TL;DR
This (long) essay focuses on “scientism” in the recent public health response to Covid-19, with a few additional observations from the public health response to AIDS. We begin with some background, in the form of a short summary of the principal asset/tools available to the public health sector, and the basic pre-Covid state of play of public health response to a viral respiratory pandemic/epidemic with features broadly similar to Covid-19 (specifically in a well-regarded 2006 paper on the subject).
Noting that none of the recommendations in that paper were actually followed, we then take a quick tour through what did happen with Covid-19: all the deviations from known best practices that occurred, specifically noting the “scientism” present here.
The post then concludes with elaborations of a few ideas touched on in the main Covid narrative: a long and detailed look at mortality rates (including some early statistics that held up remarkably well, and could/should have been used in policy-making), a look at the unfortunate governmental response to the novel use of well-understood, generic drugs in the early treatment of Covid, and a look at the dreadful consequences of placing the same people (none of whom are epidemiologists) in charge of (coercive) Federal health policy and $45B of annual medical research funding (it is here that we also take a look back at AIDS and see that much of what happened with Covid was entirely foreseeable). Also, a short postscript.
The Problem
During the 20th century, the accomplishments of medical science in general and public health in particular (in both the US and the West more broadly) were hard to miss: this century saw a 62% increase in life expectancy at birth. And yet, in May of 2021, an opinion poll observed that only 52% of Americans had substantial trust in the CDC, which was the most trusted public health institution in the poll (44% for local health, 41% for state health, 37% for NIH, 33% for US Dept. of HHS, etc.). It’s hard to know precisely how much of a decline this is, since the poll seems not to ever have been run before, but I would find it hard to believe that the same poll in the 1990s-2000s would not have produced 70-80% confidence in many of these institutions (consider, as one data point, that the 2011 movie Contagion has CDC doctors as its heroes, and an 85% fresh rating on Rotten Tomatoes). What happened? The public health response to Covid-19!
Background
My (amateur) analysis of the situation is that “public health” as an institution has one principal asset, and four principal tools. Its principal asset is the trust of the public, and its four principal tools are: persuasion, fear, shame and coercion.
In dealing with a novel health challenge, the ideal approach would be to leverage the trust of the public, and use persuasion to convince the public to adopt the behaviors that are the current best guess in terms of mitigation strategies (and note that there will generally be more than one approach possible). Then, as more information is available, modify the recommendations as appropriate, giving clear reasons for any changes (i.e., what do we know now that we didn’t know then). Honesty, clarity and respect should allow course corrections without loss of public confidence.
Problems arise, however, when persuasion is abandoned for one of the other tools (fear, shame, coercion). These burn through the asset (public trust) quickly (the public wonders why persuasion isn’t good enough anymore). Once these alternative tools are picked up, persuasion is less effective (since trust, necessary for persuasion, has been eroded). It’s difficult to move back into a more normal trust regime: the appearance (and perhaps the reality) is that the public health officials do not respect or trust the citizenry. That lack of trust is then reflected back toward the public health officials, and the cycle continues.
As I mentioned above, this is my own assessment of the situation, and I make no claim to be a public health expert. It is worth mentioning, though, that the CDC’s Field Epidemiology Manual has a very similar discussion in the chapter on “Communicating During an Outbreak,” so I don’t think my analysis is completely out of line with “received wisdom.” In addition, Guttman and Salmon wrote a paper in 2004 (in the journal Bioethics) regarding the ethics of the use of guilt, fear and stigma in Public Health communication. Their conclusions (that the use of these tools, even in well-intentioned public health messaging, should be examined carefully for unintended consequences) would seem to apply even more to coercion.
Now, consider what the public health community itself considered (around this same time – early 21st century) to be the gold standard of response to a respiratory virus pandemic: Inglesby, Nuzzo, O’Toole and Henderson* wrote a paper in 2006 for the Center for Biosecurity at UPMC in Baltimore (now the Johns Hopkins Center for Health Security). This paper details a response to a principally-airborne respiratory virus (like SARS or H5N1) with very high case fatality rates (2.5% to 50%) and presymptomatic transmission. The paper is well-written and clear, and I strongly suggest reading the whole thing. The upshot, though, is that nearly all the difficulties that happened in the US with Covid-19 in 2020 were anticipated in 2006, and the suggested mitigations were nearly the direct opposite of those actually suggested by CDC, NIAID and NIH in 2020. The only features of Covid-19 not foreseen by the 2006 paper were the heavy skew of severe disease toward elderly patients (and those with certain pre-existing conditions, all of which made focused protection easier and more effective), and the remarkable success of Operation Warp Speed in producing vaccines (the 2006 paper estimated only that vaccines against a true influenza could be produced in a timely manner, not a novel virus; even then, a tiered approach to vaccination was considered necessary for vaccine production reasons). Still, the suggested mitigation strategy is remarkably close to the “focused protection” strategy suggested by eminent epidemiologists from Stanford, Harvard and Oxford in the Great Barrington Declaration of October, 2020.
One might reasonably suspect, then, that the public health response to the novel (but largely anticipated) Covid-19 pandemic in 2020 would have been to realize that they had war-gamed exactly this eventuality in the 2006 paper (see also official strategy here), and to follow the mitigation strategies contained there (with a few tweaks, once the demographics of those vulnerable to severe illness was known, and with perhaps a few short-term modifications like “2 weeks to flatten the curve” early on when actual data about the disease was scarce).
That didn’t happen.
Covid-19 – the main argument
So, what happened instead? Note: this is neither an exhaustive nor a strictly chronological list. Furthermore, these are not “honest mistakes,” but rather deviations from established practice that could have (and should have) been known at the time to be counterproductive. This is also largely a narrative of the federal government’s actions, not individual state actions (which were quite varied).
- An early model in the middle of March 2020 from Imperial College-London (ICL) predicted that with little mitigation, there would be 2.2 million deaths in the US by August of 2020 (peak of 50,000 deaths per day around June 2020). In response, much of Europe locked down, though Sweden reversed course fairly quickly, remaining open (though with social distancing guidelines).**
- At nearly the same time as the release of the ICL model, predictions of hospital collapse were met by official calls for “two weeks to flatten the curve” or “15 days to slow the spread.” State and local officials followed suit quickly with coercive measures – in New York, the official response flipped from a mayoral “go on with your lives + get out on the town” on March 2 to a statewide “stay at home” order on March 20. Persuasion was abandoned just this quickly.
- On March 8, NIAID director Anthony Fauci recommended (in a 60 Minutes interview) that masks not be worn, saying that they were largely ineffective for the general public, echoing email comments he had made to former HHS Secretary Sylvia Burwell a month earlier.
- Then, on April 3, with no new evidence, he and the CDC recommended voluntary universal masking (when asked about the reversal, he essentially claimed that the original advice was a “noble lie” to preserve the masks for healthcare workers).
- This new advice became a mandate on Federal property on President Biden’s first day of office in 2021, but many state mask mandates had long been in place prior to this point.
- The actual science remains equivocal, in the sense that high-quality (surgical or multi-layer cloth with filter media) masks are helpful as source control in conjunction with droplet transmission in lab/mannequin studies (but not as PPE, nor in conjunction with aerosol transmission); they are not, however, detectably helpful in any randomized population studies. See here for recent population stats, here for a recent comprehensive survey of the literature on face masks, and here for a discussion of an even more recent Cochrane metastudy.
- Early discussions of the origin of the virus concentrated on the zoonotic origin hypothesis (usually horseshoe bat to pangolin to human via the “wet market” in Wuhan) as opposed to the “lab leak” hypothesis.
- Later evidence***, however, has failed to confirm this hypothesis in the way one would expect from previous zoonotic respiratory† viruses, and it is looking more and more as if there was some conflict of interest involved in convincing the scientific community early on to accept this origin as a given (and smear the “lab leak” hypothesis as a fringe conspiracy theory).
- It also turns out that, even if SARS-CoV-2 itself is zoonotic in origin, there are still related questions about the (partially) US-funded virus research going on at the Wuhan Institute of Virology – note especially that the PI at the center of the Wuhan funding controversy, Peter Daszak, was also the principal author of the Lancet letter that was largely responsible for quashing early investigation into the “lab leak” hypothesis.
- Here is a much later (March 2022) long, thorough investigative article on this whole debacle (including some apparent malfeasance by the NIH during 2021 to remove some of the evidence that would possibly help track down the viral origin definitively).
- Even later (October 2022) a number of researchers and writers (the minority oversight staff on the Senate Health, Education, Labor & Pensions Committee here, investigators with Pro Publica and Vanity Fair here, Warren Strobel of the WSJ here, Jim Geraghty, David Strom, and others) all published pieces making it clear that the probability of the zoonotic origin hypothesis is dropping rapidly. This new evidence (these essays are not independent, by any means) makes clear that what is now known was all available (but not released) by early 2020. To be clear, this is no “smoking gun” regarding the origin of the virus, but does clearly point to (at least) an intentional cover-up by China and the Federal public health apparatus in the US.
- By roughly May 2020, it was clear that something was very, very wrong with the ICL model – see here for an overall assessment, and here and here for attempts at a code review of the actual program running the ICL model.
- Conclusion of first round of code review: “All papers based on this code should be retracted immediately. Imperial’s modelling efforts should be reset with a new team that isn’t under Professor Ferguson, and which has a commitment to replicable results with published code from day one.”
- Here is a piece (written much later, but only using information publicly available in mid-2020) that details the history of poor modeling that ICL had (so much so that their model should probably never have been relied on).
- And yet, none of the policies that were sold to the public (more precisely, forced on the public) on the basis of the ICL model were retracted or modified. Note especially that the moderate-mitigation scenario (arguably similar to what the US did) predicted 1.1 million deaths in the US by the end of August 2020 – the actual number at that point was 195K.
- Basic data about the disease was compromised from the beginning. See below for details.
- By October 2020, a number of distinguished public health researchers and epidemiologists had grown very concerned about the abandonment of the traditional approach to respiratory virus pandemics. Specifically, the traditional approach (as was described in the Henderson, et al. paper) was cognizant of both the costs and limited effectiveness of coercive, lockdown-based strategies.
- Several of these epidemiologists authored the Great Barrington Declaration (named for the town in Massachusetts where the original meeting was held, with support from the AIER, located there), and many others signed on (there were around 30 original co-signers, which has now grown to more than 15,000 public health scientists, as well as nearly a million “concerned citizens”).
- The declaration took the position of traditional public health, took notice of the costs of blanket lockdowns, and also took notice of the huge skew in Covid risk (see here – these numbers are much later, though the trend observed here was also clearly observable early) between older (and otherwise vulnerable) citizens and the vast majority of citizens – the approach they advocated was called, “focused protection.” The authors of the declaration (two Americans and one Brit) have been supported by NIH, CDC, Wellcome Trust, ERC, Royal Society, and NSF grants over their multiple decades in research.
- The response of the NIH and NIAID to this declaration (detailed here) was not (as one might naïvely expect) to convene a discussion or teleconference among these eminent scientists, and other scientists favoring the lockdown approach, to see what might come out of discussions of the evidence. Here’s a short summary of what happened instead (again, read the whole thing here): the directors of NIAID and NIH conspired (as later came to light following a FOIA request) to author or publicize a “quick and devastating take-down” of the declaration, whose authors they described as “fringe epidemiologists” (but whose work they had been funding for decades!). They (and later the director of the Wellcome Trust in the UK) purposefully mischaracterized the GBD’s proposal as a “let-it-rip strategy.”
- Once this took hold in the popular press, it was difficult to convince anyone that this was not what the GBD said at all (indeed, it’s unclear whether or not the NIH, NIAID and Wellcome Trust directors ever read the GBD; certainly its actual contents seemed to be irrelevant to their vindictive response to it – it was different from the policies they were forcing on the American people, and that was enough for it to be viciously opposed). Among them, these three control (more precisely “controlled” – one of them has since retired, and another has become Chief Scientist at WHO) annual research budgets in excess of $50 billion, and they showed themselves to be willing to stoop to vicious attacks over scientific disagreement. Is it any wonder that opposition to the GBD quickly became a shibboleth in the epidemiology world? To do otherwise was clear career suicide!
- Beginning around the summer of 2020, before there were any vaccines or effective therapies for Covid-19, there were discussions surrounding a couple of different early-intervention approaches that both used well-understood, inexpensive drugs (hydroxychloroquine and ivermectin).
- The evidence in favor of these treatments was weak, but since we were so early in the pandemic, so was the evidence that they didn’t help.
- These were slammed (a more complete discussion of this debacle here) as “fake news” by the public health bureaucracy (principally the director of NIAID) in terms that were far out of contact with reality, given the lack of contemporaneous evidence one way or another.
- Since these drugs are so well-understood, there was little risk in using them – any positive effect would seem to be entirely a win, since there were no competing successful therapies at that time (and hence no opportunity cost).
- During 2021, evidence began to accumulate that the lockdowns, though exacting a terrible cost, had likely not helped reduce mortality.
- Most famously, the study (described here and here), published in January 2022, is a meta-study of 34 2021-vintage studies looking empirically at the effect of lockdowns on mortality. Now, it’s a meta-study, so all the caveats here apply, but it looks to me as though it’s pretty solid methodologically. I’ve seen a decent amount of criticism of the study, but most of it seems, oddly, to revolve around the fact that someone or other used the shorthand “Johns Hopkins study” to refer to it, when it’s not an official JHU publication, but rather a working paper in a series published by one of JHU’s institutes (and written by a JHU faculty member). I’m not sure why that matters at all – at any rate, if you don’t like the meta-study itself, just look at some of the 34 papers it aggregates.
- Despite all the accumulating evidence that any cost/benefit analysis of lockdowns shows that they’re a bad idea – even the WHO (which has rarely met an authoritarian health measure it doesn’t like) had come to this conclusion by October of 2020 – despite this, no one is going back and saying, “This was an error, we won’t do that again.” Instead, we get hints that coercive mandates will be with us forever.
- The list goes on: closing of schools (despite evidence), masking of school children (despite evidence), failure to include natural immunity alongside vaccination status in any of the stats or policies (though Europe did), etc. See here for a recent summary.
In any sort of novel circumstance, errors will be made – that’s understandable. What is not understandable, though, is the early discarding of persuasion in favor of fear, shame and coercion – that, and the fact that all the errors cut in the direction of authoritarianism and “scientism.”
Specifically, three of the characteristics of scientism mentioned here were the main drivers of these errors:
- Use of certainty instead of specified confidence (and the file drawer was a principal tool here – note that the vast majority of the papers contradicting the official narrative had their research supported and often published by the various Federal agencies; the public health authorities just didn’t include or link to them in their official policy statements, so you have to search for them yourself).
- Over-reliance on models (specifically the ICL model) without appropriate assessment.
- Politicization of public health – as just one example, officials were willing to consistently say anything that was perceived by the “ruling class” as contradicting then-President Trump and undermining his re-election chances.
I will conclude with a few elaborations of topics already touched on (and linked to) in this essay:
More than you ever wanted to know about mortality
When talking about an epidemic/pandemic, there are at least three mortality/fatality rates of concern:
- Crude mortality: the ratio of relevant deaths (usually over an entire event) to the population as a whole. This may be (and usually should be) disaggregated into various age/demographic subgroup rates, but the overall rate is both useful and easy to compute.
- Infection Fatality Rate (IFR): the ratio of relevant deaths to infections as a whole. Note that the denominator must be estimated, since not all infections are confirmed. This is perhaps the most useful rate, but the most difficult to compute with confidence.
- Case Fatality Rate (CFR): the ratio of relevant deaths to confirmed infections (or, occasionally, some other definition of “case” like hospitalizations with certain symptoms). Fairly easy to compute, but less useful than the others. One of the main reasons it’s less useful is that it is easy to confuse with the IFR.
Note that crude mortality aggregates contagion and virulence, while the other two are different attempts at separating the two (and just measuring virulence).
Quick example: in the 2017-2018 flu season (a particularly bad one), there were 45M “infections” (that is, estimated infections, called “disease burden” by the CDC), 61K deaths, and 275K positive flu tests. Since the population at the time was 325M, this gives a crude mortality of 0.019%, an IFR of 0.14%, and a CFR of 22%. Note the astonishingly high CFR for flu! This is not uncommon, since flu tests are (more precisely, were – they are performed much more often post-Covid) typically given only in very serious cases, or in immune-compromised patients – most infections are simply treated symptomatically, or at most with Tamiflu. The number of infections is estimated by past experience, and by tallying up reports of ILI (influenza-like infections) during flu season.
Now, what are these same numbers for Covid-19? Here, we are getting into murky waters – the WHO gave a “mortality rate” of 3.4% in March 2020, but what does that mean? It seems to be an early CFR associated with a particularly opaque testing regime in China, but it still gets quoted today.‡ Even putting to one side the issues surrounding Covid statistics in the US (“died with” vs. “died from”; the hospital reimbursement incentive for Covid deaths; the lack of transparency about PCR cycle numbers, etc.) here’s what we do know: official Covid deaths for the first wave (Covid Classic – through June of 2021) in the US were 623K, “cases” (i.e. positive tests with some PCR cycle number) were 34M, and the population was 333M. So, crude mortality was 0.19%, and the CFR was 1.8%.
What was the IFR for Covid-19? Here, we have no experience to guide us as in the flu scenario. What should have happened is that longitudinal antibody seroprevalence studies in several locations in the US should have been initiated to estimate what percentage of the population had recovered from (or been substantially exposed to) Covid-19. This would have allowed a solid estimate of the IFR fairly early on (at least before vaccines would have confounded some of this – later seroprevalence studies should have kept the vaccinated separate, which they didn’t as far as I can tell). Instead, the earliest papers of this kind didn’t appear until 2021 (relying on 2020 data, though). Seroprevalence studies in Europe were beginning as early as January of 2020. I have no explanation for this that doesn’t involve a possibly-overly-cynical view of the NIH. Using a population-weighted average of the 10 regional “underascertainments” (that is, the factor by which reported cases underestimate seropositive results) in the 2021 paper (which is likely not appropriate since there was such a huge variance – from 8.9 to 1178 – but it’s all we can do, since the NIH didn’t see fit to fund enough of these studies), we are led to the conclusion that there were actually 247 times as many Covid-19 infections over that period as reported.
Now, this can’t actually be correct, since the 34M reported cases, multiplied by 247 gives 8.4 billion, which is larger than the population of the Earth (and, recall that not all of those live in the US). So, something is wrong somewhere. BUT, we do get the sense that a 9x underascertainment factor (the smallest such number in the study) might be reasonable. This would say that 306M people were infected (or substantially exposed) to Covid-19 during the first wave (92% of the population). To be clear, NONE of the seroprevalence numbers in that study were anywhere near 90% – what happened was that they were in the 1-7% range at a time when there were actually very few reported cases in the catchment area. Have I mentioned that there’s just not enough data here? I will remark, though, that a much later (good quality, longitudinal, after the NIH got its act more-or-less together) blood donor study showed 85-93% (depending on age) seroprevalence at the point where the first wave finished. This would be precisely in line with my rough estimate above (though vaccination complicates this whole picture). Using this estimate (306M infections) yields an IFR of 0.2% for the US population as a whole – 43% higher than the IFR of the 2017-2018 flu season (which, recall, was a particularly bad one). So, by this reckoning, the problem with Covid-19, compared to flu, is more its contagion, not its virulence (10x the crude mortality, but with only a 43% higher IFR). It’s also worth mentioning that this single average IFR is really of little value, since the variability by age of Covid-19 mortality is so enormous.†† Later estimates (from the Financial Times, linked here on epidemiologist Vinay Prasad’s substack) indicate that as of March 2022 in the UK (with vaccines, improved therapies and less virulent variants), the IFR for Covid-19 is below flu for all age groups.
Of course, none of this was known back in March-April of 2020, when the public health decisions were being made. But, what could have been known then? I claim that there was decent (though far from comprehensive) data available, at least by April 2020, that has held up remarkably well – the data from the cruise ship, the Diamond Princess (AKA the Floating Petri Dish – they did what they could to “quarantine” but there truly wasn’t much that could be done). Here are the numbers: 3711 aboard (2666 passengers, 1045 crew), 712 infected (567 passengers, 145 crew), 13 dead (all passengers, all older – 3 unknown, 1 60s, 5 70s, 4 80s); median age 69 (passengers), 36 (crew); total IFR 1.8% but passenger IFR was 2.3% and crew IFR was 0%. Crude mortality was 0.3% – at or above age 69 was 0.9% and below age 69 was likely 0 (all deaths were described as “older” patients, presumably meaning “not below median age”). Note that, since the entire population was tested, we can say that the CFR and the IFR were identical in this case. Note also that moving the statistical cut-off down to 65 from 69 (for convenience in extrapolating to the US population as a whole) would likely only fractionally reduce the “older” crude mortality (since the denominator would increase slightly) while the “younger” crude mortality would remain at zero. We can thus continue to use the same crude mortality numbers with a different cut-off and be confident that we are likely dealing with a slight overestimate of crude mortality.
Using these numbers, now, to extrapolate to the US population, 17% of the population is age 65 or above. Crude mortality would then give an estimate of 534K deaths age 65+ and at most 116K deaths (“rounding up” from zero to 1 death on DP) age <65 – a total of 650K – before the end of the first wave (roughly comparable to the “end” of the Diamond Princess “experiment”). Actual first wave death numbers in the US were 623K, so this estimate holds up pretty well. Note also that the Diamond Princess case is roughly comparable to the ICL “do nothing” case, since there was little that the Diamond Princess could do. I claim that my crude estimate (available as of April 2020) is much closer to the US experience than the “sophisticated” ICL model estimate of 1.1M to 2.2M deaths (depending on mitigation strategy). Furthermore, it looks like the lockdowns that were done in large chunks of the US were largely ineffective in mitigating deaths, since the actual numbers were within 5% of my Diamond Princess-derived upper bound (again, in line with the 2006 Henderson paper, which predicted that all these would be ineffective, and the GBD which advocated for focusing our protection efforts on the older and vulnerable populations, not on the population as a whole).
Looking forward to the Delta wave (July 2021 through November 2021) and the Omicron wave (December 2021 through February 2022), we see that deaths in each of these waves were around 190K, so that these variants had a crude mortality around 30% of “Covid Classic” – more contagious, but less lethal.
Looking specifically at states to assess the impact of “lockdown” policies on the resulting deaths, I observe the following (note that I re-balanced the estimates based on the age demographics of the states involved, using numbers obtained here):
I am choosing three large, representative states: NY (stringent lockdown), TX (moderate lockdown, followed by opening up fairly early), FL (largely open) – all data from worldometers.info.
- NY: estimates of fatalities by wave (40K, 12K, 12K) vs. actual (54K, 4K, 10K)
- TX: estimates of fatalities by wave (47K, 14K, 14K) vs. actual (53K, 21K, 13K)
- FL: estimates of fatalities by wave (44K, 13K, 13K) vs. actual (38K, 24K, 11K)
General observations – FL and TX got hit unexpectedly hard by Delta (second wave), and the two states (NY, TX) that were largely locked down during the first wave had worse experience there than the one that was largely open (FL). I’m not asserting from these numbers that the lockdowns were harmful to Covid mortality, but there is certainly no evidence here that they helped. I would also note that the urban/rural demographic split is not taken into account in my estimates – at least in the first wave, urban experience with Covid was much worse than rural, so this may account for the NY first wave experience (note that their statewide crude mortality was 35% higher than the Diamond Princess).
I will also note for completeness that there may have been a 14th Diamond Princess fatality a few weeks after the end of the “experiment.” If that death is included, my US estimates increase to 691K in the first wave and 207K in subsequent waves (and each of my 3 state estimates is increased by 3K in the first wave and 1K in subsequent waves). I view the 14th fatality as dubious (and certainly didn’t know about it when I made these estimates back in 2020), so I didn’t include it in the main discussion here.
Novel uses of well-understood drugs
As I mentioned above, beginning around the summer of 2020, before there were any vaccines or effective therapies for Covid-19, there were discussions surrounding a couple of different early-intervention approaches (that is, approaches to prevent severe disease, not to treat severe disease) that both used well-understood, inexpensive drugs – one involving hydroxychloroquine (HCQ or plaquenil, an antimalarial and anti-inflammatory drug widely used in rheumatic disease; occasionally the Covid treatment paired it with azithromycin and zinc supplementation) and another involving ivermectin (mostly an extremely effective antiparasitic).
By this point, it was simply not possible for RCTs involving these (or any other therapies) to have been run and peer-reviewed. Due to a few positive mentions by then-President Trump about these therapies (and, to be clear, the supporters of these treatments were claiming more than they had evidence for at the time), the director of NIAID, Anthony Fauci, began††† a crusade to tar all of these therapies as “fake news.” In the process (video in linked post – my transcript), he said that, “for clinical trials, that were valid, that were randomized and controlled in the proper way, all of those trials show consistently that hydroxychloroquine is not effective in the treatment of … Covid-19.”
Let’s check the data:
The standard HCQ/Covid meta-study in November 2020 showed that by their last search point (August – a month after the flap described above), there had only been 5 RCTs on HCQ/Covid: one on hospitalized patients showing no significant differences (stopped early before peer review), another on hospitalized patients showing no significant differences (never peer reviewed that I can find), another (that was completed and peer reviewed) on outpatients that showed some positive results (specifically, the time to resolve symptoms was lower in the HCQ arm, though you wouldn’t know this from the abstract), another on outpatients that showed some positive results (specifically, the percentage requiring hospitalization was 60% lower in the HCQ arm, though again, you wouldn’t know this from the abstract), and a final one (completed but not peer reviewed) on outpatients that showed significant HCQ-related improvement in TTCR (time to clinical recovery). In addition (in the last study), 13% of the control arm developed severe Covid-19 and none of the treatment arm.
Now, the treatment being advocated was specifically an early treatment being supported to avoid the development of severe illness (not a treatment after things have become severe). Only 3 RCTs existed at the time addressing this form of the treatment, and all showed some improvement related to HCQ (if not significant improvement). That doesn’t sound to me like, “all of the valid trials…showed that HCQ was not effective.” That sounds to me like, well, Anthony Fauci saying the sorts of things that Anthony Fauci says (and maybe he didn’t read beyond the abstract with two of the RCTs – see here).
The meta-study I referenced earlier (which, to be fair, was published four months after the brouhaha under consideration), concluded that, taking into account all available evidence (43 studies and trials, not just RCTs, which were in short supply that early), “HCQ has been shown to have consistent clinical efficacy for COVID-19 when it is provided early in the outpatient setting; in general, it appears to work better the earlier it is provided. Overall, HCQ is effective against COVID-19.” Did Fauci ever correct the record on HCQ? Not that I’m aware.
The story on ivermectin is similar – here is the 2020 meta-study. Its conclusion: “Ivermectin is likely to be an equitable, acceptable, and feasible global intervention against COVID-19. Health professionals should strongly consider its use, in both treatment and prophylaxis.” On the other hand, when well-known podcaster Joe Rogan contracted Covid and (on the advice of his physician) used ivermectin, it was derided on CNN as “horse dewormer” (an episode discussed here, as well as other relevant discussion; ivermectin is an important veterinary medicine as well, but its use in humans led to, for example, the 2015 Nobel Prize in Medicine).
To be clear, evidence for both of these treatments (in the Covid-19 case) is still equivocal and weak (in particular, the ivermectin positive studies mostly come from parts of the world where certain parasites are endemic, and it may be that ivermectin knocks down those parasites, simply allowing the patient a better shot at recovery), but there was certainly no slam-dunk case against either of them in the summer of 2020. Here is a good summary of the much later (February 2022) state-of-play on HCQ.
The Fauci/Collins Conundrum
At least two highly-placed Federal government employees, Francis Collins (director of NIH) and Anthony Fauci (director of NIAID), have played prominent roles in this tragedy as described so far. Personally, I expected nothing better from Fauci, but was extremely disappointed in Collins (I suspect that, though technically Collins was above Fauci in the org chart, he allowed himself to be strongly influenced by the older and more-bureaucratically-experienced Fauci: Collins was NIH director for 12 years, Fauci has been NIAID director for 38 years so far). I keep trying to be less cynical about bureaucrats, but every attempt is met by reality “out-cynic-ing” me!
At one level, the problem here is structural: the same people (no matter who they are) should not be in charge of both public health policy and research funding – there is just too much temptation to retaliate by de-funding any researcher who dares oppose dicta from the top (and it’s nearly impossible to prove that this happened).
Furthermore, there is no reason whatsoever that these two (a geneticist and an immunologist) should be dispensing public health policy – neither one has the slightest training in the field, and, in the case of Fauci, possesses a demonstrated lack of judgment and suitability of temperament. I have just as much training in public health as either of them (i.e. none, though I may have a better statistical background), and I certainly shouldn’t have been any sort of “font of public health policy” (in fact, I think no single person should).
Fauci, in particular, has disqualified himself from any role in public health as a result of his performance in the AIDS crisis in the 1980’s/90’s. In this, he showed four traits of judgement/temperament that should disqualify him from a role in public health (see here and here for recent overall summaries, but my opposition to Fauci dates from having seen these events myself “back in the day”).
- A bias against novel uses for well-understood drugs: specifically, Bactrim as prophylaxis against pneumocystis pneumonia (PCP) in AIDS patients, which is the now-standard approach
- A bias toward stating possibilities as certainties: specifically, his discussion of the impending heterosexual AIDS outbreak, which never really happened
- A bias toward fearmongering: specifically, his discussion of the spread of AIDS by casual household contact
- A failure to read carefully, specifically reading only abstracts and summaries (there is evidence that the preceding failure [AIDS spread by casual contact] was caused by his failure to read anything beyond the summary of one paper before writing his own JAMA editorial on the subject; he demonstrated this in the Covid-19 era by his dismissive “TL;DR” response to an experienced physicist warning him not to take China’s assurances at face value).
There were a few later Covid-era developments that were not apparent in the AIDS era (like his bizarre insistence that any policy disagreement with him was really an opposition to “science”), but the early ones were enough of a red flag that he should have been ignored in any public health context thereafter.
I will mention also that some have come to the conclusion that maybe he was just trying to keep people safe when we didn’t have full knowledge of things, but that really doesn’t wash – in particular, during the time that he was publicly (and influentially) opposing Bactrim prophylaxis for PCP in AIDS patients, 16,000 patients died unnecessarily from PCP (and there was LOTS of evidence for this approach, even if it wasn’t the sort that he wanted). Furthermore, even after things were “fixed” in 1989 for Bactrim, and eventually AZT dosages and combinations were developed that reasonably threaded the needle between effect and side-effect, many AIDS patients still would not take any of the approved meds because of the lack of trust engendered by Fauci in the gay community. People I loved very much are dead today, and Anthony Fauci is, at least fractionally, culpable.
To be completely clear, Fauci was a quite good immunologist early in his career, and he ran an excellent immunology research lab that was the basis for his promotion to Director of NIAID at the age of 44. It wasn’t until his move into public health policy (the AIDS crisis was raging at the time of his appointment to his current job – it’s easy to see why an immunologist would be asked his opinion about AIDS policy) that he began to make his significant missteps. At that point, he was asked for opinions related to public health policy, and he volunteered them. Given the consequences of those opinions, he should never have been asked about public health again, but should have continued as perhaps a standard-issue bureaucrat (like the other 26 directors of National Institutes parallel to NIAID, of whom you have likely never heard), or, arguably, moved back to a research lab where he had been quite effective.
Postscript: after this post was written, this fascinating transcript was published (May 2022) of a panel discussion among the three authors of the Great Barrington Declaration, discussing many of the same ideas as this post. Here is an even later essay from one of them (Bhattacharya) about the failure of his university to provide him with his promised academic freedom.
Still more recently (July 2022), Marty Makary (Johns Hopkins) and Tracy Beth Høeg (FL Dept. of Health) wrote here (later picked up by the NY Post here) about the demoralizing effect all this had on the actual, working public health researchers at the CDC, FDA, etc. Scientism is NOT good for science!
Even later (January 2023), Kevin Bass (a MD/PhD med student) wrote about many of these same topics, specifically saying that many of these errors cost lives during the pandemic.
In addition, Glenn Reynolds has written here and here about similar topics (not later chronologically, but I inexplicably failed to include these links in the main essay).
Finally, here is a July 2023 report by epidemiologist Ashley Croft (whom I didn’t previously know) for the Scottish government, giving detailed information on coronaviruses, timelines, evidence on efficacy of just about every intervention attempted, etc. He’s not really writing about “scientism” but his information supports and reinforces my main points about the Covid-19 pandemic. Highly recommended for a variety of reasons!
* It’s worth remarking that Henderson in this paper is Donald Henderson, the late distinguished doctor, epidemiologist and educator, best known for having overseen the worldwide effort to eradicate smallpox with USAID and WHO – not a lightweight (or “fringe”) in public health!
It’s truly a tragedy that Dr. Henderson died in 2016, so he wasn’t around to help remind the public health establishment in 2020 of what they had known a few years earlier. Perhaps he had sufficient status not to be ignored!
** Sweden quickly became the “red state” of Europe, where every uptick in deaths or cases was met with a gleeful “See, we were right!” and every downtick was met with awkward silence. Ultimately, the deaths/million rate (total crude mortality) in Sweden, as I write this, was 1765 – around 60% of the US rate; among European countries, better than 29 countries, worse than 18 (also see here for a recent “excess mortality” assessment). Of course, raw comparisons like this don’t take into account population demographics, so they are of little value, except to show that Sweden didn’t have an outrageously bad Covid experience, as one would be led to believe if one were to have read the press at the time.
*** The early lack of discussion of any non-zoonotic origin is largely because of an early paper in Nature Medicine and a similar letter in the Lancet, written by people presumably in a position to know (to be clear, I bought it at the time). However, persistent searching has failed to find any reservoirs in non-human populations (and it turns out that there weren’t any pangolins at that market, nor was Patient Zero in contact with that market). That doesn’t prove that it wasn’t zoonotic, of course, but it does sharply reduce the probability of that explanation.
Subsequently, a FOIA response produced emails of 31 January 2020, among Anthony Fauci (director of NIAID) and 4 Scripps virologists (including Kristian Andersen, the corresponding author of the Nature Medicine paper). In these emails, Andersen mentions that he and 3 other virologists (including 2 of the other authors of the NM paper), believed that features of the virus were inconsistent with the zoonotic origin hypothesis. A teleconference was held among many of these same scientists on the following day (1 February), about which little is known, since transcripts are almost entirely redacted, though a few available comments indicate that at least Garry (Tulane) and Farzan (Scripps) still believed at that time that the zoonotic hypothesis was unlikely.
By the next day, 2 February, emails from Francis Collins (director of NIH, who oversees $43 billion in annual research money) complained about the “damage to science” from this “conspiracy.” It’s hard to know for sure, but this seems to have been taken as a “will no one rid me of this turbulent priest?” statement by everyone involved. From that point on, none of the participants in the teleconference had a kind word to say about any non-zoonotic origin hypothesis. To be fair, only Garry at Tulane has commented about his change of heart, and he claims that he was legitimately convinced by evidence presented by others at the teleconference. Perhaps that’s true, but if so, why wasn’t that evidence then collected and presented to the public (instead of being completely redacted to this day)?
Postscript to the footnote: after this essay was written, the Republicans took the House of Representatives, and convened a Select Subcommittee on the Coronavirus Pandemic, tasked with (among other things) looking into the role of the US Government in the origin of SARS-CoV-2. The Majority Staff of that subcommittee produced this statement containing new evidence in favor of my speculation above that Fauci and Collins leaned on the 1 February teleconference participants, essentially commissioning the Nature Medicine paper. The statement also contains evidence of the previously-unknown involvement (at least in this specific debacle) of Jeremy Farrar, then director of the Wellcome Trust, later Chief Scientist at WHO.
Postscript to the postscript to the footnote: here is an even later (August 2023) summary of the state of play regarding the evidence of misconduct surrounding the Nature Medicine paper. It’s looking more and more like I was right (and more and more prominent voices are agreeing with my assessment), though I am far from happy about it (and I have no illusions that they read my work on the subject). Here is a very thorough timeline of the whole debacle.
I should also point out the work of two apparently different (though overlapping in membership) groups of researchers, both called DRASTIC (though the acronym stands for a slightly different phrase in each case), that have been investigating all this. Their work appears useful and fairly well sourced, but both groups are at least partially anonymous, so caveat lector.
† or at least viruses with a principally respiratory transmission vector. A good case can be made that SARS-CoV-2 is not really a respiratory virus, but rather a vascular virus with a principally respiratory transmission vector.
†† Using data from here and here, I calculate the age-specific crude mortality as (aggregating all 3 waves together):
Age range | Crude mortality (%) |
0-17 | 0.0014 |
18-29 | 0.01 |
30-39 | 0.04 |
40-49 | 0.10 |
50-64 | 0.30 |
65-74 | 0.68 |
75-84 | 1.50 |
85+ | 3.64 |
In particular, the 85+ and “schoolchildren” brackets differ by a factor of 2600.
††† Anthony Fauci has a long history of opposing the use of standard, well-understood, inexpensive drugs for novel purposes, and this appears to have been yet another data point in that series. More on this here.
‡ Long after this essay was originally written, Matt Orfalea produced this video, making the case that the 3.4% “fatality rate” was the original “Covid lie.” He may well have a point.