Improving Science & Restoring Trust in Public Health | Dr. Jay Bhattacharya

Bhattacharya uses the baseball statistics analogy: you optimize what you measure. If the only statistics are home runs (publications) and there is no strikeout column (failed replications), teams will swing for the fences regardless of team-level outcomes. Adding replication to the measurement suite changes the incentive for every individual scientist, even those not doing replication work themselves — they will want their work to be replicable because that becomes a career-positive outcome.

Bhattacharya describes it as working like 'community notes on X but with formal methods sections and credentials.' The journal could eventually function like a Cochrane Collaborative that grades evidence on health questions, using the published replication work as its core. He notes it would also publish negative results — 'I tested this idea, it did not work' — converting what is currently a career-ending failure into a discoverable contribution to knowledge.

The current H-index metric perversely rewards writing review articles (which get many citations) over doing experiments. It allows a scientist who writes many marginal papers to score similarly to Watson and Crick who published one extraordinary paper. Bhattacharya argues the issue is not to replace these metrics but to add to them — a fuller set of statistics that captures what we actually want science to do.

The optimal team composition for novel research is a young scientist as first author paired with a mid-to-late career scientist as senior author. This preserves mentorship and resource access while allowing youthful intellectual risk-taking. Bhattacharya explicitly does not want to eliminate older labs — he wants to balance the ratio. He compares the current system to a VC fund that invests only in companies whose founders have 20 years of experience — you would miss Google and Apple.

Bhattacharya is careful to distinguish the original December 2020 trials — which were large, randomized, and showed real symptomatic protection for two months in previously uninfected people — from subsequent booster approvals, which were approved on antibody data from much smaller studies. He frames the mRNA vaccines as having been correctly approved for a limited indication and then overextended by public health messaging. The new framework corrects the approval standard going forward.

He uses the December 2020 data as his model: the trials showed prevention of symptomatic COVID for two months in previously uninfected people. For elderly individuals, that likely translates to prevention of death — a reasonable extrapolation given their high mortality risk from the disease itself. For young men, the known harm (myocarditis) is a real risk from a disease that rarely kills them. The calculus is different. He wrote a Wall Street Journal op-ed making exactly this argument in December 2020, before the mandates, and was blacklisted on Twitter for it.

Bhattacharya describes the current journal business model as exploitative: scientists pay journals roughly $12,000 to publish open access, the journal then paywalls the content and sells institutional subscriptions for millions annually, and individuals pay $30–100 per paper. Meanwhile universities pay millions in subscription fees so their faculty can read papers funded by taxpayer money. Nature and Elsevier generate billions in profit. Bhattacharya accelerated the open-access deadline from December to July as one of his first acts as NIH director.

Bhattacharya argues the US has a 'sick care system': biomedical advances have allowed people to stay sick longer rather than preventing disease. Compression of morbidity — the idea that people would live long healthy lives and only become ill briefly at the end — has not panned out. Instead, Americans face a long period of chronic disease. The next generation of children is projected to live shorter, less healthy lives than their parents, driven by obesity, diabetes, and mental health crises. Bhattacharya sees realigning NIH research toward these actual health needs as his central mission.

Bhattacharya explains the sociology: when you publish, peer reviewers check logical coherence but do not rerun experiments. P-values at 0.05 guarantee some percentage of results will be false positives. Negative results are rarely publishable. Scientists who retract papers face career damage. Scientists who replicate others' work cannot win large NIH grants and therefore cannot get tenure. The entire incentive structure rewards volume and influence while punishing truth-seeking behavior. Fraud by prominent scientists is a symptom of these incentives, not a moral aberration — the structure of academic science produces it systematically.

Bhattacharya identifies two ethical norms that clashed: scientific free speech (test ideas openly, admit when wrong) vs. public health unanimity of messaging (enforce consensus to ensure compliance). The public health norm has a legitimate basis for settled science like smoking, but it was applied to contested, unsettled questions — school closures, cloth masks, natural immunity — where the science was far from settled. The result was that scientists who disagreed were treated as threats to public compliance rather than as legitimate scientific voices. Drug overdose deaths, which were around 20,000/year pre-pandemic, reached 100,000/year in 2021–2022, driven partly by lockdown-caused economic dislocation.

Bhattacharya describes the public health miscommunication as a consequence of 'hope': when the vaccine results were announced, there was enormous relief and public health authorities extrapolated far beyond what the data showed, promising eradication and herd immunity. When breakthrough infections arrived, they doubled down by blaming the unvaccinated rather than acknowledging the limitation of vaccine efficacy. Evidence for myocarditis in young men emerged. Patient groups discussing vaccine injuries were pressured off Facebook by the Biden administration. Bhattacharya endorses the FDA's new framework under Marty Makary requiring COVID boosters to show clinical efficacy (not just antibody production) before approval.

The team-of-young-plus-senior-author combination is the most likely to produce novel ideas — suggesting mentorship matters but the structure that keeps young scientists subordinate to senior ones for a decade and a half before independent funding is the pathology. Bhattacharya calls for larger, longer early-career R01 grants and smaller grants for established labs. He also links this to the indirect cost structure: because universities receive overhead only on grants, they incentivize hiring faculty who can win grants — which currently means faculty with long track records, not early-career scientists.

Bhattacharya draws the parallel to the journal access problem he also addresses: in both cases, American taxpayers fund the research and then have to pay again to access the results. He notes that US drug company profits represent roughly two-thirds to three-quarters of global drug industry profits, with the dominant indications being obesity and depression — diseases more prevalent in the US. Equalizing drug prices internationally would also realign drug company R&D investment toward health problems that affect the broader developed world rather than exclusively US conditions.

Bhattacharya does not personally believe vaccines are the primary cause of rising autism prevalence, citing the large Danish MMR study that found no correlation. But he explicitly includes vaccines in the research agenda because many people disagree with him and an honest investigation will only be trusted if it was open. Other hypotheses he is aware of: gut microbiome alterations, retinoid exposure, in-utero environmental chemical exposure, ultrasound effects on cell migration in developing cortex (from Pasko Rakic's lab at Yale), nutritional factors. He emphasizes that autism covers a wide clinical spectrum and the etiology may differ across it.