Bhattacharya talks with Douthat

Source: Bhattacharya talks with Douthat
Publisher: Michael’s Substack (Free) | Author: Michael Weissman
Published: April 13, 2026 | Archived: July 18, 2026

Share

Trump’s NIH head Jay Bhattacharya recently starred in a long interview with New York Times columnist Ross Douthat. JB presented himself as an open-minded honest scientist, the quintessential reasonable person. A closer look shows that at crucial points his policies rest on falsehoods, evasions, and rigid ideology. With Bhattacharya scheduled to headline an upcoming National Academy of Science workshop on scientific integrity later this month, it seems worth posting these first reactions now without taking the time to work them into a coherent essay.

JB is happy with his boss.

“I actually have tremendous respect for Bobby…. The caricature of him I’ve seen in the press is just totally unfair.”

That’s not just an attempt to make the best of an awkward situation. JB spoke at a campaign rally for RFK Jr. when Bobby was a fringe candidate for President. As PBS reported:

A list of speakers includes Angela Stanton-King, a woman pardoned by then-President Donald Trump for her role in a car theft ring that led to a 2004 federal conspiracy conviction and two years in prison; Metta World Peace, the NBA all-star player formerly known as Ron Artest; and Dr. Jay Bhattacharya, a Stanford Medical School professor who questioned the efficacy of lockdowns during the COVID-19 pandemic and was part of Republican Florida Gov. Ron DeSantis’ presidential launch event last year.

So we have an NIH head who went out of his way to support a candidate who denies that HIV causes AIDS, pushes the “miasma theory” of disease, and has based his career largely on crazy anti-vaccine claims, already deadly even before he took office.

JB’s biggest point is that public health authorities drastically overdid interventions intended to reduce illness and death from Covid. He claims that his own predictions and recommended policies were vindicated as the pandemic played out. He’s massively distorting the history of his predictions and the consequences of the policies.

JB’s core prediction early in the pandemic was that net US mortality would be in the range of 20,000 to 40,000. He took care to publicize this estimate as widely as he could. The strategy he advocated in the Great Barrington Declaration was to allow the pandemic to burn out in a few months as most people acquired immunity from infections, while providing “focused protection” for seniors and others at highest risk of dying. He wrote that if millions of deaths were possible then other strategies would be called for.

The mainstream picture was the Imperial College (IC) report published sightly before JB and his collaborators weighed in. IC estimated that for populations like the UK or US the infection fatality rate (IFR) would be about 0.9% so long as the health care system was not overloaded with an intense wave. Given Covid’s high transmissibility it would be expected to infect most of the population before immunity stopped its spread, giving an expected mortality of about 2.2M in the US. The IC report called for a variety of non-pharmaceutical interventions (NPI) to avoid overloading health care and to delay most infections until either vaccines or improved therapies were available.

Who was closer to being right? The pre-vax IFR for Covid in the US ended up being ~0.7%, close to the standard IC estimate. It was about four times larger than the estimate Bhattacharya came up with in a notorious tendentiously flawed paper.

Almost everybody got Covid, so if the spread had not been delayed about 2.2M would have died in the US, just as in the standard estimate. Since the NPIs delayed a large fraction of the cases, especially among the most vulnerable, until after they were vaxed the current excess mortality in the US is about 1.4M. That’s a little more than 0.4% of the whole population. Even now in the nominally “post-covid” period with almost everyone having some protection from either previous infection or vaccination, it’s killing over 100,000/yr in the US, about three times as many as JB estimated it would kill over its whole course.

The standard estimates used by the mainstream epidemiologists were rather accurate. The mortality predictions of JB and his collaborators were spectacularly wrong. Yet JB is still boasting of his work underestimating the IFR and bragging that his policy proposals were correct.

To Douthat’s credit, he tried to pin JB down on what exactly he would have done to protect vulnerable people while the pandemic swept through almost all the rest of the population.

Douthat: And it is profoundly risky for older Americans, senior citizens, and so on. It might be that it’s just really hard in a big, diverse, fluid society to design a set of policies that separate off old people from the rest of society. Maybe you can do that to some degree with nursing homes, but most old people — we have a lot of old people in America, and they’re embedded in communities, towns, cities, families. They, too, want to go to Thanksgiving dinner, everything else.

JBs main proposed measure was “I wouldn’t have recommended sending Covid-infected patients back to nursing homes.”

That’s unquestionably right (and the opposite of what Cuomo did in New York), but it doesn’t come close to answering Douthat’s question. Under pressure, JB goes on to mention a trivial further measure, not having nursing home employees work at more than one home. He also mentions organizing food deliveries for elderly people living on their own, another good idea at the margins. None of his remarks show any familiarity with the staffing situation in most actual American nursing homes. Even now, six years later, JB can come up with no comprehensive measures that would have come close to providing the “focused protection” on which his policy was based.

A remarkable exchange follows.

Bhattacharya: Yeah, exactly. People died at home with heart attacks in 2020 because they didn’t go to the hospital. But also, more broadly, the economic dislocations caused by the lockdowns certainly killed vast numbers of people.

I think the fundamental error is: People think that “well, the lockdowns sort of worked.”

Douthat: Just in the sense of pushing some potential deaths into the future, past the point where we got the vaccine, that seems to me to be the strongest case.

Bhattacharya: Yeah, so that’s the argument. But I’d say a couple of things about that. So one, we didn’t know the vaccine was going to work.

Douthat: Right.

Bhattacharya: That was not a certainty. And the idea that when there’s this kind of uncertainty, you must do this extraordinary draconian measure and you take away basic civil liberties at scale for nine months or however long until you get the vaccine, that, I think, is the end of civilization. If that is our paradigm for managing these kinds of risks, we can’t have at least a free civilization. Because you can’t make plans. You don’t know if your kids are going to be able to go to school. You can’t make basic plans.

Douthat tries to get JB to acknowledge that by delaying many cases among the most vulnerable until after vaccines were available the public health measures saved many lives. (Douthat doesn’t put numbers on it but that would be roughly 2.2M-1.4M= 800,000 lives.) JB refuses to acknowledge that obvious basic point. Instead he claims “the lockdowns certainly killed vast numbers of people”, without providing any evidence or numbers. The clearest signature of the effects of the various NPIs on non-Covid mortality was the near elimination of flu deaths for one year. It’s uncertain whether such mortality-reducing effects were bigger or smaller than increases in mortality due to lock-down related effects. The net excess US mortality of 1.4M is approximately equal to the mortality caused by acute Covid plus some caused by direct post-Covid effects, mostly cardiovascular damage.

JB then justifies his early policy of letting Covid spread rapidly through the population in part by saying that “we didn’t know the vaccine was going to work.” The GBD came out in early October, 2020. By that point we all knew that several vaccines were well along in clinical trials where the preliminary results looked excellent. (E.g. I had written of excellent preliminary results for multiple vaccines in early August 2020.) The first requests for authorization in the US were submitted Nov. 20, with authorization on Dec. 11. JB’s excuse for encouraging rapid spread of the virus while offering only vague verbal protection to the vulnerable was that there was some chance that the vaccines coming online soon might not work. Does anybody take that seriously?

As it turned out, the GBD declaration and the conventional IC report shared one basic error. Both assumed that herd immunity would set in once enough people had either been infected or vaccinated. That assumption wasn’t crazy since initially breakthrough infections were rare and the virus’s immune-evasive evolution didn’t seem to be very fast. For the conventional strategy, it looked like an updated booster shot every year or two would provide better protection than flu shots, since flu evolution is fast. Even a fairly slow evolution of immune evasion would have undermined the GBD strategy since it relied almost entirely on herd immunity, but the GBD group ignored that problem. In fact, their claim was that herd immunity was probably already almost achieved and that no further big waves would occur. Once SARS-CoV-2 broke out of its initial narrow evolutionary local-optimum range and began rapidly exploring a much broader range of sequences, it became clear that any strategy based on herd immunity would fail. None of JB’s comments even hint that the assumption underlying the GBD strategy turned out to be false.

Looking ahead, it’s pretty clear that “lockdowns” or any measures approaching that are not going to be popular and for an infection like SARS-CoV-2, with a high R0 and fairly low mortality, would only be practical for short-term delays. On the other hand, an infection like SARS with a lower R0 and higher mortality can be (and was) stopped by vigorous NPIs. A SARS-like virus with a higher R0 than SARS but lower than the initial SARS-CoV-2 might be stoppable only by stronger NPIs, amounting to what would be called lockdowns. A reasonable anti-lockdown position would be that they should not be used except for more lethal viruses (like SARS) with R0‘s higher than SARS. Instead, JB makes an absolute blanket promise:

“Lockdowns were the biggest public health disaster in history and it’s true I won’t ever recommend one.”

So JB has a firm ideological commitment to allow even a stoppable highly lethal virus to propagate throughout the population if stopping it would require some sort of lockdown. This is not a far-fetched hypothetical scenario since obviously new viruses as lethal as SARS can pop up and R0 can be anywhere in the range of typical flus or coronaviruses.

\[5/15/2026 Andes hantavirus is a reminder that the category of pathogens for which lockdowns might be needed is not at the far fringe of possibilities. Its fatality is high and its transmissibility may be, if we’re lucky, just below the level at which lockdowns would be needed. Its [best-documented outbreak](https://www.nejm.org/doi/full/10.1056/NEJMoa2009040) was, in fact, stopped by a local lockdown.\]

JB has shut down all research on mRNA vaccines for respiratory infections. The administration initially refused to even look at the Moderna trial results for an mRNA flu vaccine, although they’ve reversed that under pressure. This extreme anti-mRNA policy is dangerous because there is currently no alternative to mRNA for rapidly scalable vaccines for new pathogens. Protein-based vaccines require an extra step to produce and cannot be scaled up quickly enough for initial responses to new pathogens.

JB is shutting down the only technology suited for rapid response to new pandemics while also promising to never use lockdowns no matter how severe the new challenge. In effect he’s pre-emptively announcing surrender to any bad new infection that might arise. Fortunately, viruses don’t read the news, so they don’t know we’ve already surrendered. Unfortunately, people who might be engineering new viruses do read the news.

\[6/10/2026: I may be seriously mistaken in the following paragraph and part of the next. George Davies Smith has tried a [similar regression discontinuity](https://www.youtube.com/watch?v=qlTnnQytOJ0) analysis of data from England, finding negligible effect. The reason for the discrepancy is unknown.\] JB complains that it’s hard to persuade people to fund a big RCT to check the result obtained by some of his colleagues that getting a shingles vaccination reduced progression toward dementia. That result comes from a robust regression discontinuity observational study using an abrupt change in vaccination rate in Wales. It confirms a previous study finding that an improved shingles vaccine reduced dementia rates compared to a less effective older vaccine. It is now confirmed by a careful inverse propensity weighted observational study. Contrary to the impression JB gives, the obstacle to a large RCT is not funds. It’s that a standard RCT would require giving a placebo to people who want to take a valuable shingles vax. Even if there was some way to take that unethical step, the results would take a long time to come in. So although on the surface JB is proposing research that could promote vaccine use, in reality he’s proposing not taking action on solid research \[6/19/2028- maybe not\] showing that a vaccine has a huge benefit beyond what had been previously known. What might sound good to an uninformed reader actually fits well with RFK Jr’s anti-vax campaign.

The dementia-protective effect of shingles vax does call for more research, but of a very different type. It strongly supports previous suspicions that Alzheimer’s is often accelerated by immune reactions to infections, specifically herpes viruses. To what extent do other viral infections contribute to Alzheimer’s and other neurological disease? There’s strong reason to think that’s a major public health issue, e.g. the likely role of Epstein-Barr infection in causing multiple sclerosis. Multiple studies find evidence of major neurological problems from Covid infections, including for young people. What we need from NIH is intense research on such effects. To what extent are they real? How much can they be reduced by vaccination, as with the shingles vax? What are the mechanisms? What antivirals or other treatments might reduce the effects? But none of that would fit with the MAHA agenda, which treats infectious diseases as relatively unimportant.

In fact, JB does not mention the existence of multiple post-Covid health effects on people of all ages. He hasn’t shut down the NIH-funded research on long Covid, but it’s not reassuring that he doesn’t even mention its existence in talking about the cost/benefit analysis of letting Covid spread rapidly.

JB claims that the evidence strongly points toward a lab leak. I fully agree. There are two obvious implications.

One implication is that NIH should take the lead in shutting down the most dangerous potential pandemic pathogen research, using its expertise to help write guidelines for laws and treaties to stop that sort of work. Although JB does say that such work at NIH has been paused he chose Jeff Taubenberger to replace Fauci as the new head of NIAID. Taubenberger is the researcher who for no particular reason recreated the virus that caused the 1918 flu pandemic. He coauthored one of the worst papers pretending that it was certain that Covid came from standard zoonosis. He’s the poster child for reckless research.

What was JB thinking? Who knows? Maybe it’s a lot easier to say that your political enemies should follow rules (e.g. the Biological Weapons Convention) than to follow those rules oneself once one acquires power. At any rate the promised new guidelines haven’t shown up yet. Maybe we should expect them in the same “two weeks” that have elapsed since Trump announced the imminent release of his health care plan, in 2015.

The other obvious implication is that NIH should finally release all information shedding light on the origin of Covid. Instead, NIH continues to dribble out documents redacted as if they were the ones describing Trump’s partying with Epstein. Why would JB and the rest of the Trump administration release none of the information on the lead-up to Covid, despite a unanimously passed law requiring them to do so and despite their posting a cheesy website announcing that Covid came from a lab leak? One possibility is that they don’t want any new information to get out about the years 2017-2019 in which the dangerous research was conducted while Trump was President. In 2017 under Trump the government reversed the nominal ban that had been instituted under Obama on funding dangerous gain of function research.

JB proposes a novel psychological explanation for what he considers to have been serious public health over-reaction to Covid. It’s that public health officials over-reacted out of guilt over their role in the research that led to the pandemic. That speculation can’t account for why most of the public health establishment fairly quickly supported something like the measures that were taken. (Some informed libertarian-leaning analysts supported the measures even sooner.) Very few of these people had any connection with the risky research projects. It seems most were not (and some may still not be) even aware of the likelihood that the pandemic leaked from some research project.

NIH (and currently CDC) are being run by an ideologue with an explicit near-religious commitment to never use lockdowns to fight any pandemic regardless of its infectiousness or its morbidity and mortality. He twists data and models to fit that ideology.

JB is evidently a believer in the creed described by Brad Delong:

“The Market giveth and the Market taketh away; blessed be the name of the Market.”

He is committed to making whatever sacrifice that god demands on its altar.

Share

Leave a comment