DES (Diethylstilbestrol) — A Drug Disaster Across Generations— A drug believed to work, given to mothers, brought cancer to the daughters they were carrying

01Why this case stands apart

Every drug disaster we have seen so far harmed the person who took the drug. In the Elixir Sulfanilamide case, the children who swallowed the syrup died. With thalidomide, mothers who took the drug gave birth to children with limb defects. The distance between the person dosed and the person harmed was short.

The DES (diethylstilbestrol) case breaks that pattern completely. The person who took the drug was the mother. The person who developed cancer was her daughter — and not at birth, but after puberty. Fifteen to twenty-two years passed between exposure and disease. The recipient of the drug, the person harmed, and the timing of the harm were all displaced. This case confronted medicine with a fact almost no one had seriously considered: the harm of a drug can appear across generations and across time.

02What DES was — a cheap, unpatented synthetic estrogen

DES was the first orally active synthetic estrogen, synthesized in 1938 by the British chemist Edward Charles Dodds and colleagues. It reproduced the action of the natural female hormone, yet could be manufactured cheaply at scale. Dodds did not patent it. As a result, DES became an inexpensive drug anyone could make, supplied at once by many manufacturers.

Its initial uses were menopausal symptoms and estrogen deficiency. The U.S. FDA approved DES for such limited indications in 1941. The problem was what came next — the indications were greatly expanded.

03The "prevents miscarriage" hypothesis — spread before it was proven

In the late 1940s, the Harvard gynecologists George and Olive Smith proposed an idea: "Pregnancy requires estrogen, so supplementing with DES should prevent miscarriage and premature birth." The reasoning sounded plausible. They promoted it actively, and DES came to be prescribed as a drug to prevent threatened miscarriage.

The way it spread had, by modern standards, no brakes. It was prescribed not only in high-risk pregnancies but to healthy pregnant women "just in case," and in some cases was even added to prenatal vitamins. The number of people exposed to DES in the United States between 1940 and 1971 — pregnant women and their fetuses — is estimated at 5 to 10 million.

041953 — proven not to work, yet it did not stop

In 1953, Dieckmann and colleagues at the University of Chicago ran a double-blind controlled trial of DES. Roughly 1,600 pregnant women were divided into a DES group and a placebo group — a high-quality study for its time. The result was unambiguous: DES reduced neither miscarriage nor premature birth. There was no benefit.

Scientifically, prescribing should have stopped there. In reality, DES continued to be prescribed until 1971, for some eighteen more years. A single negative trial could not overturn an entrenched prescribing habit or the individual physician's sense that "it works in my patients." Confirmation bias was operating at the scale of an entire system.

051971 — what Herbst's report connected

In 1971, at a Boston hospital, the gynecologist Arthur Herbst and colleagues noticed a strange cluster. Clear-cell adenocarcinoma of the vagina — a cancer that should appear only rarely, and in postmenopausal women — was turning up one after another in women in their late teens and early twenties. It was wildly out of pattern.

Herbst's team ran a case-control study to find what the patients shared. One thing surfaced: their mothers had taken DES during pregnancy. The finding, reported in 1971 in the New England Journal of Medicine, was the first demonstration that a drug given to the mother could cross the placenta, act on the fetus, and cause cancer more than a decade later. That November, the FDA issued a notice making DES use in pregnancy a contraindication.

06What happened to the DES children — the spread of harm

The effects went far beyond the clear-cell adenocarcinoma that first drew attention. The generation exposed in utero has been reported to suffer different harms by generation and by sex.

Clear-cell adenocarcinoma itself was uncommon among exposed daughters (on the order of 1 in 1,000). But because the exposed population numbered in the millions — and because women reached adulthood without knowing whether they had been exposed in the womb — the anxiety and the difficulty of follow-up were enormous.

07Why the harm was so hard to see

Behind the thirty years of overlooked harm lay a set of features peculiar to this drug.

• A long latency — when dosing and onset are 15 to 22 years apart, it is hard even to think of connecting them
• The harmed party differs from the dosed party — the mother's chart and the daughter's symptoms are separate medical records that never meet
• Onset is rare — a single physician noticing a cancer that strikes 1 in a few thousand is nearly accidental (Herbst could notice it only because of an exceptional cluster)
• The comfort of "widely used" — no one suspected that a drug used by millions could cause harm across generations

08Four lessons that remain today

Lesson 1 — A drug spread without proof of benefit can leave only harm

DES's miscarriage-prevention benefit was never proven. The 1953 trial showed it did not work, and later reanalyses even suggested possible harm. The benefit was near zero; what remained was harm across generations. "The reasoning sounds right" and "an authority recommends it" are no substitute for proof of efficacy. This reckoning led to the U.S. Kefauver-Harris Amendment of 1962, which required new drugs to prove they actually work.

Lesson 2 — Harm can appear across generations and across time

The person dosed and the person harmed differed; twenty years separated dosing from onset. This case showed that drug safety assessment is not enough when it only looks at the person who took the drug, over a short period. For drugs touching pregnancy and reproduction in particular, effects on the next generation must be in view — a perspective that feeds directly into today's rigor on teratogenicity assessment and prescribing in pregnancy.

Lesson 3 — Prescribing to a pregnant patient always means more than one patient

When you prescribe to a pregnant woman, there is never only one party: the mother, the fetus, and sometimes generations beyond. DES was extended even to healthy pregnant women "just in case." When the benefit is uncertain and the parties include a fetus who cannot consent, the decision should be maximally cautious. This is a question of patient rights and of the precautionary principle.

Lesson 4 — "Everyone uses it" is not proof of safety

That millions used DES guaranteed nothing about its safety. If anything, the volume of use bred the complacency that "it must be fine." The quantity of use and the proof of safety are matters of a different order. Prevalence must never be treated as evidence of safety — a pitfall that recurs in the world of promotional materials and information provision as well.

09Connections to other chapters

The DES case connects to other chapters of this site as follows.

• Drug Disasters 02 (Thalidomide) — also harm to the fetus. Where thalidomide's damage was visible at birth, DES's appeared more than a decade later; the two can be contrasted as opposite ends of "visibility"
• Drug Disasters 01 (Sulfanilamide) — where sulfanilamide produced the mandate to prove safety, DES underscored the need to mandate efficacy (the Kefauver-Harris Amendment)
• Drug Disasters 04 (SMON) — the shared problem of long latency and post-marketing follow-up
• Ethics, Part 4 — the ethics of dosing when the affected party (the fetus, the next generation) cannot consent