This week brought another spate of predictable headlines about a worthless paper (Stock, S.J., Carruthers, J., Calvert, C. et al. SARS-CoV-2 infection and COVID-19 vaccination rates in pregnant women in Scotland. Nat Med (2022), published in a prestigious, peer-reviewed journal. Although the paper’s defects are glaring, its message supports current policies.
Surprisingly, this is not the usual case of reporters misrepresenting results. The paper is explicit in its misinformation. The abstract ends with, “Addressing low vaccine uptake rates in pregnant women is imperative to protect the health of women and babies in the ongoing pandemic,” and the discussion begins with, “Our findings emphasize the need for continued efforts to increase vaccination uptake in pregnant women, especially in younger and more deprived populations.” (As an aside, note how verbose passive constructions are used to make the advice seem more authoritative and scientific. If the authors had written in simple English, “We should vaccinate more pregnant women for their own good and that of their babies,” it would have sounded more like the editorial it is.)
This is an issue beyond life-and-death, it’s double life-and-death for a pregnant woman considering getting vaccinated for CoVID-19. Solid support for the paper’s editorializing would require carefully controlled studies taking at least three years in order to assess early childhood development. It took five years to notice the severe damage to unborn children from thalidomide.
We are only now getting basic statistical data to begin assessing the consequences of vaccine in pregnancy, we are very far from having strong scientific evidence about net positive and negative effects. Yet questioning the wisdom of vaccinating pregnant women is considered “misinformation.” The American College of Obstetrics and Gynecology threatens sanctions including loss of license for doctors who disagree with the consensus. Advising pregnant women to think for themselves on the issue can get you banned on Facebook and other sites. Fact-checking sites will call you a liar.
Before going farther, let me say that I got vaccinated and boosted at the first opportunity. If my daughter had been pregnant at the time, I would have urged her to do the same. I respect the educated guesses from most medical professionals that the benefits of vaccination for the average pregnant woman outweigh the risks. In the absence of good controlled studies that follow up through early childhood, we have no choice but to rely on educated guesses.
Nevertheless, it is essential to preserve open discussion of the issues. Thalidomide was not exposed by medical authorities but by dissident doctors who had to battle the pharmacological-regulatory-public-health establishment.
CoVID-19 vaccines were not tested on pregnant women prior to approval. The public health advice to vaccinate pregnant women was based on studies in pregnant animals, and data from a few women who became pregnant inadvertently during trials—none of whom had given birth at the time. Since then we’ve accumulated information, mostly reassuring, that the vaccine is safe and effective during pregnancy, and does not lead to a very high frequency of very severe adverse outcomes. But it will take more time, and better data, to rule out less frequent and less severe outcomes that might outweigh—at least for some women—vaccine benefits.
That brings me to the paper that stimulated me to write this essay, “SARS-CoV-2 infection and COVID-19 vaccination rates in pregnant women in Scotland.” The main problem is a very common one in vaccine research that attracts news coverage. It asks only, “If you get CoVID, are you better off having been vaccinated?” not, “Are vaccinated people in general better off than non-vaccinated?”
For the general population, the negative effects of vaccination seem to be between 1% and 10% as bad as getting CoVID. Granted that’s a vague statement, how do you compare different types of outcomes?, and it has a wide margin of error. Shockingly, we don’t have good data on the question, and people who try to study it come to very different conclusions. Nevertheless, I feel confident that no one can demonstrate that it’s lower than 1%, and there’s no good evidence that suggests it’s higher than 10%.
Therefore, getting vaccinated makes sense for an average person who has more than a 10% chance of getting CoVID over the period the vaccine is effective, but not if you think your chance is less than 1%. About 20% of the US population has had CoVID in two years. If you assume vaccines protect you for six months, that means the risk of the vaccine is roughly the same order of magnitude as its benefits for the average person.
But there are huge differences among individuals. Elderly people with health issues and lots of CoVID exposure benefit tremendously from vaccines. Young, healthy people who are careful about exposures could easily have overall negative expected health value from vaccines, especially people with previous CoVID infections.
On the other hand, all vaccines are beneficial from a public health perspective. Any bad consequences you suffer from getting vaccinated are limited to you, but if you get CoVID, you can pass it on to others, even if you have no or mild symptoms (in fact, the milder your symptoms, the more chance you’ll pass it on).
This is the motivation for suppressing discussion of the personal benefits of vaccination. If people can be convinced that vaccines are clearly beneficial for all individuals (other than some rare exceptions), as well as essential for public health, then vaccine mandates are easier to sell. On the other hand, if you want to force people to accept injections with net negative personal health expectations in order to protect the lives of others, you’re on shakier political grounds.
Therefore, we see wide touting of studies demonstrating that the vaccine helps if you get CoVID, while studies that adjust for negative outcomes of all vaccinated people—including those who did not get CoVID—are downplayed, forced to retract or don’t get submitted for publication in the first place.
If “SARS-CoV-2 infection and COVID-19 vaccination rates in pregnant women in Scotland” had stopped with the common error, it would still have been worse than the usual run. Although tracking data of adverse effects of vaccines for non-pregnant people is very poor quality, at least some exists. Adverse effects for pregnant women can appear in their children years in the future, so the data cannot exist even in principle.
But this is only the beginning of the paper’s problems. It loses any claim to science by failing to control for anything. It looks at outcomes for vaccinated versus unvaccinated women with CoVID and attributes the difference to vaccination alone. But we know vaccinated women are richer and healthier than unvaccinated women, and have healthier lifestyles and better healthcare. It seems likely that vaccinated women are more likely to test for CoVID, while unvaccinated ones may only be diagnosed when they have some other health problem.
Another major control issue is both vaccination rates and CoVID infection rates varied widely over the study period. CoVID varieties changed as well. If more women got vaccinated at times of lower infection rates with milder varieties, it would overstate the value of vaccination (unlike the first control problem, this one could go either way).
These issues should have caused the paper to fail peer review, or at least to have the authors remove the strong, unqualified medical and public health advice. There is some value in publishing the raw data so more careful people could use it to form their own conclusions, but it’s not enough to justify a paper, and certainly not the editorial statements in this one.
I also have a more technical criticism of the paper. This is an observational study, not a controlled experiment. It has huge data issues because the authors try to track every pregnancy in Scotland. That necessarily means it has big data issues—records that don’t match up, duplicate records, errors, people who get pregnant in Scotland but leave and people who come to Scotland pregnant, unreported pregnancies, etc. The authors don’t know if a hospitalization or other bad outcome results from CoVID, a vaccine complication or something unrelated. Researchers sometimes accept these kinds of data problems in order to get the largest possible sample size. But the authors of this paper (and this is unfortunately a common practice) report 95% confidence intervals as if they did a controlled experiment with perfect data. This wildly overstates the accuracy and precision of their findings.
The authors gathered a large dataset and verified the unsurprising (but still good to know) fact that getting CoVID without a vaccine is not good in the short run for pregnant mothers or their unborn children. There are severe problems with the study but that result accords with common sense and seems strong enough that it would survive controls and better data.
The study makes no attempt to weigh the benefits of vaccination for those who do get CoVID against the harms of vaccination. It makes no effort to assess the effect on early childhood development—the most serious concern that dissuades many pregnant women from getting the vaccine. It considers only average women, it does not adjust for different circumstances such as age, general health or chance of getting CoVID. It does not mention these essential qualifications anywhere. So it’s not even an attempt to support the advice it gives for all pregnant women.
On top of that, the study makes to attempt to control, uses poor data and overstates its confidence. Yet it passed peer review in a prestigious journal and was widely and uncritically reported in the news—and will likely feature in official public health communications and be used to suppress open discussion and support forced vaccination.
