LINKED in a recent issue of Global Health NOW with the line: ‘Yep, a mom’s COVID shot during pregnancy protects her baby, a large study finds’ was an article from the NPR (National Public Radio) webpage.
The study comes from the medical journal Pediatrics where the article is more modestly titled: ‘COVID-19 mRNA Vaccination in Pregnancy and Risk of Infection in Early Childhood’. And indeed it should be titled modestly, as the claims amplified in the NPR article barely stand up to scrutiny.
The conclusion to the abstract of the Pediatrics article does state: ‘COVID-19 vaccination during pregnancy protects the offspring against hospital contact for COVID-19 during the first six months of life.’ But as the editor of an academic journal, I know that abstracts often make claims not wholly supported by the remainder of the article. And that is certainly the case with this study.
Sidestepping the issue of whether pregnant mothers should be vaccinated with an experimental gene therapy, the authors of the article, a Norwegian team, are not wrong in their observations, but how significant is the purported protection? Well, not very.
The data in the paper, from an impressive group of over 100,000 infants, relies on an analysis called Cox regression. This method, which can accommodate different groups of people studied for different lengths of time, produces hazard ratios. These tell you how much more quickly an event is likely to happen in one group over another.
The outcome of such an analysis can be expressed in ‘person-days’, which is the number of persons studied for a certain number of days. For example, one person observed for one day is one person-day; ten people observed for one day is ten person-days.
Crucially, however, hazard ratios do not provide information about risk, either relative or absolute. Moreover, they can look impressive even when the event being studied is rare and the absolute difference between groups is tiny.
From the data in the Norwegian study, it is not possible accurately to calculate relative or absolute risk ratios. However, they can be estimated. As ever, the relative risk reduction looks quite impressive at 35 per cent reduction in risk, but again estimated, this translates to an absolute risk reduction of approximately only 0.0215 fewer cases per 10,000 person-days. This means that for every 10,000 days of observation (across all people), the vaccinated group had an estimated 0.0215 fewer cases than the unvaccinated group.
It gets worse. The authors, it must be admitted, do take into account a range of factors which may account for their observations, and are frank about the limitations of their study. One of these is the ‘healthy vaccinee effect’ whereby healthy people are more likely, paradoxically, to seek vaccination. Using modelling (alarm bells, surely) they do try to adjust for the healthy vaccinee effect, but as its precise influence cannot be enumerated, the modelling did not fully resolve the issue.
Other factors were considered, such as the nationality of the mothers (Norwegian or not Norwegian), the level of income and education, and health-seeking behaviours. When this was done, by removing data from mothers who remained unvaccinated for the duration of the study, according to the authors, the already small effect of covid vaccination ‘was attenuated . . . suggesting that unmeasured confounding from differences in health-seeking behaviour may have biased the association’.
An additional observation regarding infections other than covid was that babies whose mothers were vaccinated were almost 5 per cent more likely to need a doctor due to an infection. Unable to accept that the covid vaccines could possibly be responsible for the higher infection rate in children of vaccinated mothers, an alternative explanation was sought. The observation had more to do with vaccinated mothers being the kind of people who were more likely to seek medical help than those mothers who were not vaccinated. Judgmental or what?
Commenting for NPR, Dr Kevin Ault, a professor of obstetrics and gynaecology at Western Michigan University, said: ‘If you’re more likely to get vaccinated during pregnancy, you’re probably more likely to take your newborn to the doctor to be checked out for those kinds of illnesses . . . There’s not really a biological mechanism to explain those findings.’ Really? I think I can.
