Possible Covid Vaccination and Infection Risks in Early Pregnancy

Recommending early vaccination is insufficiently evidence-based

A recent article on birth defects associated with Covid vaccination and infection in early pregnancy misused statistical significance testing. This is a common mistake among common methods mistakes, not proof of any bad faith or conspiracy. At the same time, such mistakes often help spin stories, with scientists frequently promoting the preferred causal narratives of powerful social networks under the umbrella of objectivity while doing a lot of hidden interpretative work under the guise of apparently neutral statistics, as statistician Sander Greenland argues (e.g., here and elsewhere).

Reporting on the article then further selectively emphasized (unproven) vaccine safety.

Selective Reporting

On Friday, JAMA Network published a “Medical News in Brief” about a recent scientific article in which the authors misused statistical significance testing to downplay possible early pregnancy Covid infection and vaccination risks.

Writer Emily Harris reported only point estimates (not confidence intervals) on only babies whose moms were vaccinated during the first trimester or not, writing:

About 1.5% of infants who were exposed to an mRNA vaccine during the first trimester were born with major structural birth defects... This compares with the roughly 1.4% of infants whose birthing parent was either not vaccinated, vaccinated before pregnancy, or vaccinated during later trimesters.

This misrepresents the article’s findings.

The article (“Covid-19 infection and vaccination during first trimester and risk of congenital anomalies: Nordic registry based study,” Magnus et al, BMJ, July 2024) found substantial possible associations between Covid vaccination and some major malformations. It also found such associations for Covid infection itself.

Misinterpretation

By misusing statistical significance testing, the article itself also got both parts of the story wrong. Evaluating “the risk of major congenital anomalies [in singletons from Mar. 2020 — or Jan. 2021, when vaccines became available — through mid-Feb. 2022] according to infection with or vaccination against covid-19 during the first trimester of pregnancy” using Swedish, Danish, and Norwegian national registries, the authors concluded “Covid-19 infection and vaccination during the first trimester of pregnancy were not associated with risk of congenital anomalies.”

Their findings, however, told a different story.

Vaccination Findings

The authors reported logistic regression on the registry data, adjusted for “maternal age, parity, education, income, country of origin, smoking, body mass index, chronic conditions, and estimated date of start of pregnancy,” found “the risk associated with covid-19 vaccination during the first trimester ranged from 0.84 (0.31 to 2.31) for nervous system anomalies to 1.69 (0.76 to 3.78) for abdominal wall defects.” (The authors did not and arguably should have reported unadjusted associations along with adjusted ones.)

These findings reflect substantial possible risks of first trimester Covid vaccination. The wide intervals including zero mean the result is not “statistically significant,” meaning there could be no true effect — or it could go either way. The possible risks estimated run from an up to 69% reduction in risk (0.31) of nervous system anomalies, to an up to 131% risk increase; and from an up to 24% risk reduction of abdominal wall defects, to an up to 278% risk increase.

These are substantial possible risks that cannot be dismissed as showing no association.

Infection Findings

Compare these numbers with the findings on possible risks of Covid infection during the first trimester: “When evaluating risk associated with covid-19 infection during the first trimester, the adjusted odds ratio ranged from 0.84 (95% confidence interval 0.51 to 1.40) for eye anomalies to 1.12 (0.68 to 1.84) for oro-facial clefts.” In other words, infection could reduce eye anomalies risk by up to 49% or raise it up to 40%; and reduce oro-facial clefts risk up to 32%, or increase it up to 84%.

Similarly, these are substantial possible risks that cannot be dismissed as showing no association.

Comparison

How do these possible risk spreads compare?

The intervals similarly hug zero. The effects (if they exist) could go either way. But the infection risk spreads might be interpreted as being both quantitatively and qualitatively less severe than those associated with first trimester Covid vaccination.

However, we might want to look at prespecified, longer-term health outcomes in affected individuals to make this kind of comparison. Even then, small numbers and quality of life considerations might still make such comparisons difficult or impossible to meaningfully make. More importantly, should these outcomes be compared?

Monocausal explanations of adverse pregnancy outcomes are implausible. If there are possible serious, substantial risks of both early Covid vaccination and early Covid vaccination, then the precautionary principle implies pregnant women might want to try to avoid both.

In reality, however, many women can’t or don’t want to isolate, mask, and do other prophylaxis to prevent Covid. So comparing the risks and benefits of early pregnancy vaccination and infection is important for a lot of women. Even if the comparisons contain uncertainties and approximations, this is something science should be doing for people.

Unless we don’t have the evidence to know… Which we don’t. Then do we take our best shot with caveats, or just throw up our hands and say we don’t know? There’s room for interpretation here.

Implications and Misinterpretations

My interpretation of these findings is that it may be safer for pregnant women to chance getting Covid than to get a first trimester Covid vaccine. We don’t really know whether first trimester infection or vaccination is riskier. But infection is uncertain while vaccination is certain. The possible risks of vaccination appear potentially much greater in magnitude, and I think they also appear worse in type. I could be wrong and, again, we don’t really know. What do we know?

These findings suggest both first trimester exposures may carry substantial risks of major birth defects. Despite public health initiatives encouraging vaccination including for pregnant women including in early pregnancy, we still don’t have enough information to establish the intervention’s safety, comparative or otherwise.

In the face of this continuing ambiguity, however, both the reporter and the study authors beat the Covid consensus drum. The authors do this in the context of misusing statistical significance testing, stating under Policy Implications:

We did not find any evidence of an increased risk of congenital anomalies after covid-19 infection… Furthermore, we did not find any indication that vaccination against covid-19 during the first trimester increased the risk of anomalies, providing additional evidence about the safety of vaccination in pregnant women. Overall, our findings support the current recommendations to vaccinate pregnant women against covid-19.

The authors may interpret their findings to support that consensus, but the data don’t have a policy opinion. Moreover, it is false to claim the authors found no evidence of increased birth defects associated with infection or vaccination. This mischaracterization of the presented evidence reflects statistical significance testing misuse.

An alternate interpretation based on the same evidence would be that the risks of first trimester Covid vaccination may outweigh the risks of first trimester Covid infection itself, especially given the observed mellowing trend in strains. But that it’s difficult to study this because of rarity, uncertainty, and time lags (pregnancy takes time; diagnosing malformations often takes longer). So we just don’t know.

The Power of Consensus Story

It’s easy enough to see why this alternate story didn’t get trumpeted. Compare Harris’s headline, “COVID-19 Vaccination in Early Pregnancy Not Linked to Birth Defects,” with the more accurate “Covid Vax in Early Pregnancy May Carry Substantial Birth Defect Risks, Maybe Not; May Be Riskier Than Catching Covid Itself, Maybe Not.” Aside from being more complex and ambiguous (like the available evidence), the latter risks being mislabeled anti-vax misinformation by breaking with the consensus on vaccination safety.

These ambiguous but concerning findings add, however, to other (similarly misinterpreted) evidence that Covid vaccines in pregnancy may carry serious safety risks. Take, for example, Denoble et al’s recent Covid vaccination-stillbirth finding (previously discussed here). If the relevant evidentiary standard for medical interventions in pregnancy is that the preponderance of evidence suggests they do no net harm, it seems worth questioning whether Covid vaccination in pregnancy meets that standard.

It’s also worth noting, at least when it comes to elective exposure to unknown Covid vaccination risks in early pregnancy, that norms about this vary…

First Trimester Experimentation: Not A Universal Norm

Here in Germany, there’s a different standard underlying a different first trimester Covid vaccinations policy and practice regime than in Scandinavia and the U.S. They don’t have analogous registry data here to do an analogous study. Because they don’t do first trimester Covid vaccination.

The Federal Ministry of Health (Bundesministerium für Gesundheit, BMG) advises:

Like other healthy adults between the ages of 18 and 59, it is recommended that pregnant people have a basic immunity. For the basic immunity, it is important that the immune system is exposed to pathogen components (through vaccination) or the pathogen itself (through infection) three times. At least one of these exposures should have occurred through vaccination. In addition to the basic immunity, pregnant people with an existing underlying illness are recommended to get a booster from the second trimester in autumn.

Pregnant people should only be vaccinated from the second trimester, and the vaccine should be Comirnaty by BioNTech/Pfizer.

The first trimester is the riskiest time vis-a-vis major malformations. Recommending later vaccination seems to apply the precautionary principle to not experiment where it’s clear we could do more harm than good; and the way we’d find out is by accidentally doing it. We don’t have the evidence to know what the net harm-benefit balance is here, and healthcare providers’ obligation is to “first do no harm.”

It’s not clear what the rationale is for the German vaccination recommendation for people who’ve had Covid 3+ times, or whether possible later vaccination risks (e.g., stillbirth) may outweigh benefits.

Uncertainty, Rarity, and Causality

That said, just because Covid vaccination in pregnancy correlates with substantial possible risks ranging from abdominal wall defects to stillbirth, doesn’t mean it’s very risky. There’s broad uncertainty in Magnus et al’s risk estimates. There could be no effect.

And since these are rare events and the calculated risks are relative, the absolute risks remain quite small.

Still, a fully preventable three-fold increase in a rare major birth defect should be prevented. Avoiding the possibly risky exposure — following Germany’s lead and not Covid vaccinating in the first trimester — seems like a no-brainer.

One could argue, however, that correlation isn’t causation. There could be a confound at work here, like more healthcare system contact, compliance with health authorities, and/or trust. This could drive both more Covid vaccination in pregnancy, and more use of other products the medical system is selling, like psychotropics. Common antidepressants and anti-anxiety drugs in the SSRI/SSNI class may pose substantial risks to pregnancies and developing offspring. Something like this could in turn explain Covid vaccination risk findings.

Similarly, better rest or nutrition in response to illness could drive associations between Covid infection in the first trimester, and some lower possible major malformation risks.

Again, there’s no reason to suspect one causal story explains the full picture here.

The Bottom Line

We don’t know what the net risks are of infection versus social isolation versus vaccination. People’s risk profiles, situational abilities, and tolerances will vary. A one-size-fits-all public health approach to Covid vaccination — recommending it early and often for pregnant women — is insufficiently evidence-based.

The precautionary principle implies avoiding first trimester exposure to potential teratogens. Perhaps this principle should apply to Covid vaccination, except where mothers have high-risk jobs or extremely risky comorbidities. Even then, it’s not clear from this study what best protects their kids.

The current consensus encouraging Covid vaccination early and often for everyone, including pregnant women, is neither universal nor supported by the best available scientific evidence.