More counter-research: The unvaccinated may not be such bad drivers after all
Another new publication is out, in which we gently take apart one of the more outrageous recent scientific propaganda pieces.
After outlining my “unexpected patterns” article and its publishing experience just a few days ago, I got word today that another article was accepted in Frontiers in Public Health. With a small group of colleagues from a variety of disciplines, we dissected the article titled “COVID Vaccine Hesitancy and Risk of a Traffic Crash” by Redelmeier et al. (2023, published electronically in early December 2022), a research team from Toronto’s Sunnybrook Health Sciences Centre and affiliated institutes.
Now, I had already written a full post about that paper, or more precisely about a number of severe criticisms circulating online; see “Reality check: Researchers did NOT show that vaccine hesitancy is in any way related to driving behaviour” for what I think was quite an entertaining take-down of the Redelmeier study. I ended that post with a blasphemous thought about “defunding (the) science”. But as some of you know, I continue working inside “the system” and have come to the conclusion that it is essential to respond to bogus science within the peer-reviewed literature. I’ve written about my first such experience, a letter-to-the-editor concerning an article by Canada’s Chief Misinformer, and about many other independent researchers’ work in “The COVID science files”.
The new response to the “traffic crash” article has been posted in an unformatted, somewhat incomplete form on the Frontiers in Public Health web site under the title “Vaccine hesitancy was not shown to be associated with traffic safety or driver behaviour“. On the initiative of one critical public health researcher, a handful of us worked on understanding the Redelmeier study and formulating a moderate response since mid-December 2022, which I submitted on behalf of the group to Frontiers in mid-April 2023.
The currently posted version is missing the following abstract and subheadings, which indicate the direction of our critique:
Abstract
We respond to the study “COVID Vaccine Hesitancy and Risk of a Traffic Crash” by Redelmeier et al. (2023), published in the American Journal of Medicine 136(2), 153-162.e5. We contend that the authors come to misleading and potentially harmful conclusions based on unsuitable study design and confounding factors that were not accounted for. The resulting association of vaccine hesitancy with driver behaviour or traffic safety is invalid.
1. Introduction
2. Study design and framework not suitable to test stated hypothesis and research question
3. Important factors influencing vaccination status and likelihood of being involved in a traffic accident were not considered
4. Assumptions about unvaccinated individuals may lead to discrimination against certain socio-demographic groups or racial backgrounds
5. Conclusion
As you may be able to tell from the above headings and the posted text body, we argue along lines that no self-respecting epidemiologist, honest public health practitioner, or any decent person could outright dismiss. The questions and issues we raise are well supported with references (the details of which aren’t included in the temporary posting yet). Due to this more thorough approach to responding, we did not fit within the 500-word limit for a response within the American Journal of Medicine that had published the original article. Fortunately, Frontiers in Public Health explicitly permits responses to work published elsewhere within its “opinion” type of article.
Some observations regarding this generally positive publishing experience:
Within 24 hours after submission, Frontiers required a conflict-of-interest statement referring to the commercial affiliation of our first author.
For two months, April-June 2023, nothing happened as the journal had trouble finding suitable reviewers.
After responding to their belated request for reviewer suggestions in mid-June, we received the first review within three days — thank you to this colleague for their extremely fast review! The reviewer lauded “this balanced article” and found “The arguments of the authors are convincing, but could do with a bit more restructuring to make the article stronger.” He/she also suggested re-running the Redelmeier analysis with the confounding factors identified; we had to decline this, as we had neither access to the data nor the resources to repeat the analysis, in particular not for the purpose of a so-called “opinion” piece. But the reviewer themselves had left the option that we could merely suggest that a re-analysis should be done. This review 1 was labelled as “Major revision is required”, which was a bit concerning since it gave the editor a lot of room in assessing our response.
We resubmitted an edited manuscript and responses to the reviewer comments within two weeks, in early July. Two days later the review was marked as “finalized” in Frontiers’ electronic “review forum”. I take this to mean that this reviewer accepted our responses and recommended the publication of the piece. However, the editor “reactivated” review 1 two times in late July and mid-August and it was marked as “finalized” again a few days later each time. No communications were sent or posted, so we don’t know what happened there, but on the surface, this back-and-forth cost over one month of time to publication.
At some point in August, I also noticed that a second review had been requested. Review 2 then was received in mid-September. It consisted of two lines: “I must say that this is a good [article]. The authors are rigorous in writing the article. The data shown are sufficient to justify hat there is no relationship between the decision not to get vaccinated and the way you drive.” And the recommendation to the editor was “Accept in current form”. While it was excellent to receive this second peer confirmation, the sequential timing of the two reviews may have cost another month of time to publication, and the fact that an “opinion” piece required two peer reviews seems overly strict.
Consequently, the article was indeed “accepted for publication” on 20 September 2023 and immediately posted in its current, unformatted appearance.
While we decided not to limit our response to the length of a letter-to-the-editor format of the American Journal of Medicine, I was pleased to discover just yesterday that there are two pertinent responses to the Redelmeier study listed under its “related articles”.
A group of Canadian public health specialists, who are well known for their reasonable, common-sense positions on a number of pandemic policy issues, “challenge this paper on its methodology, its conclusions, and its mean-spiritedness.” Schabas et al. (2023) note issues with “several obvious confounders”; the timing of the study vis-a-vis COVID-19 vaccine rollout in Ontario; definition of “vaccine hesitant”; punitive public health policies; and unfounded claims of underlying negative psychological characteristics.
And a pair of Nigerian infectious disease specialists also highlight the omission of confounding factors, issues of misclassification in the Redelmeier data, and misleading conclusions. For example, Abiola Adepoju & Imoyera (2023) point to the higher risk of essential workers; note that “The study only looked at patients in emergency departments …, which may have excluded a significant proportion of traffic accidents“; and estimate that “… the authors would only need a few hundred of the casualties to be misclassified for the difference between the vaccinated and unvaccinated groups to disappear in the analysis.”
Google Scholar suggests that the Redelmeier et al. (2023) study will soon be lost in the mists of time with currently four citations, none of which is a peer-reviewed research article. When our Frontiers “opinion” is added here, this will give future researchers interested in the study a lot to think about, and hopefully deter further attempts at “activist science” trying to ostracize COVID/mRNA vaccine decliners by wrongly associating them with negative personal attributes.