Is the U.S. Giving the Wrong Covid-19 Vaccines?
A Danish pre-print - and Martin Kulldorff - make the (weird) case for adenovirus-vector vaccines.
A couple weeks after its innocuous release, a good friend sent me a pre-print from Denmark up on the Lancet server, with the fairly banal title, Randomised Clinical Trials of COVID-19 Vaccines: Do Adenovirus-Vector Vaccines Have Beneficial Non-Specific Effects? His message was not all that friendly, more like, “Did you tell us to get the wrong vaccine, you dumb $%^&?” I was, therefore, not all that kindly disposed towards the research, which essentially advocates for the possibility that the adenovirus-vector vaccines, namely the offerings from J&J, Astra-Zeneca, and Russia’s Sputnik, might have life-saving effects beyond their protection vs Covid. It’s a rather bizarre assertion — one that was swiftly taken up by the very influential Martin Kuldorff in a piece run by the Brownstone Institute.
“Have people been given the wrong vaccine,” as per the title of Kuldorff’s article? Did I err in recommending Pfizer and Moderna instead of the J&J/Janssen vaccine? These are important questions, and it’s worth a closer look at the claims. If nothing else, the issue is a healthy case study in how to pose a question in medicine.
So — the pre-print. To be clear, since it is discussed on social media as a “Lancet pre-print,” this is simply a non-peer-reviewed paper hosted by the Lancet’s pre-print server, not conditionally accepted by The Lancet. I actually found the paper itself more cogent than I expected. The authors, a collection of Dutch and Danish researchers, one of them an infectious disease physician, take a reasonable approach to sifting through the randomized controlled trial data on the mRNA (Pfizer and Moderna) and adenovirus-vector (J&J, Astra-Zeneca, and Sputnik) Covid-19 vaccines, assessing the risk of death for those receiving the actual vaccines vs placebo injections. For example, suicides, accidental death, and death by trauma were removed, sometimes only after consulting with the study authors to clarify that deaths were not related to causes clearly unlinked to receipt of a vaccine. The concept: since death is (arguably) the most important outcome of any trial testing an intervention against a disease that has killed millions, let’s see if proper randomized data implies that one vaccine type might be better at preventing death than another.
Their finding was either shocking or absurd, depending upon your perspective. RCT participants who received the adenovirus-vector vaccines were 62% less likely to die of non-covid, non-accidental causes than those who got the placebo! This, compared to the mRNA trials, in which vaccine recipients were 17% (but not statistically significant) more likely to die of non-Covid, non-accidental causes. So, we’re giving the wrong vaccines, right?
Well… not so fast. To the authors credit, they acknowledge the novelty of their claim. After all, when Covid deaths are removed from the equation, we hardly expected these vaccines to be life-saving.
Here is where I diverge from the authors. Their assertion is not merely implausible, it is most likely impossible. Think about it. Most of the adenovirus-vector trials concluded within 2 months or so. Could a vaccine have such a profound protective effect on overall health within months? Most of the benefit arose from reduced cardiovascular mortality. Our best, most studied, specific interventions in heart disease have, put mildly, displayed a less striking impact on cardiac death. Bypass surgery? Maybe a 10-15% reduction in deaths in patients with known advanced coronary disease. Statin drugs? Possibly 10%, over years of follow-up, in high risk heart patients. A 62% mortality reduction from a Covid vaccine — given to a general adult population — in a couple months? The assertion is simply preposterous.
Again, at least the authors attempt to stake a claim for biological plausibility. They cite the overall mortality benefit of the tuberculosis BCG vaccine as an example of precedence for this sort of effect. Of note, however, any overall (i.e., non-Tb) mortality effect of the BCG vaccine is hotly contested; it was found in studies of high risk infants in the developing world, and almost entirely related to infectious disease, not broader health benefits; and, perhaps most tellingly, their citation links the work of, yes, one of the authors of this paper, Dr Peter Aaby, whose primary research focus has been on this concept of “non-specific effects of vaccines.” I will say it again: sometimes we find what we seek.
This leads us to the commentary of Dr Kuldorff. Whatever your feelings about his association with the Great Barrington Declaration and the libertarian-leaning think tank Brownstone Institute, his credentials are impressive: Harvard professor/biostatistician with experience guiding CDC and FDA committees on vaccine safety. He opens with a reasonable-sounding assertion, that the Danish pre-print is giving us good data because it derives from the best possible source, randomized controlled trials:
However, this is worth a little thought. The trials were designed primarily to study the prevention of symptomatic infections; severe disease was a secondary endpoint. Death was not one of the studied endpoints (although deaths were included in the results), primarily because so few people will die in a short study of a few ten thousand people that achieving statistical significance is a crap shoot. Picking through even an excellent RCT for results that it was not powered to produce with statistical significance is not answering “the right question with the right data;” it is answering an important question with the wrong data. Hence, confounding, poor randomization, or statistical noise far more likely led to the striking but utterly implausible findings of the Danish group than proper methodology.
I don’t expect this research will hold up well to peer review. While I don’t find that peer review always, or even often, cleans up published papers to even an acceptable degree, I do find Kuldorff’s comments below more than a little amusing; namely, this paper does not need peer review, since he (and apparently Jay Bhattacharya) have already reviewed it and they say it’s fine!
Kuldorff’s piece also includes the more troubling implication that the mRNA vaccines, with their non-improvement in mortality, might somehow be more dangerous than beneficial. It is true that they showed no survival benefit in the brief window of the studies; commenting on this fact turned out to be the final straw in pre-Elon Twitter banning Alex Berenson. This is not some condemnation of mRNA efficacy, however. When I say the mRNA trials were not powered to show a mortality benefit, it’s simply this: given that the median age of the Pfizer trial was 52, if the average participant were, say, a 52 year old hypertensive male, his risk of death from a case of Covid runs around 1/167 per a pre-Delta risk calculator. The Pfizer trial had 162 cases of Covid among the placebo group. We were never going to see a meaningful mortality benefit! That roughly the same small number of people in both placebo and vaccine groups died of all causes in the Moderna and Pfizer trials is a non-story, period.
So why report the combination of a non-story and an absurd claim like it’s serious science? Well, there’s probably a political point to be made. In this case, the Brownstone Institute has been vociferous in their opposition to U.S. vaccine mandates, and our vaccines are of the mRNA variety.
Asking for a head-to-head, randomized, controlled trial in a still-unvaccinated population pitting adenovirus-vector vs mRNA vaccines that is large and long enough to find a statistically significant mortality benefit is essentially demanding the impossible. Hence, it becomes impossible to mandate mRNA vaccines with the weight of unimpeachable science. This, presumably, is the real purpose of Dr Kuldorff highlighting this rather unimpressive study.
I wish to be clear that I am not against asking uncomfortable questions in science, especially when they are as important as challenging the mortality benefit of a widespread intervention. Devising ways to study the overall mortality effect of different Covid-19 vaccine programs is a worthwhile endeavor. However, there is no need to suggest that alternatives to the current program have miraculous benefits; nor that the vaccines in heavy use have hidden dangers that multiple studies designed to ascertain these dangers have failed to find.
I have written a lot — maybe too much — on my concerns for post-mRNA-vaccination myocarditis among adolescent males. I also share the Brownstone Institute’s aversion to mandating Covid-19 vaccines. However, there is no need to promote weird science to support these positions. This is true anywhere on the ideological spectrum. Whether it’s then-director of the CDC, Robert Redfield, stating that the pandemic would end in a month if everyone wore a mask, or professor-turned-podcaster Brett Weinstein touting fraudulent Argentinian research that ivermectin use could prevent virtually all cases of Covid-19, implausible claims require remarkable evidence. I would hope that prominent people publicly commenting on health policy take responsibility for the quality of the science they cite.
The real issue here is demanding that billions of humans take a vaccine WITHOUT solid, RCT level evidence of overall mortality benefit. It’s one thing to recommend vaccination without overall mortality evidence; it’s another thing to compel it.
It may be impossible to conduct this research now - but it wasn’t then. And that’s the scandal.
Sadly, experts (such as you are) never held regulators to account. So we’re forced to make due with inadequate RCTs that can’t answer the only question I care about: If I take this shot, will I live longer?
Again, RCTs should have been conducted to answer this definitely and unambiguously. Seems obvious.
Agreed?
More conservative people who tend to take less risks may have chosen a traditional vaccine like J&J rather than the new MRNA vaccines.