The FranceSoir site, relay of the most eccentric theses on the Covid since the start of the epidemic, published on November 18 a very relayed text whose seriousness you are asking us about. The title proclaims: “Anti-Covid-19 vaccine: a study of the Lancet shows the immune erosion of vaccinated people over the months.”
In the first lines of the text, we learn that the study in question is a scientific article published in February 2022: “Conducted in Sweden with 1.6 million individuals for nine months, [elle] showed that the protection conferred by the vaccine against symptomatic Covid-19 diminished quite rapidly” (So far, so good) “and that after six months, the most vulnerable people in the vaccinated group may be more exposed to the virus than their unvaccinated counterparts.”
The use of the conditional soon gives way to the affirmation of too much: “Immediately after the second dose, vaccinated people seem to be better protected against Covid-19 than non-vaccinated people. However, beyond six months and even more from eight months, the figures change and may show in some vulnerable people a non-significant effectiveness, or even a negative effectiveness. It is on these last words – which carry the whole alleged demonstration – that the author is mistaken and, in doing so, misleads his readers.
It is quite correct that, in this study, the efficacy of the two doses of vaccines against the circulating viral strains becomes, after several months, “statistically insignificant”. This means that after several months, the data collected no longer make it possible to distinguish whether the vaccination brings a benefit. In the tables presented in the study, this translates into results with a “confidence interval” whose limits have positive and negative values. The hypothesis according to which the vaccinated have greater protection than that of the non-vaccinated can no longer be favoured: the uncertainty attached to the results is too great (1).
If the two limits of this “confidence interval” had presented negative values, the result would again have become statistically significant… but “in the wrong way” : there would have been significantly more infected among the vaccinated than among the non-vaccinated, which would have raised legitimate questions. However, none of the results presented in the study correspond to this scenario.
In short: in this publication, the evaluation of the effectiveness of a double dose of vaccine ceases, after several months, to be statistically significant, but at no time does a deleterious effect of vaccination on immune function is highlighted.
The authors of the study denounce these errors of interpretation
Marcel Bailin, co-author of the study, confirms this point with CheckNews : “That’s right: estimates of negative vaccine effectiveness against infections of any severity at the end of the follow-up period were not statistically significant.” He thus describes – politely – “inaccurate” FranceSoir’s assertion that “the most alarming figures [de l’étude] are found in the most fragile subjects, namely people over 80 or those with comorbidities. The results presented for these ages, explains the researcher, “are subject to great uncertainty: the confidence intervals are very wide and overlap. Specifically, the vaccine efficacy was 5%, with a 95% confidence interval ranging from -53 to +41. So basically no significant effect, and a very large uncertainty. For this analysis, no efficacy was detected – which is not the same as saying there was a negative effect…”
“It should also be emphasized that this is an observational study, and that several forms of bias may explain why efficacy against infections of any severity declines, particularly over long follow-up periods”, continues Bailin. “For example, there could be a difference between vaccinated and unvaccinated in terms of health-seeking behavior, which could influence an individual’s decision to take a PCR test, which is important to consider. since the “any serious infection” result was based on the PCR test. If unvaccinated individuals were less likely to take a PCR test than vaccinated individuals, this would also contribute to the decline in vaccine effectiveness. (2).
Note that, for the same reason, if “statistically significant negative results” had appeared, they would not have been enough – on their own – to validate the hypothesis of an immune weakening of the vaccinated. The study measures, in short, the cumulative effect of vaccine protection and differences in behavior of the populations studied.
“Protection against more serious illness is maintained”
In their article published in February, the authors, far from warning against “immune erosion”, already wrote in black and white that their results “reinforce the evidence-based case for administering a third dose of vaccine as a booster, particularly to specific high-risk populations.”
Over the course of 2021, many sites relaying medical misinformation had already claimed that vaccination against Covid could cause immunodeficiency. Some have even relayed the fantasy of an “acquired immunodeficiency syndrome caused by the vaccine” (“VAIDS”), a sort of avatar of AIDS. In February 2022, Reuters returned to these accounts, and noted that the study newly published in The Lancet was already diverted from its meaning to serve this vision. Speaking to the British press agency, Peter Nordström, also co-author of the study, formulated a denial that many disinformants pretended not to hear: “Our study [montre] that protection against more severe disease is maintained, in stark contrast to any claims that our results would support claims that VAIDS exists.”
Bailin reports to CheckNews that, in the study published in February, the effectiveness of the vaccine against the serious disease “held up much better”, over the months, as the only protection against infection (2). Peter Nordström, also joined by CheckNewsnotes that this result, alongside those of other studies, “contradicts the idea that vaccines are harmful”. “With the same idea, he continues, claims sometimes encountered that recalls are harmful are contradicted by the results of the latest study we conducted on the elderlywhich show that a second reminder is associated with a reduced risk of death, compared to people who had only one reminder. […] It is clear that the overall effect of mRNA vaccines is beneficial.”
Disinformation already served by Fox News this summer
It should be noted that, to support his interpretation of the study published in February in The Lancet, FranceSoir refers to a letter published in June in the Virology Journal by a man named Kenji Yamamoto, “in which he gives his interpretation of this study of the Lancet». The readers of CheckNews have already heard of this text, which was quoted by Fox News host Tucker Carlson in a sequence of pure television disinformation. For, as we wrote then, we find in Yamamoto’s famous letter no demonstration of his interpretations.
As for the study of Lancetwe already said about him the main part of what we are rewriting today: “Authors fail [à] establish a significant difference between the over 80s vaccinated for more than six months and the non-vaccinated. It is impossible to establish that vaccinated people of this age, at this distance from their second injection, are more or less at risk of Covid than non-vaccinated people. After recalling the conclusion of the authors, favorable to a third dose of vaccine, we note “[qu’]by obscuring the elements that allow the data to be correctly interpreted […]Tucker Carlson is doing nothing but misinform – once again – his audience.” Conclusion which also applies to FranceSoir.
(1) The results are considered “statistically significant” if they are considered unlikely to have been obtained if the vaccine had, in fact, been ineffective. In a (very) schematic way, with “a 95% confidence interval”, there is only a 5% chance that by reproducing the trial on the scale of the general population, the real result will be outside of the interval. When one of the bounds of a “confidence interval” is less than zero, the result merges with what could have been observed if the vaccine had no (or no longer) effect.
(2) Bailin points out that “This result is not subject to the limitations described previously, as it was based on hospitalization. While behavioral differences may influence whether people with mild infections perform or not on a PCR test (as with the result “infection of any severity”), the same issue does not arise when considering the serious infections – if a person is seriously ill they will need to be admitted to hospital”.