The knowledge doctors trust is built to sell drugs

Lex Fridman frames the conversation as a counterweight to his earlier interview with Pfizer CEO Albert Bourla. His guest is John Abramson, faculty at Harvard Medical School, a family physician for more than two decades, and author of *Sickening*, a book about how the pharmaceutical industry broke American healthcare. Fridman asks the snap-your-fingers question: if one problem with big pharma could be fixed, which would matter most?

Abramson sets price, lobbying, and advertising aside. The core failure, he says, is that drug companies now determine the content, accuracy, and completeness of the knowledge doctors believe they need. Once commercial interests take over that knowledge, its purpose becomes maximizing returns to investors rather than optimizing Americans' health. Rebalancing that equation, he argues, is the single most important step toward aiming healthcare in the right direction.

He reaches for a basketball analogy. Trying to keep that knowledge honest without independent guardrails is like playing a professional game without referees and asking players to call their own fouls. The players are paid to win, so you cannot count on them to blow the whistle on themselves. Other wealthy countries have those referees. The United States, Abramson says, does not, and that is the largest way American healthcare is distinguished from its peers.

Abramson quotes Milton Friedman's *Capitalism and Freedom*: government has only three legitimate functions, to preserve law and order, enforce private contracts, and ensure that private markets work. That was a radical idea at the time, he notes, because it was the free-market guru's floor, not a wish list. The United States, he says, is failing on all three, and medicine shows it.

On law and order, he is blunt. Drug companies routinely violate the law in marketing and in how they present trial results. Abramson spent about ten years as a litigation expert, took what he learned in civil cases to the FBI and the Department of Justice, and that case produced what was then, as of 2009, the largest criminal fine in U.S. history. In a 2010 federal trial, he testified, and the jury found Pfizer guilty of fraud and racketeering.

From 1991 to 2017, companies paid about $38 billion in fines, and it was never enough to stop the misrepresentation of data. Rarely are the fines larger than the profits earned from the conduct. Executives have not gone to jail for misrepresenting data that involved even tens of thousands of deaths, including Vioxx and OxyContin. When parent companies plead guilty to felonies, the government usually lets a subsidiary take the plea so the parent is not one step closer to being barred from Medicare. The machinery of enforcement, Abramson says, looks like law and order. It is not working.

Fridman presses on human nature. He believes most people are good, and that many who join a company like Pfizer start out wanting to lessen suffering. How, then, do the outcomes Abramson describes take hold, and why does the narrative that a CEO is simply malevolent become so tempting? Abramson answers with sociology, not villainy.

The higher you rise in the corporation, the more you live inside a culture whose shared game is maximizing investor returns. Inside that milieu, handling scientific data in certain ways starts to look normal. Good people, he says, still evaluate themselves, but the standard of evaluation has been rewritten by the people around them. From outside the bubble, the public expects companies to make money while still playing by rules that insiders no longer treat as binding.

Fridman names the mechanism: a bubble forms in the rooms, marketing becomes part of the game, and paying fines becomes part of the game of science. Without guardrails, people keep moving in that direction. The people who survive, get raises, and keep the CEO job are the ones who play by those inside rules. If a chief executive refuses the aggressive advertising, Abramson and Fridman agree, he gets fired and the next person does it. The game solidifies itself.

Fridman had challenged Bourla on aggressive advertising. Bourla's answer was that Pfizer follows the rules, and that the rules are sometimes too strict. Abramson says both men are right. The campaigns mostly are legal. That, for him, is the indictment. From inside the CEO's culture, stating side effects and approved indications is ethical business. From outside, paying skilled media teams to leave viewers with a grossly exaggerated sense of benefit and a minimized sense of harm is sociopathic, even when it is allowed.

His example is Trulicity, an injectable for type 2 diabetes. It lowers blood sugar about as well as metformin, which costs about four dollars a month; Trulicity costs about $6,200 a year. The manufacturer ran a large trial that showed a statistically significant reduction in cardiovascular risk, then advertised that heart-and-stroke benefit. The ads do not say you have to treat 323 people to prevent one non-fatal event, at a cost of $2.7 million. They also do not say that an intensive lifestyle program reduces those risks far more than the drug does.

To be fair, Abramson notes, no head-to-head study of Trulicity versus lifestyle has been done. That is the point. The United States and New Zealand are the outliers that allow direct-to-consumer drug ads, yet they do not require companies to show that a product beats healthy living. Firms can afford huge trials so that tiny differences become statistically significant. Their question is how to sell more drug, not how to prevent cardiovascular disease in people with type 2 diabetes. If ads cannot be banned, he wants independent review of the takeaway message: chance of benefit, comparison with cheaper care and lifestyle, and the true social cost, not a $25 copay for a tens-of-thousands-of-dollars therapy.

The comparison Abramson wants is not hypothetical. In the 1990s the NIH funded the Diabetes Prevention Program, a randomized trial in people at high risk of diabetes. Participants were assigned to placebo, metformin, or intensive lifestyle counseling. Conventional wisdom among physicians, he says, is that you cannot get people to change, so you might as well skip to the drugs. The study, published in the *New England Journal of Medicine*, showed that wisdom was wrong.

People in the intensive lifestyle group lost about ten pounds, exercised five times a week, maintained it, and cut their risk of developing diabetes by 58 percent. The metformin group cut risk by 31 percent. Lifestyle won, in a well-designed trial without a drug-company imperative to prove a product was better than nothing. Why did that result not remake clinical practice? Because there is no lifestyle industry with tens of millions of dollars to advertise, write guidelines, and send sales representatives door to door.

Fridman tips his hat to voices like Joe Rogan who preach exercise and diet, but Abramson says those messages do not travel through the channels doctors are trained to treat as evidence-based medicine. Lifestyle papers appear in major journals; they still lack the gravitas of industry-funded drug-versus-placebo trials. If he ran the system, presentations of Trulicity's heart data would be stamped in red: this study did not compare the drug with lifestyle changes. Doctors, he says, are trained to forget that the missing arm is missing.

Asked how pharma affects a doctor's mind, Abramson starts with the sources physicians have been taught to trust: peer-reviewed journals and clinical practice guidelines. Doctors assume that when a paper appears in a respected journal, reviewers had the data, analyzed it independently, and corroborated the findings. They do not. Reviewers receive the manuscript, usually prepared with or by the manufacturer. They cannot vet completeness or accuracy. They can check whether the write-up is coherent.

Guidelines then become more important as the information flow speeds up and clinicians need a bottom line. The experts who write them, Abramson says, rely almost entirely on those same publications. The highest level of evidence doctors have been trained to use for evidence-based medicine has not been independently verified. Manipulation, in his litigation experience, happens at all three levels: study design, leaving data out of the aggregate analysis, and changing numbers.

The fix, he argues, is not mysterious. When a trial ends, companies produce a clinical study report, often thousands of pages, tabulating raw patient data against a pre-specified statistical plan. As a litigation expert he could work through one in 20 to 40 hours, not three months, and see whether the published analysis matched the plan. Journals could hire or share full-time statisticians with no other financial incentive, certify the methods, and only then pass papers to volunteer reviewers. After publication, anonymized data should be opened to citizen scientists, because once a finding is in print, doctors' beliefs get sticky.

Vioxx became the largest drug recall in American history when it was withdrawn in 2004, after a Merck-sponsored study showed it more than doubled the risk of heart attacks, strokes, and serious blood clots. Years earlier, in November 2000, the *New England Journal of Medicine* had published a larger Merck-sponsored arthritis trial. The paper said Vioxx was no more effective against pain than drugs such as Aleve or Advil, but caused fewer serious gastrointestinal complications. That paper was never fully corrected.

Merck left three heart attacks out of the analysis. The FDA knew, and refused to analyze the data without them. Two of the twelve listed authors were Merck employees and knew about the omitted events; the ten academic authors did not. Three heart attacks in a 10,000-person study sound trivial until you put them back. They reverse the statistics. The only conclusion Merck could then have drawn, Abramson says, is that Vioxx significantly increased heart-attack risk. The company also narrowed the endpoint from heart attacks, strokes, and clots to heart attacks alone.

This was not a technicality the FDA later blessed. The blind was broken in March 2000. That day, Merck's head scientist wrote an email, later published in the *Wall Street Journal*, calling it a shame that the cardiovascular events were there, but adding that the drug would do well and so would they. The chair of the data-safety monitoring board issued a letter stopping the count of cardiovascular events a month before the trial ended, while gastrointestinal events continued to be counted. Within two weeks of that decision, Abramson says, that person signed a consulting contract with Merck at $5,000 a day.

Fridman had asked Bourla about an FDA request for 55 years to release Pfizer COVID-vaccine documents under the Freedom of Information Act. After that interview, Pfizer petitioned the judge to join the FDA's side. Abramson puts it in one line from the episode's open: Pfizer is not battling the agency; it has joined the agency. The FDA's argument, he says, is Kafkaesque. Going faster would cost $3 million, about one hour of vaccine sales over two years. Pfizer is not offering to pay it either.

He stops to be unmistakable. He is not an anti-vaxxer. He believes the vaccines work, that everyone should get them, and that vaccination cuts the risk of dying of COVID about twenty-fold. He has had three shots. That said, he would give ten-to-one odds there is something in those hundreds of thousands of pages that will embarrass the FDA, Pfizer, or both. Emergency authorization rested on the first few months of a two-year trial; the rest of that data, he notes, was not opened up, and there was no advisory-committee meeting when full approval came.

Why does the fight over early trial data not push him into the anti-vaccine camp? Because those files cover several thousand people, while millions of real-world doses now show benefit, and so far no side effect has overcome it. Fridman adds the fog around that: anti-vaccine voices can get rich and famous too, scientists fear wrecking their reputations, and the only clearly happy actor is Pfizer, running a capitalist machine. Humira, long the world's best-selling drug, peaked around $20 billion a year. Pfizer expected to sell $65 billion of vaccine in the first two years of the pandemic.

Asked what Pfizer-BioNTech did well and badly, Abramson starts with the science. In 2016 the NIH did key infrastructure work for mRNA vaccines, a fact often left out of the story. Pfizer-BioNTech paid royalties to the NIH, not a huge sum in his telling, enough to avoid the kind of litigation Moderna faced. Operation Warp Speed, to President Trump's credit, threw money at the problem. Americans were vaccinated faster, at about five times the early per-person cost of the European Union. The vaccine works.

The federal mistake, he says, was paying guaranteed fortunes without requiring companies to join a global vaccination program. Firms followed their business model and shipped where the money was. About 85 percent of early doses went to the first world. In May 2021 the WTO, WHO, IMF, and World Bank jointly asked for $50 billion to get the third world to a 40 percent vaccination rate by year's end. Nobody answered. Africa's rate, at the time of the conversation, was about 8.9 percent.

Leave the third world unvaccinated, and you get a constant supply of variants that come back and harm Americans, exactly as Delta and Omicron did. You do not have to be a global idealist. Ordinary self-interest in American health would have funded worldwide distribution. The $50 billion those four institutions requested was, at that moment, the same sum 32 billionaires had already taken into private wealth from the vaccines. The money existed. The companies' culture, Abramson says, still treats maximizing shareholder return as the moral obligation and treats collateral damage as someone else's problem.

Fridman is about to talk with Mark Zuckerberg and is thinking about the removal of Robert Malone and Peter McCullough from platforms after they appeared on Joe Rogan's show. He leans against censoring anti-vaccine speech. Censorship, in his view, often falls to the most emotional and least able people, while the way to beat bad claims is for scientists to become better communicators and put data in public. Abramson does not fully agree, and he does not pretend to have a clean answer.

His line is the one Fridman used to open the episode: everyone is allowed their own opinion; he does not think everyone is allowed their own scientific facts. An open internet where rage earns clicks is, he says, a poor way to settle questions with large public-health stakes. He has anti-vaccine friends he loves. They send him material that takes an hour to check and usually does not hold up, and it is not a random sample of the argument. It is the most convincing slice.

He floats a middle path: not a ban, but real-time or pre-publication fact-checking by independent people who can say a study was designed to do X and concluded Y, so a spoken claim is only half true. Fridman likes a slower version, a kind of peer review for podcasts, addenda published with the episode. Abramson warns that this is not a flat-earth bar argument. COVID, with more U.S. deaths than the 1918 flu plus economic and school damage, is a problem involving millions of lives. The screaming edges, industry advocates on one side and total skeptics on the other, have wrecked the commons for the 95 percent in the middle who just want a rational way to protect their families.

Fridman raises revolving doors among Pfizer, the FDA, and the CDC. Bourla's answer, again, is that there are strict rules and the company follows them. Abramson answers with Aduhelm, Biogen's Alzheimer's drug. The development target has long been clearing amyloid plaques. Biogen showed the drug reduces plaques. Two efficacy trials found no meaningful clinical benefit, and 33 percent more people on the drug developed symptomatic brain swelling and bleeding than on placebo. Those facts, he says, are not in dispute.

An advisory committee, largely cleared of financial conflicts, voted 10 against approval and one abstention. The FDA overruled them and approved the drug. Three members resigned, asking what the point of serving was if a unanimous no did not matter. The surrogate endpoint, plaque reduction, was already known not to track clinical progression. Then the head of the FDA's Center for Drug Evaluation and Research, who had spent 16 years as a pharmaceutical executive, issued a statement: in this situation the agency should loosen the ban on financial ties so it gets less emotional responses. The ten no votes, in print, were filed under too emotional.

Abramson wants advisory members to have only the public interest, and he would ban movement between pharma and regulators even 30 years later. He has another nomination. Medicare is forbidden by law to negotiate drug prices, so a patent holder can charge whatever it wants for a monopoly utility that Medicare must buy. How did that happen? In 2003, as Medicare Part D passed, House Ways and Means chair Billy Tauzin played a key role, including the non-negotiation clause. After the bill passed he did not finish his congressional term. He took a $2 million-a-year job in pharma. A ban on that, Abramson suggests, would be a good idea.

On the NIH, Abramson is less granular than on pharma, but the satellite view is clear. The institutes do infrastructure work on targets for nearly every recent FDA-approved drug, then companies pick which products to develop. About 96 percent of U.S. clinical-trial research is about drugs and devices; only a fraction of the remaining 4 percent is preventive medicine. Meanwhile, he says, about 80 percent of health is determined by how people live. The United States ranked 38th in healthy life expectancy in 2000. After spending an extra $1.5 trillion a year, it ranked 68th. Healthcare is 17.7 percent of GDP versus 10.7 percent in peer countries, seven points of a $23 trillion economy.

His longevity advice is almost aggressively ordinary: exercise at least half an hour at least five times a week, do not smoke, and keep body weight in a reasonable range, treating a BMI above 30 as unhealthy while noting that socioeconomic status shapes who can do this. The United States inverts the usual rich-country mix, spending relatively more on medical care and less on social determinants. The market will not rebalance that, he says, because the way to move money from working people to investors is the biomedical model, not the social-health model. Other wealthy countries intervene in the quality of health data and in budgets. America is the wild west, and it is a horrendous failure.

On depression, a paper by Eric Turner in the *New England Journal* found that when all antidepressant data are pooled, the drugs do not work for people who are depressed but do not have major depression. Roughly one in four people with major depression see significant improvement. Most prescriptions, Abramson says, go to people without that diagnosis. *Prozac Nation* helped sell the idea that no one should settle for ordinary happiness, and the serotonin story became a destructive oversimplification. The first job is to respect the suffering, then ask why this person feels sadder than they want to feel: a marriage, old trauma, a job without dignity. Feeling sadder than you wish is not itself a diagnosis.

Doctors, he thinks, are not mainly complacent. They are trying to help the person in front of them, tortured by uncertainty, while industry knowledge arrives as the default. About half report burnout; his unproven judgment is that much of it is moral injury, practicing in a way they know is not working. He wrote *Sickening* so trainees can see that what looks normal is not. If you have the calling, go into medicine; if you are in it for the money, you will not be proud. Near the end Fridman asks about death and meaning. Abramson, having recently come close enough that he thought it was over, says he is not afraid. The meaning of life is to care about something that does not hurt others and go all the way, the Bukowski line he leaves in the air.