Modernize or Ossify: Why Prasad’s Randomized Controlled Trial Fetish Threatens Makary’s AI Revolution
When FDA Commissioner Dr. Martin Makary calls out our decade-long drug lag—“Why does it take over ten years for a new drug to come to market? Why are we not modernized with AI and other things?”—he’s speaking for millions of patients left waiting. Makary likens today’s regulatory process to “applying to college with a giant 100-page application after each year.” It is, he insists, “excessive.” Yet Dr. Vinay Prasad—a vocal champion of rigor— could drag us backward if the new vaccine development framework he is working on is a reflection of his puritanical zeal for vaccines, demanding larger, longer, and ever more elaborate RCTs at every stage—from early safety to Phase IV surveillance. His latest flourish: cluster randomized trials (CRTs) for mortality outcomes, an idea so absurd it exposes the lunacy beneath his rigorism.
Initial versions of Prasad’s proposed framework, which in part reflect the views of Makary’s new deputy, Tracy Beth Hoeg, could reduce the use of AI-driven screens of side effects from vaccine exploration in favor of expanded Phase I trials with hundreds of volunteers given salt water instead of a vaccine. Phase II moves to thousands; Phase III balloons into massive, multi-season studies tracking every infection under the sun. Apparently, he insists we capture any new non-target disease, from Streptococcus pneumoniae months after injection to every mild adverse event short of a sunburn. Post-licensure, Prasad demands two-year follow-ups in contemporaneous vaccinated versus placebo cohorts—despite the ethical and logistical nightmare of withholding protection for that long.
Enter the Cluster Trial—Methodological Hubris Personified
If that weren’t enough, Prasad now proposes CRTs to measure vaccine-related mortality. Picture this: entire counties randomized to “accelerated rollout” or “standard schedule,” death certificates tallied in bulk. No individual consent. No granular safety tracking. Any uptick in fatalities—say, one avoidable myocarditis death in 20,000—drowns in the statistical fog of cluster averages. Intra-cluster correlation devours your power. Cross-cluster contamination—neighbors sharing clinics, families mixing—obliterates causal clarity. CRTs were designed for community interventions like water fluoridation, not for detecting rare, high-stakes harms to individuals. To use them for vaccine mortality is not science; it is malpractice.
Make no mistake: CRTs are the bureaucrat’s darling—“practical,” “policy-friendly,” and wonderfully abstract. Assign whole cities or school districts to “fast” or “slow” vaccine rollouts, measure death rates in aggregate, and voilà: “scientific evidence.” Only a zealot could applaud this statistical malpractice. CRTs blunt every sharp edge of causality. Intra-cluster correlation cannibalizes your sample’s power; cross-cluster contamination scrambles your groups; and the rare signal of a 1-in-20,000 fatality vanishes in the miasma of anonymized averages. More insidious still, CRTs routinely bypass informed consent: a principal or mayor signs the form, and you’ve conscripted thousands into an experiment they never agreed to.
Prasad has ridiculed the nutritional science that informs some of the sensible changes in diet that RFK Jr. and the new Surgeon General, Dr. Casey Means, recommend as “the lowest form of epidemiology.” He sneers that there's a huge cottage industry preying upon people with low credibility in nutrition science, and I don't have a lot of respect for that.” He insists every nutritional or lifestyle claim must be herded into an RCT.
Contrast this with Makary’s AI vision. Imagine harnessing machine-learning models to create “digital twins”—perfect virtual controls derived from electronic health records, genomic data, and real-world outcomes. These twins stand in for placebo arms, slashing enrollment burdens and preserving patient welfare. AI algorithms detect safety signals—myocarditis clusters, off-target infections—in days, not years; they personalize risk predictions down to the individual’s HLA type. We can run smaller, adaptive RCTs for initial proof, then let AI-driven surveillance pick up the slack for rare adverse events. It’s not a fantasy; it’s the future.
Yet in Prasad’s kingdom, innovation is subordinate to ritual. He would rather smother AI-enabled efficiency under the weight of endless CRT mandates than admit that virtual controls or real-world evidence might ever suffice. If his proposal stands, a city could be randomized to receive an accelerated vaccine rollout. At the same time, its neighbor waits—**no parental consent, no individual follow-up, no chance to detect the subtle harms that matter most**. Should a preventable fatality occur, aggregated data will simply “fail to reach significance,” and life moves on.
We must choose between Makary’s AI-powered leap forward and Prasad’s self-defeating orthodoxy. Medicine needs rigor, yes, but not rigidity. We need “pluralistic evidence pathways*”—smaller, adaptive trials where ethical, virtual controls are necessary, robust AI surveillance everywhere. We must refuse the statisticians’ temptation to trade individual rights for cluster convenience.
Public health is not a sterile classroom exercise. It is a moral endeavor that honors every life, not as a data point in some municipal ledger but as a sovereign individual. To betray that principle in the name of “gold standards” is not science—it is regression. Let us heed Makary’s rallying cry for modernization and reject Prasad’s CRT absolutism before it ossifies into policy. Only then can we deliver cures faster, tailor therapies precisely, and keep the promise of science alive.