LL1-08 hindsight check: The DES story: long-term consequences of prenatal exposure#

Late lessons from early warnings: the precautionary principle 1896–2000 (EEA Environmental Issue Report No 22, 2001), Chapter 8, pp. 84–92. Authors: Dolores Ibarreta and Shanna H. Swan.

Checked September 2026. Sources were retrieved from primary documents wherever possible. These include FDA Federal Register notices, IARC and NTP carcinogen evaluations, NCI material, EU legislation, a Dutch court ruling, and peer-reviewed cohort studies, meta-analyses and a Cochrane review. Secondary sources are flagged as such where they are used.

Overview#

What holds. The chapter’s central facts and its strongest lessons have aged well.

• Causation is settled. The link between prenatal DES and vaginal/cervical clear-cell adenocarcinoma (CCA) is no longer in question. IARC confirmed DES as a Group 1 human carcinogen in 2012, and the NTP lists it as a known human carcinogen.
• The predicted harms are confirmed and quantified. The chapter said reproductive-tract harms in daughters were common and the cancer rare. That was confirmed and quantified by the pooled NCI cohorts (Hoover et al., 2011).
• The efficacy failure is confirmed. A Cochrane review (2003) found no benefit and some harm.
• A primary document sharpens the chapter’s main lesson. The chapter cites the FDA’s November 1971 notice but does not quote it. Read directly, it shows that the post-1962 federal efficacy review still classed DES as “possibly effective” for preventing pregnancy loss 18 years after the negative 1953 trial. The contraindication came only “in view of” the cancer finding. That strengthens the chapter’s key point: a proven lack of benefit went unacted on, while a vivid harm triggered action within months.

Where the record is mixed. The chapter’s forward-looking claims have a mixed record.

• Continuing harms: confirmed. The prediction that harms would keep appearing as the cohort aged is borne out:
• CCA cases up to age 55;
• raised vaginal cancer in Dutch women in their 60s;
• early menopause;
• coronary disease;
• a pancreatic cancer signal.
• Breast cancer in daughters over 40: contested. It appeared in US and French cohorts, but weakened with longer US follow-up and was absent in a 2025 Dutch national cohort.
• No overall cancer excess. Neither DES daughters nor DES sons show an overall excess of cancer.
• Third generation: unresolved. The question the chapter said would “take many years to resolve” still has no answer.

Details to correct. Several details in the chapter need correcting:

• cohorts of DES sons did exist before 2001;
• teratology as a field predates both thalidomide and DES;
• the Smiths’ DES work was based at the Free Hospital for Women, not Massachusetts General Hospital;
• the chapter’s exposure figures cannot be reconciled with later official estimates;
• “fewer than 1 in 1,000” slightly understates CCA risk.

Regulatory gaps and the endocrine-disruption lesson.

• The gaps persist. The gaps the chapter named on pp. 89–90 are well documented as persisting (Adam et al., 2011; PRGLAC, 2018; the EU valproate decision of 2018), even though formal benefit–risk assessment and active surveillance systems have since been built.
• An unforeseen consequence. The chapter did not anticipate this: the “extreme caution” it urged was partly institutionalised as the exclusion of pregnant women from research, which itself keeps knowledge thin.
• Endocrine disruption: taken up in law, still contested in science. The lesson on p. 88 was taken up strongly in EU law. But extending it from pharmacological DES doses to environmental exposures remains scientifically contested.

Annex 3 of the 2013 report (Late Lessons II, pp. 729–730) carries a one-page update by Shanna Swan, “DES: the view from 2013”. It calls the 2001 chapter “merely the tip of the iceberg”. It reports the Hoover (2011) cohort results, the sons’ urogenital findings (Palmer et al., 2009) and early third-generation birth-defect data (Titus-Ernstoff et al., 2010). It ends by pointing to bisphenol A as a synthetic oestrogen in current commerce.

• Two slips in the annex:
• it gives the unexposed comparison group as “1 027” women (Hoover et al. report 1,927);
• the annex introduction implies breast cancer is arising specifically in CCA survivors, whereas the excess reported was among exposed daughters generally.
• Annex 2 carries the 2001 errors forward. The same volume’s Annex 2 (Table A2.8) reproduces the 2001 timeline unchanged, including “November 1971 FDA withdraws approval” and “1985 Last reported use of DES by pregnant women world-wide”. Its summary table (A2.1) describes the response as “1971–1985 US, EU, global ban”, but this check found no international instrument banning DES.
• Later evidence tempers the annex. The evidence reviewed below supports the annex’s “tip of the iceberg” framing for daughters’ reproductive health. It tempers the framing on cancer and on later generations.

Claim 1: The discovery and the FDA’s response#

Original claim (pp. 84, 86; Table 8.1 p. 90). Herbst et al. (1971) found prenatal DES exposure in 7 of 8 young women with vaginal CCA versus 0 of 32 matched controls. Greenwald et al. (1971) corroborated this within months. The FDA declared DES contraindicated in pregnancy in November 1971, about seven months later. On p. 84 and in Table 8.1 this is described as the FDA having “withdrew approval”.

Subsequent developments.

• Dates confirmed. Herbst, Ulfelder and Poskanzer was published on 22 April 1971 (NEJM 284:878). Greenwald et al. followed on 12 August 1971 (NEJM 285:390), using five cases found through the New York State Cancer Registry, all five exposed to synthetic oestrogens and none of the controls. IARC’s 2012 review describes both studies in these terms and notes that the evidence was strong “primarily” because CCA is so rare in young women and the exposure relatively infrequent (IARC Monographs Vol. 100A, Diethylstilbestrol, 2012; chapter PDF archived at web.archive.org).
• What the FDA actually did. The FDA’s action was a Drug Efficacy Study Implementation (DESI) notice, dated 5 November and published on 10 November 1971 (36 Fed. Reg. 21537–21538, 10 Nov 1971).
• It records that the National Academy of Sciences–National Research Council review had classed DES preparations as “possibly effective” for “Prevention of accidents of pregnancy”.
• It then states that “in view of the fact that a statistically significant association has been demonstrated between the use of diethylstilbestrol in early pregnancy and the occurrence of adenocarcinoma of the vagina in the offspring”, DES and its congeners are “contraindicated for use in pregnancy”.
• In fairness to the efficacy review, the same notice found that these oestrogen products “lack substantial evidence of effectiveness” when labelled for one narrower pregnancy claim, “relief of pregnancy bleeding”. The main miscarriage-prevention claim was the one left at “possibly effective”. (Text checked against the Federal Register page images, Sept 2026.)
• All labelling had to carry that contraindication. Marketing for other indications continued.
• So “withdrew approval” (p. 84; Table 8.1) is loose shorthand. What happened was a mandatory contraindication imposed through a relabelling notice. The six-and-a-half-month interval is accurate.
• Causation settled. The causal inference has since been settled beyond reasonable doubt:
• IARC (2012): DES “causes CCA in the vagina and cervix of women who were exposed in utero”; Group 1.
• NTP Report on Carcinogens, 15th edition (2021): “known to be a human carcinogen”.
• The Chicago CCA registry’s 40-year update found that DES-related CCA risk by birth cohort tracked US DES prescribing over time (r = 0.98) (Huo et al., Gynecol Oncol, Sept 2017).
• Later cohort estimates of the CCA standardised incidence ratio range from about 28 to 49 (Troisi et al., Environ Mol Mutagen 2019, online 2017; Boekel et al., Eur J Epidemiol, 28 May 2025).
• For contrast, Ulfelder, one of the 1971 authors, wrote in 1980 that “the role of DES in its carcinogenesis is still unproved” (Ulfelder, Cancer, June 1980). That doubt has since gone.

Verdict: held up (the causal inference has been strengthened). The FDA’s action should be described as a contraindication rather than a withdrawal of approval.

Implication for weight. The factual core of the case is secure and can carry heavy weight. The primary FDA text adds something the chapter only asserts: the pregnancy use was ended by the harm signal, not by the efficacy system.

Claim 2: DES did not work, and the evidence was not acted on#

Original claim (p. 86, with pp. 88, 90).

• Randomised placebo-controlled trials (Dieckmann et al., 1953) showed DES did not prevent miscarriage, yet it was prescribed for about two more decades.
• Brackbill and Berendes’ 1978 reanalysis of the 1953 data, using methods available in 1953, found that DES increased the outcomes it was sold to prevent. On this basis nearly 20 years of exposure could have been avoided.
• The chapter calls this “a massive failure of the system”.

Subsequent developments.

• Cochrane confirms no benefit and some harm. The Cochrane systematic review of oestrogen supplementation, mainly DES, for preventing miscarriage concluded: “There was no benefit with the use of diethylstilbestrol in preventing miscarriages”. It found that in utero exposure “led to increased rate of miscarriage and preterm birth” and more low-birthweight babies (Bamigboye & Morris, Cochrane Database Syst Rev 2003, CD004353). The full numerical results were not retrievable in this check, so only the conclusions quoted in the abstract are relied on here.
• The formal efficacy system missed it too. The FDA’s 1971 DESI notice (Claim 1) shows that when the federal efficacy review, created after the 1962 Kefauver–Harris amendments, got to DES, it still rated the main pregnancy indication, “Prevention of accidents of pregnancy”, “possibly effective”. It did not rate it ineffective, despite the 1953 trial (it did reject the narrower claim of “relief of pregnancy bleeding”). The failure was therefore not only one of prescribers and manufacturers. The regulatory efficacy mechanism set up to catch exactly this kind of problem had not caught it by 1971. This strengthens the chapter’s charge of a system failure, and is not in the chapter itself.
• A later account from within obstetrics. Obstetrician-epidemiologist David Grimes (a secondary, reflective account in the James Lind Library, 2007) attributes continued use to authority: “The eminence of the Boston physicians trumped the evidence”. He adds that credible data were set aside because the problem was serious and DES “might work” (Grimes, James Lind Library).
• A nuance on scale. The 2017 registry analysis found that DES-related CCA risk was highest in the 1951–56 birth cohorts and tracked US prescribing closely (Huo et al., 2017), and the chapter’s own Figure 8.1 shows the same pattern. This implies that use peaked in the early to mid-1950s and then declined. “Prescribed for two decades after its lack of efficacy was clearly demonstrated” is true of continuation, not of scale.
• The counterfactual is optimistic. The claim that proper analysis in 1953 would have avoided “nearly 20 years” of exposure assumes the profession would have acted on a positive finding of harm. Yet Dieckmann’s own paper already concluded that DES was ineffective (p. 86), and that did not stop use. The counterfactual is plausible but not demonstrated.

Verdict: strengthened.

Implication for weight. The lesson that absence of benefit should itself end an exposure, and that systems fail to act on it (pp. 86, 88, 90), is the most robust lesson in the chapter. It is now backed by a systematic review and by the regulator’s own 1971 wording. The “20 years avoidable” figure should be treated as an upper bound, not an estimate.

Claim 3: A rare cancer versus common reproductive harm, and the sons#

Original claim (pp. 86–87).

• CCA occurs in fewer than 1 in 1,000 exposed daughters (Melnick et al., 1987), and the Herbst registry holds fewer than 800 cases worldwide.
• Reproductive-tract abnormalities in daughters are common.
• Ectopic pregnancy, spontaneous abortion and premature delivery are significantly increased (Swan, 1992).
• DES sons show more genital abnormalities, but “no comparable cohort of ‘DES sons’ has ever been established” (p. 87).

Subsequent developments.

• Daughters’ reproductive harm is strongly confirmed. Combining three NCI cohorts (4,653 exposed and 1,927 unexposed women), Hoover et al. (NEJM, 6 Oct 2011) reported the following cumulative risks, exposed versus unexposed:

• Risks were generally higher in women with vaginal epithelial changes, a marker of higher and earlier exposure.
• The NCI fact sheet (updated 31 Jan 2025) repeats these figures (NCI DES fact sheet).
• CCA risk is slightly higher than the chapter’s figure. Later estimates run somewhat above “fewer than 1 in 1,000”:
• The NCI cohort gave a cumulative incidence of about 1.5 per 1,000, with a standardised incidence ratio of 40.7 (Hatch et al., JAMA 1998; summarised in the NTP Report on Carcinogens, 15th ed., 2021).
• The Chicago registry update gave a cumulative risk by age 50 of 1 in 750 among women with documented exposure. The figure is 1 in 520 if probably exposed cases without documentation are included (Huo et al., 2017).
• NCI now says “approximately 1 in 1,000”.
• The registry count is consistent with the chapter: 720 CCA cases through 2014, 420 with documented DES exposure (Huo et al., 2017).
• Sons: the abnormalities are confirmed. In three US cohorts, relative risks were (Palmer et al., Environ Health, Aug 2009):
• 1.9 for cryptorchidism;
• 2.5 for epididymal cyst;
• 2.4 for testicular inflammation or infection.
• Sons: the cancer evidence is mixed.
• A meta-analysis gave a summary odds ratio for testicular cancer of 2.98 (95% CI 1.15–7.67) (Hom et al., JNCI Cancer Spectrum 2019).
• IARC (2012) had called the testicular association “now uncertain”, and NCI calls the evidence “mixed”.
• Middle-aged sons show no excess of overall or prostate cancer (overall HR 0.94) (Strohsnitter et al., CEBP, 16 July 2021).
• NCI reports no increase in infertility.
• The “no cohort of sons” statement was wrong even in 2001. The NCI Combined Cohort includes 4,101 men, about half of them exposed, drawn from four cohorts assembled between 1953 and 1994: the Dieckmann trial sons, Mayo Clinic, Horne and the Women’s Health Study. Systematic NCI follow-up began in 1994 (Strohsnitter et al., 2021, methods; PMC8492497). The claim can only be rescued if “comparable” means a clinically screened cohort like the DESAD cohort of daughters. Swan’s own 2013 update cites “three US cohorts of DES‑exposed sons”.

Verdict: strengthened for the core claim (a rare signature cancer against common reproductive harm). Two details are corrected: CCA risk is slightly higher than stated, and sons’ cohorts did exist.

Implication for weight. The pattern at the heart of the chapter is now among the best-quantified in the whole report: the most visible harm was the rarest, and most of the damage lay in common, less visible outcomes. The pattern can bear significant weight. The sons’ material should be cited from post-2001 sources, not from the chapter.

Claim 4: Scale of exposure#

Original claim (pp. 86–87, Box 8.1).

• The NIH estimated that 4.8 million pregnant women in the US took DES.
• Estimates of in-utero exposure range from 2 to 10 million, with about 10 million total exposures worldwide.
• An EC-funded survey found DES prescribed in pregnancy in all 18 European countries surveyed except Sweden and Hungary.
• The digest notes that the chapter’s figures are internally inconsistent.

Subsequent developments.

• US figures have converged on one combined estimate. Official US sources now give a single estimate of 5 to 10 million Americans, mothers and children combined, exposed between the 1940s and 1971:
• NCI fact sheet (Jan 2025);
• NTP RoC (2021), citing NIH 1999;
• IARC (2012), citing Giusti et al. 1995;
• CDC’s former DES Update site (archived April 2021), for 1938–1971.
• The chapter’s figures do not fit that estimate. None of these sources uses a figure of 4.8 million pregnant women, and this check could not trace one. If US mothers and children together number 5–10 million, the chapter’s worldwide total of 10 million is almost certainly too low. IARC states plainly that “the number of women exposed prenatally to diethylstilbestrol worldwide is unknown”.
• European figures.
• France: a 1983 French study, reported by Fillion and Torny, estimated that 200,000 women were treated and 160,000 living babies exposed (Fillion & Torny, Reprod Biomed Soc Online, online 30 Oct 2021, vol. 14, 2022). This is consistent with Box 8.1 (200,000 pregnancies).
• Netherlands: a 2025 nationwide cohort enrolled 12,249 exposed women (Boekel et al., 2025). The Amsterdam Court of Appeal records that DES preparations were marketed in the Netherlands from about 1947 and not permitted in pregnancy after 1976 (Gerechtshof Amsterdam, 1 June 2006, ECLI:NL:GHAMS:2006:AX6440). This fits Box 8.1’s 1947–75 range.
• The rest of Europe: no later primary re-survey of the 18-country picture was found.

Verdict: partly held up. The order of magnitude is right. The worldwide total is probably understated, the “4.8 million pregnant women” figure cannot be traced, and no reliable global estimate exists even now.

Implication for weight. Use “5–10 million people exposed in the US alone (mothers and children); worldwide total unknown”. Avoid the chapter’s figures. The mechanism does not depend on exact numbers, but the chapter’s arithmetic should not be reused.

Claim 5: European lag, displacement of use, and the slow exit#

Original claim (p. 89).

• Europe was up to 12 years slower than the US: Dutch advice came in 1974, French labelling in 1977, and use in Spain was reported until 1983.
• DES was used in maternity care in Brazil, Costa Rica, Kenya, Mexico, Peru, Rwanda and Zaire in 1985.
• It was still prescribed in pregnancy in Mexico, Uganda and Poland in the early 1990s.
• Eli Lilly, the last US manufacturer, stopped production in 1997.

Subsequent developments.

• The European lag is confirmed, and later work explains it.
• France: Fillion and Torny (2021/22) document that in France DES “ceased to be indicated for the prevention of miscarriage” in 1976 and that pregnancy became a contraindication in 1977.
  • They attribute the lag to “strategic ignorance”: French physicians dismissed US warnings as inapplicable.
  • They document that French institutions “never initiated an epidemiological cohort”.
  • DES daughters gained a right to maternity leave only through a 2004 law, whose implementing decrees came in 2006 (private sector) and 2010 (civil service), when “most DES daughters had already passed childbearing age”.
• Netherlands: the Amsterdam court ruling gives 1976 for the Dutch prohibition in pregnancy, two years after the 1974 ministerial advice cited by the chapter.
• The US exit was slow and piecemeal, which supports p. 89.
• Lactation suppression. An FDA advisory committee recommended withdrawing the estrogen indication for suppressing postpartum breast engorgement on 31 January 1978. The FDA published a notice of opportunity for hearing in October 1978. The final withdrawal order came only on 17 December 1998 (63 FR 69631).
• High-dose products. Oral and parenteral DES products containing 25 mg or more per unit dose were placed on the FDA list of products withdrawn for safety or effectiveness, and may not be compounded (64 FR 10944, 8 Mar 1999).
• Remaining approvals. The FDA formally withdrew approval of Lilly’s remaining DES applications (injection, tablets, suppository; NDAs 4-038 to 4-041), Bristol-Myers Squibb’s Stilbetin and Bayer’s Stilphostrol, effective 30 September 2000. The holders had told the agency the products were “no longer marketed” (65 FR 55264, 13 Sept 2000). This is consistent with the chapter’s 1997 end of production, which rests on a personal communication.
• DES did not disappear everywhere. IARC (2012) notes that DES “is available as 1 mg and 5 mg tablets for oral administration in several countries”, citing a 2007 pharmacopoeia. The NTP (2021) notes continued use in clinical trials for prostate and breast cancer.
• Developing-country use is unverified. No post-2001 primary source was found that confirms or refutes the claims of use in 1985 and the early 1990s in developing countries. They still rest on Palmlund (1996) and DES Action. The chapter’s own Table 8.1 (“1985 last reported use”) contradicts its text, and the 2013 Annex 2 repeats the table unchanged.

Verdict: partly held up. The European lag and the slow, piecemeal US exit are confirmed and now better explained. The claims about developing-country use remain unverified.

Implication for weight. The patterns of jurisdictional divergence and of piecemeal exit, one use at a time (digest insights 10 and 11), are stronger than the chapter shows. The 20-year gap in the lactation-suppression withdrawal is a telling primary-source example. The specific claims about displacement to developing countries should be flagged as resting on advocacy sources.

Claim 6 (prediction): Harms will keep emerging as the cohort ages#

Original claim (p. 87). “The known consequences of its use will continue to appear as this cohort ages”, and follow-up is needed to identify “unknown consequences, including a possibly increased cancer risk with age”.

Subsequent developments.

• Late CCA.
• In the Chicago registry, 80% of DES-related CCA occurred at ages 15–31, but cases occurred “as late as age 55”, with “a small second peak … around age 42” (Huo et al., 2017).
• US population registry data suggest CCA stays elevated into midlife and older ages, though with wide confidence intervals (White et al., CDC, Cancer Causes Control, 29 June 2022).
• The Dutch national cohort found vaginal cancer raised overall (SIR 10.5) and still raised at ages 60–69 (SIR 8.3, 95% CI 1.00–29.9). It recommended extending screening beyond 60 (Boekel et al., 2025).
• Against this, a US single-institution review of 503 exposed women aged 50 or over found only one DES-related cervical or vaginal cancer after 50 and none after 65. It proposed ending lifelong screening (Wamakima et al., J Low Genit Tract Dis, online 12 Oct 2022).
• Other outcomes that emerged with age.
• Early menopause: HR 2.35 (Hoover et al., 2011).
• Cervical precancer (grade 2 or higher): raised through age 44 (HR 2.03) but not at 45 or over (Troisi et al., Obstet Gynecol, 29 Feb 2024).
• Coronary artery disease, HR 1.74, and myocardial infarction, HR 2.20 (Troisi et al., JCEM 2018).
• Pancreatic cancer (SIR 2.43, 95% CI 1.21–4.34; the internal comparison, HR 7.16, 95% CI 0.84–61.5, is very imprecise; Troisi et al., 2019) and pancreatic disorders (Troisi et al., J Dev Orig Health Dis 2021).
• Breast cancer in daughters over 40: now contested.
• For: the first prospective analysis found an incidence rate ratio of 1.91 at 40 and over, and 3.00 at 50 and over (Palmer et al., CEBP, Aug 2006). Hoover et al. (2011) reported HR 1.82 at 40 and over, and a French cohort found an incidence rate ratio of 2.10 (Tournaire et al., Thérapie 2015). IARC (2012) did not list breast cancer in daughters among its evaluated sites, and the NCI fact sheet (2025) still states “approximately twice the risk” after 40.
• Against: with longer US follow-up the internal comparison with unexposed women fell to HR 1.05 (95% CI 0.79–1.41) (Troisi et al., 2019). The same paper still found a modest excess against general-population rates (SIR 1.17, 95% CI 1.01–1.36), and its authors called the breast and pancreatic excesses “concerning” and asked for continued follow-up, so the US evidence is weakened rather than negative. A 2026 NCI paper refers to “more recent findings of attenuated breast cancer risks” with “longer follow-up and adjustment” for standard risk factors, although in its own post hoc analysis the earlier association remained apparent (Mitra et al., Int J Cancer, 27 Jan 2026; full text PMC13359020). The Dutch national cohort found no excess (SIR 1.03; HR 0.97 versus unexposed sisters) and concluded that apart from vaginal cancer, exposed women “do not seem to be at increased risk of cancer, including breast cancer” (Boekel et al., 2025).
• No overall cancer excess. Daughters show none in the US (HR 1.03; Troisi et al., 2019; the 1998 NCI analysis also found none, rate ratio 0.96 [Hatch et al., 1998]) or in the Netherlands (SIR 0.98). Sons show none either (Strohsnitter et al., 2021).
• Some searches for unknown harms came back negative:
• depression: HR 1.1 in women, 1.0 in men (Titus et al., Epidemiology 2019);
• benign breast disease: HR 0.97 (Mitra et al., 2026).
• Mothers. The chapter does not discuss them, but DES mothers have a modestly raised breast cancer risk: RR 1.27 (Titus-Ernstoff et al., Br J Cancer 2001). IARC (2012) classes this as causal.

Verdict: held up. Harms did keep appearing as the cohort aged, and later-life surveillance proved necessary. Two parts of the picture are less certain: the one specific later-life cancer that seemed established, breast cancer in daughters, is now contested; and there is no overall excess of cancer.

Implication for weight. The general lesson that long-latency harms can surface for decades (pp. 87–88) is strong. But the evidence cuts both ways. Longer and larger follow-up also shrank some harm estimates. The 2013 annex’s framing, that harm only “expands over time”, is too one-directional and should not be carried into the lens unqualified.

Claim 7 (prediction): Third-generation effects#

Original claim (pp. 87–88). Mouse studies (Newbold et al., 1998) show more tumours in the third generation, which suggests that DES grandchildren “may also be at increased cancer risk”. The chapter says this is a question “that will take many years to resolve”.

Subsequent developments.

• Animal evidence extended.
• Male descendants of DES-exposed mice developed reproductive-tract tumours (Newbold et al., 2000, summarised in NTP RoC 2021).
• A 2024 mouse study found accelerated folliculogenesis in F2 females (Rogers et al., Sci Rep 2024).
• Human cancer evidence so far negative or imprecise.
• The first NCI third-generation report found no overall cancer increase, with more ovarian cancer than expected on small numbers (Titus-Ernstoff et al., Epidemiology 2008).
• The 2024 update found self-reported overall cancer HR 0.83 (95% CI 0.36–1.90) and no CCA. Borderline ovarian tumours gave an imprecise HR of 3.46 (0.37–32.42). Testicular cancer was not raised in grandsons (Titus et al., Cancers, 18 July 2024).
• The authors conclude there is “little evidence that DES is associated with cancer risk in third-generation females or males, but these individuals are relatively young”.
• Human non-cancer evidence: suggestive but inconsistent.
• Birth defects: raised in the NCI third-generation cohort, with ORs of 1.53 in grandsons and 2.35 in granddaughters, “although the pattern of defects did not resemble” those in the second generation (Titus-Ernstoff et al., 2010, as reported in the 2013 Annex 3).
• Menstrual and pregnancy outcomes in granddaughters: irregular menses PR 1.32, amenorrhoea PR 1.26 and preterm delivery RR 1.54 (Titus et al., Reprod Toxicol, online 27 Dec 2018).

• Hypospadias in grandsons: results vary widely.

  Study	Setting	Result
  Klip et al., Lancet 2002	Dutch cohort	PR 21.3 (based on 4 cases)
  Brouwers et al., Hum Reprod 2006	Case–referent study	OR 4.9 (95% CI 1.1–22.3)
  Palmer et al., Epidemiology 2005	US cohort	OR 1.7 (0.4–6.8), i.e. little evidence of increased risk
  Kalfa et al., Fertil Steril 2011	529 French families	Raised proportion reported

• ADHD: raised in grandchildren in the Nurses’ Health Study II (7.7% vs 5.2%; OR 1.36) (Kioumourtzoglou et al., JAMA Pediatrics 2018).

• French studies. Several French studies drawn from self-selected affected families report broader multigenerational effects (e.g. Gaspari et al., IJERPH 2023). These designs are weaker and are not given much weight here.
• Prematurity as a pathway. French registry data summarised by Prescrire (2016) report that about a quarter of DES grandchildren were born prematurely, and that cerebral palsy was raised (59 vs 6 per 10,000) (Prescrire Int 2016). Some grandchild harms may therefore run through the DES daughters’ high-risk pregnancies rather than through germline transmission.
• Mechanism. Grandchildren (F2) of a woman treated in pregnancy are not “transgenerational” in the strict sense. Their germ cells were directly exposed in utero, and F3 is the first unexposed generation (Skinner, Reprod Toxicol 2008). Human methylation data show modest differences in exposed daughters (Bodelon et al., Environ Res 2023).
• Official and legal position.
• The NCI (2025) says grandchildren “may have a slightly higher risk of cancer and birth defects, including hypospadias in DES grandsons”. That phrasing is somewhat stronger on cancer than NCI’s own 2024 cohort paper.
• The 2006 Dutch collective settlement explicitly excluded DES grandchildren, including those born prematurely to DES daughters (Gerechtshof Amsterdam, 2006, §5.19).

Verdict: unclear. The chapter was right that this would take many years. Twenty-five years on, it is still unresolved.

Implication for weight. The claim that harms are open-ended across generations should be graded strong for the second generation and unresolved for the third. It should not be presented as established for grandchildren’s cancer risk.

Claim 8: Detection as a “fortuitous accident”#

Original claim (pp. 86–87).

• The end of DES use in pregnancy was triggered by “a fortuitous accident”.
• Had the cases been diagnosed at several different centres, “rather than at Massachusetts General Hospital (where DES use had been high because the Smiths had conducted their early experiments on DES there)”, the danger “might well have gone unrecognised”.
• The common reproductive-tract harms would then probably never have been identified.

Subsequent developments.

• Primary accounts confirm how contingent the discovery was.
• Ulfelder, a co-author of the 1971 paper, recorded that the first case was seen in 1966. The first seven cases were reported in 1970 (Ulfelder, Cancer 1976).
• He also wrote that “a clue to etiology from one mother’s observation regarding DES as a pregnancy supportive medication” was then “quickly and conclusively converted into fact” by the case–control study (Ulfelder, 1980). So the hypothesis came from a patient’s mother, not from any surveillance system.
• NCI investigators take the same view of the common harms. They wrote in 2016 that the link between DES and “subsequent adverse health outcomes, most of which are fairly common, may easily have escaped detection” (Troisi, Hatch & Titus, Pediatrics, Nov 2016).
• But registries held the cancer signal. IARC (2012) notes that the New York State Cancer Registry supplied the confirming cases within months (Greenwald et al., 1971). A California Tumor Registry analysis showed an increase in vaginal tumours in girls aged 10–19 over 1950–69 (Linden & Henderson, 1972). The data to detect the cancer existed in population registries. What was missing was the question.
• For the cancer, “might well have gone unrecognised” is therefore too strong. Recognition would have been later, not necessarily never.
• For the non-cancer reproductive harms, which were identified only through screening triggered by the cancer, the claim is more plausible.
• A factual slip. George and Olive Smith’s DES work was based at the Free Hospital for Women in Brookline, Massachusetts (Little, Obstet Gynecol Surv, July 1970; only the title and bibliographic record were checked). The chapter’s explanation that the cluster appeared at Massachusetts General Hospital because the Smiths experimented “there” therefore appears garbled. High DES use in the Boston area remains plausible.

Verdict: partly held up. The counterfactual cannot be tested. The historical record supports its spirit, especially for the common harms. But registry data show the cancer signal was available to anyone who looked.

Implication for weight. The refined lesson is sharper than the chapter’s version. In this case the bottleneck was generating the hypothesis, not the absence of data. A rare, distinctive outcome, a single referral centre and a patient’s question did the work. Surveillance that records outcomes without linking them to exposures would have held the signal without seeing it.

Claim 9: What changed in drug regulation, and what gaps remain#

Original claim (pp. 89–90).

• What changed since 1970:
• efficacy evidence is now required;
• toxicity testing is more thorough;
• “the entire field of teratology was established in response to these episodes”;
• regulators are more alert.
• Remaining gaps:
• legacy drugs that were never properly evaluated;
• inconsistent surveillance that struggles with delayed or common outcomes;
• little formal risk/benefit analysis;
• surveillance limited to gross malformations and cancers.
• Recommendation: broader surveillance and formal risk/benefit analysis.

Subsequent developments.

• Correction: teratology came first. The Teratology Society was founded in 1960 (Society for Birth Defects Research & Prevention, “About”). That was before thalidomide’s teratogenicity was recognised: McBride’s Lancet letter appeared on 16 December 1961 (Vandenbroucke, James Lind Library, 2003). DES and thalidomide greatly expanded the field but did not found it. The chapter’s p. 84 citation of “James, 1965” for the thalidomide discovery is also odd, because the discovery dates from 1961.
• Formal benefit–risk analysis has since been built into regulation.
• EU: the pharmacovigilance legislation adopted in December 2010 and applying from July 2012 (“the biggest change … since 1995”). It was strengthened again in 2012 after a high-profile drug withdrawal (EMA, legal framework).
• International: the ICH Periodic Benefit-Risk Evaluation Report, effective in the EU on 1 January 2013 (EMA, ICH E2C(R2)).
• US: FDA’s final Benefit-Risk Assessment guidance, October 2023 (FDA).
• Active surveillance: FDA’s Sentinel system, launched in 2008 under FDAAA 2007 and fully operational from February 2016 (FDA Sentinel).
• Pregnancy labelling: the Pregnancy and Lactation Labeling Rule (December 2014) removed the A/B/C/D/X letter categories and required pregnancy registry information in labels (FDA PLLR resources).
• On paper, the “little formal risk/benefit analysis” gap has largely been closed.
• The pregnancy-specific gaps persist, and are well documented.
• Of 172 drugs approved by the FDA in 2000–2010, teratogenic risk in human pregnancy was “undetermined” for 97.7%, and for 73.3% there were no pregnancy data at all. For older drugs, the mean time for an “undetermined” risk rating to become a more precise one was 27 years. The authors call this “a serious public health problem” and ask for more active post-marketing surveillance (Adam, Polifka & Friedman, Am J Med Genet C 2011).
• The federal task force on research in pregnant and lactating women found that more than 90% of pregnant women take at least one medication, while information remains “extremely limited” (PRGLAC Report to HHS and Congress, Sept 2018).
• A legacy drug shows the limits of surveillance focused on malformations. In 2018 the European Commission imposed a pregnancy-prevention programme on valproate, a long-marketed medicine, and contraindicated it in pregnancy with narrow exceptions (EMA valproate referral; EC decision 31 May 2018). The decision cites about 10% malformations but 30–40% developmental disorders in exposed children. Developmental disorders are exactly the kind of non-gross, delayed outcome the chapter warned surveillance would miss (p. 90).
• Legacy DES indications were slow to go. The DES lactation-suppression indication took 20 years to withdraw, from recommendation in 1978 to final order in 1998 (Claim 5). The DESI review still rated the pregnancy use “possibly effective” in 1971 (Claim 1). Both illustrate the chapter’s point about legacy drugs.
• An unintended consequence. PRGLAC states that thalidomide and DES “illustrated gaps in safety assessment concerns, codified in part in federal regulations, that effectively excluded pregnant women from clinical drug studies”. The caution the chapter calls for (“extreme caution … before exposing pregnant women”, p. 88) was institutionalised partly as exclusion from research. That exclusion is now itself recognised as a cause of the ignorance the chapter deplores.
• Monitoring of the DES cohort itself has wound down.
• The CDC’s DES Update education site was marked “no longer maintained” by 2021, with bibliographies last revised in January 2012. Its pages returned “not found” when checked in September 2026.
• The most recent NCI follow-up questionnaire cycle reported in a 2026 NCI analysis was in 2016 (Mitra et al., 2026), although analyses continue to be published.
• France never built a cohort (Fillion & Torny, 2021/22).

Verdict: held up. The diagnosis of the remaining gaps is well supported by later primary evidence. The recommendation on formal benefit–risk analysis has been largely adopted in regulation. The historical claim about teratology is overstated.

Implication for weight. The chapter’s institutional lessons can carry substantial weight:

• surveillance tuned to gross, immediate outcomes misses most harm;
• legacy products escape re-evaluation;
• monitoring is built after the fact and then allowed to lapse.

Later evidence also adds a caution the chapter lacks. Precautionary exclusion can create its own knowledge deficit.

Claim 10: DES as the paradigm of endocrine disruption#

Original claim (p. 88). DES is “the clearest example of human endocrine disruption”. Although pharmaceutical doses were “far higher than usual environmental levels of synthetic chemicals”, DES “remains a clear warning of the consequences of perturbing the endocrine system through synthetic chemicals”.

Subsequent developments.

• The paradigm status is now standard. Swan’s 2013 update calls DES “the archetypical environmental estrogen and endocrine disruptor”, and NCI investigators in 2026 call it “a prototypical endocrine disruptor” (Mitra et al., 2026). The WHO/UNEP State of the Science of Endocrine Disrupting Chemicals (June 2012) stresses that developmental effects may be irreversible and appear later in life (WHO).
• EU law has adopted the concern.
• Pesticides: scientific criteria for identifying endocrine disruptors, applying from 20 October 2018 (Regulation (EU) 2018/605).
• Classification: new CLP hazard classes for endocrine disruption, adopted 19 December 2022 (Delegated Regulation (EU) 2023/707). The definition covers adverse effects “in an intact organism or its offspring or future generations”, which in effect codifies the DES lesson. New substances must be classified from 1 May 2025 and existing substances by 1 November 2026.
• Bisphenol A, discussed in Late Lessons II:
  • EFSA cut the tolerable daily intake to 0.2 ng/kg bw/day on 19 April 2023, about 20,000 times lower than the 2015 temporary value (EFSA);
  • the EU then banned BPA in food-contact materials, with transition periods to 2026–2028 (Regulation (EU) 2024/3190, 19 Dec 2024).
• Extending DES to environmental exposures remains contested.
• A group of toxicologists published simultaneously in several journals the argument that synthetic endocrine-disrupting chemicals are “only weak partial agonists”, with human exposure “generally negligible” compared with natural compounds. They add that it is “not surprising” that epidemiology has failed to find effects (Autrup et al., Arch Toxicol, 8 June 2020). That argument does not apply to DES, which the chapter itself notes is about five times as potent as oestradiol (p. 84). It therefore sharpens the dose caveat the chapter acknowledges.
• Regulators also diverge. The US FDA’s position remains that “BPA is safe at the current levels occurring in foods” (page current as of 20 April 2023) (FDA). EFSA published joint reports with the EMA and Germany’s BfR to “clarify and/or resolve differences” that had emerged over its method.
• The US screening programme. Its progress was not assessed in this check, because the relevant Inspector General report could not be retrieved.

Verdict: partly held up. DES as the clearest human example has held up and is now a textbook status. Its use as a warning about environmental synthetic chemicals has been strongly taken up in EU law. The scientific transfer from pharmacological to environmental doses is still contested, and the mainstream position differs by jurisdiction.

Implication for weight. Treat this part of the section as advocacy that proved influential, not as settled inference. DES shows what developmental endocrine disruption can do and how long it takes to see. It does not by itself show that low environmental exposures do the same. The chapter’s authors acknowledged this, and later critics press the point.

What the hindsight check changes about the section’s lessons#

This table maps the digest’s transferable insights (technology-neutral) to their strength after this check. Page references are to the 2001 chapter.

The following insights were not materially tested by later evidence:

• 7: comfortable assumptions persist;
• 8: promotion and peer norms;
• 9: cheap, unowned products diffuse fast;
• 14: confidence rests on not having seen harm.

Fillion and Torny’s French findings give indirect support to 7.

Lessons the hindsight record adds, stated in technology-neutral terms:

1. Harm estimates move in both directions with follow-up. Longer and larger follow-up added harms (late CCA, early menopause, coronary disease). It also removed or weakened others: breast cancer in daughters, overall cancer, sons’ cancer, depression. Pp. 87–88, and the 2013 annex’s “capacity to harm expands over time”, need that qualification.
2. Precaution can produce its own ignorance. When caution about a vulnerable group is institutionalised as exclusion from study, the knowledge needed for later decisions never accumulates (PRGLAC, 2018; relevant to p. 88).
3. Costs fall on the exposed, over decades, and compensation lags and narrows. - In 1992 the Dutch Supreme Court ruled in favour of DES daughters on alternative causation (art. 6:99 BW), easing their burden of proving which manufacturer was responsible (Hoge Raad, 9 Oct 1992, recited in the 2006 ruling). A collective settlement was declared binding only in 2006, and it excluded grandchildren. - French recognition came mainly through courts and a 2004 maternity-leave law that arrived after most daughters had passed childbearing age (Fillion & Torny). French court decisions on burden of proof are widely reported, but the primary texts were not retrieved in this check and are not relied on here. - The chapter says nothing about costs or litigation (see the digest’s omissions).
4. Formal systems can fail at their own task. The post-1962 efficacy review existed to remove unproven indications. In 1971 it still rated DES “possibly effective” in pregnancy. Having a procedure in place is not the same as the procedure working (supports pp. 86, 90).

Method notes and limits#

• Not retrievable:
• the Cochrane review’s full numerical results;
• the French Cour de cassation decisions on DES burden of proof;
• the US EPA Inspector General’s report on the Endocrine Disruptor Screening Program.

None of these is relied on for a verdict. - Not found: post-2001 primary evidence on DES use in developing countries after 1985. - Checked by bibliographic record only: Little (1970), on the Smiths’ base at the Free Hospital for Women. - Out-of-report examples: the valproate example (Claim 9) and the brief reference to a 2012 EU pharmacovigilance amendment are pharmaceutical examples outside the reports. They are used only as tests of the chapter’s own claims about drug regulation.

Sources#

Primary regulatory, legal and agency documents

• US FDA, “Certain estrogens for oral or parenteral use … Drug Efficacy Study Implementation” (DESI 740), 36 Fed. Reg. 21537–21538, dated 5 Nov 1971, published 10 Nov 1971. https://www.govinfo.gov/content/pkg/FR-1971-11-10/pdf/FR-1971-11-10.pdf
• US FDA, “Estrogens for Postpartum Breast Engorgement; Withdrawal of Approval … Final Order”, 63 FR 69631, 17 Dec 1998. https://www.federalregister.gov/documents/1998/12/17/98-33455/estrogens-for-postpartum-breast-engorgement-withdrawal-of-approval-of-the-labeled-indication-for
• US FDA, “List of Drug Products That Have Been Withdrawn or Removed From the Market for Reasons of Safety or Effectiveness”, 64 FR 10944, 8 Mar 1999. https://www.federalregister.gov/documents/1999/03/08/99-5517/list-of-drug-products-that-have-been-withdrawn-or-removed-from-the-market-for-reasons-of-safety-or
• US FDA, “Lilly Research Laboratories et al.; Withdrawal of Approval of 28 New Drug Applications”, 65 FR 55264, 13 Sept 2000 (effective 30 Sept 2000). https://www.federalregister.gov/documents/2000/09/13/00-23477-Filed/lilly-research-laboratories-et-al-withdrawal-of-approval-of-28-new-drug-applications
• US FDA, Pregnancy and Lactation Labeling Resources (PLLR published Dec 2014; page updated 10 Mar 2025). https://www.fda.gov/drugs/labeling-information-drug-products/pregnancy-and-lactation-labeling-resources
• US FDA, Benefit-Risk Assessment for New Drug and Biological Products, final guidance, Oct 2023. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/benefit-risk-assessment-new-drug-and-biological-products
• US FDA, FDA’s Sentinel Initiative (page updated 8 Mar 2024). https://www.fda.gov/safety/fdas-sentinel-initiative
• US FDA, Bisphenol A (BPA): Use in Food Contact Application (content current as of 20 Apr 2023). https://www.fda.gov/food/food-packaging-other-substances-come-contact-food-information-consumers/bisphenol-bpa-use-food-contact-application
• National Cancer Institute, Diethylstilbestrol (DES) Exposure and Cancer (fact sheet, updated 31 Jan 2025). https://www.cancer.gov/about-cancer/causes-prevention/risk/hormones/des-fact-sheet
• National Toxicology Program, Report on Carcinogens, 15th ed. (2021), Diethylstilbestrol profile. https://ntp.niehs.nih.gov/sites/default/files/ntp/roc/content/profiles/diethylstilbestrol.pdf
• IARC Monographs Vol. 100A, Pharmaceuticals (2012), “Diethylstilbestrol”. Publication page: https://publications.iarc.who.int/118 ; chapter PDF (archived): http://web.archive.org/web/20230626060446/https://monographs.iarc.who.int/wp-content/uploads/2018/06/mono100A-16.pdf
• CDC, DES Update (archived 16 Apr 2021; marked “no longer maintained”). http://web.archive.org/web/20210416201107/https://www.cdc.gov/DES/
• NICHD, Task Force on Research Specific to Pregnant Women and Lactating Women (PRGLAC), Report to the HHS Secretary and Congress, Sept 2018. https://www.nichd.nih.gov/sites/default/files/2018-09/PRGLAC_Report.pdf ; overview: https://www.nichd.nih.gov/about/advisory/PRGLAC
• European Medicines Agency, Legal framework: pharmacovigilance (legislation adopted Dec 2010, applying July 2012). https://www.ema.europa.eu/en/human-regulatory-overview/pharmacovigilance-overview/legal-framework-pharmacovigilance
• European Medicines Agency, ICH E2C(R2) Periodic benefit-risk evaluation report (effective 1 Jan 2013). https://www.ema.europa.eu/en/ich-e2c-r2-periodic-benefit-risk-evaluation-report-scientific-guideline
• European Medicines Agency, Valproate and related substances referral (EC decision 31 May 2018). https://www.ema.europa.eu/en/medicines/human/referrals/valproate-related-substances-0
• Commission Regulation (EU) 2018/605 of 19 April 2018 (endocrine-disrupting criteria, plant protection products; applies from 20 Oct 2018). https://eur-lex.europa.eu/eli/reg/2018/605/oj
• Commission Delegated Regulation (EU) 2023/707 of 19 December 2022 (CLP hazard classes including endocrine disruption). https://eur-lex.europa.eu/eli/reg_del/2023/707/oj
• Commission Regulation (EU) 2024/3190 of 19 December 2024 (BPA and other bisphenols in food contact materials). https://eur-lex.europa.eu/eli/reg/2024/3190/oj
• EFSA, “Bisphenol A in food is a health risk”, 19 Apr 2023. https://www.efsa.europa.eu/en/news/bisphenol-food-health-risk
• WHO/UNEP, State of the Science of Endocrine Disrupting Chemicals 2012 (June 2012). https://www.who.int/publications/i/item/9789241505031
• Gerechtshof Amsterdam, 1 June 2006, ECLI:NL:GHAMS:2006:AX6440 (DES collective settlement declared binding; recites Hoge Raad 9 Oct 1992). https://uitspraken.rechtspraak.nl/details?id=ECLI:NL:GHAMS:2006:AX6440
• European Environment Agency, Late lessons from early warnings: science, precaution, innovation (EEA Report No 1/2013), Annex 2 (Tables A2.1, A2.8) and Annex 3, “DES: the view from 2013” (S. H. Swan), pp. 729–730. Annex file (archived): http://web.archive.org/web/20250812193017/https://www.eea.europa.eu/publications/late-lessons-2/late-lessons-2-full-report/late-lessons-2-annexes/at_download/file

Peer-reviewed studies and reviews

• Herbst AL, Ulfelder H, Poskanzer DC. NEJM 284:878–881, 22 Apr 1971. https://doi.org/10.1056/NEJM197104222841604
• Greenwald P et al. NEJM 285:390–392, 12 Aug 1971. https://doi.org/10.1056/NEJM197108122850707
• Brackbill Y, Berendes HW. Lancet 2(8088):520, 1978. https://doi.org/10.1016/S0140-6736(78)92237-7
• Ulfelder H. “The stilbestrol-adenosis-carcinoma syndrome.” Cancer 38:426, July 1976. https://europepmc.org/article/MED/1277101
• Ulfelder H. “The stilbestrol disorders in historical perspective.” Cancer 45:3008, June 1980. https://europepmc.org/article/MED/6992969
• Little B. “George Van S. and Olive Watkins Smith and the Free Hospital for Women, Brookline, Massachusetts.” Obstet Gynecol Surv, July 1970. https://europepmc.org/article/MED/4913533
• Hatch EE et al. “Cancer risk in women exposed to diethylstilbestrol in utero.” JAMA 280:630–634, 1998. https://doi.org/10.1001/jama.280.7.630
• Titus-Ernstoff L et al. “Long-term cancer risk in women given DES during pregnancy.” Br J Cancer 84:126–133, 2001. https://doi.org/10.1054/bjoc.2000.1521
• Klip H et al. “Hypospadias in sons of women exposed to DES in utero.” Lancet 359:1102–1107, 2002. https://doi.org/10.1016/S0140-6736(02)08152-7
• Bamigboye AA, Morris J. “Oestrogen supplementation, mainly diethylstilbestrol, for preventing miscarriages and other adverse pregnancy outcomes.” Cochrane Database Syst Rev 2003, CD004353. https://doi.org/10.1002/14651858.CD004353
• Palmer JR et al. “Hypospadias in sons of women exposed to DES in utero.” Epidemiology 16:583–586, 2005. https://doi.org/10.1097/01.ede.0000164789.59728.6d
• Brouwers MM et al. “Hypospadias: a transgenerational effect of diethylstilbestrol?” Hum Reprod 21:666–669, 2006. https://doi.org/10.1093/humrep/dei398
• Palmer JR et al. “Prenatal diethylstilbestrol exposure and risk of breast cancer.” CEBP 15:1509–1514, Aug 2006. https://doi.org/10.1158/1055-9965.EPI-06-0109
• Titus-Ernstoff L et al. “Offspring of women exposed in utero to DES: benign and malignant pathology in the third generation.” Epidemiology 19:251–257, 2008. https://doi.org/10.1097/EDE.0b013e318163152a
• Skinner MK. “What is an epigenetic transgenerational phenotype? F3 or F2.” Reprod Toxicol 25:2–6, 2008. https://doi.org/10.1016/j.reprotox.2007.09.001
• Palmer JR et al. “Urogenital abnormalities in men exposed to DES in utero: a cohort study.” Environ Health 8:37, Aug 2009. https://doi.org/10.1186/1476-069X-8-37
• Adam MP, Polifka JE, Friedman JM. “Evolving knowledge of the teratogenicity of medications in human pregnancy.” Am J Med Genet C 157C:175–182, 2011. https://doi.org/10.1002/ajmg.c.30313
• Kalfa N et al. “Prevalence of hypospadias in grandsons of women exposed to DES.” Fertil Steril 2011. https://doi.org/10.1016/j.fertnstert.2011.02.047
• Hoover RN et al. “Adverse health outcomes in women exposed in utero to diethylstilbestrol.” NEJM 365:1304–1314, 6 Oct 2011. https://doi.org/10.1056/NEJMoa1013961
• Tournaire M et al. “Cancer risk in women exposed to diethylstilbestrol in utero.” Thérapie 70:433–441, 2015. https://doi.org/10.2515/therapie/2015030
• Prescrire International. “Diethylstilbestrol (DES): also harms the third generation.” 2016. https://europepmc.org/article/MED/30758926
• Troisi R, Hatch EE, Titus L. “The diethylstilbestrol legacy: a powerful case against intervention in uncomplicated pregnancy.” Pediatrics, Nov 2016. https://doi.org/10.1542/peds.2015-4268G
• Huo D, Anderson D, Palmer JR, Herbst AL. “Incidence rates and risks of DES-related clear-cell adenocarcinoma …: update after 40-year follow-up.” Gynecol Oncol, Sept 2017. https://doi.org/10.1016/j.ygyno.2017.06.028
• Troisi R et al. “A prospective cohort study of prenatal DES exposure and cardiovascular disease risk.” J Clin Endocrinol Metab 2018. https://doi.org/10.1210/jc.2017-01940
• Kioumourtzoglou MA et al. “Association of exposure to DES during pregnancy with multigenerational neurodevelopmental deficits.” JAMA Pediatrics 2018. https://doi.org/10.1001/jamapediatrics.2018.0727
• Titus L et al. “Reproductive and hormone-related outcomes in women whose mothers were exposed in utero to DES.” Reprod Toxicol, online 27 Dec 2018. https://doi.org/10.1016/j.reprotox.2018.12.008
• Troisi R et al. “Prenatal diethylstilbestrol exposure and cancer risk in women.” Environ Mol Mutagen 60:395–403, 2019 (online 2017). https://doi.org/10.1002/em.22155
• Titus L et al. “Prenatal DES exposure and risk of depression in women and men.” Epidemiology 2019. https://doi.org/10.1097/EDE.0000000000001048
• Hom M et al. “Systematic review and meta-analysis of testicular germ cell tumors following in utero exposure to DES.” JNCI Cancer Spectrum 2019. https://doi.org/10.1093/jncics/pkz045
• Autrup H et al. “Human exposure to synthetic endocrine disrupting chemicals (S-EDCs) is generally negligible …”. Arch Toxicol, 8 June 2020 (published simultaneously in several journals). https://doi.org/10.1007/s00204-020-02800-8
• Troisi R et al. “Prenatal DES exposure and risk of diabetes, gallbladder disease, and pancreatic disorders and malignancies.” J Dev Orig Health Dis 2021. https://doi.org/10.1017/S2040174420000872
• Strohsnitter WC et al. “Prenatal diethylstilbestrol exposure and cancer risk in males.” CEBP, 16 July 2021. https://doi.org/10.1158/1055-9965.EPI-21-0234 (full text: https://pmc.ncbi.nlm.nih.gov/articles/PMC8492497/)
• Fillion E, Torny D. “Like mother, like daughter, like granddaughter … Transgenerational ignorance engendered by a defective reproductive health technology.” Reprod Biomed Soc Online 14:101–110 (online 30 Oct 2021). https://doi.org/10.1016/j.rbms.2021.10.001
• White MC et al. “Risk of clear-cell adenocarcinoma of the vagina and cervix among US women with potential exposure to DES in utero.” Cancer Causes Control, 29 June 2022. https://doi.org/10.1007/s10552-022-01598-3
• Wamakima BW et al. “Postmenopausal vaginal and cervical cancer risk related to in utero DES exposure.” J Low Genit Tract Dis (online 12 Oct 2022). https://doi.org/10.1097/LGT.0000000000000713
• Gaspari L et al. “Birth outcomes in DES children and grandchildren.” IJERPH 2023. https://doi.org/10.3390/ijerph20032542
• Bodelon C et al. “In utero exposure to DES and blood DNA methylation in adult women.” Environ Res 2023. https://doi.org/10.1016/j.envres.2023.115990
• Troisi R et al. “An update on prenatal DES exposure and high-grade squamous intraepithelial lesions of the lower genital tract.” Obstet Gynecol, 29 Feb 2024. https://doi.org/10.1097/AOG.0000000000005547
• Titus L et al. “Benign and malignant outcomes in the offspring of females exposed in utero to DES: an update from the NCI Third Generation Study.” Cancers 16:2575, 18 July 2024. https://doi.org/10.3390/cancers16142575
• Rogers RE, Fowler KA, Pask AJ. “Prenatal exposure to diethylstilbestrol has multigenerational effects on folliculogenesis.” Sci Rep 2024. https://doi.org/10.1038/s41598-024-81093-8
• Boekel NB et al. “Cancer risk after in utero exposure to diethylstilbestrol.” Eur J Epidemiol, 28 May 2025. https://doi.org/10.1007/s10654-025-01234-9
• Mitra PR et al. “Prenatal DES exposure and risk of benign breast disease: the NCI DES Follow-up Study.” Int J Cancer, 27 Jan 2026. https://doi.org/10.1002/ijc.70331 (full text: https://pmc.ncbi.nlm.nih.gov/articles/PMC13359020/)

Secondary and historical sources (used only where flagged)

• Grimes DA. “Discovering the need for randomized controlled trials in obstetrics: a personal odyssey.” James Lind Library (2007). https://www.jameslindlibrary.org/articles/discovering-the-need-for-randomized-controlled-trials-in-obstetrics-a-personal-odyssey/
• Vandenbroucke JP. “Thalidomide: an unanticipated adverse effect.” James Lind Library (2003). https://www.jameslindlibrary.org/articles/thalidomide-an-unanticipated-adverse-effect/
• Society for Birth Defects Research & Prevention (formerly Teratology Society), “About” (founding year 1960). https://www.birthdefectsresearch.org/about.asp