Focused Protection, Not Lockdowns: Jay Bhattacharya on What Public Health Got Wrong

Lex Fridman asks Jay Bhattacharya, a Stanford professor of medicine, health policy, and economics, for the best current measure of how deadly COVID is. Bhattacharya says the answer does not come from counting people who show up sick. It comes from seroprevalence studies: antibody surveys of the population at large, which estimate how many people have already been infected, including those who never entered the testing system. He was a senior investigator on three of the earliest such studies, in Santa Clara County, Los Angeles County, and among Major League Baseball personnel across the United States.

COVID’s symptoms run from nothing at all to fatal viral pneumonia, so case counts miss a huge share of infections. Deaths look easier to tally, Fridman notes, yet even the numerator is messy. The CARES Act paid hospitals tens of thousands of extra dollars per COVID patient and added a 20% Medicare bump for elderly COVID cases, which created a real incentive to classify patients that way. Death certificates add a further tangle: a person with diabetes can go into heart failure, fill the lungs with fluid, catch COVID, and die. What, Bhattacharya asks, do you write as the cause?

Audits in Santa Clara and Alameda counties found that about 20 to 25% of listed COVID deaths were more likely incidental than directly caused by the virus. He does not get too excited about that, calling the cause of death almost always multifactorial — a philosophical question, like whether the bus or the brain bleed killed you. What does interest him is the psychology: for the first time in living memory, a developed country faced a disease that could kill “us,” and the lizard brain put that one risk at the front of every form, every paper, and every conversation.

Asked what historians will say about COVID’s deadliness, Bhattacharya answers that it depends on who the virus infects. The single most important fact in the seroprevalence literature, mostly from before vaccination, is a steep age gradient. Roughly, people over 70 faced about a 5% chance of dying if infected; people under 70 faced about 0.05%. There is no sharp cutoff. His rule of thumb: at age 50 the infection fatality rate is about 0.2%, meaning 99.8% survival; double it for every seven years older, halve it for every seven years younger. Severe chronic disease such as diabetes or morbid obesity is like adding seven years.

The Santa Clara and Los Angeles studies used antibody kits ordered in March 2020 that later received FDA emergency-use authorization. The false-positive rate was about 0.5%, checked against 2018 blood that should have had no COVID antibodies; the false-negative rate was around 10 to 12%. False positives matter most when true prevalence is low, so the team adjusted with a standard formula. In community-dwelling populations — not nursing homes, jails, or the homeless — both counties showed an infection fatality rate around 0.2% in April 2020.

Stanford colleague John Ioannidis later meta-analyzed a hundred or more seroprevalence studies and put the worldwide median infection fatality rate near 0.15%. That is not one number for every place. Africa’s rate was extremely low; New York City’s early rate was much higher. The virus does not have identical properties wherever it goes, Bhattacharya says, because it kills in interaction with the host. Africa is about 3% over age 65. New York’s early wave, he adds, differentially infected older people in nursing homes because of a policy choice: sending COVID patients back to those homes to keep hospital beds free. “The policy you adopt determines who is most exposed.”

The more important finding, Bhattacharya insists, was not the 0.2% fatality rate. In both Santa Clara and Los Angeles in April 2020, the studies found 40 to 50 times more infections than reported cases. For every person known to public health, dozens of others already had antibodies. Many of those people reported no symptoms. If a few percent of a county already had antibodies, test-trace-isolate and lockdowns aimed at driving the virus to zero were already too late. The virus was not going to go away.

The Santa Clara paper went to medRxiv in April 2020 as a preprint; the Los Angeles study later appeared in *JAMA*. Bhattacharya says medicine is less used to pre-peer-review debate than economics, and the blowback was enormous. Critics argued that if the fatality number was too low and later proved wrong, the social damage would be huge. He replies that they were pretty sure, and that they turned out to be right. The result mattered because the world had locked down in mid-March. If the prevalence number was right, the strategy of suppression to zero had already failed.

What pushed him to run those studies so fast was a memory of H1N1 in 2009. Early case-fatality estimates, using only people who showed up at the doctor as the denominator, ran around 3 to 4%. Two or three years of seroprevalence work later, infections outnumbered identified cases a hundredfold, and the fatality rate fell to something like 0.02%. COVID spread in a similar person-to-person way. He wanted that denominator correction in months, not years, so that who was at risk — and what strategy followed — would not wait for a slow scientific autopsy.

Fridman asks where COVID stands next to influenza, given how often people collapse the comparison into “just the flu.” Bhattacharya takes the flu seriously: it kills on the order of 50,000 to 60,000 people a year in the United States, depending on the strain. The salient difference is the age gradient. The flu is more deadly for children than COVID is, with no real controversy on that point. COVID, thank God, produces much less severe disease in kids. The 1918 H1N1 pandemic killed millions of younger people; COVID has not. Something like 70 to 80% of COVID deaths have been in people over 60.

Parental fear still centered on children, which he understands as a moral instinct and rejects as a reading of this virus. Overall, COVID would take many more lives than a typical flu season. A 0.2% infection fatality rate sounds small and is not. At the time of the conversation, official U.S. deaths were around 800,000, and he thought crossing a million was likely. Variants would keep coming. The end state, he says, is that COVID is here forever: infection produces strong protection against reinfection and against severe disease the second time, with one-year reinfection rates in follow-up studies around 0.3 to 1%, and those second infections much less likely to hospitalize or kill.

Fridman reads the October 2020 email from NIH director Francis Collins to Anthony Fauci, which had just surfaced through a Freedom of Information request. “This proposal from the three fringe epidemiologists who met with the secretary seem to be getting a lot of attention,” Collins wrote, “and even a co-signature from Nobel Prize winner Mike Levitt at Stanford. There needs to be a quick and devastating published takedown of its premises.” The proposal was the Great Barrington Declaration, co-authored by Bhattacharya, Harvard statistician Martin Kulldorff, and Oxford theoretical epidemiologist Sunetra Gupta. Bhattacharya says he felt disappointment more than surprise. He had looked up to Collins for years. He had also felt the brunt of a propaganda campaign from the day the document went up.

They posted the Declaration around October 4, 2020. Tens of thousands of doctors and hundreds of thousands of other people signed. It was not, he says, a radical document. It argued that lockdowns aimed at suppressing the virus to zero could not work and were causing catastrophic harm to the poor and vulnerable, and that there was an alternate path. What he hoped for was a discussion of how to protect older people better. What he got was a slogan. Reporters kept asking why he wanted to “let the virus rip.” Those words do not appear in the Declaration. He calls the phrase a propaganda term aimed at the fear centers of the brain, a way to preclude debate about the policy already in force.

Fridman can steel-man the email as panic, even as something he might have written in a private late-night moment. What broke his heart was doubling down on “fringe,” and the phrase “devastating takedown” from a man who sits on billions of research dollars. The published takedown, Fridman notes, would come from scientists who depend on that money. Bhattacharya adds that they were not saying something Galileo-level new. They were restating the old pandemic plan, versions of which lockdown scientists themselves had written in late February and early March 2020. The signal to everyone else was: stay silent even if you have reservations.

Bhattacharya distinguishes two pictures of science. In the one that attracted him, science is a structured conversation tempered by evidence: if I believe A and you believe B, we design an experiment, one of us buys dinner, and we go on. In the other, science is treated as truth in itself. “What does the science say to do?” is the wrong question. Science says that if we do this, that is likely to happen; human values choose which outcome we prefer. The arrogance in Collins’s email, he argues, is the same species as Anthony Fauci saying on television that to criticize him is to criticize science itself: the certainty that one policy is not only right but unchallengeable.

Fridman describes walking into a Clubhouse room that billed itself as “pro-science,” full of well-credentialed doctors mocking dissenters. He came on stage, started to say the arrogance would divide rather than unite, and was thrown off with “enough of that nonsense.” Bhattacharya hears an echo of the email. Once science is an identity, inquiry leaves the room. He puts the failure on institutions, not only individuals. The NIH’s job is to fund scientists and sometimes to steer them toward important problems with pots of money. It is not a policy organ; that is the CDC. During the pandemic, NIH leaders were pulled into public-health decision-making. Then the head of the NIH could tell scientists they must agree with his recommended policies or be fringe.

Fridman still thinks Collins is a great man and a great scientist, someone who lit up talking off-mic about the vaccine because he would get to help people. Bhattacharya does not impute bad intentions. He thinks the dual role — funder of science and setter of policy — made humility almost structurally impossible, though Fridman pushes back that great leaders with humility are remembered. The Human Genome Project, Bhattacharya notes, would not have happened without Collins. That makes it tragic that the career ended this way. Asked how to interview such people, he tells Fridman to keep doing what he does: no gotcha, sympathetic questions that paint strengths, weaknesses, and constraints, so listeners understand the person and the moment.

Bhattacharya names two mistakes of his own. In March 2020, from his experience with vaccine development, he thought a shot would take a decade or more. When preliminary Pfizer trial numbers suggested it would work, he was as happy to be wrong as he has ever been. The second mistake, in January 2021, was believing the vaccines would stop infection, because the antibodies looked neutralizing. Israeli data from a highly vaccinated population soon showed infection surges anyway. For the first two or three months after the second dose, efficacy against infection peaked around 60 to 70%. By six to eight months it dropped to about 20%; a Swedish study suggested it might fall to zero. Breakthrough cases were infectious for a shorter period, but infectivity per day was about as high.

Safety is harder to communicate. Trials of tens of thousands of people cannot map rare events once a shot is given to everyone, and once efficacy appeared it was no longer ethical to keep a placebo arm, so long-term trial safety data disappeared. He was not surprised that some people were skeptical. At population scale, though, protection against hospitalization and death has been borne out, especially in older people: something like a tenfold drop in fatality risk, which he calls a godsend. That, he says, turns focused protection from a hard social problem into a much more manageable one, because vaccinating the old *is* focused protection.

In December 2020 he and Sunetra Gupta wrote a *Wall Street Journal* op-ed with a naive title: they could end lockdowns in a month by vaccinating the vulnerable and opening up. The logic was the Declaration’s logic with a better tool. What followed instead was a transformed vaccine debate. After rollout, systems like VAERS collect raw reports with no control group, which cannot settle causality in older people who sometimes die days after a shot for other reasons. Matched-cohort systems such as VSD and BEST did pick up real, rare harms: myocarditis in young men on the order of one in 5,000 to one in 10,000, and cerebral blood clots in middle-aged women after the J&J shot on the order of one in hundreds of thousands. His messaging advice is simple: tell people the risks. Do not treat them like children.

What most undermined demand for the vaccine, he argues, was a “weird denial” that people who have recovered from COVID have extremely good immunity against infection and against severe disease. Anyone can look at the data: recovered people are not getting infected again at high rates. Deny an obvious fact, and people distrust the rest of the CDC’s message. “People are extremely good at detecting arrogance and bullshit.” Fridman sees the same tribal split in the audiences of friends like Joe Rogan and Sam Harris, smart people lining up on opposite sides of kids’ shots and mandates. Bhattacharya reads the schism as a symptom of distrust. He does not see the same fight in Sweden, where he thinks health authorities stayed more trustworthy.

Politics did the rest. Before the 2020 election he testified in the House, invited by Republicans, on how to measure vaccine safety. Democratic-invited experts with long records of keeping unsafe shots off the market openly doubted whether the FDA would do a reasonable job. After the election, he says, the rhetoric switched on a dime, and Republicans were cast as the hesitant ones. If an election can change how people talk about vaccine safety, the public notices that this is no longer science. Talking down to hesitant people — calling them rubes who want everyone to die — fuels the very hesitancy it condemns. Women considering pregnancy, for example, had legitimate questions the data could not yet answer. Pretending certainty is what people smell.

On Pfizer, Bhattacharya shares Fridman’s ambivalence. Discovering a vaccine and manufacturing it at a scale of billions in a short time is a remarkable achievement that could not have happened without companies like Pfizer. Money also distorts. No Pfizer-scale firm, he says, has an interest in cheap repurposed drugs; even Merck, which once held the ivermectin patent, has no interest in checking whether it works, and talks about the curious with the same “fringe” arrogance. The NIH-funded ACTIV-6 randomized trial of ivermectin was not due until March 2023. Meanwhile the FDA tweeted “horse paste” at desperate sick people. That, he says, was deeply irresponsible. His hope is that young scientists can detect the bullshit and refuse to repeat it.

The Great Barrington Declaration is named for the Massachusetts town where they met, not for grandeur. Kulldorff, who had designed statistical safety systems the FDA uses to track vaccines, invited Bhattacharya and Gupta to a small conference. They found they had independently reached the same strategy, and Bhattacharya, with the Port Huron Statement somewhere in the back of his head, asked the name of the town. The text is less than a page, written for the public, and rests on two scientific claims he says were not controversial: the massive age gradient, and the fact that lockdown-focused policy has devastating health consequences. The fight was never over those facts. It was over whether policy would follow them.

By October 2020 the UN was already warning that economic dislocation from lockdowns would starve tens of millions; he says hundreds of thousands of children later died of starvation in places like South Asia. Tuberculosis treatment stopped. Childhood campaigns for measles, rubella, and DPT were scaled back, leaving tens of millions of children without doses of diseases that actually kill them. In India, an order sent some 10 million migrant workers walking or crowding onto trains back to their villages overnight; about a thousand died on the way. A seroprevalence study in Mumbai found 70% positivity in the slums by mid-2020 versus 20% in the rest of the city. Lockdowns protected the relatively well-off and spread disease among the poor.

In rich countries the harms were different and still large. A July 2020 study found that one in four young adults had seriously considered suicide. Suicide rates did not spike that high, but the despair itself is a public-health alarm. Blue states closed schools for a long time; richer parents bought pods, tutors, and private classrooms, while five-year-olds were taught to read on Zoom. Health-economics research, he notes, links even short schooling disruptions to lower lifetime income, shorter lives, and worse health. Breast and colon cancer screening stopped, diabetics missed counseling, people died at home of heart attacks. Labor-force participation fell about three percentage points, much of it women staying home with children. Child-abuse *reports* dropped because schools, the main detection system, were closed; actual abuse, he says, went up.

Early in the pandemic, a reporter warned him that putting economic harms in a story would get him canceled. The March 2020 narrative was that if you care about money you are evil; you may care only about lives. The problem with that narrative, Bhattacharya says, is that the “money” is the lives of poor people. Throw a hundred million people into poverty and you will see mortality and sickness rise. Public health had so focused on COVID that it forgot every other priority. The laptop class, insulated from the costs, congratulated itself on possessing the truth.

Put a steep age gradient together with enormous lockdown harms, Bhattacharya says, and the obvious policy is to protect the people at highest risk while society keeps functioning. That is focused protection. Implementation is always local: it depends on who is at risk, where they live, and what the community can do. In the United States, many of the most vulnerable live in nursing homes. Chinese data already showed that in February 2020. Policymakers instead treated hospital beds and ventilators as the binding constraint. Northeastern states then sent recovering COVID patients back to nursing homes to free those beds, spreading the disease through the group most likely to fill the beds again. Protect the vulnerable, he says, and the hospital problem eases on its own.

He is not generally in favor of forcing a lockdown onto the old, because you will not get cooperation. You can deliver groceries, as some places did with senior-only shopping hours that still required going out. You can have a neighbor volunteer. In crowded multigenerational homes in Los Angeles County, a grandmother who worries that a grandson was exposed could call public health and get a hotel room until a test comes back. Everyone else goes on with life. People would still impose some caution on themselves. What you should not do is close businesses, churches, synagogues, and schools. Lockdowns, he notes, did not stop spread anyway.

Fridman objects that in a heterogeneous society the young will forget, government will not fund deliveries, and a good idea will collapse. Bhattacharya says that is always a risk, but March through summer 2020 still held a “we’re in this together” empathy that was spent on stay-at-home orders, mask shaming, and plexiglass — which in retrospect may have made spread worse — instead of on the people who actually needed shielding. Local public-health officers who know their communities should have done that work. For a future pathogen the same *principle* holds, even if the protected group changes: in the mid-century polio years, the group at most risk was children, and closing a school during an outbreak made sense. There is no single formula. There is a principle of concentrating protection where harm is greatest, including harm caused by the policy itself.

At the time of the taping, Stanford had just gone back to virtual teaching over Omicron. Bhattacharya calls it a failure of the educational mission, and of the last year and a half more broadly. Students are young, mostly vaccinated, and far more harmed by losing in-person teaching than by COVID. Focused protection would have offered alternative arrangements to older faculty, not shut the young out of the classroom they were promised. After the vaccine, he says, let everyone teach in person. Great universities did harm by modeling this behavior. A university should be a beacon, a place to play in the space of ideas, not a monument to pathological risk aversion.

The deeper error, he argues, is that COVID policy aimed to stop the disease from spreading rather than to reduce the harm from it. There is no technology to accomplish the first. A serosurvey of white-tailed deer found that 80% in the United States already had COVID antibodies. Dogs get it, cats get it; human-animal transmission is almost certain. Vaccines do not stop spread. Lockdowns do not stop spread. “We’re burning the earth trying to do something that’s impossible,” he says, rather than working on what is possible — including ordinary college. Universities should fight tooth and nail not to shut down. When administration, lawyers, and safety culture define the mission, the faculty and students have already lost. The mission is not safety. “The university should be dangerous places” — for ideas.

Fridman asks whether fear had value in January 2020, when the disease might have been another Black Death and societies were rusty. Bhattacharya’s answer is almost entirely on the other side. Fear should never be used as a tactic to manipulate behavior by public health. Individual fear is natural; humans facing infection already exaggerate contamination in their own minds. The job of public health is to help manage that energy into useful action, not to stoke it. In this pandemic, he says, there was a deliberate policy to make people think the disease was worse than it was. Survey after survey showed it. Young people gave up ordinary life both because they overestimated their own risk and because they feared harming the vulnerable. Those two fears fused into a powerful demand for lockdowns.

He is equally cold to the other fear business: conspiracy sellers who tell you to be afraid of the government and of “the man.” The general principle is that fear should not be a tool of public policy. Tell people accurately what the risk is, give them tools that have evidence, and level with them when you do not know. That is the adult way. Instead, public health stoked fear to gain compliance with edicts, and one consequence is a much larger movement of people who now say they have been lied to. Rebuilding that trust, he thinks, will take public health a long time.

Asked for advice to young people considering science, he calls it a wonderful profession: you get to improve lives by having fun, and it is a privilege. Then he asks them to look at the ugliness of the pandemic debate and forgive. Many who behaved poorly were acting from a sense that they needed to do good; their mistake was arrogance and power. Come in with humility and keep the love of the field that brought you there. He grew up poor after moving from India at four; his father, an electrical engineer who once worked at McDonald’s, later helped design rocket guidance systems and died when Bhattacharya was 20. He has feared COVID only once: the night before the Santa Clara study, when his wife and 19-year-old daughter were going out to help collect samples and he did not yet know the death rate. For himself, death is not the end. The meaning of life, he tells Fridman, is simple: “Love one another. Treat your neighbor as yourself.”