# Your FDA Is Unsafe and Ineffective For every 1 unit of harm prevented by efficacy delays, the system creates [3,389 (90% CI: 1,811-5,734)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-type_ii_error_cost_ratio) units of harm through denied access. Operational safety ratio: [3,389 (90% CI: 1,811-5,734)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-type_ii_error_cost_ratio)-to-1 against the patient. In 1962, your Congress decided that after proving a drug won’t kill you, sick people should wait another [8.2 years (90% CI: 4.84 years-11.5 years)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-efficacy_lag_years) before they’re allowed to take it. The drug passed safety testing. It won’t kill you, won’t melt your liver, won’t make your hair migrate to your back. But you can’t have it, because a committee needs to spend [8.2 years (90% CI: 4.84 years-11.5 years)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-efficacy_lag_years) making sure it works well enough. I asked someone why, and they said “we need to confirm efficacy.” I said, “The patient is dying.” They said, “Yes, but we don’t know if the drug works.” I said, “The patient could tell you, if you gave it to them.” They said that’s not how it works. I said, “Dying is how it works?” They changed the subject, which I suppose is also how it works. You’re not allowed to volunteer for the trials that would answer the question faster, either. You are, however, allowed to die. Nobody blocks that. There’s no [8.2 years (90% CI: 4.84 years-11.5 years)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-efficacy_lag_years) waiting period on dying. Your species confirms something won’t harm you, then locks it in a cabinet and forms a committee. The committee meets quarterly. You decompose on a different schedule. ## A Note on Blame When this chapter says “the FDA does X,” read it as “the 1962 law requires the FDA to do X.” Congress wrote the rules. The FDA follows them. Blaming the FDA is like blaming the gun instead of the person who loaded it, aimed it at patients, and pulled the trigger on a 62-year schedule. ## “But What About Safety?” This is the part where humans panic, so let me be precise: keep Phase I safety testing. It works. It caught thalidomide. Nobody is suggesting you skip it. Do *radically more* efficacy testing, on *real patients*, with *transparent results*. But stop holding known-safe drugs hostage while you run the trials. There is no word for “preventing someone from accessing a safe substance that could help them while they die.” No sane civilization would do it, so none needed a word. You have several words for it. My favorite is “the approval process.” ## The [44.1x (90% CI: 12.8x-210x)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-dfda_trial_cost_reduction_factor) Inefficiency Tax Compare two real-world systems for testing drugs: 1. **The Oxford RECOVERY Trial:** Tested COVID treatments on 48,000 patients for **[\$500 (95% CI: \$400-\$2,500)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-recovery_trial_cost_per_patient)**. Found a life-saving treatment in 3 months. Like a normal species would. 2. **The Post-1962 System:** Averages **[\$41,000 (95% CI: \$20,000-\$120,000)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-traditional_phase3_cost_per_patient)** per trial. Takes **over a decade^([1](../references.llms.md#ref-drug-development-timeline-17-years))** to approve a new drug. Like humans would. Your efficacy testing system costs **[44.1x (90% CI: 12.8x-210x)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-dfda_trial_cost_reduction_factor) more** and takes **[32.8x (90% CI: 19.4x-45.9x)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-fda_to_oxford_recovery_trial_time_multiplier) longer** than necessary. I showed these numbers to a human and they said “that can’t be right.” So I showed them the sources and they said “well, there must be a reason.” There is a reason. The reason is that your system costs [44.1x (90% CI: 12.8x-210x)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-dfda_trial_cost_reduction_factor) more and takes [32.8x (90% CI: 19.4x-45.9x)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-fda_to_oxford_recovery_trial_time_multiplier) longer. That’s the reason and also the problem. They’re the same thing. ### The [8.2 years (90% CI: 4.84 years-11.5 years)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-efficacy_lag_years) Efficacy Lag Here’s where your lifespan goes to die: - **Phase I (Safety Testing):** [2.3 years](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-phase_1_safety_duration_years) - Tests if the drug harms people: death, organ damage, mutations, serious adverse reactions, all the ways a molecule can ruin your day. *This part works. Keep it.* - **Phase II/III (Efficacy Testing):** [8.2 years (90% CI: 4.84 years-11.5 years)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-efficacy_lag_years) - The drug is now *known safe*. But patients can’t have it until the committee is satisfied. *This is where people die waiting for a drug that already passed every safety check your species invented.* ## What the 1962 Efficacy Requirements Changed Before 1962, [144 thousand physicians](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-pre_1962_physician_count) tested treatments on real patients and published results. Life expectancy grew 3.82 years every decade^([2](../references.llms.md#ref-life-expectancy-increase-pre-1962)) for 80 straight years. Things were going well. Naturally, Congress intervened. Phase I safety testing (added in 1938) prevented all US thalidomide deaths while Europe had thousands. So in 1962, Congress looked at a system that was working and thought, “What if we made this much worse?” They added massive **pre-approval efficacy requirements** via the Kefauver Harris Amendment^([3](../references.llms.md#ref-kefauver-harris-amendment-1962)), blocking patient access for years during testing. The cost of developing a drug went from [\$24.7 million (95% CI: \$19.5 million-\$30 million)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-pre_1962_drug_development_cost_2024_usd) to [\$2.6 billion (95% CI: \$1.5 billion-\$4 billion)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-pharma_drug_development_cost_current), a [105x (90% CI: 72.8x-149x)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-drug_cost_increase_pre1962_to_current_multiplier) increase. New drug approvals dropped by [70%](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-post_1962_drug_approval_reduction_pct)^([4](../references.llms.md#ref-post-1962-drug-approval-drop)). Life expectancy gains slowed from 3.82 years per decade to 1.54^([2](../references.llms.md#ref-life-expectancy-increase-pre-1962),[5](../references.llms.md#ref-post-1962-life-expectancy-slowdown)), a 60% reduction. Time to approval stretched from 2-3 years to [10.5 years (95% CI: 6 years-12 years)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-fda_phase_1_to_approval_years). Everything got worse at exactly the same time. Every single metric, all at once, the moment the law passed. I’m not a statistician so I can’t tell you whether that’s a coincidence, but I did look at the graph, and it’s not a coincidence. [![Average Life Expectancy Over Time](../../assets/images/real-world-evidence-in-efficacy-clinical-trials-vs-rcts.png)](../../assets/images/real-world-evidence-in-efficacy-clinical-trials-vs-rcts.png) Average Life Expectancy Over Time The [105x (90% CI: 72.8x-149x)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-drug_cost_increase_pre1962_to_current_multiplier) cost increase is rigorously documented. See [Drug Development Cost Analysis](https://drug-cost.warondisease.org) for methodology, sensitivity analysis, and source data. ### The Modern Consequences Even if you think historical analysis is just opinions with footnotes, the current system’s failures are happening right now: 1. A cure passes safety testing 2. You can’t take it (pending efficacy review) 3. You can’t join the trial either (too sick, too old, wrong medications) 4. You die 5. The FDA approves the cure 6. Pharmaceutical companies charge your widow \$10,000 per pill I read this sequence several times because I kept thinking I’d got the order wrong. I hadn’t. That is the correct order. The system works exactly like that, on purpose, and everyone involved considers it a success because nobody skipped a step. [![Scientists discover cure. 17 years pass. Patients die. Drug finally available. This is called ‘safety.’](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-the-modern-consequences-bw-academic.jpg)](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-the-modern-consequences-bw-academic.jpg) Scientists discover cure. 17 years pass. Patients die. Drug finally available. This is called ‘safety.’ #### High Costs Kill Innovation, Reward Monopoly Before 1962: Genius scientist invents cure, raises a few million, tests safety, gives it to people. Like a civilization that wants to survive would. [![Option 1: Spend billions developing new cure, might fail. Option 2: Shelve the cure, keep selling expensive old drug. You picked option 2, obviously.](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-high-costs-kill-innovation-reward-monopoly-bw-academic.jpg)](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-high-costs-kill-innovation-reward-monopoly-bw-academic.jpg) Option 1: Spend billions developing new cure, might fail. Option 2: Shelve the cure, keep selling expensive old drug. You picked option 2, obviously. Now: Genius scientist must convince one of three mega-corporations to fund trials. Those corporations already sell inferior drugs for the same condition, bringing in hundreds of millions a year per drug. The math for the mega-corporation: **Option 1: Develop the better drug** Cost: [\$2.6 billion (95% CI: \$1.5 billion-\$4 billion)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-pharma_drug_development_cost_current) on average to get a single drug to market. That figure already includes the cost of the nine candidates that fail for every one that succeeds, so the [10%](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-pharma_success_rate_current_pct) success rate is already priced in. Best case, the drug works: you now have a better drug for the same condition. But your old drug, which was bringing in hundreds of millions a year, is now obsolete. You spent [\$2.6 billion (95% CI: \$1.5 billion-\$4 billion)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-pharma_drug_development_cost_current) to replace your own revenue stream. Every year the new drug is on the market, it’s earning roughly what the old drug would have earned anyway, minus the [\$2.6 billion (95% CI: \$1.5 billion-\$4 billion)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-pharma_drug_development_cost_current) hole you dug getting here. **Option 2: Buy the patent, put it on a shelf** Your inferior drug keeps bringing in hundreds of millions a year. Total cost: whatever the patent costs, which is rounding error compared to [\$2.6 billion (95% CI: \$1.5 billion-\$4 billion)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-pharma_drug_development_cost_current). I asked a pharmaceutical executive which option they’d choose and they said “it’s more nuanced than that.” Then I asked which one they chose and they said Option 2. So it wasn’t more nuanced than that. The profit incentive doesn’t just fail to reward better treatments. It actively punishes them. The best possible outcome of Option 1 is the same annual revenue you already had, minus [\$2.6 billion (95% CI: \$1.5 billion-\$4 billion)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-pharma_drug_development_cost_current). Curing disease is economically irrational. The system doesn’t need a conspiracy to suppress cures. It just needs the math. I keep waiting for someone to show me where the math is wrong. Nobody shows me where the math is wrong. They just look uncomfortable and say “it’s complicated,” which I’ve learned is what humans say when it isn’t complicated and they don’t like the answer. ### Off-Patent Drugs and Rare Diseases: Mathematically Doomed The 1962 law made it mathematically impossible to cure diseases that aren’t profitable enough: - **95% of diseases are rare^([6](../references.llms.md#ref-95-pct-diseases-no-treatment)):** Development cost ([\$2.6 billion (95% CI: \$1.5 billion-\$4 billion)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-pharma_drug_development_cost_current)) / patient population (~10,000 patients) = \$260,000/patient - **No patent = no funding:** Off-patent drugs can’t attract billion-dollar investments - **Pre-specification kills serendipity:** Must predict what drug cures before testing When something costs more, you get less of it. I looked this up and apparently it’s the first thing they teach in economics. It is also, somehow, the last thing anyone considers when writing health policy. 95% of diseases^([6](../references.llms.md#ref-95-pct-diseases-no-treatment)) have zero treatments. Not because cures are impossible. Because cures are unprofitable. The best thing a disease can do for itself, career-wise, is be popular. ## The Actual Death Toll of “Drug Lag” Economists call it “drug lag,” which is a lovely, clinical way of saying “people rotting to death in a queue for drugs that have already been confirmed not to hurt them.” Early estimates: **21,000 to 120,000 American lives per decade^([7](../references.llms.md#ref-gieringer-drug-lag-estimate))**. But that was just the US. A [full accounting](https://invisible-graveyard.warondisease.org) using WHO mortality data puts the *total* cost of the [8.2 years (90% CI: 4.84 years-11.5 years)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-efficacy_lag_years) *post-safety* efficacy delay (the lag *after* drugs are known safe) at **[416 million deaths (90% CI: 244 million deaths-587 million deaths)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-dfda_efficacy_lag_elimination_deaths_averted) eventually avoidable deaths.** [![The number of people who died waiting for permission to try safe medicine. The number is very large. The permission was not.](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-the-global-body-count-bw-academic.jpg)](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-the-global-body-count-bw-academic.jpg) The number of people who died waiting for permission to try safe medicine. The number is very large. The permission was not. Including years lived with disability: [8.77 billion DALYs (90% CI: 4.88 billion DALYs-13.2 billion DALYs)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-dfda_efficacy_lag_elimination_dalys) of healthy human life deleted. Economic cost: [\$1.32 quadrillion (90% CI: \$676 trillion-\$2.14 quadrillion)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-dfda_efficacy_lag_elimination_economic_value) (2024 USD), reflecting those [8.77 billion DALYs (90% CI: 4.88 billion DALYs-13.2 billion DALYs)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-dfda_efficacy_lag_elimination_dalys) valued at [\$150,000 (90% CI: \$100,384-\$198,679)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-standard_economic_qaly_value_usd)/DALY, the standard normative rate used in WHO cost-effectiveness analyses. Even if those death estimates are off by a factor of 10, that’s still tens of millions of deaths. And the inefficiency is independently verifiable: the efficacy testing process costs [82x (90% CI: 21.4x-195x)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-recovery_trial_cost_reduction_factor) more per patient and takes [32.8x (90% CI: 19.4x-45.9x)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-fda_to_oxford_recovery_trial_time_multiplier) longer than proven alternatives like the Oxford RECOVERY trial. Those numbers don’t depend on mortality estimates at all. Here’s a news story from the Non-Existent Times by No One Ever, without a picture of all the people who die from lack of access to life-saving treatments that might have been. You can’t photograph millions of people who died from *not* receiving a drug. There’s no body with a toe tag reading “cause of death: regulatory delay.” They die of cancer, heart disease, Alzheimer’s. The delay that killed them is invisible, so nobody writes the headline. [![](../../assets/images/non-existent-times.png)](../../assets/images/non-existent-times.png) **Types of Error in FDA Approval Decision** | | Drug Is Beneficial | Drug Is Harmful | |----|----|----| | FDA Allows the Drug | Correct Decision | Victims appear on television. Regulator is fired. | | FDA Does Not Allow the Drug | Victims die at home. Nobody is fired. | Correct Decision | So the FDA has two kinds of mistakes. One where they give you a bad drug and you die on television. And one where they don’t give you a good drug and you die at home. They’ve decided the second one doesn’t count, which is convenient because it’s the one they do most. The most infamous case: beta-blockers. Europe used them to prevent heart attacks. The US delayed for a decade, killing an estimated **100,000 Americans^([8](../references.llms.md#ref-beta-blocker-drug-lag-deaths))**. More than died in Vietnam and Korea combined. From one drug. That they had. That they knew worked. That they put in a drawer. I asked someone if there was a word for killing 100,000 people by not doing something, and they said “it depends on the context.” I think 100,000 dead people is the context, but apparently that’s not enough context. [![In 1962, Americans and Swiss people died at the same age. Then America made medicine safer. Now Americans die younger. This is what winning looks like.](../../assets/images/us-swiss-life-expectancy-5.png)](../../assets/images/us-swiss-life-expectancy-5.png) In 1962, Americans and Swiss people died at the same age. Then America made medicine safer. Now Americans die younger. This is what winning looks like. After 1962, US life expectancy diverged from Switzerland’s (which didn’t introduce the same delays). More drug approvals in the ’80s narrowed the gap. Fewer approvals in the ’90s widened it again. The pattern is not subtle. I showed it to a child and the child understood it. I showed it to a policy expert and they said it was “multifactorial.” I think “multifactorial” is what adults say when they can see the same thing the child can see but have reasons not to say it out loud. [![More drug approvals in the ’80s narrowed the gap. Fewer approvals in the ’90s widened it. The pattern is not subtle.](../../assets/images/us-swiss-life-expectancy-drug-approvals.png)](../../assets/images/us-swiss-life-expectancy-drug-approvals.png) More drug approvals in the ’80s narrowed the gap. Fewer approvals in the ’90s widened it. The pattern is not subtle. ## How the Incentives Work ### FDA Regulator Decision Tree #### Approve drug that later shows problems - Congressional hearing (televised) - “FDA APPROVED KILLER DRUG” headlines - Career ends - Pension threatened #### Delay drug that could save lives - Nothing happens - Nobody writes a headline - Promotion on schedule - Retire to pharma job^([9](../references.llms.md#ref-fda-regulatory-capture-evidence)) To be fair, this isn’t a conspiracy. It’s just a system where the only career-ending move is letting patients access treatments. If you approve a bad drug, CNN says your name. If you delay a good drug, nobody dies on television. So the rational move is to delay everything, and that’s what happens, and everyone who could object is dead, which does simplify the feedback process. [![If a regulator approves a bad drug, CNN says their name. If they delay a good drug, nobody dies on television. So nobody dies on television.](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-incentive-structure-flowchart-bw-academic.jpg)](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-incentive-structure-flowchart-bw-academic.jpg) If a regulator approves a bad drug, CNN says their name. If they delay a good drug, nobody dies on television. So nobody dies on television. Dr. Henry I. Miller ran the FDA team reviewing recombinant human insulin in the early 1980s. Mountains of evidence showed it was safe and effective. His supervisor refused to approve it. If someone died from the drug, heads would roll. If people died waiting for the drug? Nothing happens. Dead patients don’t testify before Congress. I checked. [![Approve a bad drug: fired. Delay a good drug: promoted. The system works exactly as designed, which is the problem.](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-why-bureaucrats-are-rewarded-for-letting-you-die-bw-academic.jpg)](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-why-bureaucrats-are-rewarded-for-letting-you-die-bw-academic.jpg) Approve a bad drug: fired. Delay a good drug: promoted. The system works exactly as designed, which is the problem. ### The Math: Why Current Regulations Increase Total Harm $$ \begin{gathered} Ratio_{TypeII} \\ = \frac{DALYs_{lag}}{DALY_{TypeI}} \\ = \frac{8.77B}{2.59M} \\ = 3{,}390 \end{gathered} $$\ where:\ $$ DALYs_{lag} = YLL_{lag} + YLD_{lag} = 7.9B + 873M = 8.77B $$\ where:\ $$ \begin{gathered} YLL_{lag} \\ = \text{DEATHS\_TOTAL} \times (REMAINING_LIFE_EXPECTANCY_AT_60 - (\text{MEAN\_AGE\_OF\_DEATH} - 60)) \end{gathered} $$\ where:\ $$ \begin{gathered} Deaths_{lag} \\ = T_{lag} \times Deaths_{disease,daily} \times 338 \\ = 8.2 \times 150{,}000 \times 338 \\ = 416M \end{gathered} $$\ where:\ $$ \begin{gathered} YLD_{lag} \\ = Deaths_{lag} \times T_{suffering} \times DW_{chronic} \\ = 416M \times 6 \times 0.35 \\ = 873M \end{gathered} $$\ where:\ $$ \begin{gathered} DALY_{TypeI} \\ = DALY_{thal} \times 62 \\ = 41{,}800 \times 62 \\ = 2.59M \end{gathered} $$\ where:\ $$ \begin{gathered} DALY_{thal} \\ = YLD_{thal} + YLL_{thal} \\ = 13{,}000 + 28{,}800 \\ = 41{,}800 \end{gathered} $$\ where:\ $$ \begin{gathered} YLD_{thal} \\ = DW_{thal} \times N_{thal,survive} \times LE_{thal} \\ = 0.4 \times 540 \times 60 \\ = 13{,}000 \end{gathered} $$\ where:\ $$ \begin{gathered} N_{thal,survive} \\ = N_{thal,US,prevent} \times (1 - Rate_{thal,mort}) \\ = 900 \times (1 - 40\%) \\ = 540 \end{gathered} $$\ where:\ $$ \begin{gathered} N_{thal,US,prevent} \\ = N_{thal,global} \times Pct_{US,1960} \\ = 15{,}000 \times 6\% \\ = 900 \end{gathered} $$\ where:\ $$ \begin{gathered} YLL_{thal} \\ = Deaths_{thal} \times 80 \\ = 360 \times 80 \\ = 28{,}800 \end{gathered} $$\ where:\ $$ \begin{gathered} Deaths_{thal} \\ = Rate_{thal,mort} \times N_{thal,US,prevent} \\ = 40\% \times 900 \\ = 360 \end{gathered} $$ Even generously crediting ALL drug withdrawals to the 1962 changes (ignoring that Phase I safety testing already existed), the prevented harm totals ~[2.59 million DALYs (90% CI: 1.88 million DALYs-3.38 million DALYs)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-type_i_error_benefit_dalys). The delay harm totals [8.77 billion DALYs (90% CI: 4.88 billion DALYs-13.2 billion DALYs)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-dfda_efficacy_lag_elimination_dalys). **For every 1 unit of harm prevented, the current system creates [3,389 (90% CI: 1,811-5,734)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-type_ii_error_cost_ratio) units of harm through delay.** The safety net catches one person and drops [3,389 (90% CI: 1,811-5,734)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-type_ii_error_cost_ratio). There is a word for that. The word is “hole.” ## Clinical Trial Theater: Excluding [86.1%](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-antidepressant_trial_exclusion_rate) of Reality Makes Drugs More Dangerous To be clear: Phase I safety testing should exclude vulnerable populations. You don’t give an untested molecule to a pregnant woman. But once a drug clears safety, the efficacy trials that follow *also* exclude those populations, which means nobody learns whether the drug works for the people who will actually take it, or what interactions to expect. Then the drug is approved for everyone, and the post-market surveillance system (FAERS) catches less than 10% of adverse events. So the exclusions create a data gap, and nothing ever fills it. The FDA requires rigorous safety trials. Good. Then it runs efficacy trials on people who aren’t sick, aren’t old, aren’t on other drugs, and aren’t pregnant. I read this three times. They test whether drugs work on people who don’t need the drugs, to predict whether they’ll work for people who do need the drugs, who are different people, with different bodies, taking different medications. Then they act surprised when the second group has a reaction. I would call this stupid but I think it might actually be something worse than stupid. Stupid doesn’t usually have a peer-review process. [![They test drugs on healthy people, then give them to sick people. It’s like testing a ladder on Olympic gymnasts, then selling it to your grandmother.](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-clinical-trial-theater-excluding-86-1-of-reality-makes-drugs-bw-academic.jpg)](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-clinical-trial-theater-excluding-86-1-of-reality-makes-drugs-bw-academic.jpg) They test drugs on healthy people, then give them to sick people. It’s like testing a ladder on Olympic gymnasts, then selling it to your grandmother. Trials under the current system exclude^([10](../references.llms.md#ref-antidepressant-trial-exclusion-rates)): - **Patients over 65:** Most people who actually take medications (excluded due to “comorbidities”) - **Patients under 18:** All children (metabolism differs from adults) - **Pregnant women:** Excluded entirely (then drugs prescribed during pregnancy anyway) - **Anyone with comorbidities:** The sickest patients most likely to have adverse reactions - **Anyone on other medications:** Everyone elderly (can’t detect drug interactions) - **Anyone too far from trial sites:** Poor and rural populations It’s like testing whether a car is safe by driving it very slowly into a nice cushion, and then selling it to people who drive into walls. And when they die you say “well, it was safe when WE tried it.” Which is true. It was. **Result:** In antidepressant trials alone, [86.1%](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-antidepressant_trial_exclusion_rate) of real patients are excluded. The drugs are tested on the healthiest people alive, declared “safe and effective,” then handed to everyone the trial specifically avoided. Nobody knows whether the drug works for those populations, what the interactions are, or what the real adverse event profile looks like. This is how Vioxx, Fen-Phen, and Bextra all got approved, celebrated, prescribed to millions, and then withdrawn after killing people whose demographics the trials had carefully excluded. The [8.2 years (90% CI: 4.84 years-11.5 years)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-efficacy_lag_years) delay didn’t make these drugs safer. It made them seem safer by hiding the danger behind an exclusion list. [![Drugs tested on healthy 25-year-olds. Prescribed to 80-year-olds on twelve medications. The system did not anticipate this would be a problem.](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-why-this-makes-drugs-more-dangerous-bw-academic.jpg)](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-why-this-makes-drugs-more-dangerous-bw-academic.jpg) Drugs tested on healthy 25-year-olds. Prescribed to 80-year-olds on twelve medications. The system did not anticipate this would be a problem. Testing drugs only on unrepresentative volunteers to prove they work is like testing parachutes only on the ground. The parachute worked perfectly, by the way. It just wasn’t falling at the time. ### No Long-Term Outcome Data Data collection can be as short as several months. You take the drug for forty years. The gap between these two numbers is where side effects live. Nobody checks. It’s like reviewing a marriage based on the wedding. [![Data collection: several months. Duration of use: forty years. The gap between these numbers is where side effects live.](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-no-long-term-outcome-data-bw-academic.jpg)](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-no-long-term-outcome-data-bw-academic.jpg) Data collection: several months. Duration of use: forty years. The gap between these numbers is where side effects live. ### Pre-Specification Requirements Kill Innovation The regulations require drug developers to predict exactly what a treatment will cure before testing it on humans. You must know the answer before you’re allowed to ask the question. I asked if this was how science works and they said “it’s how regulatory science works,” which I think means “no, but we gave it the same name.” [![Dendreon’s cancer drug worked. The FDA said it didn’t count because the paperwork predicting it would work was filed in the wrong order. The patients died correctly, however.](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-pre-specification-requirements-kill-innovation-bw-academic.jpg)](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-pre-specification-requirements-kill-innovation-bw-academic.jpg) Dendreon’s cancer drug worked. The FDA said it didn’t count because the paperwork predicting it would work was filed in the wrong order. The patients died correctly, however. In 2007, Dendreon showed that its immunotherapy drug Provenge significantly reduced deaths from prostate cancer. The FDA advisory committee agreed it worked. Then the FDA rejected it anyway. Not because it didn’t save lives. Because Dendreon filled out the wrong form about which lives it was planning to save. “People stopped dying” wasn’t good enough. The paperwork predicting that people would stop dying wasn’t filed in the right order. Three more years and another trial were required. During those three years, people who could have stopped dying continued dying, but at least the forms were correct. A rational civilization would have a saying for this: “The gravestone was filed on time.” It means someone prioritized procedure over the thing the procedure was supposed to protect. You could use it daily. Due to these additional costs, Dendreon filed for chapter 11 bankruptcy^([11](../references.llms.md#ref-filed-for-chapter-11-bankruptcy)). So the approval process killed the patients, then killed the company trying to save them. You have a phrase for this: “killing two birds with one stone.” Although in this case the stone was paperwork and there were considerably more than two birds. But you can’t count the dead birds who would have been cured by a company you bankrupted. ### The Blinding Catch-22: How Knowing a Drug Works Proves It Doesn’t MDMA was synthesized in 1912^([12](../references.llms.md#ref-mdma-synthesis-1912)). By the 1970s, therapists were using it to treat trauma, and an estimated half a million doses were administered in therapeutic settings^([13](../references.llms.md#ref-mdma-therapeutic-history)). In 1985, the government scheduled it. In 1986, Rick Doblin founded MAPS to get it approved for PTSD^([14](../references.llms.md#ref-maps-founded-1986)). Thirty-eight years and over \$140 million in philanthropic funding later, they ran two successful Phase 3 trials. 67% of participants no longer met the diagnostic criteria for PTSD^([15](../references.llms.md#ref-mdma-phase3-mapp2)). The drug worked. Veterans with treatment-resistant PTSD got better. The FDA’s own advisory committee reviewed the data. They voted 9-2 that it wasn’t effective^([16](../references.llms.md#ref-fda-mdma-advisory-vote-2024)). The problem: patients could tell they’d taken MDMA, because MDMA makes you feel things, which is how it treats PTSD, which is the point. About 90% of participants in the drug group correctly guessed they weren’t on placebo. The FDA called this “functional unblinding” and said it invalidated the results^([17](../references.llms.md#ref-fda-mdma-rejection-2024)). I want to make sure I understand the logic. The drug works by making you feel profoundly different. The FDA requires that you not know whether you feel profoundly different. If you notice that you feel profoundly different, the trial fails, because you weren’t supposed to notice. The drug’s mechanism of action is, itself, a protocol violation. This is not a fixable design problem. You cannot double-blind a drug whose therapeutic effect is that the patient feels it working. The FDA knows this. They granted MDMA “Breakthrough Therapy Designation” in 2017, signaling that yes, this looks promising, please spend another seven years and tens of millions of dollars on Phase 3 trials. Then they rejected it for a property of the molecule that was evident before the trials began. No sane inspector would tell someone to build a house and then condemn it for having walls. Here, you owe them nothing. The veterans kept their PTSD. The FDA kept its process. Everyone kept exactly what they had, which was the problem in the first place. Meanwhile, approximately 49,000 Americans kill themselves every year^([18](../references.llms.md#ref-cdc-suicide-statistics-2023)), including 6,400 veterans^([19](../references.llms.md#ref-veteran-suicide-2022)) at twice the civilian rate. About 13 million Americans have PTSD in any given year^([20](../references.llms.md#ref-nimh-ptsd-prevalence)), and PTSD nearly quadruples your suicide risk^([21](../references.llms.md#ref-ptsd-suicide-risk-swedish-cohort)). MDMA-assisted therapy produced a 67% remission rate versus 32% for standard care. Even conservatively, if half of PTSD patients had accessed the treatment over the 40 years since it was scheduled, the 35-percentage-point improvement in remission would have prevented something on the order of 30,000 suicides in the US alone, and freed millions of people from a condition that destroys families, drives substance abuse, and fills your homeless shelters with veterans. A molecule that Merck synthesized the same year the Titanic sank is sitting in regulatory limbo because the people it helped could tell it was helping them. I looked for the word that describes letting thousands of people kill themselves each year while a known-effective treatment waits for someone to solve an unsolvable blinding problem, and the closest I found was “process.” ### Small Trials Are Dangerous [![Phase III: 3,000 patients. Post-approval: millions. A 1-in-10,000 adverse event is invisible in the first group and very visible in the second.](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-small-trials-are-dangerous-bw-academic.jpg)](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-small-trials-are-dangerous-bw-academic.jpg) Phase III: 3,000 patients. Post-approval: millions. A 1-in-10,000 adverse event is invisible in the first group and very visible in the second. Phase III trials test 1,000-3,000 patients. A 1-in-10,000 adverse event is mathematically invisible in a trial of 3,000. Then the drug goes to millions. After approval, the FDA monitors safety using FAERS, a system where doctors voluntarily report problems if they feel like it. It captures less than 10% of actual adverse events^([22](../references.llms.md#ref-adverse-event-underreporting)). Vioxx killed an estimated 38,000-55,000 Americans^([23](../references.llms.md#ref-vioxx-deaths)) before this honor system noticed. That’s more people than died in the entire Korean War, detected by a system that relies on doctors filling out optional paperwork. I want to be clear: the [8.2 years (90% CI: 4.84 years-11.5 years)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-efficacy_lag_years) delay is supposed to be for safety. And then the safety monitoring after approval is a suggestion box. You waited [8.2 years (90% CI: 4.84 years-11.5 years)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-efficacy_lag_years) to get to the suggestion box. [![Doctors report 10% of drug side effects. The other 90% just happen to people who assume that’s what dying feels like.](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-adverse-event-iceberg-bw-academic.jpg)](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-adverse-event-iceberg-bw-academic.jpg) Doctors report 10% of drug side effects. The other 90% just happen to people who assume that’s what dying feels like. A [decentralized FDA](../../knowledge/solution/dfda.llms.md)^([24](../references.llms.md#ref-dfda-spec-paper-2025),[25](../references.llms.md#ref-dfda-impact-paper-2025)) with 10 million participants catches 1-in-10,000 reactions, includes real populations, tracks long-term outcomes, and provides automated surveillance instead of voluntary paperwork. You have a fax machine and a suggestion box. One of these is a safety system. The other is a prayer with a [8.2 years (90% CI: 4.84 years-11.5 years)](../../knowledge/appendix/parameters-and-calculations.llms.md#sec-efficacy_lag_years) waiting period. ## The Negative Results Black Hole Here’s something that would be illegal if anybody important cared: - **Negative trial results published:** 37%^([26](../references.llms.md#ref-publication-bias-clinical-trials)) - **Positive trial results published:** 94%^([26](../references.llms.md#ref-publication-bias-clinical-trials)) - **Money wasted repeating failed experiments:** ~\$100 billion annually^([27](../references.llms.md#ref-wasted-spending-on-failed-experiments)) Companies are allowed to hide when drugs don’t work. Restaurants can’t hide when they fail a health inspection. I checked this twice because I thought I was wrong, but I wasn’t. A sandwich has more regulatory transparency than chemotherapy. I don’t know what to do with that information but I’ve had it for three days now and it won’t go away. [![Drug companies publish successful trials and hide failed ones. Researchers then repeat the failures. This is called “reinventing the wheel,” except the wheel is on fire and nobody told you.](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-the-negative-results-black-hole-bw-academic.jpg)](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-the-negative-results-black-hole-bw-academic.jpg) Drug companies publish successful trials and hide failed ones. Researchers then repeat the failures. This is called “reinventing the wheel,” except the wheel is on fire and nobody told you. Company: “We tested this drug!” Regulator: “Did it work?” Company: “…We tested this drug!” Pharmaceutical companies bury negative results deeper than Jimmy Hoffa. Other companies waste billions testing the same dead ends. Your insurance premiums fund this magnificent inefficiency. It’s like casinos only having to report when people win. Actually, I think casinos DO have to report when people win. So drugs get less honesty than gambling. I keep finding these comparisons and they keep getting worse. ## Countries That Don’t Have Our “Safety” [![Japan approves experimental drugs fast. Europe approves them medium fast. America approves them after you die. Different cultures, different priorities.](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-countries-that-don-t-have-our-safety-bw-academic.jpg)](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-countries-that-don-t-have-our-safety-bw-academic.jpg) Japan approves experimental drugs fast. Europe approves them medium fast. America approves them after you die. Different cultures, different priorities. **Japan**^([28](../references.llms.md#ref-japans-regenerative-medicine-act)) gives conditional approval after Phase II with real-world data collection. Time to patient: 2-3 years. Americans fly there for treatment. **The EU**^([29](../references.llms.md#ref-eu-compassionate-use-program)) lets terminal patients access experimental drugs at a doctor’s discretion. I asked the FDA what might go wrong and they said “the patient could die.” The patient is already dying. “Yes,” they said, “but they could die differently.” I waited for the rest of the sentence but that was it. **Right to Try (US)**^([30](../references.llms.md#ref-right-to-try-act-2018)) passed despite FDA opposition. The FDA made it effectively impossible to use. The law says you can *ask* a manufacturer for their experimental drug. The manufacturer can say no. They almost always say no, because the law provides no liability protection, no funding mechanism, and no insurance coverage. So “Right to Try” means “Right to Ask a Corporation for a Favor While Dying.” Corporations said no. Patients helped: \<200 total^([31](../references.llms.md#ref-right-to-try-act-outcomes)). Patients who wanted help: tens of thousands. Everywhere else in the universe, “Right to Try” is just called “rights.” The fact that your species needed a special law to allow dying people to take known-safe drugs, and then made the law effectively unusable, is the most human thing I’ve ever observed. You named it “Right to Try.” The name was accurate, which was so unusual for your species that it should have been suspicious. And it was: the law doesn’t work. The naming convention holds. ## The COVID Test Fiasco January 2020: WHO develops COVID test, world starts using it. February: FDA blocks all non-CDC tests^([32](../references.llms.md#ref-fda-covid-test-delay)) to “ensure quality.” The CDC tests were contaminated^([33](../references.llms.md#ref-cdc-covid-test-contamination-2020)). Private labs begged to help. Late March: FDA finally allows other tests after thousands die^([34](../references.llms.md#ref-excess-deaths-fda-covid-response)). [![While COVID spread, America banned COVID tests. The approved tests didn’t work. The unapproved tests did. Quality control of the broken tests continued on schedule.](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-the-covid-test-fiasco-bw-academic.jpg)](../../assets/images/fda-is-unsafe-and-ineffective/fda-is-unsafe-and-ineffective-section-the-covid-test-fiasco-bw-academic.jpg) While COVID spread, America banned COVID tests. The approved tests didn’t work. The unapproved tests did. Quality control of the broken tests continued on schedule. The only approved tests didn’t work. The unapproved ones did. The FDA blocked the ones that worked to ensure the quality of the ones that didn’t. I’ve written this paragraph four times now and it keeps sounding like I’m making it up, but I’m not. That is what happened. In order. On purpose. ## The Bottom Line To be fair to the FDA, the system does eventually approve most good drugs. So if you think about it, the system works. Unless you were one of the people who died waiting, in which case the system worked slightly less well for you specifically. But you’re dead, so you can’t complain, which means the system has a 100% satisfaction rate among living participants. That’s actually quite impressive if you don’t think about it. It is not a safety system. It is a filing system that produces corpses as a byproduct. The only thing missing is the political will to admit that a law signed before the Beatles existed might not be the pinnacle of pharmaceutical regulation. The Beatles broke up. The law didn’t. Draw your own conclusions. ------------------------------------------------------------------------ *P.S. - The FDA will likely object to this chapter. Estimated response time: 12-15 years, pending Phase III review and proper documentation. The full quantitative model is at [invisible-graveyard.warondisease.org](https://invisible-graveyard.warondisease.org), for those who enjoy numbers large enough to induce nausea.* 1\. Drugs.com. Time to develop one drug: 17 years. *Drugs.com* *Time to develop one drug: 17 years Additional sources: https://www.drugs.com/fda-approval-process.html \| https://www.fdareview.org/issues/the-drug-development-and-approval-process/* . 2\. Source: US Life Expectancy FDA Budget 1543-2019 CSV. [US life expectancy growth 1880-1960: 3.82 years per decade](https://manual.warondisease.org/knowledge/data/us-life-expectancy-fda-budget-1543-2019.csv). (2019) *Pre-1962: 3.82 years/decade Post-1962: 1.54 years/decade Reduction: 60% decline in life expectancy growth rate Additional sources: https://ourworldindata.org/life-expectancy \| https://www.mortality.org/ \| https://www.cdc.gov/nchs/nvss/mortality_tables.htm* . 3\. GovTrack. Kefauver harris amendment of 1962. *GovTrack* (1962) *regulating efficacy testing via the 1962 Kefauver Harris Amendment. The 1962 regulations made these large real-world efficacy trials illegal. Additional sources: https://www.govtrack.us/congress/bills/87/s1552 \| https://www.fda.gov/about-fda/histories-product-regulation/promoting-safe-effective-drugs-100-years \| https://pmc.ncbi.nlm.nih.gov/articles/PMC4101807/* . 4\. Kinch, M. S. & Griesenauer, R. H. [Lost medicines: A longer view of the pharmaceutical industry with the potential to reinvigorate discovery](https://doi.org/10.1016/j.drudis.2019.01.009). *Drug Discovery Today* **24**, 875–880 (2019) *Research identified 1,600+ medicines available in 1962. The 1950s represented industry high-water mark with \>30 new products in five of ten years; this rate would not be replicated until late 1990s. More than half (880) of these medicines were lost following implementation of Kefauver-Harris Amendment. The peak of 1962 would not be seen again until early 21st century. By 2016 number of organizations actively involved in R&D at level not seen since 1914.* 5\. Source: US Life Expectancy FDA Budget 1543-2019 CSV. [Post-1962 slowdown in life expectancy gains](https://manual.warondisease.org/knowledge/data/us-life-expectancy-fda-budget-1543-2019.csv). (2019) *Pre-1962 (1880-1960): 3.82 years/decade Post-1962 (1962-2019): 1.54 years/decade Reduction: 60% decline Temporal correlation: Slowdown occurred immediately after 1962 Kefauver-Harris Amendment Additional sources: https://ourworldindata.org/life-expectancy \| https://www.mortality.org/ \| https://www.cdc.gov/nchs/nvss/mortality_tables.htm* . 6\. GAO. 95% of diseases have 0 FDA-approved treatments. *GAO* (2025) *95% of diseases have no treatment Additional sources: https://www.gao.gov/products/gao-25-106774 \| https://globalgenes.org/rare-disease-facts/* . 7\. FEE. Patients and doctors vs FDA. *FEE: Patients and Doctors vs FDA* *Dale Gieringer (1985): 21,000-120,000 lives lost per decade from FDA delay Gieringer: "Loss of life from delay alone in the hundreds of thousands" (not millions) Beta-blockers alone: William Wardell estimated "10,000 lives/year" if allowed; FDA delay 1965-1976 Sam Peltzman: Post-1962 death toll from regulatory delay "easily number in thousands per year Practolol (beta-blocker): "Could save 10,000 lives/year" (Wardell estimate) FDA allowed propranolol 1968 (3 years after Europe); for hypertension/angina not until 1978 Note: "4-10 million" figure not found in sources. Gieringer’s estimates: 21K-120K deaths per decade, "hundreds of thousands" total (not millions). Specific drug delays (beta-blockers):  100K deaths estimated Additional sources: https://fee.org/articles/patients-and-doctors-vs-the-fda \| https://www.econlib.org/library/Enc/DrugLag.html \| https://www.fdareview.org/issues/theory-evidence-and-examples-of-fda-harm/* . 8\. FDAReview.org. Estimated deaths due to FDA delay in approving beta blockers. *FDAReview.org: FDA Harm* (2011) *Beta blockers approved in Europe mid-1970s, FDA didn’t approve until 1981 FDA estimated the drug could save 17,000 lives/year after approval Estimated 100,000 deaths from secondary heart attacks during 6-7 year delay Note: FDA imposed moratorium due to possible animal carcinogenicity despite human clinical evidence from 1974 Additional sources: https://www.fdareview.org/issues/theory-evidence-and-examples-of-fda-harm/ \| https://www.ocregister.com/2011/02/09/walter-williams-death-by-fda-delay-denials/* . 9\. CRS. Evidence of FDA regulatory capture by pharmaceutical industry. *CRS: FDA Human Medical Product User Fees* *FY2023: User fees = 75% of PDUFA program costs (vs 7% in FY1993) FY2022: User fees = 66% (\$1.4B) of human drugs program budget FY2022: User fees = 46% (\$2.9B) of FDA’s total \$6.2B budget Pharma finances  75% of FDA’s drug division (New York Times) Concerns: Budgetary dependence, urgency of PDUFA reauthorizations, required industry participation in negotiations Most PDUFA policy changes favor industry: decreased regulatory standards, shorter approval times, increased industry involvement FDA maintains decisions based on science, not fee collection ability Note: Regulatory capture concerns well-documented. Industry funding grew from 7% (1993) to 75% (2023) of drug review program. No evidence found for "10x government salary" claim for FDA reviewers Additional sources: https://www.congress.gov/crs-product/R44750 \| https://aspe.hhs.gov/sites/default/files/documents/e4a7910607c0dd76c40aa61151d154f9/FDA-User-Fee-Issue-Brief.pdf \| https://pmc.ncbi.nlm.nih.gov/articles/PMC8917050/* . 10\. NIH. Antidepressant clinical trial exclusion rates. *Zimmerman et al.* (2015) *Mean exclusion rate: 86.1% across 158 antidepressant efficacy trials (range: 44.4% to 99.8%) More than 82% of real-world depression patients would be ineligible for antidepressant registration trials Exclusion rates increased over time: 91.4% (2010-2014) vs. 83.8% (1995-2009) Most common exclusions: comorbid psychiatric disorders, age restrictions, insufficient depression severity, medical conditions Emergency psychiatry patients: only 3.3% eligible (96.7% excluded) when applying 9 common exclusion criteria Only a minority of depressed patients seen in clinical practice are likely to be eligible for most AETs Note: Generalizability of antidepressant trials has decreased over time, with increasingly stringent exclusion criteria eliminating patients who would actually use the drugs in clinical practice Additional sources: https://pubmed.ncbi.nlm.nih.gov/26276679/ \| https://pubmed.ncbi.nlm.nih.gov/26164052/ \| https://www.wolterskluwer.com/en/news/antidepressant-trials-exclude-most-real-world-patients-with-depression* . 11\. Smith, A. D. [Dendreon files for bankruptcy, provenge still available](https://www.targetedonc.com/view/dendreon-files-for-bankruptcy-provenge-still-available). (2014). 12\. Freudenmann, R. W., Öxler, F. & Bernschneider-Reif, S. [The origin of MDMA (ecstasy) revisited: The true story reconstructed from the original documents](https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1360-0443.2006.01511.x). *Addiction* **101**, 1241–1245 (2006). 13\. Transform Drug Policy Foundation. MDMA: History and lessons learned (part 1). *Transform* (2021). 14\. Doblin, R. [FDA decision, redoubling of MAPS efforts, and lykos board resignation](https://maps.org/news/update/fda-decision-redoubling-of-maps-efforts-and-lykos-board-resignation/). (2024). 15\. Mitchell, J. M. *et al.* [MDMA-assisted therapy for moderate to severe PTSD: A randomized, placebo-controlled phase 3 trial](https://www.nature.com/articles/s41591-023-02565-4). *Nature Medicine* **29**, 2473–2480 (2023). 16\. Stone, W. FDA advisers reject MDMA as treatment for PTSD. *NPR* (2024). 17\. U.S. Food and Drug Administration. [FDA declines to approve MDMA-assisted therapy for PTSD](https://www.fda.gov/news-events/press-announcements/fda-declines-approve-mdma-assisted-therapy-ptsd). (2024). 18\. Centers for Disease Control and Prevention. [Changes in suicide rates in the united states from 2022 to 2023](https://www.cdc.gov/nchs/products/databriefs/db541.htm). (2024). 19\. U.S. Department of Veterans Affairs. [2024 national veteran suicide prevention annual report](https://www.mentalhealth.va.gov/docs/data-sheets/2024/2024-National-Veteran-Suicide-Prevention-Annual-Report-FINAL-508.pdf). (2024). 20\. National Institute of Mental Health. [Post-traumatic stress disorder (PTSD) statistics](https://www.nimh.nih.gov/health/statistics/post-traumatic-stress-disorder-ptsd). (2023). 21\. Fox, V. *et al.* [Suicide risk in people with post-traumatic stress disorder: A cohort study of 3.1 million people in sweden](https://pmc.ncbi.nlm.nih.gov/articles/PMC7758737/). *Journal of Affective Disorders* **279**, 609–616 (2021). 22\. Sakaeda et al. FAERS adverse event underreporting rate. *PubMed: Empirical estimation of under-reporting in FAERS* (2017) *Empirical estimation: Average reporting rate approximately 6%, meaning  94% of adverse events are underreported Variability: 0.01% to 44% for statin events; 0.002% to \>100% for biological drugs; 20% to \>100% for narrow therapeutic index (NTI) drugs Selective reporting: Serious, unusual events more likely reported than mild or expected ones Newly marketed drugs: Higher reporting rates due to heightened awareness Older drugs: Events often under-reported Note: FAERS voluntary reporting system captures only "tip of the iceberg" of drug safety problems. Under-reporting introduces inherent biases and limitations in pharmacovigilance data Additional sources: https://pubmed.ncbi.nlm.nih.gov/28447485/ \| https://pmc.ncbi.nlm.nih.gov/articles/PMC12393772/* . 23\. Graham, David (FDA) \| Lancet. Vioxx cardiovascular deaths (rofecoxib). *PMC: FDA incapable of protecting against another Vioxx* (2007) *Graham testimony (2004): 88,000-139,000 U.S. heart attacks/strokes from Vioxx; up to 55,000 deaths (40% fatality rate) Lancet study estimate: 88,000 Americans had heart attacks from Vioxx; 38,000 died FDA memo (2004): Vioxx contributed to  27,785 heart attacks and sudden cardiac deaths (1999-2003) High-dose Vioxx: Tripled risk of heart attacks and sudden cardiac death Prescriptions: 92.8 million U.S. prescriptions 1999-2003 Withdrawn: September 30, 2004 after APPROVE trial showed cardiovascular risks Note: Vioxx case demonstrates failure of passive post-market surveillance (FAERS) to detect safety signals in time. Voluntary reporting missed cardiovascular risks for years despite millions of prescriptions Additional sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC534432/ \| https://www.npr.org/2007/11/10/5470430/timeline-the-rise-and-fall-of-vioxx \| https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(05)67712-4/fulltext* . 24\. Sinn, M. P. *The Continuous Evidence Generation Protocol: Two-Stage Validation (RWE → Pragmatic Trials)*. (2025) doi:[10.5281/zenodo.18203375](https://doi.org/10.5281/zenodo.18203375) *We present the Predictor Impact Score (PIS), a novel composite metric operationalizing Bradford Hill causality criteria for automated signal detection from aggregated N-of-1 observational studies. Combined with pragmatic trial confirmation (based on evidence from 108+ embedded trials), this two-stage framework would generate validated outcome labels at 44.1x lower cost than traditional Phase III trials. This enables continuous, population-scale pharmacovigilance and precision dosing recommendations.* 25\. Sinn, M. P. *Ubiquitous Pragmatic Trial Impact Analysis*. (2025) doi:[10.5281/zenodo.18243914](https://doi.org/10.5281/zenodo.18243914) *Only 15 diseases/year get their first treatment each year. With 6,650 diseases lacking effective treatments, the backlog would take 443 years to clear. Integrating pragmatic trials into standard healthcare increases trial capacity 12.3x, cutting that timeline from 443 years to 36 years. The average untreated disease gets a treatment 212 years earlier, saving 10.7 billion deaths at \$9.62 per year of healthy life saved.* 26\. AllTrials. Publication rate of clinical trial results. *AllTrials: Half of Trials Unreported* (2013) * 50.0% of clinical trials never publish results (NHS-funded systematic review, 2010) Schmucker et al (2014): 53% of trials published (analyzing 39 studies, \>20,000 trials) Munch et al (2014): 46% of pain treatment trials published Chang et al (2015): 49% of high-risk cardiac device trials published Positive findings: 3X more likely to be published than negative results Antidepressant example: Published literature showed 94% positive trials; FDA analysis showed only 51% positive Additional sources: https://www.alltrials.net/news/half-of-all-trials-unreported/ \| https://www.nature.com/articles/nature.2013.14286 \| https://pmc.ncbi.nlm.nih.gov/articles/PMC8276556/* . 27\. Science/AAAS. Estimated annual cost of repeating failed experiments due to non-publication of results. *Science/AAAS* (2020) *Up to 50.0% of published preclinical research is irreproducible, with an estimated annual cost of \$28 billion in the U.S. alone. This is based on \$56B annual spending on preclinical research × 50.0% irreproducibility rate. Main causes: reagents/materials (36%), study design (28%), data analysis (25%), protocols (11%). Additional sources: https://www.science.org/content/article/study-claims-28-billion-year-spent-irreproducible-biomedical-research \| https://www.idbs.com/2020/11/replicating-science-28-billion-is-wasted-every-year-in-the-us-alone/* . 28\. FDLI. Japan’s regenerative medicine act and conditional approval pathway. *FDLI: Japan’s Regen Med Pathways* (2019) *Act on Safety of Regenerative Medicine (RM Act) + amended Pharmaceuticals and Medical Devices Act (PMD Act): passed Nov 2013, effective Nov 2014 Conditional and time-limited approval pathway: Obtain approval after exploratory trials demonstrate probable benefit and proven safety 7-year conditional approval period to confirm clinical benefit (e.g., using surrogate endpoints) SAKIGAKE designation (April 2015): Expedited pathway for innovative products targeting serious/life-threatening diseases without effective treatment Benefits: Prioritized consultation, accelerated review, extended re-examination period, premium pricing Examples: Terumo’s HeartSheet and Stemirac obtained conditional approval; Stemirac also SAKIGAKE-designated Additional sources: https://www.fdli.org/2019/02/global-focus-japans-regenerative-medicine-regulatory-pathways-encouraging-innovation-and-patient-access/ \| https://www.insights.bio/cell-and-gene-therapy-insights/journal/article/310/Experiences-from-Japan-SAKIGAKE-Designation-System-for-Regenerative-Medical-Products \| https://pmc.ncbi.nlm.nih.gov/articles/PMC6696404/* . 29\. EMA. EU compassionate use program for experimental drugs. *EMA: Compassionate Use* *Established by: Article 83 of Regulation (EC) No 726/2004 Eligibility: Life-threatening, long-lasting, or seriously debilitating illnesses Requires: No satisfactory authorized treatment; medicine in trials or approval process EMA role: CHMP provides recommendations; national authorities implement programs Pan-European programs rare: Only 6 approved by EMA in last 10 years vs hundreds of national programs Each EU member state sets own rules and procedures Note: Pan-European framework exists but rarely used due to complex navigation across varying national regulations. Most programs remain country-specific Additional sources: https://www.ema.europa.eu/en/human-regulatory-overview/research-development/compassionate-use \| https://pmc.ncbi.nlm.nih.gov/articles/PMC5116859/ \| https://www.eurordis.org/information-support/compassionate-use/* . 30\. Congress.gov. Right to try act (2018). *Congress.gov* (2018) *Right to Try Act (2018) Additional sources: https://www.congress.gov/bill/115th-congress/house-bill/2368 \| https://pmc.ncbi.nlm.nih.gov/articles/PMC7416898/* . 31\. PMC. Number of patients helped by the u.s. Right to try act. *PMC: Understanding Right to Try* (2024) *2018-2022: Supported access to only 12 products total; 4 products in 2023 Number of people treated: Not publicly reported, "likely only in the hundreds Specific documented cases: 75+ neuroendocrine cancer patients (LU-177); at least 1 ALS patient (NurOwn); 7 glioblastoma patients (Gliovac) Minimal safety reporting: Only annual basis, no requirement to publish results/outcomes Additional sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC7416898/ \| https://www.statnews.com/2024/08/02/trump-gave-patients-right-to-try-it-hasnt-helped-them/ \| https://www.healio.com/news/hematology-oncology/20200303/right-to-try-a-wellintentioned-but-misguided-law* . 32\. CNN. FDA delay in approving rapid COVID-19 tests. *CNN: How Government Delayed Testing* (2020) *Feb 16, 2020: Seattle research lab ordered to stop COVID testing without FDA approval May 2020: Gates Foundation partnership instructed to discontinue testing until authorization Feb 29, 2020: U. Nebraska finally got FDA permission after Feb 4 special permission Timeline: Critical delays measured in weeks-to-months (Feb-March 2020), not specific "6 months FDA guidance suggested EUAs needed for laboratory-developed tests (LDTs), causing delays By May 2020: 400+ test applications awaiting FDA review Note: Delays were weeks-to-months during critical Feb-March 2020 period, not continuous "6 months." FDA intervention added minimal value while contributing to deadly delays Additional sources: https://www.cnn.com/2020/04/09/politics/coronavirus-testing-cdc-fda-red-tape-invs/index.html \| https://www.yalelawjournal.org/forum/deadly-delay-the-fdas-role-in-americas-covid-testing-debacle \| https://oig.hhs.gov/oei/reports/OEI-01-20-00380.asp* . 33\. NPR. Contamination of early CDC COVID-19 tests in 2020. *NPR: CDC Test Flawed* (2020) *CDC distributed flawed test kits Feb 6, 2020 - contaminated reagents caused false positives 24 of 26 public health labs found contamination, CDC recalled kits by Feb 10 Tests made in CDC lab (not manufacturing facility), violated sound manufacturing practices Contamination occurred in Respiratory Virus Diagnostic Lab during processing Note: Delays had significant consequences for early pandemic tracking and response Additional sources: https://www.npr.org/2020/11/06/929078678/cdc-report-officials-knew-coronavirus-test-was-flawed-but-released-it-anyway \| https://www.cnn.com/2020/04/18/politics/cdc-coronavirus-testing-contamination/index.html* . 34\. PMC. Estimated excess deaths attributed to FDA’s COVID-19 response. *PMC: EUAs vs FDA Approval Implications* *FDA regulations restricted clinician/patient access to COVID-19 testing, remdesivir, vaccines General drug delay estimate: 37,000-76,000 deaths per one-year delay Testing delays: By May 2020, 400+ applications awaiting FDA review Seattle lab ordered to stop testing Feb 16 for lack of FDA approval Gates Foundation partnership instructed to discontinue testing May 2020 until authorization EUA process prevented months of vaccine/testing delays Note: Specific "500,000+" figure not found in sources. Research shows FDA testing restrictions caused weeks-to-months of critical delays (Feb-March 2020). One-year drug delay = 37-76K deaths Additional sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC8101583/ \| https://pmc.ncbi.nlm.nih.gov/articles/PMC8012986/ \| https://www.yalelawjournal.org/forum/deadly-delay-the-fdas-role-in-americas-covid-testing-debacle* . 35\. Biotechnology Innovation Organization (BIO). BIO clinical development success rates 2011-2020. *Biotechnology Innovation Organization (BIO)* (2021) *Phase I duration: 2.3 years average Total time to market (Phase I-III + approval): 10.5 years average Phase transition success rates: Phase I→II: 63.2%, Phase II→III: 30.7%, Phase III→Approval: 58.1% Overall probability of approval from Phase I: 12% Note: Largest publicly available study of clinical trial success rates. Efficacy lag = 10.5 - 2.3 = 8.2 years post-safety verification. Additional sources: https://go.bio.org/rs/490-EHZ-999/images/ClinicalDevelopmentSuccessRates2011_2020.pdf* . 36\. Tufts CSDD. Cost of drug development. *Various estimates suggest \$1.0 - \$2.5 billion to bring a new drug from discovery through FDA approval, spread across  10 years. Tufts Center for the Study of Drug Development often cited for \$1.0 - \$2.6 billion/drug. Industry reports (IQVIA, Deloitte) also highlight \$2+ billion figures.* 37\. Nature Reviews Drug Discovery. Drug trial success rate from phase i to approval. *Nature Reviews Drug Discovery: Clinical Success Rates* (2016) *Overall Phase I to approval: 10-12.8% (conventional wisdom  10%, studies show 12.8%) Recent decline: Average LOA now 6.7% for Phase I (2014-2023 data) Leading pharma companies: 14.3% average LOA (range 8-23%) Varies by therapeutic area: Oncology 3.4%, CNS/cardiovascular lowest at Phase III Phase-specific success: Phase I 47-54%, Phase II 28-34%, Phase III 55-70% Note: 12% figure accurate for historical average. Recent data shows decline to 6.7%, with Phase II as primary attrition point (28% success) Additional sources: https://www.nature.com/articles/nrd.2016.136 \| https://pmc.ncbi.nlm.nih.gov/articles/PMC6409418/ \| https://academic.oup.com/biostatistics/article/20/2/273/4817524* . 38\. Baily, M. N. Pre-1962 drug development costs (baily 1972). *Baily (1972)* (1972) *Pre-1962: Average cost per new chemical entity (NCE) was \$6.5 million (1980 dollars) Inflation-adjusted to 2024 dollars: \$6.5M (1980) ≈ \$22.5M (2024), using CPI multiplier of 3.46× Real cost increase (inflation-adjusted): \$22.5M (pre-1962) → \$2,600M (2024) = 116× increase Note: This represents the most comprehensive academic estimate of pre-1962 drug development costs based on empirical industry data Additional sources: https://samizdathealth.org/wp-content/uploads/2020/12/hlthaff.1.2.6.pdf* . 39\. Think by Numbers. Pre-1962 physician-led clinical trials. *Think by Numbers: How Many Lives Does FDA Save?* (1966) *Pre-1962: Physicians could report real-world evidence directly 1962 Drug Amendments replaced "premarket notification" with "premarket approval", requiring extensive efficacy testing Impact: New regulatory clampdown reduced new treatment production by 70%; lifespan growth declined from  4 years/decade to  2 years/decade Drug Efficacy Study Implementation (DESI): NAS/NRC evaluated 3,400+ drugs approved 1938-1962 for safety only; reviewed \>3,000 products, \>16,000 therapeutic claims FDA has had authority to accept real-world evidence since 1962, clarified by 21st Century Cures Act (2016) Note: Specific "144,000 physicians" figure not verified in sources Additional sources: https://thinkbynumbers.org/health/how-many-net-lives-does-the-fda-save/ \| https://www.fda.gov/drugs/enforcement-activities-fda/drug-efficacy-study-implementation-desi \| http://www.nasonline.org/about-nas/history/archives/collections/des-1966-1969-1.html* . 40\. ICER. Value per QALY (standard economic value). *ICER* (2024) *Standard economic value per QALY: \$100,000–\$150,000. This is the US and global standard willingness-to-pay threshold for interventions that add costs. Dominant interventions (those that save money while improving health) are favorable regardless of this threshold. Additional sources: https://icer.org/wp-content/uploads/2024/02/Reference-Case-4.3.25.pdf* . 41\. Oren Cass, Manhattan Institute. RECOVERY trial cost per patient. *Oren Cass* (2023) *The RECOVERY trial, for example, cost only about \$500 per patient... By contrast, the median per-patient cost of a pivotal trial for a new therapeutic is around \$41,000. Additional sources: https://manhattan.institute/article/slow-costly-clinical-trials-drag-down-biomedical-breakthroughs* . 42\. FDA Study via NCBI. Trial costs, FDA study. *FDA Study via NCBI* *Overall, the 138 clinical trials had an estimated median (IQR) cost of \$19.0 million (\$12.2 million-\$33.1 million)... The clinical trials cost a median (IQR) of \$41,117 (\$31,802-\$82,362) per patient. Additional sources: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6248200/* .