PEAK PERFORMANCEMonths to result

The Evidence-Based Vaccine Evaluation Protocol

A four-pillar standard for evaluating and communicating vaccine safety and efficacy.

Problem it solves

Individuals whose communication in peak performance contexts creates friction, misunderstanding, or missed opportunities for connection and alignment.

Best for

Regulators (FDA, CDC), public health agencies, independent scientific review panels, and journalists covering medical products.

Not ideal for

Rapid-response scenarios like initial pandemic vaccine rollout under emergency use, where some trade-offs between speed and thoroughness are unavoidable.

Overview

Why this framework exists

This protocol establishes a rigorous, transparent standard for evaluating vaccines (and by extension, any medical intervention) to restore public trust. It moves beyond surrogate endpoints (like antibody levels) and short-term trials to demand evidence on outcomes that matter to people: prevention of disease, hospitalization, and death. It insists on proper control groups even in post-market surveillance to assess long-term effects. Crucially, it requires honest communication that strictly delineates what is known from what is extrapolated or unknown. The protocol was developed in response to the COVID-19 experience, where vaccines were approved based on short-term symptomatic infection data, promises of herd immunity were extrapolated, and long-term safety was assumed, leading to a crisis of confidence when reality diverged from promises.

Core principles

4 total
  1. Evaluate vaccines on clinically meaningful endpoints (disease, hospitalization, death), not just biological surrogates (antibodies).
  2. Maintain rigorous control groups for long-term safety surveillance; never sacrifice this for political or commercial convenience.
  3. Communicate precisely: distinguish proven effects from extrapolations, and acknowledge evidence gaps explicitly.
  4. Treat vaccine evaluation with the same skeptical, benefit-harm rigor as any other drug or medical intervention.

Steps

5 steps
  1. Define and Power for Clinical Endpoints
    Design trials to detect differences in outcomes people care about: symptomatic disease, severe disease, hospitalization, and death. Ensure trials are large enough and last long enough to have statistical power for these endpoints, especially in high-risk groups.
    Pro tipFor pandemic vaccines, pre-define adaptive trial designs that can pivot to measure new variants' impact on clinical efficacy.
    WarningAvoid the trap of using easy-to-measure surrogate endpoints (like antibody titers) as a substitute for hard clinical outcomes unless a validated, long-established correlation exists.
  2. Preserve Long-Term Control Groups
    Do not unblind or vaccinate placebo groups immediately after trial completion. Maintain a blinded, long-term follow-up cohort to assess delayed adverse events, waning immunity, and long-term efficacy differences.
    Pro tipPlan and fund this long-term follow-up as a non-negotiable part of the initial trial protocol, not an afterthought.
    WarningSacrificing the control group destroys any chance of rigorously identifying long-term signals, forcing reliance on flawed observational studies.
  3. Conduct Stratified Safety & Efficacy Analysis
    Analyze trial and post-market data by age, sex, race, and prior infection status. Publish these subgroup analyses transparently to inform personalized benefit-harm assessments.
    Pro tipMandate sex-specific analyses, as adverse events like myocarditis disproportionately affect one sex, and these can be missed in pooled data.
    WarningHiding or failing to conduct subgroup analyses leads to blanket recommendations that harm specific subpopulations and erode trust.
  4. Implement Active, Hypothesis-Driven Surveillance
    Move beyond passive adverse event reporting. Actively study specific, biologically plausible long-term risks (e.g., autoimmune issues, fertility impacts) in dedicated cohorts, comparing vaccinated to properly matched unvaccinated controls.
    Pro tipLeverage large, detailed healthcare databases (like the VA system) to conduct rigorous, retrospective matched-cohort studies for specific concerns.
    WarningPassive surveillance (VAERS) alone is prone to both under-reporting and false signals; it cannot establish causality, only generate hypotheses.
  5. Adopt Precision in Public Communication
    All public statements and labels must precisely match the evidence. Example: 'Reduces risk of symptomatic COVID-19 for at least two months' not 'Stops COVID.' Explicitly list what is not known (e.g., impact on transmission, duration of protection, long-term safety).
    Pro tipUse a standardized 'Evidence Facts' box for all vaccine communications, modeled on drug labels, listing proven benefits, known risks, and major unknowns.
    WarningVague, hopeful language ('path back to normal') is interpreted as promise; when promises break, trust shatters.

Checklist

Saved in your browser

Examples

2 cases
FDA's New COVID-19 Booster Framework

Under new leadership, the FDA revised its framework for evaluating COVID-19 booster shots. Previously, boosters could be approved based on demonstrating an immune response (antibodies). The new protocol requires manufacturers to show evidence of clinical efficacy—actual reduction in COVID-19 cases, hospitalizations, or deaths—before approval.

OutcomeThis raises the evidence bar, aligning regulatory approval with outcomes that matter to patients and the public, and prevents the approval of boosters that may produce antibodies but provide little real-world benefit.
The MMR Vaccine and Autism Research

Dr. Bhattacharya cites a "massive Danish study" that tracked vaccinated and unvaccinated children for years to investigate a potential link to autism. This represents the protocol's ideal: a long-term, large-scale, controlled observational study designed to test a specific safety hypothesis.

OutcomeThe study failed to find a link, providing high-quality evidence to address a major public concern. It stands in contrast to areas where such rigorous, hypothesis-driven long-term safety studies are lacking.

Common mistakes

4 traps
Relying on Surrogate Endpoints for Approval
Approving boosters based solely on antibody response, without evidence they prevent actual illness, assumes the surrogate perfectly predicts clinical benefit—an assumption often wrong.
Sacrificing the Control Group Prematurely
Unblinding trials and vaccinating the placebo group early for ethical reasons destroys the only source of high-quality long-term comparative safety and efficacy data, a catastrophic loss for science.
Treating Vaccines as a Moral Category
Framing vaccines as 'good' and skepticism as 'bad' removes them from the realm of scientific evaluation and turns questions of evidence into ideological litmus tests, poisoning discourse.
Making Population-Wide Extrapolations from Limited Data
Asserting that a vaccine will lead to herd immunity or stop transmission based on short-term, individual-level symptom prevention data is a massive, unsupported extrapolation that sets up public disappointment.

Origin story

How this framework came to be

The protocol crystallized from the post-2020 analysis of COVID-19 vaccine communication and policy failures. Dr. Bhattacharya observed that public health authorities "extrapolated far beyond what the data actually showed"—promising freedom from disease and transmission based on 2-month symptom prevention data. When boosters were later approved based solely on antibody response without clinical endpoint data, it highlighted a regulatory gap. The final catalyst was the need to dissociate vaccine evaluation from ideological 'pro-vax/anti-vax' battles and re-anchor it in neutral, evidence-based science. The protocol is what Bobby Kennedy Jr.'s inquiry and the new FDA framework for COVID boosters are essentially asking for: a return to demanding clinical proof of benefit and rigorous, ongoing safety monitoring.

Source

Traced to primary
Source · PODCAST
Improving Science & Restoring Trust in Public Health | Dr. Jay Bhattacharya
Andrew Huberman · 2025
Open source →