RFK Jr: The Anatomy of a Grifter

HHS Secretary, Robert F. Kennedy Jr., lies about the Influenza vaccine to the blind applause of 109k people on Instagram

The propagation of pseudoscientific rhetoric has been extensive in recent years. We have seen a resurgence in HIV-denialism, a steadily growing $11B per year U.S. supplement industry, and “biohackers” with an exhaustive panel of unproven (yet, expensive) tests and miracle-cures for sale. As of January 23rd we have the first week in over 60 years where the C.D.C. has not published a Morbidity and Mortality Weekly Report according to the schedule, a direct consequence of the new agenda from the executive branch. Many contemporary public health topics were politicized during the COVID-19 pandemic, and often discussed within the confines of internet echo chambers. Legitimate scientific concerns have been extracted in isolation and used to strong arm the uninformed into holding positions that may be directionally correct on one issue, but become inappropriately generalized to a wide range of topics which are often unrelated. An example of this can be seen regarding the paucity of sufficiently powered prospective randomized controlled studies on the COVID-19 vaccine boosters compared to the original vaccines. Statements like these make their rounds online, but frequently get co-opted by someone with an agenda who ever-so-slightly tweaks the phrasing or insinuates an unfair conclusion. This has likely played a role in the spread of scientific misinformation and support for things like defunding the FDA. Indeed, public belief that science is an overall net positive force in our society has dropped to just 57% according to Pew polling data from 2023. Shunting resources away from vital institutions leaves people increasingly vulnerable to bad information from people with an agenda, ringing faintly of Lysenkoism; Under Stalin's backing in the 1930s-1960s, agronomist Trofim Lysenko rejected genetics as "bourgeois science" and promoted pseudoscientific agricultural techniques aligned with Soviet ideology—resulting in crop failures, famines, and the imprisonment or execution of over 3,000 scientists who defended evidence-based biology.

In this post, I want to walk through a video that I came across on Instagram, in which R.F.K. Jr. makes several factually incorrect claims about the Influenza vaccine as part of his decades long disinformation campaign on healthcare and medicine. In particular, I would like to use this video clip as an example of how “source amnesia”–the “aw shucks” moment where a manipulator conveniently forgets or misremembers the source of the information supposedly underlying their entire claim–can be wielded by bad actors to inoculate the public against reason, evidence, and even decisions in their own best interest.

R.F.K. Jr. Claim 1: Just Trust Me, Bro

It is unfortunately common practice for people to cite, vaguely, that “studies show” [fill in the blank plausibly true statement]. It’s my hope that we collectively move towards a culture of information literacy where the mere utterance of this statement is enough for consumers to scroll to the next reel. This is essentially what Kennedy does in the video right off the bat, claiming that the revered Cochrane Library published a review article finding that the flu vaccine actually increases your odds of getting sick. Since he doesn’t provide a clear reference, let’s take a look at the most recent Cochrane review on the flu vaccine (Demicheli V et al., 2018) available prior to his filming of the video in question. The study, interestingly, concluded the exact opposite of what Kennedy claims it to have; the review found that the flu vaccine not only reduced lab-confirmed influenza (RR ≈ 0.41), but also reduced influenza-like illness modestly (RR ≈ 0.84); there was roughly a NNT≈71 to prevent one case of influenza and ≈29 for ILI. Fortunately, this information from the Cochrane Library is widely available for free and written without excessive scientific jargon; however, there seems to have been insufficient impetus to verify the claims, as evidenced by the hundreds of thousands who positively engaged with Kennedy’s social media statement.

R.F.K. Jr. Claim 2: The Boy Who Cried “Wolfe”

Kennedy proudly cites a retrospective review of Department of Defense personnel during the 2017-2018 influenza season by “W-o-l-f-e”, claiming it demonstrates that influenza vaccine recipients were more likely to become infected with COVID-19 compared to those who did not receive the influenza vaccine. I could not locate this “Wolfe” study, but I believe he was referring to the 2020 retrospective study by Greg Wolff which is oft cited by the “Children’s Health Defense” organization, a non-profit indistinguishable from parody. At baseline, it should be plainly evident that the R.F.K. claim is impossible given that this study was performed on data years prior to the advent of COVID-19 (2017-2018). However, for the purpose of this exercise, let’s take a closer look at the study. The study finds that when comparing to non-flu vaccine recipients, the odds ratio for non-flu viruses was 0.97 (95% confidence interval 0.86, 1.09; p=0.6). For those who are unfamiliar with statistics, this may not mean a lot at first glance. An odds ratio is a way to measure how likely an event is to happen in one group compared to another. For example, an odds ratio of 2 means that an event is twice as likely in a given group when assessed against its comparator group; an odds ratio of 0.5 means that the event is half as likely. A confidence interval is the average of your estimate, plus and minus the variation in that estimate; all measurements have some degree of error, but the confidence interval basically says that it is 95% likely that the true result (the odds ratio, in this instance) is between 0.86 and 1.09. Since the confidence interval in this study includes values between 0.86 and 1.09, containing the value of 1, the actual affect of vaccination has no impact on the acquisition of non-flu respiratory infections. The Wolff study found that the influenza vaccine conferred a protective effect against infection with influenza [OR: 0.48; (95% CI: 0.43-0.52)], indicating that the vaccine reduces by half your odds of contracting an infection with influenza.

Astute readers will note that the study does indicate that upon univariate analysis of individual ILI infections–influenza-like illness, caused by several different viruses known to manifest as the colloquially named “common cold”–people who got the influenza vaccine were more likely to have had infection with the seasonal coronavirus [OR: 1.36; (95% CI: 1.14-1.63)] and human metapneumovirus [OR: 1.51; (95% CI: 1.20-1.90)]. Without digging much further, it makes sense why someone might see this and from it, interpret that getting the influenza vaccine causes a 36% increase in coronavirus infections or a 51% increase in human metapneumovirus infections. Upon further inspection, however, this conclusion doesn’t hold water. Even if we were to look beyond the fact that the study was not a randomized prospective trial, it was performed over a single season, the odds ratios for the secondary endpoints (seasonal coronavirus and human metapneumovirus) were modest at best, and that causality generally cannot be determined in a retrospective study, we are still left with issues.

The primary endpoint of this study was infection with influenza, with secondary endpoints being univariate analysis of individual ILI microorganisms of which this study investigated 18. The main issue with interpreting the findings for coronavirus and human metapneumovirus is the issue of multiple comparisons. Consider the following scenario:

Imagine I come to you and offer a deal: I bet you that I can make a half-court basketball shot, and if I do, then you owe me $100; if I cannot make the half-court shot, and I owe you $100. Anyone familiar with my basketball ability, will likely take that bet. So what’s the catch? The catch is that I get to take an unlimited number of shots. That’s basically what’s going on when we do loads of different statistical tests without adjusting how “strict” we are about calling something “significant.” One way that researchers can account for this is the Bonferroni Correction. If such a correction were done for the Wolff study, it may not be that the secondary endpoint findings would remain significant. Sadly, the paper only reports its findings to “<0.01” which is insufficient information to properly perform the correction.

Still not sold? Fantastic, because I am not yet done! A secondary analysis of the Wolff paper found major methodological flaws which account for these unexpected findings. The issue is that Wolff used a test-negative design, specifically intended to compare vaccinated vs. unvaccinated among people who present with the same symptoms (e.g., ILI), ensuring that the only difference is whether they test positive for influenza or negative for influenza. This principle was violated when Wolff retained the influenza-positive specimens in the non-influenza respiratory virus test negative control groups. Obviously the flu vaccine, which Wolff’s own data indicate to be effective, would mean that influenza-positive specimens are systematically more likely to have not received the influenza vaccine; this violates the core-principal of the test negative design and would artificially inflate the odds ratios for the non-influenza respiratory virus univariate analyses.

Finally, Wolff’s own data show that vaccination status strongly correlates with age group: older adults, active-duty members, etc., are more likely to be vaccinated, while younger populations are less likely. While some analyses (e.g., the combined “non-influenza virus” vs. “pan-negative” approach) involved adjusting for age, univariate analyses of coronavirus and human metapneumovirus did not–even though the data indicated it was an important confounder. This strongly suggests that the “increased risk” findings for coronavirus and human metapneumovirus are explained, at least in part, by age imbalances rather than biologically real viral interference.

R.F.K. Jr. Claim 3: Sanity is not statistical

Although Kennedy declines to cite the other “six major studies finding the same thing”, I will do him a professional courtesy by considering a randomized prospective study from 2012 by Cowling et al. which finds an increased risk of non-influenza respiratory virus infections associated with receipt of inactivated influenza vaccine. In this study, 115 children from Hong Kong were randomly assigned to either a trivalent inactivated influenza vaccine or placebo. The patients were followed for a median of 9 months and monitored for laboratory confirmed influenza and non-influenza respiratory infections. This study too is fraught with errors, to include self-reporting of respiratory infections, failure to perform RT-PCR confirmatory testing on 69 of the 134 respiratory infection episodes (specimens were collected from only 65 episodes), multiple comparisons without correction, enormous confidence intervals ranging from 1.31 to 14.8, single season design, and inappropriate statistical power to detect true effects in these secondary endpoints.

Conclusion

I hope that this article was helpful to some who are uncertain about where to turn in an era where misinformation–and at times, disinformation–is propagated without a second thought. The purpose was not to advocate for or against any particular vaccine, but to highlight a pervasive issue where people will blindly rally for claims made by people with whom they align ideologically. And to be clear, there are arguments based on scientific merit which can be used to critique the recommendation to get an annual influenza vaccine; I do not happen to find these arguments particularly compelling from a public health standpoint, but they do exist and should be acknowledged by public health officials where appropriate. Scientific merit, integrity, and good faith, however, are clearly not considered by Kennedy in his relentless assault on public health via anti-vaccination advocacy. When making claims that have the potential to significantly impact the health or safety of citizens, it is mandatory that such claims are made with proper reporting of the substantiating evidence and subjected to reasonable skepticism. While it excites me that people are beginning to take initiative with respect to their own health, I strongly urge all people to carefully consider the validity of your own findings when you “do your own research”; it may be deceptively challenging to interpret the results of studies within the appropriate context of the greater scientific literature and statistical reporting.

TLDR: When someone makes a big claim and they don’t include a source, the odds that they are not providing a complete story may be higher than many people would otherwise anticipate. Even when a source is referenced, the mere utterance of a citation isn’t enough to place your trust in someone’s words, especially when that someone has no medical training or expertise. Unfortunately, “vibes” alone are often insufficient to vet whether or not someone is giving you quality information.