Sold As Safe

Cocaine

Verified 8 Sep Vin Mariani · Parke-Davis · Coca-Cola · Freud

Sold as A tonic, an anaesthetic, a cure for melancholy, and — the recurring theme of this whole register — a cure for morphine addiction. Freud's Über Coca (1884) recommended it for exactly that. Parke-Davis sold cocaine toothache drops marketed for children. Coca-Cola carried roughly 9mg a serving from 1886 until 1903, and cocaine toothache drops for children were advertised from 1885.

What happened Freud's friend Ernst von Fleischl-Marxow — Freud's friend and mentor — began cocaine treatment for morphine addiction on 7 May 1884, developed cocaine psychosis, returned to morphine, bankrupted himself pursuing the drug, and died in 1891 at 44, in worse pain than before (verified 8 Sep). Mass dependency followed the mass marketing. (Correction: I named Merck as the dominant producer. The source I can verify identifies Parke-Davis as cocaine's largest manufacturer — and it also published the Therapeutic Gazette, where many of the favourable case reports appeared.)

Ended The US Harrison Narcotics Tax Act, signed 17 December 1914; the UK Dangerous Drugs Act 1920.

Who gained Merck, which was the world's dominant cocaine producer, and Parke-Davis, which sold it in every delivery form including a syringe kit. Vin Mariani and Coca-Cola built consumer brands on it. Freud's early reputation was made partly on the cocaine papers.

Heroin

Verified 8 Sep Bayer, Elberfeld

Sold as Diacetylmorphine, trademarked Heroin, launched commercially in 1898 as a cough suppressant for bronchitis, tuberculosis and asthma — and explicitly as a non-addictive substitute for morphine and codeine. Sold over the counter, including in preparations for children. Bayer launched aspirin in the same period; the two sat in the same catalogue.

What happened Heroin is metabolised to morphine and is more lipid-soluble, so it crosses into the brain faster — it is more addictive than the drug it was sold as a cure for. Reports of dependency appeared within a few years and were available well before the marketing stopped.

Ended Bayer stopped producing it in 1913 — as addiction numbers climbed and prohibition in several countries looked imminent, rather than because the evidence changed (verified 8 Sep). The US Harrison Narcotics Tax Act restricted it in 1914 and it was banned outright in 1924; Britain brought it under the Dangerous Drugs Act 1920, and it remained prescribable long after.

Why it matters The exact structure recurs in this register three more times: a new opioid is marketed as the safe answer to the last opioid. Morphine for opium. Heroin for morphine. Methadone for heroin. OxyContin for everything.

Who gained Bayer, which held the trademark. The commercial damage was nil: Bayer lost the Heroin and Aspirin trademarks in 1919 as war reparations, not as a penalty, and is today one of the largest pharmaceutical companies on earth. The idea gained more than the company did: heroin was sold as a cure for morphine addiction, which required believing that addiction was a property of the molecule rather than of the person — and therefore chemically substitutable. That belief was not Bayer's invention and did not die with the product. It sold methadone, and it is audible in the opioid marketing of ninety years later.

Opiate soothing syrups for infants

Verified 8 Sep Mrs Winslow's Soothing Syrup · Godfrey's Cordial · Street's Infants' Quietness

Sold as Teething and colic remedies for babies, advertised with illustrations of contented mothers. Mrs Winslow's — first sold in 1849 from Bangor, Maine — contained morphine and alcohol at roughly 65mg of morphine per fluid ounce, and the label did not say so (verified 8 Sep). The printed dosage was 6 to 10 drops for a child under one month, half a teaspoon at three months, and a teaspoonful three or four times a day from six months. A single teaspoonful contained enough morphine to kill an average child.

What happened Infant deaths from respiratory depression, and dependency in babies. The American Medical Association listed it under Baby Killers in Nostrums and Quackery in 1911, and JAMA ran headlines including the slaughter of the innocents continues (verified 8 Sep). The death toll is unknown and unknowable — parents did not know the medicine was dangerous and the deaths were routinely attributed to the teething or diarrhoea it had been given for. It went on selling into the 1930s. Godfrey's Cordial was implicated in infant deaths across industrial northern England, where working mothers used it to quiet children through a shift.

Ended Labelling forced by the US Pure Food and Drugs Act 1906 and the UK's later poisons legislation; sales continued into the 1930s.

Who gained The patent-medicine trade, which was the largest single advertiser in Victorian and Edwardian newspapers — and therefore also the press that carried the advertising and declined to investigate it. Indirectly, mill and factory employers, since the syrups let mothers work a full shift.

Arsenic tonics

Verified 8 Sep Fowler's Solution · arsenic complexion wafers

Sold as A general tonic. Fowler's Solution — 1% potassium arsenite in water — was proposed in 1786 by Thomas Fowler of Stafford, entered the London Pharmacopoeia in 1809, and stayed in use over 150 years, into the 1940s: for asthma, eczema, psoriasis, rheumatism, chorea, syphilis, tuberculosis and ulcers (verified 8 Sep). Dr Campbell's Arsenic Complexion Wafers were sold to women from 1889, promising to transform the most sallow skin into a radiant health.

What happened Chronic arsenicism, hyperkeratosis, and skin, bladder and lung cancers with latencies of decades. Fowler's is one of the earliest documented human carcinogens.

Who gained Apothecaries and, importantly, the profession itself: Fowler's Solution was in the official pharmacopoeia. This was not fringe quackery being tolerated — it was orthodoxy.

Mercury teething powders

Verified 8 Sep Calomel preparations · Steedman's Powders

Sold as Standard treatment for teething pain and as a purgative for infants, sold freely in British chemists.

What happened Pink disease (infantile acrodynia) — pain, photophobia, peeling hands and feet. Correction (verified 8 Sep): I conflated two figures. About 1 in 500 exposed children developed the disorder — it appears to be an idiosyncratic hypersensitivity rather than a dose effect — and that is an incidence, not a death rate. On deaths, the official English record shows 585 deaths from acrodynia between 1939 and 1948, falling to single figures by the mid-1950s. Mercury was identified as the cause in 1945, when 12 of 14 diagnosed children were found to have raised mercury levels; calomel was removed from British teething powders in 1954, and the disease disappeared.

Why it matters The gap between a plausible mechanism being published and the product leaving the shelf is roughly a decade here. That interval is the single most repeatable number in this register.

Who gained Steedman's and the British patent-medicine trade, whose products were sold in registered pharmacies with professional endorsement. What pharmacy gained was the reverse of what it supposed: stocking these products was what distinguished a registered pharmacy from a hawker, so the profession lent its authority to compounds it had not tested in exchange for the standing that came from selling them. And pink disease was not recognised as mercury poisoning until the 1950s — the diagnosis was missing, so no individual pharmacist ever had to weigh a known risk. Authority was extended on credit and the bill was presented forty years later.

Radium

Verified 8 Sep Radithor · US Radium Corporation · Radiendocrinator

Sold as Radioactivity as vitality. Radithor was radium dissolved in water, sold as an energy tonic. There were radium suppositories, radium face creams, and a radium-lined jockstrap marketed to men for virility.

What happened The industrialist Eben Byers drank more than a thousand bottles of Radithor on his doctor's recommendation and died on 31 March 1932, aged 51, having lost most of both jaws, with holes in his skull (verified 8 Sep). Separately, at the US Radium factory in Orange, New Jersey, opened in 1917, young women painting luminous dials were told the paint was harmless and taught to shape their brushes between their lips — "lip, dip, paint". The first died in 1922 at 25; by 1924 fifty were ill and a dozen were dead, of radium necrosis of the jaw and bone sarcoma.

Ended Five women — Grace Fryer, Edna Hussman, Katherine Schaub, Quinta McDonald and Albina Larice — brought suit in January 1928. None was strong enough to raise her arm to take the oath; two were bedridden. Fearing they would not survive a trial, they settled in June 1928 for $10,000 each plus $600 a year for life and their medical and legal costs. Byers's death in 1932 collapsed the radium fad and tightened regulation of radioactive products. Their case established in US law an employee's right to sue over occupational disease — one of the register's rare instances where harm produced a durable legal protection rather than a recall.

Who gained US Radium Corporation and Bailey Radium Laboratories. The largest customer for luminous dials was the US military. Radium's respectability came directly from the Curies' Nobel prizes — the scientific prestige was the marketing.

Thorotrast

Verified 8 Sep Heyden Chemical

Sold as A superb X-ray contrast agent — thorium dioxide gave far better cerebral angiograms than anything available, and was well tolerated. Correction (verified 8 Sep): I wrote that "millions of patients received it worldwide". The figure is more than 50,000, before it was abandoned around 1955. I was wrong by roughly two orders of magnitude.

What happened Thorium is an alpha emitter and the body never clears it. It lodges in liver, spleen and marrow and irradiates them for life. Liver angiosarcoma — otherwise vanishingly rare — appeared in recipients twenty to forty years later, with one documented case at 65 years, along with leukaemias. The German cohort study followed 2,326 exposed patients against 1,890 matched controls and recorded 454 primary liver cancers in the exposed group against three in the controls; the liver-cancer rate ratio is about 36 (verified 8 Sep).

Why it matters The best single argument in the register for long-latency surveillance: the drug performed exactly as promised, was genuinely superior, and the harm was undetectable for a generation.

Who gained Heyden Chemical, and the young speciality of radiology, which gained a genuinely superior diagnostic tool and used it in good faith.

Shoe-fitting fluoroscopes

Verified 8 Sep Pedoscope (UK) · X-Ray Shoe Fitter (US)

Sold as A scientific shoe fitting — children stood on a live X-ray machine in the high street so parents could see the bones in their feet. At their 1950s peak they were in more than 10,000 shops in the United States, and in Britain they were known as Pedoscopes after the St Albans firm that made them (verified 8 Sep).

What happened Repeated unshielded exposure to children and, worse, to the shop assistants beside them all day — a salesman standing ten feet away took more than 1 mSv an hour. One saleswoman who operated a machine 10 to 20 times a day for a decade developed radiation dermatitis of the hands; a shoe model suffered a burn severe enough that her leg was amputated (verified 8 Sep).

Ended Pennsylvania banned them first, in 1957; by 1970, 33 states had, and the rest heavily restricted them. In Britain they stayed in use into the mid-1970s — later than I first wrote (verified 8 Sep).

Who gained High-street shoe retailers, who used it purely as a sales device. There was never a clinical claim behind it — the machine's authority was borrowed wholesale from medicine.

Lysol as feminine hygiene

Verified 8 Sep Lehn & Fink

Sold as Advertised for decades in women's magazines for intimate douching, with copy implying — never stating, because it was illegal to state — contraceptive effect. Endorsed in advertisements by named European physicians.

What happened The pre-1953 formulation was cresol-based. Chemical burns, inflammation and systemic poisoning: by 1911 doctors had already recorded 193 Lysol poisonings and five deaths from uterine irrigation — that is, four decades of documented harm before the product's peak as a douche in the late 1920s and 1930s (verified 8 Sep). It was not an effective contraceptive.

Who gained Lehn & Fink. Structurally, the beneficiary was also the law. Contraception remained illegal in the United States until 1965 for married couples, so feminine hygiene was the only permitted channel for a contraceptive claim — the prohibition created the market for a dangerous substitute and made accurate information about it unlawful to print.

Tobacco, on doctor's orders

Verified 8 Sep R. J. Reynolds · Philip Morris · the tobacco industry

Sold as Medically endorsed. R. J. Reynolds ran More doctors smoke Camels than any other cigarette from 1 January 1946, claiming a survey of 113,597 physicians — with no methodology described, and no record of whether free Camels had been posted to those doctors shortly beforehand (verified 8 Sep). The companies advertised in most weekly and monthly medical journals, JAMA above all. Menthol brands were sold for throat irritation; cigarettes were issued in military rations and handed out in hospitals.

What happened Doll and Hill published the lung cancer association in the BMJ in 1950. The industry's response came on 14 December 1953, when the chief executives of the major American tobacco companies met the public relations firm Hill & Knowlton at the Plaza Hotel in New York and founded the Tobacco Industry Research Committee — a body whose function was to manufacture the appearance of scientific doubt (verified 8 Sep). The strategy was set out plainly in a 1969 Brown & Williamson memorandum: Doubt is our product, since it is the best means of competing with the "body of fact" that exists in the mind of the general public, the aim being to neutralize the influence of academic scientists. (Correction: I had implied that line dated from the 1953 period; it is from 1969.) Note also which firm is in this record — Hill & Knowlton reappears in Pillar II staging the Kuwaiti incubator testimony in 1990.

Why it matters This is the template case for the whole study: not a wrong belief, but a deliberately sustained one. It also belongs in the control set, because the warning was correct from 1950 and was resisted for forty-five years.

Who gained The tobacco industry, its advertising agencies, and — the part that goes unmentioned — the medical journals themselves. The Journal of the American Medical Association and others carried cigarette advertising for years and lent it their mastheads. Governments took the duty. This is the strongest single case for the study's thesis and it implicates the professional bodies directly.

Thimerosal: how its safety was established

Verified 13 Sep Eli Lilly · US Food and Drug Administration · US Centers for Disease Control and Prevention

Sold as A safe mercury-based preservative, patented in 1929 and sold by Eli Lilly as Merthiolate, and used in multi-dose vaccines for about seventy years. Its removal as a precaution is recorded in the counter-register below, under Removing thimerosal as a precaution. This record is the other half: how its safety came to be accepted, and how the question was handled when it was reopened. It is scored on conduct, not on whether thimerosal causes autism.

The founding evidence Around 1930, thimerosal was injected intravenously into 22 patients with meningococcal meningitis. It did not save them, and all of them died. The published account by Powell and Jamieson reported that these large doses did not produce any anaphylactoid or shock symptoms, conceded that the drug's beneficial effects were not definitely proven, and called for a wide range of toxicity tests. There is no evidence those tests were done.

What Congress found A 2003 report by the House Government Reform subcommittee chaired by Dan Burton, a leading proponent of the vaccine–autism hypothesis, concluded that the introduction of thimerosal appears to have been based on a single uncontrolled and poorly reported human study, and that the FDA never required the pharmaceutical industry to conduct extensive safety studies on thimerosal or ethylmercury. It also recorded that an FDA advisory panel judged thimerosal unsafe for over-the-counter topical use in 1982, that the ban was not finalised until 1998, and that the finding did not trigger a review of vaccines (verified 13 Sep against the report as entered in the Congressional Record, 21 May 2003). The report's author had a position; the quotations from the 1930s record are not his.

When the question reopened In 1999–2000 the CDC epidemiologist Thomas Verstraeten ran thimerosal exposure against outcomes in the Vaccine Safety Datalink, the CDC's database of health-insurer records. In June 2000 the CDC convened specialists from government, the WHO, universities and vaccine manufacturers at the Simpsonwood retreat centre in Georgia to review the results in private. The released transcript records Verstraeten reporting statistically significant relationships between exposure and unspecified developmental delay, tics, attention deficit disorder, and language and speech delays. For neurodevelopmental disorders as a whole, the trend test was significant with a P value below 0.01. Asked about biological plausibility, he said: I was actually stunned by what I saw because I thought it is plausible. The reviewer Dr Weil said: You can play with this all you want. They are linear. They are statistically significant. Another reviewer said the data won't go away and would be captured by the public. Another, whose grandson had been born that morning, said: I do not want that grandson to get a Thimerosal containing vaccine until we know better what is going on. The reviewers were not unanimous: several rated the evidence for causation weak, citing the quality of the diagnoses and the size of the effect. Participants were asked to treat the material as embargoed and protected until it is made public at the Advisory Committee on Immunization Practices two weeks later. One remarked that it had so far been kept out of less responsible hands.

What was published The study appeared in Pediatrics in November 2003 as a two-phase analysis. In phase I, cumulative exposure was associated with tics at one health insurer (relative risk 1.89) and with language delay at another (1.13 at three months, 1.07 at seven). Phase II, at a third insurer, found no significant associations, and in no analyses were significant increased risks found for autism or attention-deficit disorder. Its conclusion: No consistent significant associations were found between TCVs and neurodevelopmental outcomes. The June 2000 presentation had reported a significant relationship with attention deficit disorder, and the published paper reports none. By publication Verstraeten was working for GlaxoSmithKline, a vaccine manufacturer. He answered his critics in a 2004 letter to Pediatrics titled Thimerosal, the Centers for Disease Control and Prevention, and GlaxoSmithKline.

Access to the data Parents' groups and outside scientists spent years seeking the underlying data so they could test the analysis themselves. The Institute of Medicine found in 2005 that, for researchers outside the CDC's network, the opportunities to use the database had been at best, unclear and narrow. The 2003 House report said the process for independent access remains inadequate.

Scored as conduct The later studies that follow individual children do not support a link between thimerosal and autism (see the counter-register record), and this record does not claim one. What it records is conduct. A preservative was accepted on a single study of dying patients and never required to be tested further. An early signal was reviewed behind closed doors, with manufacturers present and the public excluded until a policy meeting. The database was held so that the people raising the question could not check the answer. Each of those would be condemned in this register if a company had done it, and the symmetric standard applies to agencies as well.

Who gained Eli Lilly, which sold a preservative for seventy years without having to prove it safe. Vaccine manufacturers, for whom thimerosal made cheap multi-dose vials possible. And the vaccination programme itself. An agency whose mission depends on public confidence has a structural reason to settle a safety question in private before it faces it in public. The Simpsonwood transcript shows that reason spoken aloud: the concern was not only what the data meant but who might get hold of it.

Eugenic sterilisation and Buck v Bell

Verified 13 Oct Eugenics Record Office · Harry Laughlin · US Supreme Court · state boards · Carnegie, Harriman and Rockefeller money

Sold as Applied science in the service of public welfare. Feeblemindedness, criminality, pauperism and promiscuity were held to be inherited, and preventable by preventing reproduction. Indiana passed the first sterilisation law in 1907. Harry Laughlin's model law of 1914 was written to survive legal challenge, and about eighteen states adopted versions of it, Virginia's in 1924.

Who paid for the science The Eugenics Record Office at Cold Spring Harbor opened in 1910 on a gift from Mary Harriman, was taken over by the Carnegie Institution, and also drew Rockefeller Foundation money. Carnegie withdrew its funding and the office closed in 1939. These were the most respectable institutions in American science, and the Carnegie Institution apologised for the funding in 2020.

What the court held In Buck v Bell (2 May 1927) the Supreme Court upheld Virginia's law by eight votes to one. Holmes wrote that three generations of imbeciles are enough. Carrie Buck's case was built on a diagnosis of feeblemindedness in her, her mother and her seven-month-old daughter Vivian, the last after a cursory examination. Paul Lombardo's history, Three Generations, No Imbeciles (2008), finds little evidence that Carrie had any cognitive disability and describes the case as a sham, with her own lawyer working against her. The judgment has never been formally overruled; Skinner v Oklahoma (1942) superseded it only in part.

What happened Roughly 60,000 to 70,000 Americans were sterilised without meaningful consent, in more than thirty states, into the 1970s. California accounted for the largest share. North Carolina's board sterilised about 7,600 people up to 1974, and the state voted compensation in 2013. Germany's 1933 sterilisation law was closely modelled on Laughlin's; estimates of those sterilised under it run from 200,000 to 400,000. In 1936 Heidelberg gave Laughlin an honorary degree, which he accepted in writing as doubly valued coming from Germany. The programme that followed, Aktion T4, killed an estimated 275,000 to 300,000 disabled people. Its own ledger at Hartheim recorded the money saved by "disinfecting" 70,273 people, counted in Reichsmarks of food not eaten.

Why it belongs at the centre of this study Nothing here was sold. No product, no patent, no marketing. The claim that complex social traits were simply heritable ran far ahead of the evidence and was held with total confidence by universities, courts, charities and medical boards, for forty years, across two continents. It is the study's purest case of a false claim with no commercial beneficiary and an enormous body count, and this register did not contain it until the owner's own gap search found it missing.

Who gained A profession, a budget and a worldview. Psychiatry, public health and the new science of genetics gained jurisdiction over who could reproduce, and careers were built on it. States gained a cost argument: the Hartheim ledger is the extreme form, but American advocates also argued sterilisation would save the cost of institutions, and that reasoning ran through welfare administration everywhere. Philanthropies gained the standing of funding modern science. And the respectable classes gained an explanation of poverty, crime and illegitimacy that located the cause in the poor themselves rather than in anything society might owe them. No one needed to profit. The idea paid in authority, in budgets and in self-regard, and it kept paying for forty years.

Elixir Sulfanilamide

Verified 8 Sep S. E. Massengill Company

Sold as A raspberry-flavoured liquid form of a genuinely effective new antibacterial, made palatable for children. The company's chief chemist dissolved it in diethylene glycol — antifreeze — and tested it for flavour, appearance and fragrance. Not for toxicity.

What happened 105 deaths — 71 adults and 34 children — from renal failure, after 633 commercial lots shipped from 4 September 1937 (verified 8 Sep). Massengill had broken no law: none required a safety test. The company's president said publicly he could see no responsibility on his part. The chemist who formulated it, Harold Cole Watkins, killed himself on 17 January 1939.

Ended Directly produced the US Federal Food, Drug and Cosmetic Act of 1938.

Who gained Massengill, briefly — and then a regulator. The company sold a palatable liquid formulation into a market that wanted one, which is an ordinary commercial gain however short-lived. The larger beneficiary was the 1938 Act, which acquired the authority it had been denied for thirty years. Note the penalty. Massengill pleaded guilty to 112 counts of adulteration and misbranding and was fined $26,100 — the largest penalty ever levied under the 1906 Act, and about $250 per death (verified 8 Sep). The only legal hook was the word elixir, which implies alcohol the product did not contain. Selling an untested poison as medicine was not itself an offence, and not one of the 105 deaths was.

Lobotomy

Verified 8 Sep Egas Moniz · Walter Freeman · UK asylum services

Sold as A cure for schizophrenia, depression, anxiety and — increasingly — inconvenience. Moniz received the Nobel Prize in Medicine in 1949. Freeman toured America performing transorbital lobotomies through the eye socket with an instrument modelled on an ice pick, sometimes a dozen a day, without a surgical theatre.

What happened Estimates of 40,000 to more than 50,000 procedures in the United States, most between 1949 and 1952, and 17,000 in Britain (verified 8 Sep). Freeman himself performed transorbital lobotomies on some 2,500 patients across 23 states before his death in 1972. No controlled trial ever supported it. Outcomes included apathy, incontinence, seizures, personality obliteration and death. Rosemary Kennedy was left permanently incapacitated at 23.

Ended Displaced by chlorpromazine from the mid-1950s — replaced by a drug, not refuted by evidence. The Nobel has never been rescinded.

Who gained Walter Freeman, personally, in fees and fame. But the institutional beneficiary is the one that matters: asylum systems, which were catastrophically overcrowded and underfunded, and for which a procedure that made patients manageable was an operational and budgetary solution. The Nobel committee gained a prize it has never revisited.

Supplemental oxygen for premature babies

Verified 8 Sep Neonatal units, UK and US

Sold as Obvious good practice. Premature infants breathe poorly; incubators were therefore run at high oxygen concentrations, routinely and without trial.

What happened Retrolental fibroplasia — an epidemic of blindness in premature infants, described by Terry in 1942, peaking in 1951, and blinding on the order of 10,000 children between 1942 and 1954, with some estimates above 12,000 affected (verified 8 Sep). It became a leading cause of childhood blindness in the United States and elsewhere. Stevie Wonder is among those affected.

Ended It took roughly fifteen years from Terry's first description to the identification of oxygen; the cooperative controlled study reported in the mid-1950s and use was then restricted immediately. The practice had run that whole time on inference alone.

Why it matters The cleanest possible answer to but it was obviously beneficial. Nothing about the reasoning was stupid; it simply was not tested.

Who gained No company — and a beneficiary all the same: clinical authority itself. What was preserved was the entitlement to act universally on mechanistic reasoning without testing it. Oxygen was given because it was obviously right, and the obviousness was the credential. The trial that settled it cost no manufacturer anything; what it cost was the standing of unaided clinical judgement, which is why it was resisted and why neonatology now treats this as the founding case for trials in newborn care.

Routine X-ray of pregnant women

Verified 8 Sep Obstetrics, UK and US · radiology

Sold as Good modern antenatal care. Pelvimetry to measure the birth canal, and X-ray to check fetal position, multiple pregnancy and placental site. Presented to women as a reassuring routine, and in some hospitals given to a majority of expectant mothers, sometimes repeatedly. It was never trialled, and the dose was never justified against a measured benefit.

What happened Alice Stewart's Oxford Survey of Childhood Cancers reported in 1956 that a single antenatal X-ray roughly doubled the child's risk of leukaemia and other childhood cancers. It was the first clear demonstration that a dose previously considered trivial could be harmful to a fetus — and that there may be no safe threshold.

What happened next She was disbelieved. Her funding was cut, her findings were disputed by the radiological establishment for years, and the practice continued for roughly a quarter of a century after she published. Brian MacMahon replicated her result in the US in 1962 and it still did not stop it. Routine antenatal X-ray was not abandoned in Britain until the 1970s, once ultrasound offered an alternative — displaced by a technology rather than refuted by evidence, exactly as lobotomy was.

Why it matters This is the register's cleanest single case, and it should probably lead the study. Every element is present at once: a practice adopted with no trial; a correct warning published early; an institution that rejected the warning because accepting it meant accepting that the profession had been harming children; a delay measured in decades; and a harm that fell on the fetus while the decision was taken by the clinician. The information was not missing. The willingness to act on it was.

Scale Genuinely uncertain and worth your own work — exposure was routine across two continents for decades, and the excess-risk estimate is per-exposure and small in absolute terms, so the total is a modelled figure rather than a counted one. Do not repeat a headline number you cannot trace.

Who gained Radiology and obstetrics — in standing, not in money. Radiology gained standing and equipment utilisation, and obstetrics gained the appearance of technical rigour. The strongest interest at work was not money but professional self-image — the finding was resisted hardest by the people it implicated, which is the same mechanism that destroyed Semmelweis a century earlier and delayed the recognition of pink disease.

Diethylstilbestrol (DES)

Eli Lilly and c. 200 other manufacturers

Sold as A synthetic oestrogen given to pregnant women to prevent miscarriage. Millions of prescriptions; estimates of exposure in the US alone run to several million mothers and children.

What happened Two separate failures. First, it never worked — Dieckmann's 1953 controlled trial at the University of Chicago, some 2,000 women on alternate allocation to DES or placebo, found no beneficial effect whatever on preventing miscarriage, and prescribing continued for eighteen more years (verified 8 Sep). Second, in 1971 Herbst reported seven cases of clear cell adenocarcinoma of the vagina in young women in the NEJM: seven of the eight mothers of affected daughters had taken DES in the first trimester, against one of thirty-two controls. Also implicated: reproductive tract malformations, infertility, and effects reported in a third generation.

Ended FDA contraindicated it in pregnancy in 1971. It remained in use in some countries into the 1980s.

Why it matters The eighteen-year gap is the finding. This is not a case of unknowable risk; it is a case of a negative trial being ignored because the drug was already established.

Who gained Eli Lilly and roughly 200 other manufacturers — DES was never patent-protected, so the whole industry sold it. Obstetrics gained something to offer women who had miscarried, which is why the negative 1953 trial had no effect: the drug met a demand that evidence was not the source of.

Amphetamines as diet pills and mood lifters

Verified 8 Sep Smith, Kline & French (Benzedrine) · Obetrol

Sold as Smith, Kline & French sold the Benzedrine inhaler over the counter from 1932 — a cotton strip soaked in amphetamine oil, which users quickly learned to prise out and swallow or inject (verified 8 Sep). Pills followed from the mid-1930s for sleep disorders, low mood and weight loss, prescribed heavily to women. In Britain, Drinamyl — purple hearts — combined amphetamine with a barbiturate.

What happened The scale is the point: around 750 million Benzedrine pills a year by the end of 1945, and 8 billion a year by 1962 (verified 8 Sep). Widespread dependency, amphetamine psychosis and a large diversion market followed. The FDA did not make the inhaler prescription-only until 1959 — prescribing peaked before restriction, rather than after evidence.

Who gained Smith, Kline & French above all. Also employers and armed forces on every side of the Second World War, which issued amphetamine deliberately and at scale. Underneath both is a belief with no owner: that fatigue and appetite are defects medicine ought to abolish — performance as a legitimate medical target. That conviction preceded the drug, survived its restriction, and is the reason the same market re-forms around each successor compound.

Barbiturates

Verified 8 Sep Bayer (Veronal) · multiple

Sold as Safe hypnotics and sedatives for anxiety and insomnia, prescribed on a very large scale for half a century.

What happened A narrow therapeutic index — the effective dose and the fatal dose are uncomfortably close, and lethally so with alcohol. Severe dependency, with a withdrawal syndrome that can kill. The comparison that quantifies it (verified 8 Sep): a UK study of fatal overdoses from sedatives between 1983 and 1999 found 146.2 fatal toxicities per million barbiturate prescriptions against 7.4 per million for benzodiazepines — a roughly twenty-fold difference.

Ended Displaced by benzodiazepines from the 1960s — which were themselves marketed as the safe alternative, and became the next dependency problem. Again the pattern.

Who gained Bayer, Abbott and others for fifty years — and then Roche and Wyeth, who took the same market with benzodiazepines marketed as the safe replacement, and held it through a second dependency epidemic. The durable beneficiary is neither firm but the premise they both sold into: that distress is a pharmacological problem with a pharmacological answer. Note what that premise survived — barbiturates failed and it re-sold benzodiazepines; benzodiazepines failed and it re-sold the z-drugs. The molecule changed three times. The belief did not change once.

Chloramphenicol

Verified 8 Sep Parke-Davis

Sold as A powerful broad-spectrum antibiotic, prescribed for minor and self-limiting infections including sore throats and colds.

What happened Idiosyncratic aplastic anaemia, largely irreversible and frequently fatal. Introduced in 1948 as Chloromycetin and linked to its first reported case of aplastic anaemia in 1952. Correction (verified 8 Sep): I quoted an incidence of "one in 25,000–40,000 courses" and cannot support it. The figure I can source is a conservative estimate of 1 in 60,000 for fatal aplastic anaemia, from a study of deaths between 1957 and 1961, whose authors added that the true risk was probably greater. Rare either way — but the drug was being given for conditions that needed no antibiotic at all, so the risk fell on people with nothing to gain.

Why it matters Risk is meaningless without indication. A one-in-30,000 death rate is acceptable for typhoid and indefensible for a cold.

Who gained Parke-Davis, for whom it was the single largest-selling product of the period. The over-prescription that created the risk was the commercial case for it — and also the era's medical common sense. Post-war antibiotic optimism held that infection was a solved problem and that broad-spectrum cover was close to costless, so prescribing it for trivial illness was not a lapse but good practice. The commercial interest and the professional conviction pointed the same way, which is why neither had to argue with the other.

Clioquinol · the SMON epidemic

Ciba-Geigy (Entero-Vioform), Japan

Sold as An over-the-counter remedy for diarrhoea and traveller's stomach, sold widely and used long-term.

What happened Subacute myelo-optic neuropathy — an epidemic of peripheral neuropathy, paralysis and blindness in Japan, commonly given as around 10,000 cases, though some sources put diagnoses between 1955 and 1970 far higher (verified 8 Sep; treat the total as contested). It was investigated for years as a virus; circumstantial evidence in 1969–70 pointed to the drug, sales were suspended in 1970, and Tsubaki and colleagues published in the Lancet in 1971. Cases ceased after withdrawal, and many survivors carry the sequelae still.

Why it matters A drug-induced epidemic that looked exactly like an infectious one. Worth holding in mind whenever a cluster is attributed to a pathogen.

Who gained Ciba-Geigy. The Japanese health ministry's decision to permit unrestricted over-the-counter sale is the specific regulatory act that turned a drug risk into a national epidemic — and what the ministry gained by it was a decade of not having been wrong. When SMON cases appeared, the official investigation pursued a viral cause. A new virus would have been nobody's fault; a licensed drug sold without prescription was the ministry's. The wrong hypothesis was not the cheap one to hold, it was the comfortable one, and it was held until the evidence was overwhelming.

Stalinon

Verified 8 Sep Laboratoire Feuillet, France

Sold as An oral capsule containing 15mg of diethyltin diiodide, for boils, acne, staphylococcal skin infection, osteomyelitis and anthrax. The agent that actually did the damage was an impurity — roughly 1.5mg of triethyltin iodide per capsule (verified 8 Sep).

What happened Figures vary by source and both are worth recording: one contemporary account gives 102 dead and 150 blinded or crippled, another 217 poisoned and 111 dead — from interstitial oedema of the cerebral white matter, an effect later reproduced in laboratory animals with triethyltin (verified 8 Sep). The pharmacist, Feuillet, was tried for involuntary homicide by an association of survivors and bereaved families. France's equivalent of the sulfanilamide disaster, and far less remembered outside it.

Who gained A single small laboratory, with no international reach — which is precisely why almost nobody outside France has heard of it. Scale of manufacturer, not scale of harm, determines whether a case enters the canon.

MER/29 (triparanol)

Verified 8 Sep Richardson-Merrell

Sold as The first cholesterol-lowering drug, and a commercial sensation.

What happened About 400,000 people took it, and by 1967 490 cases of cataracts caused by it had been reported (Toole v. Richardson-Merrell, California Court of Appeal, 1967; verified 13 Sep). Hair loss, skin damage and other injuries added to that: Paul Rheingold, who coordinated the plaintiffs' litigation, is cited as putting the total injured at about 5,000. (Correction, 13 Sep: this record said cataracts and blindness in more than 5,000 people, which attached the total-injury figure to cataracts alone. Rheingold's 1968 article could not be retrieved, so his figures are given as others cite them.) The company had falsified the animal data submitted to the FDA: its application reported that four of eight rats had died when in fact all eight had, and gave wholly fictitious body weights, organ weights and blood results for animals that were dead. In January 1961 a patient reported vision problems and another manufacturer reported cataracts in test animals; neither was passed to the FDA. Richardson-Merrell and its Merrell company were indicted for false and fraudulent statements and fined $80,000 between them. About 1,500 civil suits followed, on accounts citing Rheingold. (Also corrected 13 Sep: I had written some 500 suits and about $200m in damages. Neither figure traces to a primary source, and the total paid is reported inconsistently, so none is given here.) (The executives were not convicted at trial. Harold Werner, Evert Van Maanen and William King were indicted alongside the two companies; every defendant pleaded no contest; the fines of $60,000 and $20,000 fell on the Merrell company and Richardson-Merrell; and each executive's sentence was suspended and he was placed on probation. Source: Toole v. Richardson-Merrell, California Court of Appeal, 1967; verified 13 Sep.)

Why it matters Richardson-Merrell is also the company that applied to license thalidomide in the United States, and later marketed Bendectin. All three appear in this register. Track companies as well as drugs.

Who gained Richardson-Merrell — the same company that applied to license thalidomide in the US and later marketed Bendectin. Three of this register's entries belong to one firm. The second beneficiary is the cholesterol hypothesis itself. A drug that lowered cholesterol was self-evidently good, so its harms were weighed against a benefit that was assumed rather than demonstrated — the same assumption that, in the same decades, produced the dietary fat guidance recorded elsewhere in this register on an evidence base its own authors described as absent.

Thalidomide

Chemie Grünenthal (Contergan) · Distillers, UK (Distaval)

Sold as A sedative so safe it was impossible to overdose on, sold over the counter in West Germany and specifically recommended for morning sickness in pregnancy. British advertising showed a toddler at a medicine cabinet with the line that Distaval could be given with complete safety to mother and child alike.

What happened Phocomelia and other severe malformations. Roughly 10,000 affected births worldwide, of whom around half survived infancy; about 2,000 born in the UK, and the Thalidomide Trust today has 414 UK beneficiaries (verified 8 Sep). A wider figure also circulates — some 100,000 pregnancies affected in total, with around 90,000 lost in utero or at birth — but its sourcing is much softer than the 10,000 and it should be attributed, not asserted. McBride in Australia and Lenz in Germany independently identified it in 1961.

Ended Withdrawn November 1961. Grünenthal's criminal trial collapsed into a settlement; the company did not apologise until 2012. Distillers settled after a Sunday Times campaign that itself had to fight an injunction to the European Court of Human Rights. The UK government apologised in 2010.

Not approved in the US because of one FDA reviewer, Frances Kelsey, who declined to approve it on inadequate evidence under sustained pressure from Richardson-Merrell. She belongs in the control set: the system worked, once, because of a person.

Who gained Chemie Grünenthal, and in Britain Distillers Company (Biochemicals) — the drinks conglomerate, whose successor businesses were absorbed into what is now Diageo. Fourteen countries' regulators licensed it. Grünenthal did not apologise for fifty-one years, and the company remains in business. The fourteen regulators gained the thing a permissive standard always gives: speed, and no work. There was no requirement to test in pregnancy because there was no established belief that the placenta transmitted harm — the barrier was thought protective. That assumption cost nothing to hold, spared every agency the expense of demanding evidence, and was not anybody's commercial property.

Putting babies to sleep on their fronts

Verified 8 Sep Dr Benjamin Spock, Baby and Child Care · paediatric consensus

Sold as Advice from the most trusted childcare book of the twentieth century. Two reasons were given, both from armchair reasoning: a baby on its back might choke on vomit, and a back-sleeping baby turns its head to one side and may flatten it. No trial supported either. A precision worth having: the 1955 first British edition actually favoured back sleeping — it was from the 1956 edition until the late 1970s that Spock argued against it (verified 8 Sep). Front sleeping became the majority practice; a 1987 survey in Avon found 60% of infants placed on their fronts.

What happened Prone sleeping is the single largest identified risk factor for sudden infant death. A 2005 analysis by Gilbert and colleagues concluded that evidence sufficient to reverse the advice existed by around 1970, and estimated that the delay cost on the order of 10,000 infant lives in the UK and tens of thousands more across the developed world.

Ended The UK Back to Sleep campaign, 1991. Within two years cot deaths fell nationally by 70% — from around 1,600 a year to under 500, about twelve babies a week (verified 8 Sep).

Why it matters No manufacturer, no profit motive, no fraud. A trusted expert reasoning from mechanism, at scale, for thirty-five years. If the study is about institutions and money it will miss this — and this may be the largest single entry on the page.

Who gained The authority of expert advice as a genre. Spock's royalties are trivial beside the scale; what gained was the proposition that a paediatrician could tell a mother what to do with her own infant and that she should comply. The advice's confidence was its appeal, and the confidence was unearned — he cited no trial because none existed. Note the direction: it advised against the traditional practice, and superseding what mothers' mothers had done was the point of it. Roughly ten thousand British infants, and many times that elsewhere. This is not the counter-example to the study's thesis. It is the thesis with the money taken out.

Primodos

Verified 8 Sep Schering

Sold as A hormone pregnancy test — two tablets; a bleed meant not pregnant. Given to an estimated 1.5 million British women. The dose was many times that of a contraceptive pill.

What happened Alleged association with severe birth defects. A UK warning was issued in 1975 and the product remained on the market until 1978, by which time 1.5 million women had taken it (verified 8 Sep). The Commission on Human Medicines' Expert Working Group concluded in 2017 that the data did not support a causal relationship. The Cumberlege Review in 2020 reached the opposite conclusion — that Primodos caused avoidable harm — and criticised the earlier review's conduct. Two official bodies, three years apart, on the same evidence, disagreeing.

Where it stands The litigation is over: Mrs Justice Yip heard the claim at the Royal Courts of Justice from 2 to 5 May 2023 and dismissed it (verified 8 Sep). My first draft said litigation continues; that is out of date.

Status Genuinely unresolved. Leave it unresolved in the study — a register that quietly upgrades contested cases to proven ones is doing the same thing it accuses others of.

Who gained Schering. But note the second beneficiary: the review process itself, which took forty years and whose slowness protected no one and was never anybody's stated intention.

Practolol (Eraldin)

ICI

Sold as A cardioselective beta-blocker, extensively tested and considered a model of the new post-thalidomide licensing regime.

What happened Oculomucocutaneous syndrome — dry eyes progressing to corneal scarring and blindness, deafness, and sclerosing peritonitis. No animal model predicted it and no trial of feasible length would have found it. Correction (verified 8 Sep): I had "on the order of 7,000 people affected in Britain" and cannot support it. What is on the record: the drug had been taken by more than 250,000 people for a year or more; by September 1976 the Committee on Safety of Medicines had received about 1,000 reports of suspected adverse reactions — 670 skin, 510 eye and 64 sclerosing peritonitis, categories that overlap; and ICI's voluntary compensation scheme had received around 800 claims by March 1977, of which roughly 200 were settled out of court. Reported cases understate true incidence in any voluntary scheme, but 7,000 was my number, not anyone else's.

Why it matters The strongest case against a purely regulatory reading of the register. Practolol was properly tested. Pre-marketing trials cannot detect an effect that takes years to appear in a minority of users; only post-marketing surveillance can, and surveillance was voluntary then and is largely voluntary now.

Who gained ICI commercially, briefly — but this is one of the few entries where the company behaved well: ICI withdrew it and paid compensation without being compelled to. Worth including precisely because it complicates the pattern.

Dalkon Shield

A. H. Robins

Sold as A safer, more effective intrauterine device, advertised with a pregnancy rate the company knew was derived from a study of women also using spermicide. More than 2.5 million women in the United States and abroad, including through US-funded overseas population programmes (verified 8 Sep).

What happened The multifilament withdrawal string wicked bacteria into the uterus. Pelvic inflammatory disease, infertility, uterine perforation and septic mid-trimester abortions. Internal memos showed the wicking problem was known early. On deaths I had "at least 18" and cannot confirm it (verified 8 Sep) — reference sources give "at least four", higher counts circulate, and I have not found a figure I would stand behind. Treat the death toll as unresolved; the injury scale is not in doubt.

Ended Withdrawn 1974; A. H. Robins filed for Chapter 11 in 1985 under more than 327,000 lawsuits and claims, and a $2.3bn trust fund was established (verified 8 Sep) — among the largest product-liability cases in US history.

Who gained A. H. Robins, and — the part that belongs in a study of global bodies — international population-control programmes, which distributed unsterilised Shields in bulk across the developing world after US sales had stopped, on the reasoning that the risk was acceptable relative to the birth rate.

Sodium valproate (Epilim)

Sanofi · UK regulators

Sold as A first-line epilepsy drug, and later a mood stabiliser, prescribed to women of childbearing age for forty-five years.

What happened Fetal valproate syndrome — major malformation in roughly 10% of exposed pregnancies and neurodevelopmental disorder in roughly 40%. Around 20,000 UK children affected — the figure the Cumberlege review used (verified 8 Sep). Teratogenicity was evident in animals before UK launch and in humans through the 1980s. Warnings existed but were not reliably reaching women; surveys found large proportions of patients unaware.

Ended Pregnancy Prevention Programme, 2018. The Cumberlege Review (2020) covered it alongside Primodos and mesh, and its title is the finding: First Do No Harm.

Who gained Sanofi. But the structural beneficiary is the warning itself — a label change discharges liability while changing nothing at the point of prescription. The register's most repeatable evasion: the information existed, the transmission did not, and the paperwork recorded a warning given.

Contaminated blood products

UK Department of Health · NHS · US commercial plasma industry

Sold as Factor VIII concentrate, a transformative treatment for haemophilia. Britain imported it from the United States, where plasma was bought from paid donors including prisoners and drug users, and pooled from thousands of donations per batch.

What happened Around 30,000 people in the UK infected with hepatitis C, HIV or both; over 3,000 dead and still rising. The risks of paid-donor pooled plasma were understood, and the UK's failure to reach self-sufficiency in domestic blood products was a known and documented policy shortfall. Patients were treated without being told, some were tested without consent, and records were destroyed.

Ended The Infected Blood Inquiry under Sir Brian Langstaff reported on 20 May 2024 (verified 8 Sep). Its findings, in its own words: the disaster could largely, though not entirely, have been avoided and it should have been; patients were failed not once but repeatedly and knowingly exposed to unacceptable risks; and the government and the NHS were found to have hidden the truth from them.

Who gained The US commercial plasma industry, which had a paid-donor supply chain it declined to reform, and the UK government, which avoided the capital cost of domestic self-sufficiency. This is the register's clearest case of a state body being the beneficiary of its own bad information.

Cadaver-derived human growth hormone

Verified 8 Sep UK Medical Research Council · US National Pituitary Agency

Sold as The only treatment for growth hormone deficiency, extracted from pituitary glands harvested at autopsy and pooled from thousands of cadavers per batch.

What happened Transmission of Creutzfeldt–Jakob disease, with incubation periods of decades. At least 1,848 UK patients were treated under the national scheme; 80 UK cases of iatrogenic CJD have been recorded, 78 of them fatal, out of more than 200 worldwide (verified 8 Sep). Cases were still appearing thirty years on.

The detail that changes the record The hormone was extracted from pituitary glands taken from close to a million corpses. And the product had never been licensed by the Committee on Safety of Medicines — the doctors prescribing it within the NHS did not know that. This was not a licensed medicine that failed; it was an unlicensed one that nobody had to license because the state was making it itself.

Ended The first UK death was reported in April 1985 and use was suspended the following month, with 865 people then in treatment; recombinant hormone replaced it.

Who gained The programme's own continuation. No company, but not no interest: pooling was the only way to make enough, and enough was the measure by which these programmes judged their success. An operation that rates itself by how many children it treats has a structural reason not to look hard at a risk whose acknowledgement would reduce the number. The gain was the ability to keep going at scale.

Benzodiazepines as non-addictive

Verified 8 Sep Roche (Librium, Valium) · Wyeth (Ativan)

Sold as The safe replacement for barbiturates — non-addictive minor tranquillisers for anxiety and insomnia, prescribed overwhelmingly to women. Valium was the best-selling drug in the Western world through the 1970s.

What happened Physical dependence at therapeutic doses, with a withdrawal syndrome that can last months or years. Britain reached an estimated 1.5 million long-term users — a figure quoted by the All-Party Parliamentary Group in 2007–08, though the group itself noted hard evidence was lacking, and a GP-practice extrapolation gives over a million (verified 8 Sep; treat 1.5m as an estimate, not a count). The Committee on Safety of Medicines did not recommend limiting treatment to two to four weeks until 1988. Dependence at therapeutic doses was reported from the early 1960s and was widely recognised by the late 1970s, so the delay from clear recognition to restriction is around a decade rather than the quarter-century I first claimed.

Why it matters Third iteration of the register's core pattern, and the one that should be hardest to excuse: opium, then morphine, then heroin, then barbiturates, then benzodiazepines, then OxyContin. Each was marketed as the safe answer to its predecessor. The claim was novel each time to the people making it.

Who gained Roche, on a scale that funded much of its modern research base. And general practice, which acquired a prescribable answer to distress it had no other tool for — the same demand-side mechanism as DES.

Benoxaprofen (Opren)

Verified 8 Sep Eli Lilly

Sold as An anti-arthritic promoted as disease-modifying, marketed aggressively to the elderly, who were the least studied group in its trials.

What happened More than 500,000 people in Britain took it between 1980 and 1982. The Committee on Safety of Medicines received over 3,500 adverse reaction reports and 61 deaths, most of them elderly, from photosensitivity, onycholysis and fatal hepatorenal failure; the drug was cleared slowly in the old and the dosing did not account for it. Kenneth Clarke suspended sales with immediate effect on 4 August 1982 (verified 8 Sep).

Two facts that make it worse Lilly's own published data showed photosensitivity in about 10% of patients and nail separation in about 10%, and it was approved anyway, on animal toxicology that was inadequate. And when independent researchers established that the drug accumulated in elderly patients, Lilly attempted to prevent the study being published.

Who gained Eli Lilly. The instructive detail for the study: the harm fell on a group deliberately excluded from the trials and then deliberately targeted by the marketing. That gap is a design choice, not an accident of evidence.

Antiarrhythmic drugs after heart attack

Cardiology consensus · CAST trial

Sold as Irregular heartbeats after a heart attack predict sudden death; drugs suppress the irregular beats; therefore the drugs prevent death. Encainide and flecainide were given to large numbers of post-infarction patients on that reasoning.

What happened The CAST trial was stopped early in 1989 because the treated group was dying at more than twice the rate of placebo. Thomas Moore's book Deadly Medicine claimed more than 50,000 Americans died from antiarrhythmic drugs over a two-year period, a figure repeated ever since. That figure is contested and should not be used as settled (verified 8 Sep). A 1997 analysis in the American Journal of Cardiology, explicitly testing Moore's claim, found national coronary artery disease mortality fell linearly through the 1980s with no deviation during the introduction, rise and fall of class IC drug sales — which is hard to reconcile with a death toll of that size. The trial result itself is not in doubt: CAST was stopped early because the treated group was dying at more than twice the placebo rate. The national toll is an extrapolation, and the one in circulation has a published rebuttal that is almost never cited alongside it.

Why it matters The definitive case against surrogate endpoints. The drugs did exactly what they were designed to do to the measurement, and killed people. Every claim of the form it improves the marker, therefore it improves the outcome should be read against this entry.

Who gained The manufacturers, and — more honestly — cardiology's own theoretical model, which was elegant, mechanistically satisfying, and wrong. Nobody had to be corrupt for this to happen.

The low-fat dietary guidelines

Verified 8 Sep US Senate Select Committee (McGovern, 1977) · UK COMA/NACNE (1983–84) · WHO

Sold as National dietary advice for entire populations: cut total and saturated fat to prevent heart disease. Applied to roughly 220 million Americans and 56 million Britons by 1983 (verified 8 Sep).

The evidence base, in the institutions' own words The American Medical Association warned in 1977 that the evidence for assuming that benefits are to be derived from the adoption of such universal dietary goals is not conclusive, and there is potential for harmful effects. Challenged on the science, Senator McGovern replied that Senators do not have the luxury that the research scientist does of waiting until every last shred of evidence is in. And Britain's COMA committee wrote in its own 1984 report — the document setting the target of cutting fat from 42% to 35% of energy, saturated fat to no more than 15% — that there has been no controlled clinical trial of the effect of decreasing dietary intake of saturated fatty acids on the incidence of coronary heart disease nor is it likely that such a trial will be undertaken.

Why that quotation is the record This is not a critic's reconstruction. The committee issuing population-wide advice stated in the same document that no trial supported it, and issued it anyway. Nothing else in this register is quite so explicit.

What happened Fat calories were replaced largely by refined carbohydrate and sugar, and the food industry reformulated en masse toward low-fat products with added sugar. Obesity and type 2 diabetes rose steeply across the same period. Causation is contested and should stay contested — but the guidance was issued as settled when it was not, which is the defensible claim.

Who gained The sugar industry, directly and demonstrably: Kearns and colleagues showed in 2016 that the Sugar Research Foundation paid Harvard researchers in 1967 for a review that steered blame toward fat, and the funding was not disclosed. Also the processed-food industry, which gained an entire low-fat product category, and the vegetable-oil industry.

Trans fats — the recommended substitute

Verified 8 Sep Public health bodies · margarine manufacturers

Sold as Margarine and hydrogenated vegetable shortening as the healthy alternative to butter and lard. Actively promoted by heart-health campaigns; in the US the Center for Science in the Public Interest and the National Heart Savers Association campaigned hard against saturated fats and endorsed trans fats as the healthier alternative — CSPI's best-known success being the campaign that moved McDonald's from beef tallow to vegetable oil for frying (verified 8 Sep).

What happened Industrial trans fats raise LDL and lower HDL — worse for cardiovascular risk than the saturated fat they replaced. Harvard School of Public Health estimates put the American toll at 72,000 to 228,000 heart attacks a year including roughly 50,000 fatal, with a separate Harvard estimate of about 30,000 premature coronary deaths annually; the WHO attributes more than 500,000 cardiovascular deaths worldwide each year to industrial trans fat and launched its REPLACE elimination programme in 2018 (verified 8 Sep).

The qualifier that belongs here The regulators were not simply ignoring evidence. FDA-commissioned reviews in 1976 and 1985 both concluded that partially hydrogenated oils and trans fat were not harmful — so the official position had a documented evidence base at the time, even though preliminary work in those same decades pointed the other way. That makes this a slower and more ordinary failure than the fraud cases in this register, and a better argument for surveillance than for blame.

Why it matters The register's cleanest example of public health advice causing the harm it was issued to prevent, and it took four decades to reverse. If you cite one nutrition case, cite this rather than eggs — it is stronger and much less contested.

Who gained Unilever, Procter & Gamble (Crisco) and the hydrogenation industry. And the advocacy organisations, who gained a clear public campaign with a simple message — and were sincere.

Fen-phen

Verified 8 Sep Wyeth-Ayerst

Sold as A weight-loss combination that became a mass-market phenomenon: more than 18 million US prescriptions for fenfluramine and phentermine in 1996 alone (verified 8 Sep). The combination itself was never approved — it was an off-label pairing of two separately approved drugs. (Correction, 13 Sep: I had said it rested on a single small unblinded study. The study that launched it was Michael Weintraub's long-term weight control study, published in 1992 in Clinical Pharmacology and Therapeutics as a seven-part series. It enrolled 121 participants and was double-blind and placebo-controlled for its first 34 weeks, then open-label from weeks 34 to 104, and ran to week 210. So it was one study, but it was not unblinded.)

What happened Mayo Clinic researchers identified valvular heart disease in 24 women with no cardiac history, reported publicly on 8 July 1997 and published in the NEJM that August; primary pulmonary hypertension was also implicated. Fenfluramine and dexfenfluramine were voluntarily withdrawn on 15 September 1997 (verified 8 Sep).

The scale of the liability, which I understated I wrote that settlements "ran into billions". Wyeth's initial settlement in 1999 was $3.75bn; by 2002 its reserves exceeded $14bn; and by 2006 the company had set aside $22bn to pay some 600,000 claimants. That is among the largest product-liability provisions in pharmaceutical history and belongs in the record as such.

Who gained Wyeth, and a very large private weight-loss clinic sector that grew up specifically to prescribe it — an entire service industry built on an off-label combination. And the idea that made both possible: obesity reconceived as a disease requiring pharmacological correction rather than a condition. That reframing was contested at the time, was advanced by people who believed it on the evidence as they read it, and long outlived the drug — it is a live professional and commercial position today, which is why this record is not closed.

Hormone replacement therapy — reversed twice

Wyeth · Women's Health Initiative · NICE

Sold as First, from Robert Wilson's Feminine Forever (1966, and funded by the manufacturer), as protection against ageing and heart disease for essentially all menopausal women — a claim resting on observational data.

What happened — first reversal The Women's Health Initiative arm was halted in 2002 reporting increased breast cancer and cardiovascular risk. Prescribing collapsed by roughly half worldwide within a year.

What happened — second reversal The trial population's average age was 63, over a decade past menopause. Re-analysis supports a timing hypothesis: for women starting in their fifties the risk profile is substantially better and mortality may be reduced. Sarrel and colleagues estimated in the American Journal of Public Health in 2013 that over the ten years from 2002, oestrogen avoidance caused between 18,601 and 91,610 excess deaths among hysterectomised women aged 50–59 (verified 8 Sep). Note the width of that range, and note that WHI investigators published a rebuttal in the same journal disputing the analysis — it is a serious estimate, not a settled number, and citing only its upper bound would be the exact failure this register documents. UK guidance has since moved back toward treatment.

Why it matters The most important entry in the register for your bias question, because the harm ran in both directions from the same institution. Overclaiming hurt women; the correction hurt women; and at each point the official position was stated with more confidence than the evidence carried. A study that only counts harm from the original claim will get this case exactly half right.

Who gained Wyeth, on the way up. On the way down the beneficiary is less visible but not absent: the specialty's own conviction. HRT had become constitutive of a model of menopause as a deficiency state, and of gynaecology's role in correcting it. That is why the reversal was fought rather than absorbed, why the WHI investigators were attacked rather than answered, and why the partial rehabilitation for younger women took another decade to be heard. Harm does not require a profit. It has not yet been found to require no beneficiary at all.

The swine flu vaccination programme

US CDC · President Ford

Sold as An imminent pandemic on the scale of 1918, inferred from a single outbreak at Fort Dix. Around 45 million Americans were vaccinated in ten weeks.

What happened The pandemic never came — the virus did not spread beyond the base. About 45 million Americans were vaccinated and the vaccine was associated with roughly 450 cases of Guillain–Barré syndrome; the programme was suspended on 16 December 1976 and the CDC director resigned (verified 8 Sep). Death counts attributed to the vaccine vary between sources and I have not been able to confirm a reliable figure — do not quote one.

Why it matters A precautionary action taken on a genuine worst case, where the precaution was the only thing that caused harm. It is also the origin of a durable American vaccine-scepticism, which means the cost did not end in 1976 — the reputational damage was itself a long-run harm.

Who gained Manufacturers, who secured a federal indemnity against liability as the price of participating — a precedent that outlived the programme. Politically, an administration in an election year gained a demonstration of decisiveness.

OxyContin

Verified 8 Sep Purdue Pharma · the Sackler family

Sold as A long-acting opioid whose slow release made addiction rare — Purdue's sales training put the risk less than one per cent. That figure traces to a letter of about 101 words — variously described as four or five sentences — to the New England Journal of Medicine in 1980 by Porter and Jick, reporting that among some 11,880 hospitalised inpatients given opioids, only a handful developed documented addiction (verified 8 Sep). Researchers later identified 608 citations of it, rising sharply after OxyContin's launch, of which 439 — 72% — cited it as evidence that addiction is rare in patients treated with opioids. It was not a study of chronic outpatient use and never claimed to be.

What happened Purdue pleaded guilty to misbranding in 2007 — admitting it had falsely marketed OxyContin as less addictive, less open to abuse and less likely to cause withdrawal than rival painkillers — and again in 2020 to a three-count felony information, with damages of $8.34bn (verified 8 Sep). US overdose deaths involving opioids exceed 500,000 since 1999. The epidemic moved from prescription pills to heroin to illicit fentanyl, so the later deaths are not Purdue's product — but the pathway was.

Why it matters The professional infrastructure is the story, not the marketing. Purdue funded pain societies, continuing medical education, patient advocacy groups, and the campaign to make pain the fifth vital sign. The Joint Commission's pain standards and the practice of scoring hospitals on pain-control satisfaction created institutional pressure to prescribe. The message reached doctors as professional consensus, not as an advertisement — which is exactly why it worked.

Who gained The Sacklers, to the order of $10bn withdrawn from the company. But the beneficiary list your study wants is longer: pain societies, medical schools, accreditation bodies and patient advocacy groups funded by the manufacturer, plus McKinsey, which advised on boosting sales and later paid nearly $600m to settle. Also the museums and universities that took the philanthropy and lent their names back.

Vioxx (rofecoxib)

Merck

Sold as An anti-inflammatory as effective as older ones without the stomach bleeding. Around 80 million people took it worldwide.

What happened The VIGOR trial in 2000 showed a five-fold excess of heart attacks; it was presented as the comparator being protective rather than Vioxx being harmful, and three cardiovascular events were omitted from the NEJM paper, which later issued an Expression of Concern. Merck withdrew it in 2004. The FDA's own David Graham estimated 88,000 to 139,000 excess US heart attacks and strokes, of which he put roughly 40% — about 55,000 — as fatal (verified 8 Sep), and testified to the Senate that the agency was incapable of protecting America.

Who gained Merck, at $2.5bn a year. Also the journals: NEJM sold 900,000 reprints of the VIGOR paper, mostly to Merck as sales aids, for a reported $697,000 to $836,000 (verified 8 Sep). Worse than I first recorded: the journal's Expression of Concern did not appear until December 2005 — five years after publication — and its executive editor learned of a July 2000 internal Merck memo showing the authors knew of the heart attacks before the paper appeared only when he was deposed in the product-liability litigation. Add academic authors on payment, and ghostwriting firms — internal documents showed manuscripts drafted by Merck with academics recruited as named authors afterwards.

Troglitazone (Rezulin)

Verified 8 Sep Warner-Lambert · FDA

Sold as A breakthrough diabetes drug, given accelerated review.

What happened 94 cases of liver failure were reported — 89 of them acute, of which only 11 recovered without a transplant (verified 8 Sep). Britain withdrew it in December 1997; the FDA did not ask for withdrawal until 21 March 2000 — over two years later.

The reviewer story, which is worse than I first wrote The FDA's chief medical reviewer for the drug, Dr John Gueriguian, concluded in autumn 1996 that it was not safe to market. He was removed from the review and replaced, and his successor recommended approval. That successor, Dr Robert Misbin, later became the internal whistleblower on the drug's dangers — and when he wrote to Congress urging withdrawal, the FDA opened an internal investigation into him, which Congress intervened to stop as retaliation against a whistleblower.

Why it matters A natural experiment. Same drug, same data, two regulators, a two-year divergence and a measurable body count in the gap. Use this one whenever someone argues that regulators simply follow the evidence.

Who gained Warner-Lambert, roughly $2bn in US sales during the period Britain had already banned it. And the FDA's own commitment to accelerated approval, which had been presented as a policy success and which the drug's withdrawal would embarrass. The reviewer who objected was removed from the review — an institutional act with no commercial motive whatever, protecting a reputation rather than a revenue.

Paroxetine in adolescents — Study 329

SmithKline Beecham / GSK

Sold as Published in 2001 as showing paroxetine generally well tolerated and effective for adolescent depression. The paper was ghostwritten.

What happened The trial missed both primary endpoints. Suicidal thoughts and self-harm were substantially more common on the drug and were recorded under the term emotional lability. An independent reanalysis of the underlying data, published in the BMJ in 2015 as the first trial restored under the RIAT initiative, reversed the conclusions entirely (verified 8 Sep): neither paroxetine nor high-dose imipramine beat placebo, suicide attempts were significantly more numerous than the original paper reported, other serious adverse events had gone unreported, and the reanalysis concluded paroxetine was ineffective and unsafe in this study. GSK paid $3bn in 2012 — then the largest healthcare fraud settlement in US history — covering this, Wellbutrin and Avandia.

Why it matters The single best demonstration that peer review does not detect this. The paper passed review, sat in the literature for fourteen years, and was only overturned when outsiders obtained the underlying data. Peer review checks whether a paper is plausible, not whether it is true.

Who gained GSK; the ghostwriting agency; the academic authors who took credit; and the journal. The $3bn fine was a fraction of the revenue.

The CDC's MMR timing study and the Thompson statement

Verified 13 Sep US Centers for Disease Control and Prevention · DeStefano, Bhasin, Thompson, Yeargin-Allsopp, Boyle

Sold as Reassurance on the timing of MMR. The CDC's case-control study in metropolitan Atlanta (Pediatrics, 2004) compared 624 children with autism with 1,824 matched controls. It reported that similar proportions of both groups had been vaccinated by 18 months and by 24 months, before autism is usually recognised. It has been cited ever since as evidence against the MMR hypothesis.

What the paper itself shows For vaccination before 36 months, the whole sample gave an odds ratio of 1.49 (95% CI 1.04–2.14), which is statistically significant. In the subsample with Georgia birth certificates it was 1.23 (0.64–2.36), which is not. The authors put the association down to catch-up vaccination: children already identified with developmental delays needed immunisation to enter early-intervention programmes, and the effect was strongest among 3- to 5-year-olds.

What a co-author said On 27 August 2014 William Thompson, a CDC senior scientist and one of the authors, issued a statement through his lawyer: I regret that my coauthors and I omitted statistically significant information in our 2004 article. The omitted data suggested that African American males who received the MMR vaccine before age 36 months were at increased risk for autism, and decisions were made regarding which findings to report after the data were collected, and I believe that the final study protocol was not followed. In July 2015 Representative Bill Posey read into the Congressional Record Thompson's account that four co-authors met with a big garbage can to discard study documents. Thompson said he kept hard copies and computer files because he believed destroying them was illegal. The CDC's position is that restricting the analysis to children with Georgia birth certificates gave more complete information on race and other characteristics (verified 13 Sep against the statement, the Congressional Record and the 2004 abstract).

What the same statement also says Vaccines have saved and continue to save countless lives. Thompson would never suggest that any parent avoid vaccinating children of any race. He wrote that he had faced no pressure or retaliation, and had received a performance award since the story broke. He offered to help any independent reanalysis. Campaigning accounts quote the first paragraph and seldom these.

The reanalyses Brian Hooker, who had recorded his conversations with Thompson without Thompson knowing, published a reanalysis in Translational Neurodegeneration in August 2014. It was retracted that October over undisclosed conflicts of interest and concerns about its methods. As far as this study can find, no independent, pre-registered reanalysis of the original data has been published, although Thompson says he kept the data and offered to help. The 2016 film Vaxxed, directed by Andrew Wakefield and produced by Del Bigtree, was built around the affair.

Scored as conduct The disputed finding is one subgroup, one vaccine and one age window, and it is not significant in the subsample the authors preferred. On its own it cannot show that MMR causes autism, and the large cohort studies do not support that. What stands is conduct. An author of a CDC study says reportable results were dropped after the data were seen, and that documents were discarded. The agency's answer has been an explanation, not an independent reanalysis. A company in the same position would be expected to release the data for independent analysis, and the agency should be held to the same expectation.

Who gained The CDC's immunisation programme, which gained a published study on MMR timing at the height of the MMR controversy and whose credibility rested on the answer. The authors, whose reputations rest on the paper as published. And, on the other side, campaigners and filmmakers, for whom "the CDC whistleblower" has been worth more as an unresolved grievance than a reanalysis that might settle it either way.

Rosiglitazone (Avandia)

Verified 8 Sep GSK

Sold as A blockbuster diabetes drug, approved on its effect on blood glucose — a surrogate endpoint again — rather than on cardiovascular outcomes.

What happened Nissen and Wolski's 2007 meta-analysis of 42 trials found a significant excess of myocardial infarction — odds ratio 1.43 (95% CI 1.03–1.98), about a 43% increase (verified 8 Sep). A two-year US Senate Finance Committee investigation concluded GSK had known of the signal for years and had pressed doctors to withdraw warnings; GSK disputed the report. The EMA suspended the marketing authorisations across the EU in 2010; the FDA kept the drug available under a stringent restricted-access programme.

And then it went the other way In 2013 the FDA lifted the prescribing restrictions, and later removed the remaining ones, on the basis of a re-adjudication of the RECORD trial that found no elevated cardiovascular risk. The final scientific verdict is genuinely muddier than the scandal narrative, and any use of this case has to carry both halves.

Who gained GSK, at roughly $3bn a year at peak. Also the FDA advisory committee system, whose members' industry ties in this case were substantial and disclosed only under pressure.

Transvaginal mesh

Verified 8 Sep Johnson & Johnson (Ethicon) · Boston Scientific · regulators

Sold as A quick, minimally invasive fix for prolapse and stress incontinence. Cleared in the US through the 510(k) pathway — clearance by declared equivalence to an existing device, requiring no clinical trial. A systematic review of the FDA database found no clinical trial evidence at all for 61 transvaginal mesh devices at the point of approval, with randomised trials appearing a median of five years after the device was on the market; one manufacturer relied on a 90-day study in rats (verified 8 Sep).

What happened Chronic pain, mesh erosion through tissue, autoimmune symptoms, loss of sexual function, and removals that are often impossible because the mesh integrates with tissue. Between April 2008 and March 2017 there were 100,516 tape insertions for stress incontinence and 27,016 mesh procedures for prolapse in England; a 2017 Guardian investigation found roughly one in fifteen women who had a mesh implant later needed surgery to remove it (verified 8 Sep). The NHS paused use in July 2018 on Cumberlege's recommendation; her review reported in 2020 that women had been dismissed and disbelieved for years.

Why it matters The devices gap. Everything the register earlier attributes to lax drug regulation applies with much greater force to implants, which in many jurisdictions still reach patients on an equivalence declaration. If the study covers only medicines it will miss the larger current exposure.

Who gained The manufacturers, and the surgeons and private clinics for whom a 30-minute procedure replaced a complex operation. The 510(k) pathway itself is the institutional beneficiary — it is fast, cheap, and its speed is counted as a regulatory success metric.

Metal-on-metal hip replacements

Verified 8 Sep DePuy (Johnson & Johnson) · Smith & Nephew

Sold as Longer-lasting joints for younger, more active patients. About 93,000 DePuy ASR devices implanted worldwide (verified 8 Sep).

What happened — with a correction to my own figure (verified 8 Sep) Metal debris caused local tissue destruction and raised blood cobalt and chromium. The registry data that triggered the recall showed revision rates of 12–13% at five years against an expected 5–6% — roughly double, and unacceptable for a joint meant to last decades. I had written "roughly 40% at six years", which is not supported; higher figures appear in some later single-centre series but they are not the registry finding that caused the withdrawal. DePuy recalled the ASR system on 24 August 2010.

Who gained The manufacturers. Note also the mechanism that caught it: the National Joint Registry. Britain detected the failure rate because the registry counted outcomes systematically and independently of the surgeons implanting the devices. Registries are the single most effective countermeasure in this entire register and are the cheapest thing on the page. The non-commercial beneficiary was surgical enthusiasm itself — the appeal of offering younger patients a bearing that might outlast them, and the standing that attaches to adopting a new technique early. Note what that implies about the countermeasure: the registry is a check on judgement, not on greed, and its independence from the implanting surgeon is the whole of its value.

PIP breast implants

Verified 8 Sep Poly Implant Prothèse · TÜV Rheinland

Sold as CE-marked medical-grade silicone implants. Around 300,000 women received implants containing the substituted silicone, with wider estimates of more than 400,000 affected across over 65 countries, including about 47,000 in Britain (verified 8 Sep).

What happened The company had substituted industrial-grade silicone intended for mattresses, and concealed it from inspectors by scheduling around announced audits. The implants were roughly twice as likely to rupture as other brands, and about a quarter of those later removed were found to have ruptured (verified 8 Sep). PIP's founder, Jean-Claude Mas, was prosecuted and convicted; a French court subsequently ruled that the certifier, TÜV Rheinland, must compensate thousands of women.

Why it matters European device approval is performed by private certification companies paid by the manufacturer they certify. A journalistic sting later obtained a CE mark for a hip implant using a device known to have been withdrawn for harming patients. The conflict is structural, not incidental.

Who gained PIP, on the price difference. And the notified-body system, which competes for manufacturers' business — the customer is the regulated party.

Tamiflu stockpiling

Verified 8 Sep Roche · WHO · UK Department of Health

Sold as Reduces complications, hospitalisation and transmission in pandemic influenza. Governments stockpiled it on that basis; the UK spent almost £424m on a stockpile of some 40 million doses (verified 8 Sep). WHO listed it as essential.

What happened Roche resisted releasing the full clinical study reports for four years — announcing disclosure in December 2009, supplying redacted reports in September 2010, and finally delivering some 160,000 pages covering 83 studies in 2013. The Cochrane review published in 2014 on that full evidence found Tamiflu shortens symptoms by about half a day, with no good evidence that it reduces hospital admission or complications (verified 8 Sep). Some of the earlier favourable literature had authors paid by the manufacturer.

Why it matters This is WHO in the beneficiary column, which your study will want. It is also the case that produced the AllTrials campaign and the EMA's data-transparency rules — a genuine institutional reform driven from outside.

Who gained Roche, in the billions, from public money. Governments gained a visible pandemic-preparedness action that could be announced. WHO gained a concrete recommendation to make; some contributors to its guidance had industry ties that were not fully declared at the time.

Pandemrix and narcolepsy

GSK · European regulators

Sold as Swine flu vaccine, authorised rapidly under pandemic procedures with a novel adjuvant and a truncated evidence package, and given to millions of European children.

What happened A clear excess of narcolepsy with cataplexy in children and adolescents — an incurable, life-long condition — first detected in Finland and Sweden, at rates that differ by country and by method (verified 8 Sep): about 1 case per 16,000 vaccinated aged 4–19 in Finland, roughly 1 in 52,000–57,500 doses in England, an almost seven-fold increase in vaccinated Swedish children, and a pooled estimate of about 1 per 18,400 doses. Its use in the under-20s has been restricted across Europe since July 2011. The UK has paid compensation in a number of cases, some only after litigation.

Why it matters Handle this one carefully and exactly, because it is where the register's two failure modes meet: a real, confirmed, serious vaccine injury, of a kind that is genuinely rare — and the fact that it is real is precisely why it should be stated accurately rather than either buried or generalised.

Who gained GSK, which sold under advance-purchase agreements with liability indemnities negotiated with governments — the taxpayer, not the manufacturer, carried the injury cost. That indemnity structure is the durable finding here and it recurred in 2020.

Advice to avoid peanuts in pregnancy and infancy

Verified 8 Sep UK Committee on Toxicity (1998) · American Academy of Pediatrics (2000)

Sold as Precautionary official guidance: mothers at risk should avoid peanuts in pregnancy and breastfeeding, and children should not be given peanut before age three. Issued explicitly on precautionary reasoning, with no trial behind it.

What happened Peanut allergy prevalence rose sharply over the following decade. The LEAP trial — 640 high-risk British infants aged 4 to 11 months, randomised either to eat peanut protein at least three times a week or to avoid it entirely — found that early introduction cut peanut allergy at age five by more than 80% (Du Toit et al., NEJM, 2015; verified 8 Sep). A follow-up, LEAP-ON, showed the protection persisted after a year of avoidance. The advice had been the opposite of correct, and had been given to a generation of parents.

Ended Guidance withdrawn 2008–09; reversed to recommend early introduction from 2015.

Why it matters The most important recent entry for the study's argument, because it is pure precaution with no commercial actor anywhere in it. It shows that official advice can cause large-scale harm through caution alone. Any theory that locates the danger in industry capture cannot explain this case, and it is the strongest single answer to anyone who says the failures are all about money.

Who gained The committee. No product, no manufacturer, no funding — and the asymmetry that governs every advisory body: over-warning is career-safe, under-warning is career-ending. Advise avoidance and be wrong, and the result is a decade of avoidable allergy with no name attached to it. Advise exposure and be wrong, and it is a dead child with your signature on the guidance. Precaution is not the absence of interest; it is the option that protects the adviser — and it was chosen with no trial supporting it in either direction.

Essure permanent contraception

Verified 8 Sep Conceptus · Bayer

Sold as Permanent sterilisation without surgery — nickel-titanium coils placed in the fallopian tubes in an outpatient appointment.

What happened Chronic pelvic pain, device expulsion, uterine and tubal perforation, bleeding requiring hysterectomy, unintended pregnancy and miscarriage. The FDA received 26,272 adverse event reports between 2002 and January 2018 (verified 8 Sep). Bayer announced in July 2018 that it would end US sales by the year's end, and in 2020 settled more than 90% of claims for $1.6bn.

Who gained Bayer, which acquired Conceptus in 2013 for about $1.1bn, $31 a share, announced 29 April 2013 (verified 13 Sep against Bayer's tender filings). Also health systems, for which an outpatient procedure is far cheaper than surgical sterilisation — the cost saving was part of the case for adoption.

The serotonin theory of depression

Verified 8 Sep SSRI manufacturers · prescribing guidance · public information

Sold as Depression as a chemical imbalance — a serotonin deficiency corrected by SSRIs. Communicated to patients as established biology, in advertising and in clinical explanation, for three decades.

What happened Moncrieff and colleagues' 2022 umbrella review in Molecular Psychiatry reported only very limited evidence for a direct role of serotonin levels or transport in major depression, and no consistent evidence of a relationship (verified 8 Sep). It does not show the drugs do not work; the argument over effect size continues.

The rebuttal I omitted, and it is substantial The paper drew formal published replies, including from King's College London. Critics argued it departed from umbrella-review method — summarising existing reviews rather than extracting and analysing data — that it read receptor-binding findings without accounting for serotonin receptor pharmacology, and, most pointedly, that the serotonin theory was never advanced by researchers as a claim that depression is caused by a deficiency of one chemical; that formulation belongs to marketing and to public explanation rather than to the literature. If so, the failure recorded here is of science communication rather than of science — which is exactly the distinction Pillar III's tier rule exists to enforce, and I did not apply it to my own entry.

Why it matters The harm is subtle and worth articulating precisely: being told you have a permanent chemical deficiency changes what you believe about your own prognosis and your ability to stop. Withdrawal effects were also minimised for years, and UK guidance did not properly acknowledge their severity until 2019.

Who gained Manufacturers, obviously. But also psychiatry as a discipline, which gained a biological account that positioned it alongside the rest of medicine, and — genuinely — patients and campaigners, for whom a biochemical explanation reduced stigma and blame. A good-faith beneficiary can still be a beneficiary.

What happened after eugenics was discredited

Verified 13 Oct Sweden · Japan · Peru · US state prisons · European gender-recognition law

The claim That eugenics ended in 1945, discredited by what Germany had done with it. The programmes say otherwise: the word was dropped and the practice continued, under welfare, public health and population policy.

Sweden A government commission found that about 63,000 people, mostly women, were sterilised between 1935 and 1996, around half of them under compulsion. Parliament voted compensation of 175,000 kronor each in 1999.

Japan The Eugenic Protection Law of 1948 remained in force until 1996. The government acknowledges about 16,500 forced sterilisations; plaintiffs' lawyers put it nearer 25,000. On 3 July 2024 the Supreme Court held the law unconstitutional, rejecting the twenty-year limitation period the state had relied on, and the Prime Minister apologised. A compensation law took effect in January 2025, paying ¥15m to each surviving victim.

Peru Under Alberto Fujimori's family-planning programme of 1996–2000, more than 280,000 people were sterilised, overwhelmingly poor, rural and indigenous women; health ministry figures cited by the Center for Reproductive Rights are higher still. Prosecutions have been repeatedly stalled on jurisdiction and limitation grounds. Fujimori died in September 2024.

The United States, after the laws were repealed California's state auditor found in June 2014 that 144 women were sterilised in state prisons between 2005 and 2013, and that consent was not lawfully obtained in 39 cases: physicians failed to certify competence in 27, and required waiting periods appear to have been breached in 18. Records for seven more had been destroyed in routine purging. California voted reparations in 2021, at $15,000 or more; by 2023 about 101 people had been approved.

Europe, by law, until 2017 Several European states required sterilisation before they would recognise a person's gender. Sweden required it from 1972 until 2013 and voted compensation of 225,000 kronor in 2018. In A.P., Garçon and Nicot v France (6 April 2017) the European Court of Human Rights held, by six votes to one, that conditioning legal recognition on an operation carrying a high probability of sterility violated Article 8.

The pattern Each programme was run by a health system, authorised by law, and aimed at people with the least standing to refuse: the poor, the disabled, the indigenous, prisoners, and trans people. None of them needed the word. Each ended only when the people it was done to were believed, which took between twenty and sixty years, and each ended in an apology and a payment rather than a prosecution.

Who gained The budget and the professional judgement. States gained cheaper welfare, prisons and health systems, and administrators gained the authority to decide who should not have children. In Peru the programme produced targets met and international population goals satisfied. In the prisons, a doctor's judgement that a woman would be back and pregnant again substituted for her consent. The gain is the same one the 1920s programmes delivered — authority over other people's reproduction, exercised by professionals and paid for in someone else's dignity.

Routine antibiotic over-prescription

Verified 8 Sep Prescribing culture · agriculture · WHO

Sold as Antibiotics for viral illness, for reassurance, and as routine growth promoters in livestock — the agricultural use being by volume the larger.

What happened Antimicrobial resistance. The GRAM study published in The Lancet estimated that in 2019 resistance was directly responsible for 1.27 million deaths and associated with 4.95 million — more than HIV/AIDS or malaria — across 23 pathogens and 88 pathogen-drug combinations in 204 countries (verified 8 Sep). Fleming warned of it in his Nobel lecture of 11 December 1945: The time may come when penicillin can be bought by anyone in the shops. Then there is the danger that the ignorant man may easily underdose himself and by exposing his microbes to non-lethal quantities of the drug make them resistant. (Verified 13 Sep. NobelPrize.org refused automated access, so the check was made against the lecture as quoted, with citation, in a 2025 article in PMC.)

Why it matters The largest ongoing entry, and the one where the harm is spread thinly enough that no individual prescription looks wrong. Include it so the register is not only about identifiable catastrophes.

Who gained Manufacturers, intensive agriculture, and — the uncomfortable one — individual doctors and patients, each of whom got a small local benefit while the cost was borne collectively and in the future. Not capture. A commons problem.

Bendectin / Debendox

Verified 8 Sep Merrell · US tort litigation

Sold as Then alleged as: the treatment for morning sickness, taken by an estimated 33 million women, accused in mass litigation of causing birth defects.

What happened The epidemiology did not support the claim — successive studies and meta-analyses found no teratogenic effect. The manufacturer withdrew it in 1983 anyway, because litigation costs exceeded the product's revenue. Hospitalisations for hyperemesis gravidarum in the United States trebled after the withdrawal — I had recorded a doubling, and understated it (verified 8 Sep). The same doxylamine–pyridoxine combination was re-approved by the FDA in April 2013 as Diclegis, thirty years later, and had been available in Canada continuously throughout.

Why it matters The mirror of thalidomide, and the study needs it for exactly that reason. Pregnant women were left with untreated hyperemesis for thirty years by a false alarm. It also shows the withdrawal decision was economic rather than scientific — the same mechanism that keeps bad drugs on the market can take good ones off.

Who gained The plaintiffs' bar, and the expert witnesses who testified to causation the literature did not support. This case is a major reason US courts adopted the Daubert standard for admitting scientific evidence.

MMR and autism

Andrew Wakefield · The Lancet · the British press · the Department of Health

What was claimed A 1998 Lancet case series of twelve children proposing a possible link between MMR, a bowel condition and developmental regression. The paper stated it had not proved an association; at the accompanying press conference its lead author went considerably further and called for single vaccines in place of MMR.

What was established against him Brian Deer's investigation for the Sunday Times and later the BMJ found that Wakefield had been paid £435,643 in fees plus £3,910 expenses from the legal aid fund, via a solicitor preparing litigation against vaccine manufacturers, and had not disclosed it; that some children were litigants in that action; and that he had filed a patent relating to a measles vaccine. Deer compared the paper's descriptions against the children's own medical records and reported discrepancies in all twelve cases. The GMC found charges including dishonesty proven in 2010 and struck him off; The Lancet retracted fully the same year.

The epidemiology since Very large cohort studies — Madsen's 537,000 Danish children (2002), Jain's 95,000 (2015), Hviid's 657,000 (2019) — and meta-analysis over roughly 1.27 million children have found no association. Overturning this requires data, not argument.

Added 8 Sep — a fraud conviction touching that literature On 1 September 2026 Poul Thorsen pleaded guilty in the Northern District of Georgia to wire fraud, having stolen more than $1m of CDC grant money (verified 8 Sep). He had been on the HHS Inspector General's most-wanted list since 2011, was arrested in Germany in June 2025 and extradited in May 2026; sentencing is set for 1 December. From 2002 he was the principal investigator administering the CDC grants — over $11m awarded between 2000 and 2009 — funding Danish research into autism and vaccines, and he is the sixth of eight named authors on Madsen 2002, the first study cited in the paragraph above.

What that does and does not establish It is embezzlement, not falsification. The plea is to theft of grant money; no finding has been made that any data were fabricated, and he was neither lead author nor corresponding author. But this register should not reach for that distinction as a comfort. In any ordinary audit, discovering that the officer responsible for the money was stealing it triggers re-examination of everything he touched — and no such public re-examination of these studies has taken place, while they continued to be cited without caveat throughout the fourteen years he spent as a fugitive. That is an institutional-conduct question and it is open.

And a caveat I had not previously stated The two Danish studies are less independent of one another than the paragraph above implies: Anders Hviid and Mads Melbye are co-authors on both Madsen 2002 and Hviid 2019. They are the same group examining a later birth cohort, not a separate team replicating a result. The genuinely independent replication in that list is Jain 2015 — a different country, a different population, an unrelated team — together with the wider meta-analytic literature. The epidemiological conclusion does not rest on Madsen 2002 and would survive its withdrawal. That is what replication across populations is for, and it is the reason a single author's misconduct cannot settle the question in either direction.

Note how this reached me, and how it is being framed The owner raised it; I did not have it. Coverage divides exactly as this study would predict: the outlets giving it most prominence describe Thorsen as "the CDC researcher whose studies debunked the vaccine-autism link" — which overstates a sixth author on one of several papers — while mainstream coverage has been thin relative to a Justice Department conviction after a fourteen-year manhunt, which understates a real story about institutional conduct. Both distortions are present, they run in opposite directions, and the reader is being served by neither.

The harm UK MMR uptake fell from around 92% to below 80%. Britain lost WHO measles elimination status in 2019. In Samoa the same year, an outbreak in a collapsed-vaccination population killed 83 people, mostly infants.

Where the official account is deficient — 1: the process In 2012 Mitting J in the High Court quashed the GMC's findings and erasure against John Walker-Smith, the paper's senior clinical author, holding the panel's determination could not stand because its reasoning was inadequate and superficial and wrong in a number of instances, partly through failing to distinguish research from clinical care. Wakefield did not appeal and Mitting's ruling did not concern him. But the disciplinary process that produced the public verdict has been judicially criticised, and accounts that present it as clean are not accurate.

Where the official account is deficient — 2: why parents had a prior Britain introduced three MMR vaccines in 1988. Two of them, Pluserix and Immravax, contained the Urabe mumps strain and were withdrawn in September 1992 after Department of Health-commissioned research at Nottingham demonstrated a link with aseptic meningitis, at a reported risk in the region of 1 in 143,000 doses. Correction, 13 Sep: I had repeated a report that a Urabe-containing vaccine was withdrawn in Canada before the UK introduced these products in 1988. The Institute of Medicine's 1994 review gives a different timeline. A Urabe-strain MMR was released in Canada in 1986, and cases of mumps meningitis began to appear soon after. Sale there was suspended in May 1990, once virus isolated from eight patients had been sequenced as the vaccine strain. That was two years after Britain licensed the same strain, and two years before Britain withdrew it. A July 1988 recall in Ontario circulates only in secondary accounts and is not relied on here. Parents in 1998 were therefore not being irrational in principle: a real MMR safety problem had occurred, been denied at the time by some officials, and led to a withdrawal within four years. Wakefield's claim landed in ground the institutions themselves had prepared, and almost no account of the scare says so.

Who gained from the claim Wakefield, in fees and later a career built on it. The litigation solicitors. The newspapers, for which it ran as a circulation story for six years.

Who gained from his destruction Merck, Aventis Pasteur and GlaxoSmithKline, the three defendants in a UK group action covering more than 1,000 children — legal aid was withdrawn in September 2003 on the ground the case was very likely to fail, and it collapsed with about £15m of legal aid spent and no compensation paid. The Department of Health and the JCVI, which had staked themselves publicly on MMR and specifically refused to license single vaccines. The Lancet, for which one man's fraud is a far cheaper account than twelve years of failed peer review and ethics oversight found by an outside journalist. The GMC, which demonstrated capability. Brian Deer and his publishers — career-defining and award-winning work, which is a real interest in the finding standing up and does not make him wrong. And the profession generally, for which a single villain costs less to admit than a systemic failure — the same cost-of-admission mechanism as the Post Office record in Pillar IV.

My own bias, on this record Until challenged, this entry's beneficiary line named only Wakefield, the solicitors and the newspapers. Across roughly 120 records it appears to be the only place I applied the beneficiary test in a single direction — and it is the record where the wrongdoer is an individual challenging an institutional position. I did not notice; the owner of the study did, by turning my own method on me. Logged as prediction P8 confirmed. Note also what this does not do: my ozone record makes the argument against over-reading it, since DuPont held the patents on CFC replacements and the science was still right. A beneficiary can be aligned with a true finding. The record discrepancies, the undisclosed funding and 1.27 million children are not answered by showing that Merck benefited.

Why it is the register's most instructive entry Because every failure mode in the study is present at once. A fraudulent claim killed people. An institution's own earlier safety failure created the credulity it landed in. The disciplinary response was found deficient by a court. And the account that settled into public memory served the interests of several parties who were never asked what they had gained from it. All four of those are true simultaneously, and any version that keeps only one is propaganda for someone.

Who gained Wakefield, in the £435,643 and the patent — and then, in the other direction, everyone who needed the question closed. A critic who was demonstrably paid and demonstrably dishonest is an asset of great and lasting value: it licenses treating every subsequent question about vaccine safety as the same kind of thing. That licence is still being drawn on, and this register's own entries on thimerosal, Pandemrix and the 1976 swine flu programme show what it costs. Also: the Department of Health, which got a story in which the failure was one man's fraud rather than any question about its own communication; the press, in both directions, having sold the scare and then sold the exposure; manufacturers, for whom a discredited critic is worth a great deal; and an investigative reporter whose career was made by it — honestly earned, and still a gain. These are the parties the previous paragraph said were never asked. This is the asking.

Removing thimerosal as a precaution

Verified 8 Sep US Public Health Service · American Academy of Pediatrics

What happened A joint statement by the US Public Health Service and the American Academy of Pediatrics on 9 July 1999 called for the immediate reduction and elimination of the ethylmercury preservative from infant vaccines as a precaution, while stating there was no evidence of harm; by 2001 it was gone from all US childhood vaccines except some multi-dose influenza preparations (verified 8 Sep).

What the evidence shows — corrected 13 Sep This record called the Danish and Swedish experience "the natural experiment that settled it". It did not settle it, and I had rested the case on the weakest evidence available. Denmark removed thimerosal in 1992 and recorded autism went on rising (Madsen et al., Pediatrics, 2003), and Stehr-Green et al. (2003) found the same pattern in Sweden and California. But these are ecological comparisons of population trends, and the Danish register changed underneath them. Diagnostic criteria moved from ICD-8 to ICD-10 in 1994, and from 1995 outpatient contacts were counted too, when outpatients outnumbered inpatients four to six times. The Danish paper names both as reasons the rise may be exaggerated. It reports an inpatient-only analysis showing the same trend, but as data not shown. California's figures also kept rising after thimerosal left US childhood vaccines by 2001 (Schechter and Grether, 2008), with the same kind of limitation. The evidence that carries the weight follows individual children. Hviid et al. (JAMA, 2003) followed 467,450 Danish children and found a relative risk for autism of 0.85 (95% CI 0.60–1.20), with no dose-response. Andrews et al. (Pediatrics, 2004) followed 109,863 UK children and found no evidence of neurodevelopmental harm, with the possible exception of tics. The Institute of Medicine concluded in 2004 that the body of epidemiological evidence favors rejection of a causal relationship between thimerosal-containing vaccines and autism (verified 13 Sep). I had quoted it as reviewing "more than 200 studies" that "consistently provided evidence of no association", and I cannot find that wording in the 2004 report.

The counter-case, and what it establishes Critics argue the Danish ecological paper was shaped to reach its conclusion: Brian Hooker, the parents' group SafeMinds and, in 2025, Anthony Colpo. Three of their points stand. First, emails released by the CDC under Freedom of Information in 2006, heavily redacted, include a co-author at Aarhus writing in November 2002 that the manuscript's figures do not include the latest data from 2001 and that incidence and prevalence are still decreasing in 2001. The published paper ends at 2000. Second, José Cordero, a senior CDC official, wrote to Pediatrics in December 2002 supporting expedited review. The paper itself thanks Cordero, Coleen Boyle and Diana Schendel of the CDC for comments and advice during preparation of the manuscript. Third, the co-author Poul Thorsen went on to steal more than $1m of the CDC grants that funded the Danish programme (see the MMR record). Two of their points do not stand. The paper did not hide a fall: it states that rates were lower in 2000 than in 1999, and a dip in the most recent year is what incomplete recent diagnosis produces, so the 2001 email fits that at least as well as a real decline. And the claim that correcting for the register changes shows autism falling four-fold after 1992 comes from a 2004 letter by Hooker and Trelka in the Journal of American Physicians and Surgeons, not a study; the individual-level cohorts above contradict it. (The wording of the email and the letter is taken from the advocacy groups that published the release. Critics of those groups quote the same wording and do not dispute it. I could not retrieve the released documents themselves, 13 Sep.) Scored as conduct: an agency with a stated position helped shape and fast-track a paper supporting that position, and the paper left out the most recent year. That is a legitimate criticism whatever the science shows, and it is the conduct this study criticises everywhere else.

Why it matters A precautionary act read by the public as an admission. It fuelled the movement it was meant to defuse, and in some lower-income countries it disrupted vaccine supply. Precaution is not free — it carries an information cost that official bodies routinely price at zero. That principle generalises far beyond vaccines and is one of the study's most transferable findings.

Who gained The appearance of responsiveness, bought at a moment when looking dismissive was unaffordable. That is a real asset and it was paid for with someone else's credibility: the removal itself became the strongest available evidence that there had been something worth removing. What was gained was the avoidance of a headline. What was spent was the force of every reassurance issued afterwards.

Laetrile

Verified 8 Sep Alternative cancer clinics · Committee for Freedom of Choice in Cancer Therapy

What was claimed An apricot-kernel extract suppressed by organised medicine to protect chemotherapy revenue. Promoted through a political freedom-of-choice campaign; over twenty US states legalised it.

What happened The NCI-supported trial reported by Moertel and colleagues in the New England Journal of Medicine in January 1982 treated 178 patients and found no benefit whatever — no cure, no improvement, no stabilisation, no symptom relief, no extension of life — while several patients showed symptoms of cyanide toxicity or blood cyanide levels approaching the lethal range (verified 8 Sep). The FDA banned it. The real harm was the treatment abandoned or delayed in its favour. (That more than twenty states legalised it during the 1970s is confirmed by the National Cancer Institute's PDQ summary on laetrile, verified 13 Sep.)

Who gained The clinics, and the political organisation that raised funds on the suppression narrative. Note the structural symmetry with the pharmaceutical cases: a financial interest promoting a claim beyond its evidence. The direction differs; the mechanism does not.

AIDS denialism in South Africa

President Thabo Mbeki · Health Minister Manto Tshabalala-Msimang

What was claimed That HIV does not cause AIDS, that antiretrovirals are toxic, and that the epidemic was better addressed with garlic, beetroot and lemon. Advanced from the head of state, drawing on a small group of dissident scientists.

What happened Antiretroviral rollout was obstructed for years. Chigwedere and colleagues, in JAIDS in 2008, estimated more than 330,000 lives — about 2.2 million person-years — lost, plus 35,000 babies born with HIV for want of a mother-to-child prophylaxis programme, benchmarked against what Botswana and Namibia achieved with comparable resources (verified 8 Sep).

Why it matters The highest death toll in this register attributable to a single position — and it was an anti-establishment position, taken by a government against the international consensus, partly in reaction to a pharmaceutical industry that had genuinely behaved badly over drug pricing in Africa. Both things are true simultaneously. That is the study's hardest and most valuable case.

Who gained The state, in the short term, avoided the cost of a national treatment programme it had said was unaffordable. The dissident scientists gained a national platform unobtainable anywhere else.

Semmelweis and handwashing

Verified 8 Sep

The warning Doctors moving from the autopsy room to the delivery room were killing mothers; washing in chlorinated lime, introduced at the Vienna General Hospital in 1847, cut maternal mortality in his clinic.

A correction to my own figures (verified 8 Sep) I wrote "roughly 18% to about 2%". Those are the best and worst months, and using them is cherry-picking. The defensible comparison is average monthly maternal mortality falling from 10.65% to 1.98% — a risk ratio of about 5.4. Still overwhelming; still ignored. But a register that objects to dramatic numbers has no business using the dramatic version of its own best case.

What happened Rejected by the profession, partly because it implied physicians were the vector. Semmelweis lost his post and died in an asylum outside Vienna in 1865, of an infected wound, after being beaten by guards. Acceptance came only with germ theory, decades later.

Who gained Nothing commercial existed to gain, which is why this record is the study's control case. What was protected was the self-image of the physicians who would have had to accept that they had personally carried death from the autopsy room to the delivery bed, and had killed the women in their care. That is not a proposition a person can weigh dispassionately about himself. Protected alongside it: the authority gradient — the finding came from a junior Hungarian outsider and implicated senior men — and miasma theory itself, an entire explanatory framework under which a result with no mechanism behind it was easy to dismiss. The notion that a gentleman's hands could be dirty was socially unavailable before it was medically unavailable. Zero money, decades of resistance, and a man dead in an asylum. Every commercial case in this register was settled faster.

Asbestos

Verified 8 Sep

The warning Merewether and Price reported asbestosis to the British government in 1930. Mesothelioma's link was established by Wagner in 1960.

What happened Britain did respond in part: the Asbestos Industry Regulations followed in 1931 and came into force on 1 March 1932 — a fact I omitted from my first draft, and it matters, because "nothing was done for fifty years" is not accurate (verified 8 Sep). What took fifty-five years was prohibition: blue and brown asbestos were not banned until 1985, and white asbestos, still in active use, not until November 1999. Correction, 13 Sep: I had flagged my "around 2,500 UK mesothelioma deaths a year" as unverified. HSE's Mesothelioma statistics for Great Britain, 2026 records 2,146 deaths in 2024 and 2,255 in 2023, against an average of 2,508 a year over 2011–2020. The figure was right for the last decade and is now falling. HSE expects deaths among men to keep falling through the 2020s, but says a decline among women may not be clear until near 2030.

Who gained The asbestos industry, which ran a doubt campaign closely modelled on tobacco's — and, importantly for the study, the same consultancy and public-relations firms served both industries. The technique is transferable and was transferred.

Alice Stewart and X-rays in pregnancy

Verified 8 Sep

See also The practice itself is recorded in full under Routine X-ray of pregnant women in Era II. It appears twice deliberately: it is both a harm caused by a promoted practice and a true warning refused, and separating the two is what the study is for.

The warning Stewart's Oxford Survey of Childhood Cancers, begun in 1956, reported in 1958 that children who died of leukaemia or cancer had been X-rayed in the womb roughly twice as often as healthy children (verified 8 Sep).

What happened Dismissed. Brian MacMahon corroborated her in 1962 with a study of 734,243 children born in 37 north-eastern US maternity hospitals between 1947 and 1954, using medical records rather than mothers' recall — removing the recall-bias objection her critics had leaned on. Despite that vindication, Stewart's Oxford unit never received another Medical Research Council grant (verified 8 Sep). She ran the survey until 1981, ending with 15,276 matched case-control pairs. She was right.

Why it matters A woman, working outside the establishment consensus, correct against it, and ignored for a generation. Any account of institutional failure that treats dissent as the problem cannot accommodate her — and neither can any account that treats dissent as automatically correct, given Wakefield sits three entries above.

Who gained Obstetrics and radiology in the short term — and then something much larger. Her Oxford unit never received another MRC grant, which protected the standing of the committees that had called the practice safe. But the durable interest was elsewhere: from 1974 Stewart worked with Thomas Mancuso and George Kneale on the US government's own Hanford nuclear-worker data, and their 1977 analysis of 3,710 deaths found radiation-induced cancer running roughly ten times higher than the protection standards assumed. Mancuso's contract was terminated that year (verified 8 Sep). What was being defended by then was not a diagnostic practice but a dose limit — and the assumption of safety underwriting an entire industry and its workforce. She was inconvenient to obstetrics; she was expensive to the atomic establishment.

Leaded petrol

Verified 8 Sep

The warning Public health scientists objected at the outset in 1925. Clair Patterson demonstrated in the 1960s that atmospheric lead was overwhelmingly industrial.

What happened For more than forty years the safety of leaded petrol was studied almost entirely by Robert Kehoe and his assistants, in work funded, reviewed and approved by the companies that made the additive (verified 8 Sep). Patterson was refused contracts by ostensibly neutral bodies including the US Public Health Service, and in 1971 was excluded from a National Research Council panel on atmospheric lead contamination despite being the world authority on measuring it. Britain removed lead from petrol in 1999; the global phase-out completed in July 2021, Algeria last. Population IQ loss and the association with violent crime are still being estimated.

Who gained General Motors, DuPont and Standard Oil, who jointly owned the additive company — and the scientist whose research they funded, who set the safety standard used for forty years. But Kehoe is misread as bought. He held, and defended to the end, a coherent position: that lead was natural, that a body burden was therefore normal, and that the burden of proof lay on anyone claiming harm. He was wrong, and he was sincere, and the sincerity is what made him effective — a purchased advocate can be discredited by the receipt, whereas a believer must be defeated on the evidence. Patterson needed the geological record of the ice cores to do it, because nothing less would have been enough.

Marshall, Warren and H. pylori

Verified 8 Sep

The warning Peptic ulcers are caused by a bacterium and curable with antibiotics, not by stress and acid requiring lifelong suppression.

What happened Rejected for years — even the 1984 Lancet paper struggled, its editors unable to find reviewers who thought it important enough. In the summer of 1984 Marshall drank a culture and gave himself acute gastritis to prove the point. The evidence became impossible to ignore only in the early 1990s; the NIH consensus panel recommended antimicrobial treatment in 1994, a decade after the self-experiment, and the Nobel came in 2005 (verified 8 Sep). In the interim, patients took acid suppressants indefinitely for a curable infection — and those were not merely profitable drugs: cimetidine was arguably the first blockbuster in pharmaceutical history, and omeprazole was the biggest-selling drug in the world for years.

Who gained The manufacturers of H2 blockers and proton-pump inhibitors, whose market was lifelong management of a condition that turned out to need two weeks of antibiotics. Nobody needed to suppress the finding for the incentive to point the wrong way. But the drug companies are the smaller beneficiary. The larger one was gastroenterology's own model — ulcers as a chronic constitutional disease of stress and acid, requiring lifelong management by a specialty organised around exactly that. A bacterial cause did not merely change a treatment; it dissolved a disease category and the professional identity built on it. That is why argument alone could not move it, and why Marshall had to drink the culture. No money changed hands in the resistance. It lasted longer than most of the commercial cases in this register.