Question asked by Adrien on June 25, 2022.
You ask us about an article published on the morning of June 25 on the website of the Free lunchwhose title was: “Pfizer, Moderna, Astra Zeneca: anti-Covid vaccines increase cardiovascular risks in young people.” This was widely shared on social networks, in particular by Florian Philippot, Francois Asselineau, Nicolas Dupont-AignanWhere Laurent Mucchielli.
The article in question – since amended – stated that “Covid vaccines Moderna, Pfizer and Astra Zeneca are all said to have adverse effects on cardiovascular health“, referring to three “studies published in prestigious Anglo-Saxon scientific journals between April and June 2022”. And to suggest to the reader to do “the point on this [que ces publications] underline»to know “the interest of vaccines”but also “its potentially harmful effects on the cardiovascular health of patients”.
Problem: for each of the studies mentioned, the account that was made of them was erroneous or concealed an element of context essential to their understanding.
A shell… which multiplies the risk of myocarditis by 1,000
The first publication cited was a retrospective study, presented on June 11 on the website of the Lancet, involving just over 15 million vaccinated. She observes that the number of myocarditis and pericarditis is higher than expected in the population studied, particularly in men aged 18 to 25 within a week of a second dose of a Moderna or Pfizer vaccine. Free lunch correctly stated that, according to the article of the Lancetthe incidence of these pathologies “was rare”. And to specify that “411 myocarditis or pericarditis having been detected out of the more than 15,000 people aged between 18 and 64 observed by scientists”. Here, the error is typographical (fifteen thousand instead of fifteen million), since the text suggests events a thousand times more frequent than they really are. Netizens were quick to peddle the claim that the risk of myocarditis following a second dose of mRNA vaccine was “by 2.7%” (411 /15,000). After being widely reported, this gross error has been corrected by Free lunch Saturday afternoon, following numerous reports.
The typographical error seems all the more obvious since, when it was published, the article correctly related the authors’ point of view, according to whom “the study results, as well as the benefit-risk balance, continue to support vaccination with either of the two messenger RNA vaccines”.
A study that the publisher himself calls to be taken with a grain of salt
Midi Libre then reports a study published at the end of April in Scientific Reportspresented as “one of the most important open access journals on the planet”. Study which, explains the daily, reports “[d’]an increase of more than 25% in the number of calls [pour une aide médicale d’urgence] in 16-39 year olds for cardiac arrest or acute coronary syndromes during the deployment of anti-Covid vaccination in Israel”citing similar findings in Germany and Scotland. Free lunch precise “[qu’]on the other hand, no link has been established between these cardiovascular problems and Covid infections”.
The regional daily, however, omits to report an important detail, highlighting the study in the form of a “editor’s note” : “Readers are cautioned that the findings of this article are subject to criticism, which is currently being reviewed by the editors. An editorial response will follow as soon as all parties have had the opportunity to respond fully.” In fact, many voices were raised regarding the interpretation of the reported data. For example, observers note that the number of calls for cardiac arrest during the vaccination campaign does not differ from that observed in the same period the year preceding the Covid crisis, but only from the year 2020 when they were abnormally low. “At the start of the Covid-19 pandemic, emergency call centers asked the population to limit the use of these platforms in order to allow better care for patients with severe signs of Covid. -19″, observe the authors of a critical analysis pre-published on the OSF platform. The same authors note that without recording the vaccination status of the callers, or further description of the symptoms, it is impossible to draw conclusions. Statistical analysis issues were also noted.
Doubtful results… according to the authors themselves
The third study cited by Free lunch was published in the JAMA Network Open journal. For the regional daily, it would conclude with an increase in the number of “hospital contacts” (consultations, etc.) “due to coagulation disorders and cerebrovascular diseases, in particular for thrombocytopenia and cerebral venous thrombosis, after vaccination with Moderna, Pfizer and especially Astra Zeneca”. On Twitter, a user remarks that, in the study, the link with venous thrombosis is not observed for Moderna and Pfizer. But a superficial reading of the study could also suggest that the authors conclude an association between vaccination and the other pathologies mentioned. However, in order to ensure the robustness of their conclusions, the authors carried out “sensitivity analyses”, which make it possible to assess whether minor differences in the analysis criteria retained (or in the initial data set) lead or not to similar results. For example, one might expect the results to be similar when taking into account all hospitalizations or when focusing on those lasting more than one day. The fact that this is not the case raises questions, conversely, about the strength of the correlation.
After five separate sensitivity analyses, the results obtained for AstraZeneca’s vaccine are considered robust – confirming the warnings already established for this vaccine. In contrast, for messenger RNA vaccines from Pfizer and Moderna, these analyzes “were not consistent”they write. “Therefore, the overall and combined increase in risk [identifiés] after vaccination with [ces deux vaccins] should be interpreted with caution.” According to them, an analysis intended to confirm “by other methods” the results obtained for these two vaccines appear to them to be “justified”. An important precision totally absent from the article of Free lunch.
“The most read article of the week”
Following multiple alerts from Internet users, the title of the daily article was changed to the conditional on Saturday afternoon. As a result, the article remained unchanged until 3:56 p.m. Monday. The version currently online is extensively revised, and clarifies “[qu’]due to inaccurate and incomplete information, this article was modified on Monday, June 27. We apologize to our readers.”
Asked by CheckNewsthe digital division assistant of the newspaper and the author of the article return to the chronology of the events which led to the publication of this text, and the current delay until this last update. “The subject was proposed on Friday afternoon, drafted Friday evening, and scheduled for Saturday morning”explain our interlocutors. “Usually, the articles are proofread by the department heads, over the water. When written by more experienced journalists, proofreading of online articles sometimes takes place after the fact, and the changes made to them are generally minor.. However, in this specific case, “there was no proofreading, and the article escaped the vigilance of the person in charge then present at the newspaper”notes the assistant to the digital pole of the Free lunchherself “away for the weekend”.
If two minimum corrections were made to the article on Saturday (the correction of the erroneous figure and the conditionalization of the title), the article “needed to be thoroughly overhauled”she explains. “There was no point deleting the article because it had already been shared a lot, and a lot of screenshots had been taken. We prefer that people viewing the page can read that the article was published incomplete, and present them with the update. As soon as the author of the article returned to the editorial staff of the newspaper on Monday afternoon, he set about completing the text and providing details for each of the studies cited.
The author, who tells us that he has no training in science journalism, acknowledges “that such a subject could have been treated with much more time and rigor”. The deputy head of the pole notes that “On Covid subjects, we usually have a journalist who is more specialized in these issues, and our articles normally take the time to give the elements that allow us to take stock. Here, we plead the error, because the published text was clearly incomplete”.
She herself laments that because of the weekend, “the article had a very long exposure time”even though “the antivaxers rush on the slightest breach to make their butter”. In fact, she notes,this is the most read article of the week… The updated version will probably not be shared as much, but we will distribute it on social networks, making it clear that the initial version was incomplete”.