1273 Vaccine Recipients.” Clinical Infectious Diseases 74 (4): 715-18.

Pfizer. 2021-2024. Protocol C4591015, “A Phase 2/3 Study to Evaluate the Safety, Tolerability, and Immunogenicity of BNT162b2 Against COVID-19 in Healthy Pregnant Women 18 Years of Age and Older.” NCT04754594, with final clinical study report synopsis.

Prahl, Mary, Yarden Golan, Arianna G. Cassidy, et al. 2022. “Evaluation of Transplacental Transfer of mRNA Vaccine Products and Functional Antibodies during Pregnancy and Infancy.” Nature Communications 13: 4422.

Santos, Ana, Meredith Sauer, Alexander J. Neil, et al. 2022. “Absence of SARS-CoV-2 Spike Glycoprotein Expression in Placentas from Individuals after mRNA SARS-CoV-2 Vaccination.” Modern Pathology 35 (9): 1175-80.

Shimabukuro, Tom T., Shin Y. Kim, Tanya R. Myers, et al. 2021. “Preliminary Findings of mRNA Covid-19 Vaccine Safety in Pregnant Persons.” New England Journal of Medicine 384: 2273-82. Correction published October 14, 2021.

Sun, Haiyan. 2021. “Adjustment Is Required to Calculate the Risk of Early Pregnancy Loss with COVID-19 Infection or Vaccination.” American Journal of Obstetrics and Gynecology, August 3.

Tampa Free Press. 2026. “Newly Released Fauci Texts Reveal Internal Early Discussions on COVID Vaccines and Pregnancy.” August 11.

The Hill. 2026. “Paul, Johnson Release Fauci Texts.” August 11.

Valdes, Gloria, L. A. A. Neves, L. Anton, et al. 2006. “Distribution of Angiotensin-(1-7) and ACE2 in Human Placentas of Normal and Pathological Pregnancies.” Placenta 27 (2-3): 200-207.

Wrapp, Daniel, Nianshuang Wang, Kizzmekia S. Corbett, et al. 2020. “Cryo-EM Structure of the 2019-nCoV Spike in the Prefusion Conformation.” Science 367 (6483): 1260-63.

Zambrano, Laura D., Sascha Ellington, Penelope Strid, et al. 2020. “Update: Characteristics of Symptomatic Women of Reproductive Age with Laboratory-Confirmed SARS-CoV-2 Infection by Pregnancy Status, United States, January 22 to October 3, 2020.” MMWR Morbidity and Mortality Weekly Report 69 (44): 1641-47.