LL2-13 — Ch13 Ethinyl oestradiol in the aquatic environment#

Late lessons from early warnings: science, precaution, innovation (EEA Report No 1/2013), in the part headed “Emerging lessons from ecosystems”. Report pages 279–307 (chapter text pp. 279–297; Table 13.1 p. 298; references pp. 299–307). PDF pages 281–309.

Reading record: I read the whole text extract in order, through the last page marker (PDF 309 / report p. 307). I also read report pp. 279–298 visually in the PDF to check the layout, Figures 13.1–13.4, Box 13.1 and Table 13.1. I checked the authors’ biographies in the report’s contributor list (pp. 692, 696) and the places where other chapters cite this one (pp. 211, 644, 649, 660). The chapter has no panels or commentaries.

Authors and standpoint#

Susan Jobling. Her report biography (p. 692) gives her as head of the Brunel University London Institute for the Environment. She works on how environmental contaminants interfere with hormones and with the reproductive and endocrine systems of aquatic wildlife, “from the molecular level, through to the individual and population levels”. She has sat on government expert committees on chemical test methods in Europe and the US. The chapter itself names her as an actor in the story: “Some detective work by Susan Jobling (then a student of John Sumpter) at Brunel University” linked nonylphenol in effluent to effects on fish (p. 288). Much of the core evidence is her own work: Jobling et al. 1998 (widespread intersex, p. 285), 2002a/b (fertility), 2006 (exposure models match field effects, p. 282), 2009 (anti-androgens, p. 290) and Baynes et al. 2012 (sand filtration versus granular activated carbon, p. 293). She also co-edited the EEA’s Weybridge+15 report on endocrine disrupters (p. 289 fn 7; refs p. 300).

Richard Owen. His biography (p. 696) gives him as holding the Chair in Responsible Innovation at the University of Exeter Business School, and as “originally an environmental scientist by training”. Earlier he was “Head of the Environment and Human Health Programme at the Environment Agency (UK), which included research supporting policy development regarding endocrine disrupting chemicals and the UK National Demonstration Programme” (p. 696). So his Environment Agency programme included research supporting the National Demonstration Programme (NDP), which produced the chapter’s treatment and cost data. The bio does not say he ran the NDP itself; the chapter describes the NDP as a water-industry programme run “in collaboration with the Environment Agency and the UK government” and financed mostly by the water companies (p. 291). The report’s synthesis chapter (Ch. 27, p. 660) says that “Richard Owens [sic], co‑author of the Chapter 13 on the ‘birth pill’” has taken the ideas of public involvement and choice of technology further in a forthcoming book on responsible innovation.

Joint sources. The England-and-Wales cost figure (EUR 32–37 billion) is attributed to “provisional estimates by the UK government” but cited to the authors’ own Nature commentary, Owen and Jobling 2012, ‘The hidden costs of flexible fertility’ (p. 292; refs p. 304).

Standpoint. These are participant-historians. One author produced much of the scientific evidence; the other headed the Environment Agency programme whose research supported policy on endocrine disrupters and the NDP. That gives them inside knowledge: personal communications from Roger Sweeting and Derek Tinsley (pp. 282–283), details of unpublished Thames Water studies (p. 283), and the remark that the demonstration programme “was not welcomed by the UK water industry” (p. 291). It also gives them a stake in how the story is told. Their tone towards colleagues is warm: Geoff Brighty “passionately championed” the case (p. 290), and the BBC documentary was “beautifully presented” (p. 288).

Evident stance. - Pro-precaution on the evidence. “the evidence suggests that it is entirely reasonable to invoke the precautionary principle and introduce regulation to limit aquatic exposure to EE2” (p. 280). - Self-critical about precaution itself. This is the chapter’s distinctive move. The authors say the reasons regulation is hard “raise serious questions for the precautionary principle itself” (p. 280). The Rio formulation’s “cost effective” wording is precaution’s “Achilles heel” (p. 296), and defined that way precaution “can also paradoxically become a perfect excuse for inaction” (p. 296). - Gives the benefits their full weight. The pill “has allowed women to control their fertility reliably on a global scale. But it has come at a price to the environment” (p. 294; also p. 281). - Ends on questions, not a fixed programme. The lessons are framed as questions (“is the price of being precautionary simply too high?”, p. 294; “Is this option preferable?”, p. 295; “Do we care?”, p. 296). The firmest recommendations are for open public debate (p. 297) and for considering redesign of the drug at source as “a constructive precautionary approach” (p. 295).

Panels and commentaries. None. No industry or regulator responds in the chapter. Industry positions come only as the authors report them: - the chemical-industry paper Webb et al. 2003 (pp. 282, 295, 297); - EUREAU’s cost estimates and the “representations to the EC from both the water and pharmaceutical industries (EUREAU, 2012)” (pp. 293–294).

Other notes. - The word “innovation” never appears in the body of the chapter, only in the running headers (“innovative” appears once, for “innovative modelling approaches”, p. 290), even though the redesign argument (p. 295) is in substance an argument about innovation. - Other chapters cite this one: the DBCP chapter (p. 211) and the synthesis chapter (p. 644, where the EEA mislabels EE2 as “the pregnancy pill”; p. 649; p. 660).

Section-by-section notes#

Chapter summary box (p. 279)#

• Framing: “Many decades of research” show that oestrogens, both synthetic and natural, harm wildlife, including intersex in male fish that “diminishes fertility and fecundity. Although often sublethal, such impacts may be permanent and irreversible.”
• Subject: ethinyloestradiol (EE2), the active ingredient of the pill, first developed in 1938, which reaches water through wastewater treatment plants. Wildlife is exposed to “a cocktail of endocrine disrupting chemicals”, but there is “reasonable scientific certainty that EE2 plays a significant role, and at vanishingly low levels”.
• Regulatory arc:
• 2004: the Environment Agency of England and Wales (EA) accepts the evidence as enough to warrant considering risk management.
• 2012, “nearly 75 years after its synthesis”: the European Commission (EC) proposes EE2 as an EU-wide ‘priority substance’ under the Water Framework Directive (WFD).
• The proposal is then amended, “delaying any decision on a regulatory ‘environmental quality standard’ until at least 2016”.
• Why: control “will come at a significant price”. Meeting the limits means removing part-per-trillion levels from effluents “at considerable expense”.
• Framing questions: “Is this a price we are willing to pay? Or will the price of precautionary action be simply too high — a pill too bitter to swallow?” How far is society, which has enjoyed “decades of flexible fertility” and will pay for the clean-up, involved in the decision? What does this mean for the “many thousands of other pharmaceuticals” in the environment?

13.1 Introduction (p. 280)#

• Evidence claim. “A large body of scientific data built up over many decades indicates a cause‑effect link” between exposure to a “complex cocktail of chemicals” and the feminisation and demasculinisation of wildlife, especially aquatic species (citing Lyons 2008, a CHEM Trust report).
• Human-health associations. “Some scientists also relate” endocrine disruptors (alkylphenols, bisphenol A (BPA), phthalates, flame retardants) to falling sperm counts, male genital abnormalities, testicular, breast and prostate cancer, breast growth in young men and early puberty in girls. Note the hedge: “some scientists”.
• Carson lineage. Since Silent Spring (1962) we have learned that chemicals in food, household products, medicines and cosmetics “can be and are harmful to wildlife at very low levels”. Many chemicals at parts-per-trillion concentrations can together have additive effects, “producing ‘something from nothing’” (Silva et al. 2002). This has “awoken scientists and the public to an era of possible harms from low‑level, chronic chemical pollution”.
• Contrast with TBT (tributyltin). These chemicals may lack the obvious catastrophic effects of TBT, which “devastated commercially important oyster populations”. They can still cause “irreversible harm to individual organisms”. That raises “difficult questions” about how sub-lethal individual harm connects to population-level and human-health effects.
• Thesis. The precautionary principle applies. The chapter’s p. 280 paraphrase of Rio (UN 1992) reads: “where there is evidence of damage and irreversible harm, lack of full scientific certainty should not excuse inaction”. Analyst note: this paraphrase drops the “cost‑effective measures” wording (and turns “threats” into “evidence of damage”); that is the wording the authors later call precaution’s “Achilles heel” when they quote Rio in full (pp. 295–296). The thesis continues: “the evidence suggests that it is entirely reasonable to invoke the precautionary principle and introduce regulation to limit aquatic exposure to EE2. But this is no trivial matter, and the reasons for this raise serious questions for the precautionary principle itself” (p. 280).
• Scope. The story “as it has unfolded in the United Kingdom from the 1970s to today” (p. 280). The case is explicitly UK-centred.

13.1.1 The contraceptive pill (pp. 280–281)#

• Invention.
• EE2 is “the first orally active synthetic steroidal oestrogen”. It was synthesised in 1938 by Hans Herloff Inhoffen and Walter Hohlweg at Schering AG, Berlin.
• It came out of an era of hormone research inspired by insulin, thyroxine and the discovery of progesterone in 1934.
• Parallel discoveries. Edward Charles Dodds found the oestrogenic properties of diethylstilboestrol (DES, 1938) and of bisphenol A and 4-nonylphenol (1936). BPA and nonylphenol “never found uses as drugs” but went into plastics and detergents. Dating inconsistency: footnote 6 (p. 288) instead dates Dodds’s identification of BPA to 1938 (“Dodds, 1938”), and the reference list has no “Dodds et al., 1936” entry (it has Dodds & Lawson 1935 and 1938, and Dodds et al. 1938).
• Mental model of the era. “for many medical practitioners and scientists there seemed no limit to the extent to which artificial oestrogens could be put to good medical use” (p. 280).
• First market and key property. Schering marketed EE2 as Estinyl in 1943 for menopausal symptoms and female hypogonadism. It “appeared to be readily absorbed orally and very resistant to degradation and metabolism by the gut”. This stability “paved the way” for oral contraceptives in the 1950s–60s.
• Analyst note (not drawn by the authors): the same stability may matter environmentally. On p. 292 the authors say the ethinyl group “is thought to hinder EE2 sorption and metabolism” in treatment works. The property that made the drug useful may be part of why it is hard to remove.
• Symbolism. The pill was “powerfully intertwined” with concerns about overpopulation and with the 1960s “sexual revolution”. “Symbolically, it was much more than a tool for contraception”, being linked to hopes of curbing population growth and bringing “world stability” (p. 280).
• The first pill, Enovid, contained mestranol (the methyl ether of EE2) and norethynodrel.
• Confidence of the pioneers. Dr John Rock, who ran early trials: “If taken as it should be … it will stop ovulation 100 per cent” (p. 281). Dated “1967” in the text but cited to “CBC digital Archives, 1964”, an internal inconsistency.
• Mechanism. Synthetic oestrogens and progestogens bind oestrogen and progesterone receptors and alter gene expression. The pill stops ovulation, thickens cervical mucus and thins the uterine lining (p. 281).
• Diffusion.
• The FDA approved the pill “in early 1961” (p. 281). Likely error: the FDA approved Enovid for contraception in 1960 (verify).
• Enoch Powell announced NHS prescription on 4 December 1961 at “a subsidised price of two shillings per month”.
• UK users rose from about 50,000 to one million between 1962 and 1969, out of about 10 million worldwide. Time magazine cover, April 1967.
• Credited with a major role in women’s liberation and sexual freedom (Asbell 1995; Goldin & Katz 2002). It gives “unprecedented control over fertility” and avoids the mortality and morbidity of pregnancy, childbirth and termination (p. 281).
• Present scale (c. 2012).
• 32 different forms; more than 100 million users worldwide.
• A quarter of women aged 16–49 in Great Britain use a pill, against 3% in Japan, the last developed country to legalise it (1999) (p. 281).
• This contrast is used later (p. 289) as a natural comparison.

13.1.2 Evidence of environmental harm from ‘the pill’ (pp. 281–282)#

• 1970s speculation. In the 1970s “some scientists began speculating” that the pill might cause environmental problems (Tabak & Bunch 1970). Their reasoning: oral dosing is inefficient, so most of the drug passes through the body into wastewater. Treatment plants “were not designed to remove pharmaceuticals (or indeed other man‑made chemicals)”, so “the contents of our medicine cabinets” were probably reaching the environment and “eventually, even into drinking water supplies” (p. 281).
• Chemistry.
• EE2 is excreted as sulphate and glucuronide conjugates, alongside the natural oestrogens oestrone (E1), 17β-oestradiol (E2) and oestriol (E3).
• They share a mode of action, so their effects “in combination are additive”. “EE2, the synthetic oestrogen, is however by far the most potent of the four” (p. 281).
• Strength of evidence.
• Data supporting risk management for EE2, E2 and E1 “are now comprehensive and compelling” (Gross‑Sorokin et al. 2006, an EA paper; Caldwell et al. 2008).
• The mechanism of toxicity is “well understood”. The substances are released continuously and widely.
• Half-life in fresh water ranges from under a day to about 50 days in aerobic conditions (p. 281).
• Lines of evidence (p. 282).
• Effects at environmentally relevant concentrations in lab studies (Lange et al. 2001; Nash et al. 2004).
• Caged fish downstream of treatment works (Harries et al. 1996, 1997).
• A whole-lake dosing experiment (Kidd et al. 2007).
• Field extent.
• “In some UK rivers, 100 % of wild male fish” of roach (Rutilus rutilus), “a common freshwater fish of significance to anglers”, sampled 1995–2002, “had female characteristics”. Cited to Jobling et al. 1998; the citation date does not match the 1995–2002 sampling window.
• Intersex has been reported in fish in more than 10 countries.
• Exposure models “correlate well” with observed impacts (Jobling et al. 2006).
• Impacts reduce fertility and fecundity, and some are “irreversible” (Rodgers‑Gray et al. 2001) (p. 282).
• Scale of research. Risk characterisation of EE2, E2 and E1 “is possibly one of the most comprehensive for any chemical pollutant. Many millions of euro have been spent” (p. 282).
• Industry acknowledgement.
• The chemical industry itself concluded that “endocrine disruption is undoubtedly occurring in wild fish populations” and that the evidence of adverse impact is “extensive” (Webb et al. 2003).
• “The pharmaceutical industry, in particular, has carried out and funded some of the key lab studies showing that EE2 plays a key role” (Lange et al. 2001, a Schering study) (p. 282).
• Analyst note (not the authors’ framing): this is a notable feature of the case, since the producer contributed to the incriminating evidence.
• Regulatory milestones (p. 282).
• 2004: the EA concludes risk management is needed.
• 2012: the EC proposes EE2 as a Priority Substance under the WFD, “one of the most important pieces of legislation for protecting European surface and ground waters”.
• “At present, this is only a proposal.”
• July 2012 amendment: delays review of a limit until 2016 (instead of 2012), with compliance by 2027.
• Framing questions. “nearly 75 years after its initial development, during which time some of the most comprehensive and compelling evidence of environmental impact for any chemical has been amassed, a decision to regulate of EE2 is finally under serious consideration. Why has this taken so long? And what lessons can we learn?” (p. 282).
• Generalisation. EE2 is “in many ways a test case for many thousands of low‑level pollutants”, many with chronic impacts that “go beyond the acute polluting effects of past industrial chemicals” (p. 282).

13.2 Early warnings — 13.2.1 Early warnings from wildlife: the UK experience (pp. 282–284)#

• First observation, 1978.
• Tony Dearsley, a Thames Water area biologist, found eggs developing in the testes of 5 of 26 male roach during “a routine health check on a small sample of fish from the River Lea” (p. 282).
• Table 13.1 (p. 298) dates this to 1982, which contradicts the text.
• Recognition took a year. Roger Sweeting, Dearsley’s manager and Senior Scientist at Thames Water: “It was amazing to see macroscopically hermaphrodite fish that were both male and female all at the same time” (personal communication).
• Only “a year later”, after reading Jafri & Ensor (1979), did Sweeting phone David Ensor, who confirmed the finding was highly unusual.
• The normal incidence reported by Jafri & Ensor was 1 in 1,000, “more than 100 times lower” than in the Lower Lea (p. 282).
• Analyst note: Jafri & Ensor’s baseline paper was itself published in 1979, so the year’s lag partly reflects the baseline not yet being in print, not only slow recognition. The chapter does not comment on this.
• Figure 13.1 (p. 283) shows three histological sections of intersex roach gonads: an ovarian cavity in an otherwise normal testis; a severely intersex gonad after effluent exposure; and a wild intersex fish from the River Aire.
• Confirmation and a pattern. Larger samples confirmed the finding. Incidence varied with age, highest (20%) in fish over six years old. Sweeting reported this suggested a “cumulative effect with time” (Sweeting 1981) (pp. 282–283).
• Public-health concern (p. 283).
• Thames Water, “at the time a government‑owned company, was proactive” in investigating relevance to public health.
• Sweeting reasoned that “some risk of endocrine disturbance in human consumers of the water could be implied” (Sweeting 1981).
• The Thames receives discharges from many works (352 according to Williams et al. 2008) and supplies drinking water to North London; water is abstracted, used, treated and returned several times (“the water has already passed through several pairs of kidneys”).
• Unpublished rat studies (p. 283).
• Thames Water commissioned studies at Liverpool University (“never published”).
• River Lea water taken at the drinking-water abstraction point was sent by train to Liverpool. Derek Tinsley, then a PhD student and later at the EA, collected it.
• Twelve months of drinking it “induced persistent oestrous” in female rats.
• But water taken part-way through drinking-water treatment had no effect, and autopsies of water voles and wild rats near the works showed “no obvious gross reproductive abnormalities”.
• Tinsley recalls “relief that there were no signs of any effects of the treated drinking water on the rat reproductive system, and therefore no apparent risks to the consumer” (personal communication).
• Rapid reassurance for humans (p. 283).
• Thames Water presented the results to a standing committee of the Department of Health. After further studies, it “agreed that there was enough information to discount any possibility of risk to human consumers of water abstracted from the Lower Lea”.
• Authors: “It is striking that such a rapid decision was made on the basis of this small set of studies. This, as we will go on to discuss, is in stark contrast to decision‑making regarding protection of aquatic wildlife” (p. 283).
• The date of the committee decision is not given.
• The link to the pill was never made (pp. 283–284).
• “The possibility that the contraceptive pill hormone might well be causing intersex in fish was not formally stated in any of the scientific reports circulated between the government and the water industry.”
• Nor was a link made to published reports of steroid oestrogens in wastewater and drinking water (Tabak & Bunch 1970; Tabak et al. 1981; Aherne et al. 1985).
• Aherne et al. 1985 (p. 284). The authors introduce it favourably: “At least one of those studies had recognised that steroidal oestrogens could potentially have adverse effects, even if none were identified”. Aherne’s conclusion: “the concentrations found are far below therapeutic doses and there appears to be no evidence of adverse effects from reused water resources”.
• Aherne’s team “could not have known about the Thames Water studies as they were confidential”.
• Why the significance was missed (p. 284).
• “the human health agencies responsible for international drug regulation at the time usually had limited expertise in environmental issues; consideration of these was not formally required.”
• “The situation is today very different”, citing the European Medicines Agency’s 2006 guideline.
• Footnote 5: under later EMA guidance, if the predicted surface-water concentration is below 0.01 ppb and “no other environmental concerns are apparent”, no further environmental risk assessment is needed, “i.e. no action needed by a pharmaceutical company”.
• Analyst note: 0.01 ppb is 10 ng/L, 100 times the EE2 no-effect concentration of 0.1 ng/L given later (p. 291). The authors do not point this out, and they do not say whether the “no other environmental concerns” caveat would catch endocrine-active substances such as EE2.
• Measurement limits (p. 284). “until the 1990s, any concerted chemical analytical efforts to look for drugs in the environment achieved limited success”, because tools with enough separating power and detection limits at ng/L were “not commonly available”.
• Forward concern (p. 284). There is growing concern about pharmaceuticals in the environment (Kümmerer 2004, 2007; Apoteket; EEA 2010). The UK Office for National Statistics “predicts that the country’s medicine usage will more than double by 2050” (cited to a 2011 Nature letter).

13.2.2 More evidence of an environmental problem (pp. 284–285)#

• A second, independent signal.
• In the mid-1980s, fisheries scientists at the UK Ministry of Agriculture, Fisheries and Food (MAFF) — Colin Purdom, Vic Bye and Alex Scott — were asked to comment on the evidence.
• “Quite independently”, John Sumpter had found high levels of vitellogenin (VTG), a female-specific yolk protein, in the blood of male fish at an experimental fish farm rented by MAFF.
• VTG is under strict oestrogen control, so males produce it only if exposed to an oestrogen (p. 284).
• The scientists wondered whether effluent upstream, and oestrogens in the water, also explained the River Lea fish.
• Caged trout trials, late 1980s (Department of the Environment (DoE) funded). Male trout held for two weeks in effluent from Rye Meads works (which flows into the Lea) showed “a 100 000 fold increase” in VTG, reaching levels of mature females (p. 284).
• National survey, 1987–1990. Carried out by Brunel (Sumpter, Tyler) and MAFF (Purdom, Bye, Scott, Hardiman), DoE-funded. It “proved beyond doubt that oestrogenic effluents entering rivers were widespread throughout England and Wales” (Purdom et al. 1994) (p. 284).
• Knowledge held back (p. 284).
• “By the late 1980s, therefore, there was already evidence that wastewater from sewage treatment plants was having harmful effects on aquatic wildlife, and that at least one possible culprit was EE2.”
• “This information was not widely circulated beyond government and industrial organisations. Indeed, the results were not published until 1994 (Purdom et al., 1994), because of the contractual agreement between the DoE, MAFF and Brunel University.”
• Standard of proof (p. 284). “There was little action: policymakers of that era perhaps preferred to wait until the level of proof linking cause with effect was beyond reasonable doubt, reflecting a wider resistance at the time to precautionary action in the absence of higher levels of proof.” This is hedged (“perhaps”); no documentary evidence of policymakers’ reasoning is given.
• Later confirmation (p. 285).
• Research has since “proven EE2’s presence” and established that it is “highly likely” to contribute significantly (cited “Caldwell et al., 2009”; the reference list has 2008).
• Canadian whole-lake study (Kidd et al. 2007): EE2 at 6.1 (±2.8), 5.0 (±1.8) and 4.8 (±1.0) ng/L over three years “provoked a large surge in fish plasma VTG levels, followed by complete collapse of the fish population”. The species is not named in the chapter.
• Continuity claim (p. 285). “The fundamental conclusion, that EE2 and other hormones in wastewater, both natural and synthetic, are harming aquatic wildlife, particularly downstream of wastewater plants with low dilution, has not changed since the late 1980s. Only the level of uncertainty has reduced.” See the bias check: the chapter’s own §13.3 shows the culprit hypothesis moving away from hormones in the early 1990s.

13.2.3 Widespread endocrine disruption in wild fish and other wildlife (p. 285)#

• Field extent. More caged-fish trials showed oestrogenic pollution further downstream (Harries et al. 1996, 1997). UK field studies in 1995–2000 “showed unequivocally” that intersex in wild roach was widespread. It was especially common up to 10 km below medium-to-large works (serving 50,000–675,000 people) where effluent was diluted less than 10-fold (Jobling et al. 1998). Effects extended into estuaries (Lye et al. 1997, 1998; Allen et al. 1999).
• Other wildlife (reviewed in Lyons 2008): feminisation and sub-fertility reported in:
• amphibians (male VTG, ovotestes);
• reptiles (smaller phallus in alligators, shorter penis in turtles, reduced hatching);
• birds (male VTG, reproductive tract deformities, egg-shell thinning, poor parenting);
• otters and mink (shorter penile bone, smaller testes);
• seals and sea lions (implantation failure, sterility, abortion);
• cetaceans (low testosterone, hermaphrodite organs);
• polar bears (intersex features, deformed genitals, reduced cub survival).
• Explicit caveat. “In none of these cases was a link with exposure to EE2 investigated and/or proven” (p. 285). The list illustrates endocrine disruption in general, not EE2. Many of these effects are usually linked to persistent organochlorines, though the chapter does not say so.

13.2.4 Wildlife as sentinels for human reproductive health (pp. 285–288)#

• The sentinel argument. Feminising syndromes in wildlife “appeared to mirror” human male infertility, genital abnormalities and testicular cancer, grouped as Testicular Dysgenesis Syndrome (TDS). The reasoning: “If testicular dysgenesis syndrome was occurring in humans due to environmental pollutants, then genital disruption should have been found in wildlife exposed to those pollutants. This did indeed seem to be the case” (p. 285).
• Logic note: this shows consistency, not causation (affirming the consequent).
• DES precedent. From the 1940s to the early 1970s, pregnant women were prescribed DES “in the mistaken belief that it could prevent miscarriage”. Some sons had low sperm counts, undescended testicles and penile deformities (p. 285; cross-reference to EEA 2001).
• Timing. The steepest declines in male reproductive health appear to fall between the 1960s and 1980s. That coincides with pill uptake, but also with the entry of many other synthetic chemicals, the US Clean Water Act (1972) and upgrades to secondary sewage treatment (pp. 285–287).
• Candidate pathways (p. 287).
• Inadvertent pill-taking in the first trimester (Smithells 1981, “although refuted by Joffe, 2002”).
• Drinking-water residues of pill hormones and “a whole plethora of industrial chemicals”.
• Skin, inhalation and diet.
• Margel & Fleshner 2011: an ecological association between pill use and prostate cancer.
• Dose history. Early pills contained up to 100 μg of oestrogen, “at least five times higher than current formulations” (p. 287).
• Cross-country comparison.
• Allowing a 30–35-year lag, historical pill use looks higher where testicular cancer and hypospadias rates are higher (Denmark, Germany, Hungary, UK, USA) than where they are lower (Bulgaria, Finland, Italy, Poland, Portugal, Romania, Spain, Japan) (Leridon et al. 2006).
• Authors’ own caution: “These associations may well be coincidental, however, as the available evidence in support of the oestrogen theory is not entirely convincing” (p. 287).
• Anti-androgens. Recent animal studies suggest industrial anti-androgens “are more likely culprits” and could act together with oestrogens. But, quoting Sharpe (2009), “there is no human or experimental animal data to support this” (p. 287).
• Epistemic argument (pp. 287–288).
• “The lack of cause‑effect, especially in human epidemiological studies, could easily be wrongly interpreted as ruling out involvement of endocrine disruptors in TDS.”
• Complicating factors: long latency; mixtures in which each chemical alone may be insufficient. The text has a double negative (“may not be insufficient”), clearly a typo.
• “This is also why Bradford Hill’s criteria, when invoked in the 2002 WHO report on endocrine disruption (WHO, 2002), rejected the hypothesis of an impact on male reproductive function. These criteria are not suited for determining causation by environmental toxicants because such exposures are simply too complex.”
• “Perhaps we should abandon the unrealistic hope of achieving more certainty prior to policymaking about human male reproductive health and use wildlife as sentinels instead?”
• These are strong epistemological claims, asserted rather than argued. The characterisation of what WHO 2002 concluded should be checked against that report.
• Media and funding (p. 288).
• The case was “beautifully presented” in the BBC Horizon documentary ‘Assault on the male’ (Deborah Cadbury, 1993). “The world began to sit up and take notice.”
• Between 1998 and 2007 the EC invested “over EUR 150 million” in endocrine-disruption research. This underpinned test methods for existing and new chemicals, work on mixtures and vulnerable life stages, and showed “that a wide range of chemicals could have endocrine disrupting effects”.

Box 13.1 Human health concerns (pp. 286–287). No separate author is given, so it is treated as the chapter authors’. It leans heavily on Skakkebaek. - 1974. Kinloch Nelson & Bunge found only 7% of men awaiting vasectomy had sperm concentrations above 100 million/ml, against 65% in MacLeod & Heim (1945). They speculated about “an environmental factor to which the entire population has been exposed”. - Pushback. MacLeod & Wang (1979) “discredited” the study, citing analytical errors, while acknowledging some decline in fertile men since the 1930s. They proposed 20 million/ml as the lower limit of normal, adopted by WHO in 1980. - 1992. Carlsen, Skakkebaek et al.: a meta-analysis of 61 studies found average sperm density fell from 113 million/ml (1940) to 66 million/ml (1990). “Debate on this issue was intense and opinion was divided” (Swan 1997; Paulsen 1996; Fisch 1996). - Cohort approach. In seven European countries and more than 4,000 young men, in most countries 20% or more had counts below 20 million/ml, with averages of 45–65 million (Jorgensen et al. 2006). This is “consistent with sperm counts having fallen” and suggests subfertility “is likely to be a common issue for current and future generations”. - TDS link. Semen quality is linked to rising testicular cancer and to cryptorchidism and hypospadias, which often co-occur and may share a foetal-origin pathology (Sharpe & Skakkebaek 1993). Figure 13.2 is a schematic: environmental factors including endocrine disrupters, plus genetic defects, lead to testicular dysgenesis, then disturbed Sertoli and Leydig cell function, then reduced semen quality, testicular cancer, hypospadias and maldescent. - Unresolved causes. “Unlike in the case of intersex in fish, there has never been one widely accepted theory regarding the cause of the decline in male reproductive health”, only “a bewildering array of hypothetical culprits” (p. 287). Research is hampered by: - lack of human exposure data; - absence of past endocrine testing of industrial chemicals; - poor animal-to-human translation (“a chemical may reduce sperm counts by 80–90 % in rats before male fertility is affected”). - Strongest (advocacy) claim. “no systematic effort has been made to prevent infertility”. “the official WHO response to male infertility has been to redefine the ill people as ‘normal’” by lowering the reference value from 60 million/ml (1940s) to 20 (1980) to 15 (WHO 2010), “making it of little use in helping society recognise that there is a problem” (Skakkebaek 2010) (p. 287). - Analyst note (background knowledge, not in the chapter): this is a contestable reading. WHO reference limits are statistically derived from distributions in fertile men. The box offers no counter-view, though it does cite studies that disputed the decline (Paulsen 1996; Fisch 1996, “no decline in quality”). - Chronology conflicts with Table 13.1. The table lists the Nelson & Bunge study under 1979, with “low sperm counts in 93 % of men”. It lists the 1992 meta-analysis as showing a “50 %” decline; the box’s own figures (113 to 66) give about 42%.

13.3 The hunt for the culprit chemicals (pp. 288–289)#

• Doubt about hormones. Lab studies showed male fish “extremely sensitive” to EE2 at low ng/L (Sheahan et al. 1994). “Until the mid‑1990s, however, few were convinced that this or any other hormone was present in wastewater in sufficient amounts to cause the effects seen in fish” (p. 288).
• The industrial-chemicals turn.
• Information from the US showed that everyday industrial chemicals could mimic oestrogens. Dodds had shown this in the 1930s; John McLachlan’s first meeting in 1979 spread awareness.
• “It now seemed possible that effects seen in aquatic and other wildlife were actually more likely to be a result of exposure to cocktails of ‘endocrine‑disrupting’ industrial chemicals … than to the contraceptive pill hormone” (p. 288).
• Theo Colborn (1988, Great Lakes): persistent chemicals pass from top-predator mothers to offspring.
• The 1991 Wingspread meeting (21 scientists, 15 disciplines) coined “endocrine disruption”.
• Nonylphenol.
• Ana Soto (1991) showed nonylphenol was oestrogenic.
• Jobling’s “detective work” found that Walter Giger had measured these chemicals in effluent, sludge and river water (1984). She confirmed the concentrations were oestrogenic to fish (1993, 1996).
• Dodds had reported 4-nonylphenol as oestrogenic in 1936. “The detergent industry may have been unaware of this literature when it embarked on the largescale manufacture and sale of nonylphenol ethoxylates as detergents in the 1940s” (p. 288). An example of knowledge that existed but was not connected.
• Resolution.
• Nonylphenolics were “causing at least part of the problem in wild fish in some, but not all, areas” (Sheahan et al. 2002a,b).
• Fractionating effluents and screening the fractions in the lab showed that the natural and synthetic steroid oestrogens (EE2, E2, E1) “were in fact the most potent oestrogenically active substances present in domestic effluents” (Desbrow et al. 1998; Routledge et al. 1998; Snyder et al. 2001).
• “Of these, EE2 was by far the most potent: the pill was once again under the spotlight” (p. 289).

13.4 Government and industry action in the 1990s (p. 289)#

• Alkylphenols. “Enormous efforts” went into assessing and managing their risks, for endocrine effects and wider aquatic toxicity.
• UK voluntary agreement. On the recommendation of the UK Chemicals Stakeholder Forum, the government negotiated an agreement with suppliers of nonylphenols, octylphenols and their ethoxylates. Suppliers agreed:
• not to promote octylphenol as a substitute (avoiding a like-for-like harmful substitution);
• not to make or import new formulations containing them;
• to reformulate existing products “as a matter of urgency”.
• Authors’ verdict. “This was a constructive application of the precautionary principle (Lokke, 2006), which has not been extended to EE2” (p. 289).
• Dating note: the Forum was set up around 2000, so “1990s” may be loose (verify).
• EU. An EU risk assessment led to restrictions on nonylphenol across Europe (p. 289). Table 13.1 says nonylphenol and its ethoxylates were “banned” in 2003, stronger wording than “restrictions” in the text.
• Policy fora.
• The Weybridge workshop, December 1996, sponsored jointly by the EC, EEA, European Centre for Environment and Health and WHO, highlighted “continued uncertainty”.
• The EU strategy on endocrine disrupters (1999) followed, “but still no action was taken on EE2” (p. 289).
• Footnote 7: a 2006 tenth-anniversary Weybridge conference (Academy of Finland, EC DG Research, EEA), whose papers the EEA updated and published (EEA 2012). The reference list (p. 300) shows that this Weybridge+15 report was “edited by Sue Jobling and Jorma Topari”; the footnote itself does not mention her.

13.5 The last decade of research (pp. 289–290)#

• Global extent. Oestrogenic effluents have been found in Denmark, Germany, the Netherlands, Portugal, Sweden, Switzerland, China, Japan and the US. Endocrine disruption in fish has been found in rivers, estuaries and oceans (p. 289).
• Determinants. Incidence and severity depend mainly on the size of the works (treatment type also matters) and on dilution. Effects are lesser in large, high-dilution rivers (common in the US and parts of Europe) than in small, low-dilution rivers (common in the UK) (p. 289).
• Japan as a comparison. Pill use is “probably the lowest anywhere in the developed world” and fish endocrine disruption is “reported to be quite rare”. Per-capita dilution is five times the UK’s. Even with English levels of pill use, widespread disruption “would still not be predicted” (Johnson et al. 2012; the reference list has 2011) (p. 289).
• Irreversibility (pp. 289–290).
• Early-life exposure produces fish with both an oviduct and a sperm duct. “Depuration in clean water does not correct this condition”, so feminised ducts are “likely to be a permanent feature” (Rodgers‑Gray 2001).
• Prevalence and severity rise with age (Jobling et al. 2006).
• In extreme cases there is a 100% female population after three years of continuous exposure to effluent or EE2 (Lange et al. 2009).
• Population-relevant effects (p. 290).
• A full fish life-cycle test “carried out by the original manufacturers of EE2, Schering” (Lange et al. 2001) showed breeding inhibited at concentrations “exceeding the environmentally relevant concentration of 2 ng/L” and “full sex reversal of males producing an all female population at 4 ng/L”. Note that the chapter itself calls 2 ng/L “environmentally relevant”; it gives no figures for typical river concentrations.
• This was “corroborated” by Kidd et al. (2007): population collapse at about 5–6 ng/L.
• Table 13.1 (p. 298) wrongly gives the Schering sex-reversal threshold as “>2 ng/L” and dates Kidd to 2008.
• Better measurement, more doubt (p. 290).
• Better analysis and modelling found other steroid oestrogens, such as equine oestrogens from hormone replacement therapy (Tyler et al. 2009), and gave more accurate measurements and “credible modelled estimates” (Hannah et al. 2009). From my background knowledge, not stated in the chapter, this exposure assessment had many co-authors from pharmaceutical companies.
• “This has caused a mixture of doubt and amazement at the possibility that EE2 could be causing endocrine disruption at the very low concentrations at which it is present.”
• Analyst reading: this is a candid admission of a gap between measured concentrations and the lab effect thresholds. The sentence does not say whose doubt it is, or give the measured concentrations.
• Mixtures (p. 290). Risk assessments “are not adequately protective if based on data for individual substances”. EE2, E2 and E1 each at or below their individual ‘no effect’ level are “particularly potent when present in combination” (Silva 2002; Thorpe 2003; Brian 2005, 2007), including with anti-androgens. “These findings confirmed concerns originally mooted by Rachel Carson in 1962.”
• Anti-androgens (p. 290). They occur almost everywhere alongside oestrogens in UK effluent. Statistical modelling suggests these chemicals, “although not yet identified”, “could play a pivotal role” in feminisation. “Until now it was thought that such effects were caused only by oestrogens” (Jobling et al. 2009).
• Progestins (p. 290). Synthetic progestins are found in natural waters and affect fish at ng/L. “Ironically, despite the history of their combined use, oestrogens and progesterones have yet to be tested in fish ‘in combination’, despite the fact that human females have been carrying out this ‘test’ for 50 years.”

13.6 From risk assessment to risk management (pp. 290–294)#

• UK acceptance, 2004 (p. 290). “some 25 years after initial observations of intersex in fish”, the UK government “accepted the weight of evidence” that EE2, E2 and E1 in combination posed significant risk to aquatic life (Gross‑Sorokin et al. 2006).
• “The long journey to this point was passionately championed by Geoff Brighty, Science Manager at the Environment Agency”, whose team “built and defended the evidence‑based case”.
• Brighty, 2004: “We now have enough data to act as a policy trigger for taking action” (ScienceBlog 2004).
• Choosing where to intervene (p. 290).
• “End‑of‑pipe treatment of effluent by water companies was chosen as the risk management approach”, rather than alternatives “such as pharmaceutical industry action to develop substitutes for the active ingredient (EE2) in the pill”.
• The reasons, hedged as “may have partly reflected”: the pill’s public-health benefits; the fact that natural oestrogens also had to be removed; and the co-benefit of removing other priority hazardous chemicals.
• Consequence: “This would place the responsibility for risk management on the water industry and (ultimately) the cost of treatment of the tax paying public” (pp. 290–291). Analyst note: water services in England and Wales are paid for through water bills, so “tax paying public” is the authors’ shorthand.
• Standard-setting (p. 291).
• In 2007 the EA began a draft environmental quality standard (EQS) for EE2 based on a predicted no-effect concentration (PNEC) of 0.1 ng/L (Young et al. 2004).
• The WFD was identified as the vehicle: either a national ‘specific pollutant’ or an EU-wide ‘priority substance’. The WFD had “already been used to regulate other endocrine disrupting chemicals such as TBT and nonylphenol”.
• “At this point, no other country in the world had Environmental Quality Standards for any of these substances.”
• Footnote 8: in the US the FDA regulates pharmaceutical manufacture and the EPA regulates discharges; these limits “may not be strict enough to protect the environment because they are technology-based and do not rely on environmental data”, and “there is no treatment technique currently available that comprehensively deals with EE2”.
• Figure 13.3 (p. 291) maps English and Welsh rivers where the combined steroid oestrogens (in EE2 equivalents) exceed the 0.1 ng/L PNEC (red) or fall below it (green). Visually, red dominates much of the Midlands, the North West, the London/Thames area, much of eastern England and many urbanised catchments; green covers mainly upland and rural Wales, the South West and the North. The caption says “UK rivers”, but the map shows only England and Wales.
• Risk evaluation (p. 291). Before managing the risk, “risk evaluation” weighs the wider costs to society against the benefits, for example through a regulatory impact assessment. For the oestrogens, removal efficiency had to be quantified “in terms of both financial and carbon costs”.
• National Demonstration Programme (NDP) (p. 291).
• From 2004 the UK water industry worked with the EA and government, “under the watchful eye of the independent auditor Ofwat”.
• Budget GBP 25–40 million. It “was not welcomed by the UK water industry, which financed most of the costs”. All ten water companies in England and Wales took part.
• Phase 1: 14 works testing conventional treatment (primary sedimentation; secondary treatment by nitrifying or non-nitrifying activated sludge or biological filtration; sometimes tertiary sand or biologically aerated filtration).
• Phase 2: two full-scale plants testing granular activated carbon (GAC).
• Phase 1 results (mid-2008; UKWIR 2009) (p. 292).
• Conventional treatment removed E1 and E2 well, especially nitrifying activated sludge, but “EE2 was far harder to remove”.
• The best process removed about 54%. Tertiary steps removed a further 0–38% of what remained.
• This is consistent with the Racz & Goel (2010) review: long retention times (over 10 days) and nitrification help because slow-growing bacteria and a more diverse microbial community develop; membrane bioreactors are effective but expensive.
• Why EE2 persists (p. 292).
• “The ethinyl group of EE2 is thought to hinder EE2 sorption and metabolism.”
• EE2 is often present below the concentration needed to support bacterial growth.
• Biodegradation studies have been run at higher-than-environmental concentrations “due to limitations in analytical chemistry techniques” and “may therefore lead to erroneous conclusions”.
• Result: the 0.1 ng/L PNEC “is exceeded in many UK final effluents … irrespective of conventional treatment type”.
• Phase 2 results (May 2010) (p. 292). GAC produced effluent below the PNEC, “with no significant induction in fish VTG or intersex” (Filby et al. 2010; Baynes et al. 2012). One study found reproduction “slightly, but significantly impacted” by GAC-treated effluent. “GAC suffers from a fundamental problem: it is expensive to implement.”
• GAC costs (p. 292).
• Capital: over EUR 3 million for a 50,000 population-equivalent works; over EUR 8 million for a 250,000 works (Swindon was a test site).
• Operating: EUR 800,000 a year for a 250,000 works, depending on carbon life.
• Carbon: about 14 kg CO2 per person per year.
• England and Wales total: EUR 32–37 billion for about 1,360 works needing extra treatment. Attributed to “Provisional estimates by the UK government” but cited to Owen & Jobling 2012. The time basis (capital only, or capital plus operation over some period) is not stated.
• Racz & Goel quoted: advanced treatments “may be effective, their current capital and operation costs may make them not viable options”.
• Implementers’ reaction (p. 292). They “increasingly began to worry about two important considerations: technical feasibility and disproportionate cost. It confirmed what they had suspected for many years: EE2 was potent and hard to get rid of.”
• Benefits side (pp. 292–293).
• Costs “must be balanced against the (often intangible) benefits of risk management to key stakeholders, such as the angling community, and the public, who will ultimately have to pay for it”.
• “This has been a highly contested and often acrimonious area of debate, with highly charged discussions about what constitutes harm at the level of individual fish and what it means for (more ecologically important) fish populations.”
• Attribution of cost (p. 293). “Adding to this debate is the fact that the costs associated with removing EE2 and E2 should not be seen in isolation”: GAC would also “quite possibly” remove other substances “posing risks to the environment and requiring control”. “Why should such removal costs not be considered for all such chemicals as a whole: is it scientifically incorrect to blame just EE2 for the costs?” In context this is a co-benefits point: charging the whole treatment cost to EE2 overstates the price of controlling it. It is not presented as an argument by opponents of regulation.
• EU-wide cost (p. 293). The EC estimated EUR 11–18 per person per year (EC 2012). EUREAU, the European federation of national water and wastewater associations, estimated 25–50% of current annual sewerage charges. “There is clearly considerable uncertainty here.”
• Country variation (p. 293).
• Costs vary with population density, existing treatment and river size, and may be “substantially lower” than in the UK elsewhere (EU project Neptune; Swiss Micropoll; Eawag 2009).
• Mainland Western European plants are generally better. Germany mostly has three stages, some four; the UK most commonly two.
• A third stage (GAC, mild ozonation or “simple sand filtration”) could cut oestrogen concentrations and effects dramatically.
• Baynes et al. (2012), “using biological effects rather than chemical concentrations as measures of effective risk reduction”, found sand filtration after activated sludge “almost as effective as GAC at preventing the feminisation of male fish albeit it was two thirds cheaper” (p. 293). Jobling is a co-author.
• Stall, then the EU proposal (p. 293).
• “For a number of years the decision of whether or not to regulate EE2 appeared to stall in the still waters of cost‑benefit analysis.”
• In January 2012 the EC proposed a revised priority-substances list including E2 and EE2.
• The proposed EE2 EQS was 0.035 ng/L as an annual average in inland surface waters.
• If adopted (first European Parliament Environment Committee vote in November 2012), it could enter the 2015 river basin management plans, with enforcement by 2021.
• Figure 13.4 (p. 293) (Taylor et al. 2005, EA survey). Intersex was present at 44 of 51 English river sites (86%), with aggregate incidence of 23% of sampled males. The map covers the North East, Anglian, Midlands, Thames and Southern regions.
• Industry representations and deferral (p. 294).
• The decision “remains a proposal requiring agreement”. It “may be stalled by recent representations to the EC from both the water and pharmaceutical industries (EUREAU, 2012)”.
• July 2012 amendment (draft report by MEP Richard Seeber, footnote 9): “It is appropriate not to specify the EQS for certain substances of pharmaceutical relevance that have been added to the list of priority substances”. The EQS should be proposed “in the next review of the list in 2016”, “with the aim of meeting the EQS by 2027”.

13.7 Late lessons (pp. 294–297)#

Framing (p. 294). - EE2 “has allowed women to control their fertility reliably on a global scale. But it has come at a price to the environment.” - Mixtures of oestrogens seriously affect wildlife. The impacts are “associated with early life exposure but manifest themselves later in adult life”, “often sub‑lethal but may be permanent and irreversible”, and “may also serve as sentinels for impacts on human health”. - “Since wildlife is exposed to a cocktail of endocrine disrupting chemicals, it is naive to conclude that EE2 alone is the culprit.” But EE2 is the most potent, occurs widely, and “can frequently exceed” the 0.1 ng/L PNEC. - “There is reasonable certainty, based on sufficient scientific evidence, that EE2 plays a significant role.” - The need for risk management has been accepted by the EA and, through its proposal, the EC. - “A critical question is whether we as a society are willing to pay a potentially very high premium to be precautionary.” Or “would we, as a society, prefer to live with the impacts …; are they acceptable risks?” And “To what extent is society, which ultimately bears the benefits of flexible fertility but also the costs of cleaning up its unintended consequences … having a say on this decision?” - The lessons are said to apply to “many other low‑level chemical pollutants … with sublethal effects, alone or in combination”.

Lesson 1: “for low-level pollutants in the environment, is the price of being precautionary simply too high?” (pp. 294–295) - The EQS of 0.035 ng/L applies “in the water body itself, not in the final effluent” (p. 294). But the NDP shows that reducing effluent levels enough to comply, especially where dilution is low, “will be extremely difficult”. - GAC “might achieve this, but it is expensive and may have a potentially large carbon footprint”. Mild ozonation may also work but has high capital costs (p. 294). - Footnote 10 (p. 295), Switzerland. Some countries are upgrading anyway, “even in the absence of regulation of EE2 and E2”, because of pharmaceuticals, pesticides and endocrine disruptors in general. Full-scale treatment is planned at more than 100 Swiss plants treating about 80% of municipal wastewater. The first full-scale ozonation plant (spring 2013) will cost “only 5 EUR/person/year” including capital and operation, “because sand filtration already exists on this plant”. - Analyst note: this footnote cuts against the “price too high” framing. It shows how much cost depends on existing infrastructure and on treating micropollutants together rather than one substance at a time. The main text does not fully bring this in. - Monitoring (p. 295). - Measuring compliance at such low levels is “hardly routine”. Enzyme-linked immunoassay (ELISA) methods have been “fraught with issues around selectivity and sensitivity”. - “it will be costly to remove EE2 to enable compliance with the target EQS, and costly to monitor and demonstrate legal compliance itself. So while it is technically possible to develop an EQS, in practice it may be very hard to implement.” - Widespread contamination makes this a Europe-wide cost. - Substitution at source (p. 295). - Because EE2 is both the most potent oestrogen and the hardest to remove, “could there be an argument for substituting this as the active ingredient in the contraceptive pill, leaving the remaining oestrogens to be removed conventionally at less cost? But what would replace it?” - “If there were sufficient demand or need, the pharmaceutical industry could continue to provide women with access to birth control and reproductive choice while substantively altering the design of pharmaceuticals to protect the environment from unnecessary harm. This would represent a constructive precautionary approach.” - Footnote 11: research on oestrogen-free formulations such as the progestogen-only pill (Lakha et al. 2007). - Closes: “Is this option preferable?” - Gap: the chapter’s own evidence that progestins affect fish at ng/L (p. 290) complicates the substitution idea, and it is not reconciled here.

Lesson 2: “low-level pollutants with sublethal effects present fundamental issues for the precautionary principle” (pp. 295–296) - Rio definition quoted: “Where there are threats of serious or irreversible damage, lack of full scientific certainty shall not be used as a reason for postponing cost‑effective measures to prevent environmental degradation” (p. 295). - Footnote 12: the EC 2000 Communication (“reasonable grounds for concern”; cost–benefit “is not simply an economic cost-benefit analysis”; “transparent”). - Cross-reference to the report’s precautionary-principle chapter and the EEA working definition. - Seriousness. “It is not in dispute that intersex in fish represents both serious and irreversible damage to fish and that exposure to environmentally relevant levels of EE2 has adverse effects on fish reproduction, an ecologically relevant measure of impact.” - “Uncertainties remain, particularly around fish population level effects”, but precaution exists to act despite them. - “the level of scientific certainty concerning the risks … posed by oestrogens as mixtures … is extremely high” (p. 295). - The “so what?” response. “the water and chemical industries (and indeed some scientists) have increasingly asked the question, ‘so what?’ There may be male fish with eggs in their testes and this might be unpleasant, irreversible and widespread. But does it seriously damage fish populations? (Webb et al., 2003). Why pay potentially vast sums …?” (pp. 295–296). - “These populations might collapse in the future, as indicated in the Great Lakes study of EE2 (Kidd et al., 2007), but maybe they will not. Is intersex an unpalatable but acceptable harm?” (p. 296). - Error: Kidd’s study was at a Canadian experimental lake (correctly described on p. 285), not the Great Lakes. - Source note: the only citation for the “so what?” position is Webb et al. 2003, the same chemical-industry paper the chapter cites on p. 282 for the concession that endocrine disruption is “undoubtedly occurring”. So in the cited evidence the concession and the challenge to its significance come together in one 2003 paper, not as successive stages. The attribution to the water industry and “some scientists” is not separately sourced. - Achilles heel. “even if the seriousness of the harmful threats is accepted, the precautionary principle’s Achilles heel, however, as defined in the Rio Declaration, lies in the words ‘cost effective’” (p. 296). - The authors add that for oestrogens (and “quite possibly” many other chemicals with sublethal effects at very low levels) any risk management “is likely to be very costly, and as we have discussed above, this may be a price too high to pay” (p. 296). They leave open that the cost may genuinely be too high. - The core argument (p. 296). - High costs and “protracted debates about what constitutes acceptable harm” have seriously delayed decisions. This has been “compounded by a precautionary principle whose definition includes issues of disproportionate cost and cost effectiveness, either explicitly or implicitly”. - “Defined in this way the precautionary principle may be logical and rational, but it can also paradoxically become a perfect excuse for inaction or, at best, seriously delayed action.” - Footnote 13: the EC 2000 “pros and cons” examination “should include an economic cost benefit analysis where this is appropriate and possible” but “cannot be reduced to an economic cost-benefit analysis”. The EEA’s working definition uses “pros and cons” to include things that cannot be quantified, such as the “melt down of public trust” in the BSE saga. - Delay. Eight years after UK recognition and “some 30 years after the first observations”, regulation is still undecided and would not take effect “until at least 2015”. “Such long delays have, one might argue, been completely within the spirit of the precautionary principle at least as defined in the Rio declaration.” Decisions “should expedite decision‑making in the face of uncertainty. But in reality decision‑making has been painfully slow … This is a bitter pill to swallow” (p. 296).

Lesson 3: “the need for an open debate on precaution and decision-making” (pp. 296–297) - Scale of the chemicals problem (p. 296). - The European Inventory of Existing Commercial Chemical Substances (EINECS) lists more than 100,000 compounds, with “little known about the toxicity of about 75 %”. - Several hundred new substances are marketed each year after basic testing; the European List of Notified Chemical Substances (ELINCS) holds about 2,000. - REACH (2007) is addressing about 30,000 substances. Assessing the whole “‘chemicals universe’ is unfeasible”, and “we do not have the tools to fully analyse how mixtures of these chemicals behave”. - Precaution as exposure reduction. “The only logical way forward seems to be to reduce exposure as much as possible — to be precautionary. But this comes at a price and raises ethical questions of where responsibility should lie” (p. 296). - Water industry through treatment, or pharma through “new, less harmful contraceptive pills”? “Either way, are we as a society, prepared to pay for it? Do we care?” - Body burden (p. 296). “The average fish in a stream or person in the street now has hundreds of novel compounds in their bodies that were not there 60 years ago … We have changed the chemical environment of the developing organism. EE2 is a perfect case study of how we are responding as a society.” - Silent public (p. 297). “A key observation from this case study is that the public has been and continues to be silent witnesses.” - Previous lessons (EEA 2001; Gee 2006; Lokke 2006) say precaution “must encourage public participation” so that costs of action and inaction are debated and value judgements made “in an open and democratic way”. - Cited in support: the EC 2000 Communication (“All interested parties should be involved to the fullest extent possible …”); the US National Academy of Sciences (2009) on stakeholder involvement at all stages; the Royal Commission on Environmental Pollution (RCEP) 1998, Setting Environmental Standards (early understanding of people’s values; transparency). - The chemical industry shares this view (Webb et al. 2003). - Regulator–polluter deals. RCEP: “it is no longer acceptable for decisions to be negotiated privately between the regulator and polluter”. The UK government fully endorsed this, “But this endorsement is yet to translate into action.” Existing opportunities (footnote 14: WFD river basin plans) “are insufficient” (p. 297). - Closed process. EE2 decisions have been taken “in a poorly understood, closed process that has little engaged the public”. Decision-makers need to understand “acceptability of risk, appetite for precaution and the willingness to pay”. Public views “have to date however gone largely undocumented” (p. 297). - Concession and conclusion (p. 297). “Without public support the costs … may be seen by policy makers as disproportionate. It might be that the weight of evidence suggests that public opinion is not on the side of risk management, that we are prepared to live with endocrine disruption in the environment as collateral damage associated with flexible fertility in our own species. But what is very clear is that without asking the public it will be far easier to come to a conclusion based largely on costs alone. This loads the dice before they are thrown.”

Table 13.1 Early warnings and actions (p. 298)#

A chronology from 1938 to 2012. It is useful, but contains several discrepancies with the chapter text:

These do not change the overall story, but they matter for anyone citing dates or numbers. Use the text, and ideally the primary sources, rather than the table.

References (pp. 299–307)#

• The references are heavy on the authors’ Brunel and Exeter circle (Jobling, Sumpter, Tyler, Filby) and on EA reports (Gross-Sorokin, Young, Williams, Taylor).
• Industry sources: Webb et al. 2003 (chemical industry, Pure and Applied Chemistry) and Lange et al. 2001 (Schering), as the chapter says. Hannah et al. 2009 and Caldwell et al. 2008 (a paper deriving an aquatic PNEC for EE2) also had largely industry co-authors, according to my background knowledge rather than the chapter; verify. The chapter treats both as neutral sources. Note that the chapter’s own PNEC (0.1 ng/L) is cited to the EA’s Young et al. 2004, not to Caldwell et al. 2008.
• NGO source: Lyons 2008 and Sharpe 2009 (CHEM Trust reports).
• Several items are grey literature or hard to trace: personal communications, “Sweeting, 1981, Report to Thames Water”, “Science Blog 2004”, “EUREAU initial position paper”, and the Mistrapharma newsletter.

Case timeline#

Early warnings (who, what, how strong).

Responses. - Water industry. Thames Water, while government-owned, was proactive about human health (p. 283). The industry later financed most of the GBP 25–40 million NDP but “not welcomed” it (p. 291). Through EUREAU it gave higher cost estimates (25–50% of sewerage charges) and made representations to the EC (pp. 293–294). - Department of Health. Quickly discounted risk to drinking-water consumers on a small set of studies (p. 283). - DoE and MAFF. Funded key 1980s research, but contracts held back publication until 1994; “little action” (p. 284). - Drug regulators. Limited environmental expertise; environmental assessment not required until later (EMA 2006) (p. 284). - Chemical industry. Acknowledged endocrine disruption in wild fish and questioned its population significance (“so what?”), both cited to the same paper, Webb et al. 2003 (pp. 282, 295). Accepted voluntary phase-out of alkylphenols in the UK (p. 289). - Pharmaceutical industry. Schering funded and conducted a key life-cycle study (pp. 282, 290). Made representations to the EC alongside water companies (p. 294). Substitution of EE2 was not required of it (p. 290). - Scientists. Brunel, MAFF and EA scientists built the evidence. Some scientists joined the “so what?” questioning (p. 295). Human-health scientists were divided (Box 13.1). - Media. BBC Horizon, 1993, “The world began to sit up and take notice” (p. 288). - UK regulators. The EA accepted the need for risk management in 2004, championed by Brighty. A draft EQS was prepared in 2007 on a 0.1 ng/L PNEC (pp. 290–291). - EU. Research funding of more than EUR 150 million, 1998–2007 (p. 288). Weybridge 1996 and the 1999 strategy (p. 289). Nonylphenol restrictions (p. 289). The 2012 priority-substance proposal (EQS 0.035 ng/L), then the July 2012 amendment deferring the EQS to 2016 and compliance to 2027 (pp. 293–294). - Public. “silent witnesses”; views “largely undocumented” (p. 297). - Courts. None mentioned.

When action came. As of writing (2012), no binding EE2 standard in the UK or EU. The chapter says that when the EA began drafting its EQS (2007), “no other country in the world had Environmental Quality Standards for any of these substances” (p. 291); it does not make a worldwide claim for 2012. The UK had accepted the risk (2004) and run a demonstration programme; the EU proposal was pending and deferred. Action on the co-contaminants (alkylphenols) came much earlier through UK voluntary agreement and EU restrictions (p. 289). Switzerland was upgrading treatment for micropollutants generally (p. 295 fn 10).

Harms and costs. - Harm to fish. - Intersex at 86% of 51 English sites, 23% of sampled males (Fig. 13.4, p. 293); up to 100% of males in some rivers (p. 282). - Reduced fertility and fecundity (p. 282); permanent duct feminisation (p. 290). - Population collapse in the experimental lake at about 5–6 ng/L (p. 285). Real-world population effects are uncertain (p. 295). - Other wildlife. Effects listed, but no EE2 link investigated (p. 285). - Human health. Hypothesised only; the oestrogen theory is “not entirely convincing” (p. 287). - Research costs. “Many millions of euro” (p. 282); EC more than EUR 150 million (p. 288); NDP GBP 25–40 million (p. 291). - Mitigation cost estimates. - England and Wales: EUR 32–37 billion for about 1,360 works (p. 292). - EU: EUR 11–18 per person per year (EC) against 25–50% of sewerage charges (EUREAU) (p. 293). - Switzerland: about EUR 5 per person per year at one plant with existing sand filtration (p. 295). - Carbon: about 14 kg CO2 per person per year for GAC (p. 292).

What was known when (per the chapter). - By the early-to-mid 1980s: the anomaly in fish; that effluent-affected river water was oestrogenic to rats (study undated); that steroidal oestrogens were in water (Tabak 1970, 1981; Aherne 1985). The link to the pill was never formally made (pp. 282–284). - By the late 1980s: oestrogenic effluent was widespread and harmful to fish, with EE2 “at least one possible culprit” (p. 284), though in the early 1990s many thought industrial chemicals the more likely cause (p. 288). - By 1998: steroid oestrogens identified as the main cause, with EE2 the most potent (pp. 288–289). - By 2001–2007: permanence, life-cycle effects and population collapse at ng/L shown (pp. 285, 290). - By 2008–2010: it was known that conventional treatment could not reliably reach the PNEC and that GAC could, at high cost; sand filtration looked nearly as effective and cheaper (pp. 292–293).

Lags (computed from the chapter’s own dates).

The authors’ own lessons and conclusions#

Lessons derived from the evidence 1. EE2, with other oestrogens as mixtures, has caused serious, often sub-lethal but permanent and irreversible reproductive harm to fish. Effects arise from early-life exposure and show up in adulthood (p. 294; evidence pp. 282, 289–290). 2. EE2 is not the sole culprit (“naive to conclude that EE2 alone is the culprit”), but there is “reasonable certainty” that it plays a significant role, as the most potent steroid oestrogen and one that frequently exceeds the PNEC (p. 294). 3. The basic finding (wastewater hormones harm fish) has been stable since the late 1980s; “Only the level of uncertainty has reduced” (p. 285). 4. Conventional wastewater treatment removes EE2 poorly (about 54% at best). Advanced treatment works but is costly, and monitoring compliance at the proposed EQS is technically hard (pp. 292, 295). 5. Costs vary greatly between countries and technologies, and cheaper options (sand filtration) can deliver most of the biological benefit (p. 293). 6. The combination of high costs, disputes about “acceptable harm”, and a precautionary principle that includes cost-effectiveness has caused serious delay (p. 296). 7. Decisions have been made in “a poorly understood, closed process that has little engaged the public”, with the public as “silent witnesses” (p. 297). The chapter cites RCEP’s rejection of decisions “negotiated privately between the regulator and polluter” as a principle, but does not document any specific private negotiation over EE2. 8. Wildlife may serve as sentinels for human reproductive health where human causal evidence cannot be obtained (pp. 287–288, 294). This is presented as a question (“Perhaps we should …?”).

Recommendations and advocacy - Regulation of EE2 is justified (stated as a judgement, not a direct call to regulate; the authors then ask whether the price is too high). It is “entirely reasonable to invoke the precautionary principle and introduce regulation” (p. 280); the evidence is “sufficient … to justify applying the precautionary principle” (p. 295). - Consider redesign at source. Substituting EE2 in the pill and redesigning pharmaceuticals would be “a constructive precautionary approach” (p. 295). Posed as a question; not argued through. - Frame precaution as “pros and cons”. Do not let cost-effectiveness or proportionality reduce precaution to an economic cost-benefit test (implicit in p. 296 and fn 13, which approvingly cite the EC 2000 and EEA wording). - Hold an open public debate. Involve the public early on acceptability of risk, appetite for precaution, willingness to pay, and where responsibility lies (water industry and bill-payers, or pharma). End private regulator–polluter negotiation (p. 297). - Reduce exposure generally. “The only logical way forward seems to be to reduce exposure as much as possible — to be precautionary” (p. 296), because substance-by-substance assessment of the chemicals universe is unfeasible. - Drop the demand for more certainty. Abandon “the unrealistic hope of achieving more certainty prior to policymaking” on human male reproductive health (pp. 287–288). Posed as a question.

Distinguishing the two. The authors’ own formal lessons are the three in §13.7.1–13.7.3 (see section notes above); items 1–8 above collect conclusions from the whole chapter. Items 1–5 rest directly on the chapter’s evidence, though item 3 is contested by the chapter’s own account of the early-1990s shift to industrial chemicals (see bias check). Item 6 is an interpretive argument supported by the chronology, but the counterfactual (would a precaution rule without a cost clause have acted faster?) is not tested. Item 7 is asserted from the authors’ insider experience; they document no specific closed negotiations. The recommendations are advocacy, although notably even-handed. The authors openly allow that the public might choose to accept the harm (p. 297).

Mechanisms and dynamics#

1. How the warning arose. - The first signal came from routine field monitoring by a water-company biologist, not from a research programme (p. 282). - Its significance was only recognised when compared with a published baseline (1 in 1,000), and that took “a year”, until Sweeting read Jafri & Ensor’s paper (p. 282). That paper was published in 1979, so part of the lag was the baseline not yet being in print. - A second, independent biomarker signal (VTG in males at a fish farm) came from a separate research line (p. 284). - The combination of a visible anomaly and a sensitive biomarker, plus caged-fish experiments, turned the observation into evidence (p. 284).

2. Containment of knowledge. - Confidentiality kept early findings inside government and industry: - Thames Water’s rat studies were “never published” (p. 283). - The 1987–90 national survey was withheld until 1994 by a DoE–MAFF–Brunel contract (p. 284). - Aherne’s team “could not have known” of the Thames Water work (p. 284). - Separately, the pill link “was not formally stated” in reports between government and the water industry (p. 283). - Knowledge was fragmented across institutions (water utility, health department, fisheries ministry, universities, US researchers). That fragmentation, not outright suppression, is what the chapter documents.

3. Remit gaps between regulators. - The route from harm to consequence (medicine, then excretion, then sewage works, then river, then fish) crossed regulatory domains. - Drug regulators “had limited expertise in environmental issues”, and environmental assessment “was not formally required” (p. 284). - Water treatment “was not designed to remove pharmaceuticals” (p. 281). - The chapter attributes the non-recognition of early published signals (Tabak, Aherne) specifically to drug regulators’ limited environmental expertise and remit (p. 284). The broader framing, that nobody’s remit covered the whole route, is the analyst’s.

4. Asymmetric evidentiary standards. - Speed. Human-health reassurance was reached quickly on “this small set of studies” (p. 283), though the chapter notes the Department of Health committee agreed only “after conducting further studies”, which it does not describe. Protective action for wildlife waited for proof “beyond reasonable doubt” (p. 284) and took decades. - Direction. The chapter frames this as a contrast of speed. The deeper asymmetry is about which way the decision went: a small amount of evidence was enough to conclude no action needed, while a very large amount was not enough to conclude act. - Target. Action came sooner on alkylphenols, but was “not … extended to EE2” (p. 289). Analyst inference: the chapter does not say substitutes existed for alkylphenols or that they lacked major benefits; it reports only that suppliers agreed to reformulate.

5. Moving targets in the science. - The causal hypothesis moved from pill hormones to industrial chemical cocktails (early 1990s), back to steroid oestrogens (1998), and then to mixtures of oestrogens, anti-androgens and progestins (2002–2012) (pp. 288–290). - Each turn was scientifically legitimate. But the accumulating complexity: - (a) delayed the identification of EE2; - (b) later qualified singling it out (“naive to conclude that EE2 alone is the culprit”, p. 294), though the authors still claim “reasonable certainty” that it plays a significant role; - (c) complicated substance-by-substance cost accounting. The authors themselves ask “is it scientifically incorrect to blame just EE2 for the costs?” (p. 293), a co-benefits point: one treatment removes many contaminants, so charging it all to EE2 overstates the cost of controlling it. - Analyst inference: the more realistic the science became, the harder it was to attach harm, or the cost of remedies, to any one substance. The chapter does not show that this complexity was used to argue against regulation.

6. Measurement as a limit on both knowledge and governance. - Discovery. Analytical capability at ng/L was lacking until the 1990s (p. 284). - Test design. Biodegradation studies were run at unrealistic concentrations because of analytical limits, and “may … lead to erroneous conclusions” (p. 292). - Enforcement. Compliance monitoring at 0.035 ng/L is “hardly routine” (p. 295). - Endpoints. Choosing chemical concentration or biological effect as the compliance measure changes which technology looks adequate and at what cost (p. 293).

7. Contestation shifts along the chain of evidence. - As evidence accumulated, the contest moved from “is it happening?” to “is it significant?” to “who pays, and is it worth it?”. - Industry accepted that endocrine disruption was “undoubtedly occurring” (p. 282) and asked “so what?” about population effects (p. 295). Both are cited to the same 2003 paper (Webb et al.), so for the chemical industry this was concession and challenge at once, not a later retreat; the chapter says only that the “so what?” question has been asked “increasingly”. Implementers later, after the NDP results (2008–2010), worried about “technical feasibility and disproportionate cost” (p. 292). - Only “the level of uncertainty has reduced” (p. 285). More evidence did not settle the decision, because the decision had moved to ground (significance, cost, values) that evidence alone could not resolve.

8. Cost, feasibility and the cost-effectiveness clause. - Costs were concrete, large and quantified: EUR 32–37 billion; GBP 25–40 million just to find out (pp. 291–292). Benefits were “often intangible” (p. 292). - Cost estimates came from different parties, one of them the water-industry federation, and diverged sharply (EC EUR 11–18 per person against EUREAU 25–50% of charges, p. 293; the units differ, so direct comparison is hard). - The UK was a worst case (low dilution, two-stage treatment; pp. 289, 293), and the chapter’s own figures are UK figures. Analyst inference: whether UK figures anchored the EU debate is not shown. - The Rio “cost-effective” wording let delay count as consistent with precaution (p. 296). The decision “appeared to stall in the still waters of cost‑benefit analysis” (p. 293).

9. Who pays: the choice of intervention point. - Choosing end-of-pipe treatment, rather than substituting the drug, put responsibility on the water industry and “(ultimately)” the bill-paying public (pp. 290–291). - The chapter’s reasons for the choice are sympathetic: the pill’s benefits, natural oestrogens needing removal anyway, and co-removal of other chemicals. - The effect was that the product’s manufacturer faced no requirement. The chapter never uses the phrase “polluter pays”. Its RCEP quotation treats the polluter as the party negotiating with the regulator (p. 297), which sits awkwardly with the diffuse source (millions of users). - Burden (p. 279; p. 294): society “enjoyed decades of flexible fertility and will also ultimately pay”; the harm falls on fish (and anglers, p. 292).

10. Industry roles cut both ways. - The pharmaceutical producer generated key incriminating evidence (Schering life-cycle study, pp. 282, 290). The chemical industry conceded the effect (p. 282). - Both the pharmaceutical and water industries then made “representations to the EC”, which the authors say “may” have stalled the decision; the next sentence (“Indeed in July 2012 …”) introduces the deferral amendment, a draft report by MEP Richard Seeber (p. 294 and fn 9). The content of the representations is not given. - Taking part in the science did not visibly translate into support for regulation, though the chapter does not report what the pharmaceutical industry’s representations said. The chapter does not describe manufactured doubt of the tobacco or lead kind. The resistance it describes works through arguments about significance and cost.

11. Institutions, champions and public-sector science. - Publicly owned or funded bodies drove the early science: government-owned Thames Water (p. 283), MAFF, DoE (p. 284). - A named champion inside the regulator (Brighty) “built and defended the evidence‑based case” (p. 290). - The move from evidence to a policy “trigger” depended on individuals and on the EA’s willingness to take a position (Brighty’s “policy trigger” quote, p. 290). - The chapter does not analyse privatisation, though it notes Thames Water was government-owned “at the time” (p. 283).

12. Media and research funding as amplifiers. A documentary made the issue public, “The world began to sit up and take notice”, and was followed by more than EUR 150 million of EU research (p. 288). The causal link is asserted, not shown.

13. Irreversibility, latency and cumulative exposure. - Harm is set early in development and is permanent (ducts do not recover in clean water, p. 290). It worsens with age and duration (pp. 283, 290). It shows up long after exposure (p. 294). - In humans, latency of 30–35 years and mixtures defeat standard causal criteria (p. 287). - Irreversibility is the chapter’s main reason to act despite uncertainty about population effects (p. 295).

14. Mental models and blind spots. - Pharmaceutical optimism. “there seemed no limit to the extent to which artificial oestrogens could be put to good medical use” (p. 280). “it will stop ovulation 100 per cent” (p. 281). - Therapeutic-dose reference frame. Aherne et al. judged concentrations “far below therapeutic doses” and so saw no adverse effects (p. 284). The human pharmacological dose was the yardstick, while fish respond at ng/L (pp. 288, 290). Analyst inference. The authors themselves present Aherne favourably, as a study that “had recognised that steroidal oestrogens could potentially have adverse effects, even if none were identified” (p. 284). - Human-centred risk frame. The original inquiry asked about “human consumers of the water” (p. 283). Once humans seemed safe there was “relief” (p. 283). Analyst inference: the wildlife problem then lost urgency; the chapter implies this through its contrast between the two decisions but does not say it. - Proof threshold. Policymakers “perhaps preferred to wait until … beyond reasonable doubt” (p. 284). - Infrastructure frame. Sewage works were designed for other purposes (p. 281). The NDP results “confirmed what they [implementers] had suspected for many years: EE2 was potent and hard to get rid of” (p. 292). Analyst inference: this suggests expected difficulty and cost shaped how readily they engaged. - Individual versus population harm. A “highly charged” dispute over what counts as harm (p. 293): intersex as “unpleasant” but perhaps “acceptable” (pp. 295–296).

15. Framing and language. - Borrowed phrase: “something from nothing” (p. 280) is Silva et al.’s (2002) phrase, quoted by the authors (“in the words of Silva et al.”). - The authors’ rhetoric: “a pill too bitter to swallow” / “a bitter pill to swallow” (pp. 279, 296); “flexible fertility” (pp. 279, 294, 297); “silent witnesses” (p. 297); “loads the dice before they are thrown” (p. 297); “Achilles heel” (p. 296); “perfect excuse for inaction” (p. 296); “collateral damage” (p. 297). - Reported opposing frames: “so what?” (p. 295); “unpalatable but acceptable harm” (p. 296). - Box 13.1 advocacy: “redefine the ill people as ‘normal’” (p. 287). - Media: “Assault on the male” (p. 288). - Reassurance language: “no apparent risks to the consumer”, “discount any possibility of risk” (p. 283). - The chapter recasts the pill’s success as “flexible fertility”, a phrase that sets its benefits beside its hidden costs.

16. Lock-in and substitutes. - EE2 is “an active ingredient” of the pill (p. 279), and the pill is used by more than 100 million women (p. 281). The chapter does not say what share of pills contain EE2 (the first pill used mestranol, its methyl ether, p. 280; some UK users take progestogen-only pills, p. 281). Its clinical entrenchment is implied but not analysed. - Substitutes are raised only as questions (p. 295), with one citation to oestrogen-free research (fn 11). Clinical, commercial and regulatory barriers to substitution are not discussed. - For alkylphenols, where substitutes existed, a voluntary phase-out happened, with a clause against harmful substitution (octylphenol) (p. 289). - The contrast suggests that the availability of acceptable substitutes shapes whether source control happens. This is the analyst’s inference from the chapter’s juxtaposition.

17. Innovation effects. - Regulatory attention prompted systematic testing of treatment options (NDP trials of conventional treatment and GAC; evidence on ozonation and sand filtration, pp. 291–293). “Mounting concern” prompted EU research funding that underpinned test methods (p. 288). - The chapter’s redesign suggestion (p. 295) echoes the “green pharmacy” idea of designing drugs to be benign in the environment. That term appears only in the title of Kümmerer 2007 in the references; the text cites Kümmerer 2007 only on p. 284, for general concern about pharmaceuticals. The chapter does not claim the EE2 case generated the idea. - The NDP itself was an innovation-by-demonstration mechanism financed by regulated industry (p. 291). - The chapter does not assess whether regulatory uncertainty (the 2012 deferral) held back investment in either treatment or drug redesign.

18. Systemic and complex dynamics. - Mixtures and additive effects (p. 290), multiple sources (natural and synthetic hormones, industrial chemicals, progestins), and hydrology (dilution, pp. 285, 289) make impacts depend on context. - Modelling suggests that even if Japan took up the pill to English levels, its greater dilution (five times the UK’s per capita) means widespread disruption “would still not be predicted” (p. 289). This is a modelled counterfactual; actual Japanese pill use is very low. Harm is a property of the whole system (consumption, infrastructure and receiving environment), not of the molecule alone. - Governance organised substance by substance fits this poorly (p. 296).

Transferable insights (technology-neutral)#

1. When a technology’s main benefits are large and socially valued, the response to its side effects tends to move downstream to end-of-pipe controls and away from redesign at source. That shifts costs to intermediaries and the general public. Evidence: end-of-pipe was chosen partly out of awareness of the pill’s public-health benefits, placing responsibility on the water industry and the public (pp. 290–291). Substitution was raised only as a question (p. 295). By contrast, alkylphenols, for which the chapter claims no comparable public-health benefit, were restricted at source (p. 289). Strength: moderate. Explicit in the text but hedged (“may have partly reflected”), and the chapter also gives technical reasons (natural oestrogens need removing anyway; co-removal of other chemicals); a single case.

2. Evidentiary thresholds can be asymmetric: little evidence suffices to conclude “no action needed”, while much more is demanded before protective action. Evidence: the Department of Health discounted human risk on “this small set of studies” (p. 283), against decades of evidence for wildlife and a “beyond reasonable doubt” preference (p. 284). Strength: moderate. The contrast is explicit in the text; but the DoH decision is undated, its “further studies” are not described, its reasons are not given, and the two decisions concerned different receptors.

3. Confidentiality clauses and unpublished commissioned research can delay collective recognition of a hazard by years, and can prevent other researchers from connecting their findings. Evidence: Thames Water studies “never published” (p. 283); the national survey held back until 1994 by contract (p. 284); Aherne “could not have known” (p. 284); information “not widely circulated beyond government and industrial organisations” (p. 284). Strength: moderate to strong for the fact of delay; how much it contributed to the overall lag is not quantified.

4. Hazards whose pathway crosses the remits of several institutions tend to go unrecognised, because no institution is responsible for the whole route. Evidence: drug regulators lacked environmental expertise and remit (p. 284); treatment works not designed for pharmaceuticals (p. 281); early published signals “not recognised” (p. 284). Strength: moderate. A plausible mechanism, supported by the later remit change (EMA 2006), but asserted rather than documented through institutional records.

5. Measurement capability sets the boundary of both knowledge and enforceability. Hazards at the edge of detection are found late, studied under distorting conditions and hard to regulate. Evidence: ng/L analysis unavailable until the 1990s (p. 284); biodegradation studies at unrealistic concentrations “may … lead to erroneous conclusions” (p. 292); compliance monitoring “hardly routine” at 0.035 ng/L (p. 295). Strength: strong within this case; three independent points in the chain.

6. Judging risk against the reference frame of intended use (for example the dose that affects the target) can hide effects on non-target systems that are far more sensitive. Evidence: “far below therapeutic doses … no evidence of adverse effects” (p. 284), against fish effects at low ng/L (pp. 288, 290). Strength: suggestive. An analyst inference from one quoted source; the authors do not frame it this way.

7. As evidence accumulates, the contest over a hazard moves along the chain (existence, significance, cost, responsibility), so more evidence alone does not produce action. Evidence: industry concedes effects (p. 282) and asks “so what?” (p. 295); implementers turn to “technical feasibility and disproportionate cost” after the NDP (p. 292); “Only the level of uncertainty has reduced” (p. 285); stall “in the still waters of cost‑benefit analysis” (p. 293). Strength: moderate. The shift from existence to cost is clear across the chronology. But the industry concession and the “so what?” challenge are both cited to one 2003 paper (Webb et al.), so the “existence then significance” step is not documented as a sequence over time.

8. Growing scientific realism about multiple causes and mixtures makes it harder to attribute harm, and the cost of remedies, to any single substance. It can delay identifying the key agent, and it strains substance-by-substance assessment, while strengthening the case for reducing exposure in general. Evidence: the early-1990s turn to industrial chemicals before steroid oestrogens were identified in 1998 (pp. 288–289); “naive to conclude that EE2 alone is the culprit” (p. 294); the authors’ co-benefits question, “is it scientifically incorrect to blame just EE2 for the costs?” (p. 293); the chemicals universe is “unfeasible” to assess substance by substance, so “The only logical way forward seems to be to reduce exposure as much as possible” (p. 296). Strength: moderate for delayed identification. The chapter does not show that multi-causality was used to resist regulation or that it drove the deferral.

9. Precautionary formulations that include cost-effectiveness or proportionality can turn precaution into a justification for delay when mitigation costs are concrete and benefits hard to quantify. Evidence: Rio’s “cost effective” as “Achilles heel”; “perfect excuse for inaction” (p. 296); costs quantified against “often intangible” benefits (p. 292); fn 13 on “pros and cons”. Strength: moderate. Well argued and consistent with the chronology, but the counterfactual is untested and other causes of delay (lobbying, EU process) are also present.

10. Cost estimates are contested evidence, not neutral facts. They vary widely with who produces them, the existing infrastructure, the technology assumed and the compliance measure chosen. Evidence: EC EUR 11–18 per person against EUREAU 25–50% of charges (p. 293); UK worst case against lower-cost countries (p. 293); Swiss EUR 5 per person where sand filtration already exists (p. 295 fn 10); sand filtration nearly as effective as GAC and two-thirds cheaper when judged on biological effect (p. 293). Strength: strong for variability. Only suggestive for “interested parties give higher estimates”, since there is a single comparison and the two estimates use different units.

11. Outcome-based measures (does the harm stop?) can reveal cheaper adequate solutions than proxy-based measures (does the concentration fall below X?). Evidence: Baynes et al. 2012, using biological effects, found sand filtration nearly as effective as GAC at two-thirds the cost (p. 293). Strength: moderate. One study, co-authored by a chapter author; not independently replicated within the chapter.

12. Where in the causal chain an intervention sits decides who bears the cost and who gets off. Evidence: the end-of-pipe choice places cost on the water industry and public (pp. 290–291); the ethical question of “where responsibility should lie” between water industry and pharma (p. 296); benefits enjoyed by users, costs spread across society, harm borne by wildlife (pp. 279, 294). Strength: strong. Explicit in the text and logically robust.

13. Regulation tends to move first against tractable contributors (substitutes available, weak claims of benefit), leaving high-benefit contributors unaddressed even when they are the most potent. Evidence: voluntary phase-out and EU restrictions on alkylphenols, “which has not been extended to EE2” (p. 289). Strength: moderate. Clear contrast in the text; the explanation in terms of tractability is the analyst’s.

14. A proposal to regulate can be put off by deferring the step that sets the standard, while the substance stays formally listed. Evidence: the July 2012 amendment (a draft European Parliament report, fn 9) proposing not to specify the EQS “for certain substances of pharmaceutical relevance” until the 2016 review, with compliance by 2027 (p. 294); the authors say the decision “may be stalled” by representations from the water and pharmaceutical industries (p. 294). Strength: documented for the deferral itself; the link to industry representations is hedged (“may”); suggestive as a general pattern.

15. Early warnings often come from routine practitioner monitoring. Recognising them depends on a baseline to compare against and on someone connecting the observation to its significance, and that step can itself take time. Evidence: Dearsley’s routine health check, and a year’s lag until Sweeting read Jafri & Ensor’s baseline, itself published only in 1979 (p. 282); the independent VTG signal at a fish farm (p. 284). Strength: moderate. Well documented through first-hand accounts; a single case.

16. Getting from evidence to a policy “trigger” can depend on individual champions inside regulators and on publicly funded science. Evidence: Brighty “passionately championed”, “built and defended the evidence-based case” (p. 290); proactivity of government-owned Thames Water (p. 283); DoE- and MAFF-funded research (p. 284). Strength: suggestive. A single named example, told by insiders with an admiring tone.

17. Irreversible harms set early in development, which appear late and build up over time, weaken the case for waiting for more certainty, and defeat causal frameworks designed for simpler exposures. Evidence: permanent duct feminisation (p. 290); severity increasing with age (pp. 283, 290); critique of Bradford Hill criteria for complex exposures (p. 287); “abandon the unrealistic hope of achieving more certainty” (p. 288). Strength: strong for the biology in fish; asserted for the general epistemological claim about causal criteria.

18. Treating a decision as a technical cost question between regulator and regulated leaves out the value judgement (what harm is acceptable, who should pay) and tilts the outcome towards whatever is quantified. Evidence: “silent witnesses”; “closed process”; public views “largely undocumented”; “without asking the public it will be far easier to come to a conclusion based largely on costs alone. This loads the dice” (p. 297). Strength: asserted. A normative argument drawing on RCEP, EC and NAS authority; no evidence is given about what public engagement would have changed, and the authors concede the public might accept the harm.

19. Producers can generate key incriminating evidence and still lobby over the regulatory consequences. Taking part in the science does not necessarily mean supporting action. Evidence: the EE2 producer’s own life-cycle study (Schering; pp. 282, 290), then pharmaceutical-industry representations to the EC, which the authors say “may” have stalled the decision (p. 294). Separately, the chemical industry both conceded disruption and questioned its significance (Webb et al. 2003; pp. 282, 295). The “so what?” position is attributed to the water and chemical industries, not the pharmaceutical industry. Strength: suggestive. The chapter gives no detail of what the representations said, so “resistance” by the EE2 producer is inferred, not documented.

20. Infrastructure built for an earlier problem can become the conduit for a new one, and the cost of retrofitting depends on the legacy standard. Evidence: works “not designed to remove pharmaceuticals” (p. 281); UK two-stage against German three- and four-stage treatment (p. 293); Swiss low cost where sand filtration already existed (p. 295). Strength: moderate.

Limitations, contestation and bias check#

Participant authorship. Both authors were central actors, as scientist and as regulatory programme head (pp. 692, 696). Much of the key evidence is Jobling’s own (1998, 2002, 2006, 2009; Baynes 2012). The cost figure comes from their own commentary (Owen & Jobling 2012). This brings first-hand detail, but: - no opposing voice appears (no industry panel); - colleagues are praised (“passionately championed”, “beautifully presented”); - the “closed process” critique comes from an author who was himself inside the regulator.

A reader should treat the narrative of why things were delayed as an informed insider account, not an independent history.

UK-centred worst case. The UK combines high pill use, low dilution and mainly two-stage treatment (pp. 281, 289, 293), so UK harms and costs are at the high end. The authors acknowledge this fairly (p. 293), but the lesson that precaution may be “too high” a price (p. 294) leans on UK costs. Their own Swiss footnote (EUR 5 per person per year) partly undercuts it.

Population-level evidence is the weak point, and the chapter mostly acknowledges it. - The strongest population evidence comes from dosing studies (the Kidd lake at about 5–6 ng/L; Schering at 2–4 ng/L). The chapter calls 2 ng/L “environmentally relevant” but gives no typical river concentrations. It admits “a mixture of doubt and amazement” that EE2 could act at “the very low concentrations at which it is present” (p. 290). Analyst inference: this implies typical exposures are below the dosing-study levels. - It calls certainty about oestrogen mixtures “extremely high” (p. 295) while conceding “Uncertainties remain, particularly around fish population level effects” (p. 295). - It says individual-level harm is “not in dispute”, then reports that industry and some scientists question its population-level significance (p. 295). The two claims concern different levels, so they are not strictly contradictory, but the chapter does not resolve the population question. - The “so what?” position is presented fairly as a question but not answered with evidence in Lesson 2. The chapter does cite wild-roach fertility studies earlier (Jobling et al. 2002a,b; Harris et al. 2011, “The consequences of feminisation in breeding groups of wild fish”, p. 282), but does not bring them into its reply. - The “Great Lakes” mislabelling of Kidd’s study (p. 296) is a factual slip.

Human-health material is mostly context, and some of it is advocacy. Section 13.2.4 and Box 13.1 bring in the contested debate over sperm counts and TDS. - The authors themselves say the oestrogen theory is “not entirely convincing” (p. 287) and that anti-androgens are more likely. Its inclusion still adds rhetorical weight to the EE2 case. - The sentinel logic (“If … then … This did indeed seem to be the case”, p. 285) is consistency reasoning, not causal evidence. - The WHO “redefine the ill people as ‘normal’” claim (p. 287, attributed to Skakkebaek 2010) is, in the analyst’s judgement, a one-sided reading of how reference values are set. - The wildlife list (p. 285) is mostly unrelated to EE2, as the chapter admits.

Hindsight sharpening. - “The fundamental conclusion … has not changed since the late 1980s” (p. 285) sits uneasily with the chapter’s own account. Until the mid-1990s “few were convinced” hormones were present in sufficient amounts, and industrial chemicals were then seen as “more likely” culprits (p. 288). Steroid oestrogens were identified as the main agents only in 1998 (p. 289). - The claim that “at least one possible culprit was EE2” by the late 1980s (p. 284) is true as a hypothesis. It should not be read as established knowledge that policymakers ignored. - The “policymakers … perhaps preferred to wait” explanation (p. 284) is speculative.

Internal inconsistencies and factual slips. See the Table 13.1 comparison above. Also: - John Rock quote dated 1967 but cited 1964 (p. 281). - FDA approval “early 1961” (likely 1960) (p. 281). - Caldwell 2009/2008 (p. 285); Johnson 2012/2011 (p. 289); “Harries, Hamilton et al., 2011” for Harris et al. (p. 282). - 1995–2002 sampling cited to a 1998 paper (p. 282). - “Great Lakes study” (p. 296). - “may not be insufficient” (p. 287). - Dodds’s BPA finding dated 1936 (p. 280) but 1938 in fn 6 (p. 288). - “breast growth in young men (Henley et al., 2007)” (p. 280), but the reference title is about prepubertal gynecomastia linked to lavender and tea tree oils. - “Ibaretta and Swan, 2001” in the text (p. 285) against “Ibarreta and Swan, 2002” in the references.

These suggest light editorial checking. They do not change the core argument, but they argue for checking any specific date or number against primary sources.

The costs are the least transparent numbers. EUR 32–37 billion is attributed to “provisional estimates by the UK government” but cited to the authors’ own commentary, with no stated basis (capital or capital plus operating costs, time horizon, discount rate, number of works needing upgrade). The EC figure is per person per year; EUREAU’s is a percentage of charges. They are hard to compare.

Under-developed alternatives. - The substitution idea (p. 295) ignores the clinical pharmacology (why EE2 is used), regulatory and commercial barriers, and the chapter’s own finding that progestins affect fish (p. 290), so an oestrogen-free pill might shift rather than remove the problem. - Other options go undiscussed: producer-funded treatment (polluter or producer pays), treatment at source (hospital or urine separation), lower-dose formulations. The phrase “polluter pays” never appears. - Agricultural and livestock oestrogen sources are not discussed.

The public-participation lesson is asserted, not evidenced. No data on public views are presented (they are “largely undocumented”, p. 297). How participation would change outcomes, or who “the public” is (pill users, anglers, bill-payers), is not specified. To their credit, the authors concede that the public might accept the harm (p. 297). This is more even-handed than many advocacy framings.

Fairness in both directions. - Towards industry. The chapter credits pharma and chemical industry contributions to the evidence (p. 282), reports the water industry’s cost concerns as legitimate uncertainty (“There is clearly considerable uncertainty here”, p. 293), and gives the attribution question a fair hearing (p. 293). - Towards the pill. It never downplays the pill’s benefits (pp. 281, 294). - Towards precaution. Unusually for the report, it turns critique on the precautionary principle itself (p. 296). - Where the pro-precaution framing still shapes conclusions: - by treating delay as failure rather than as a reasonable response to real uncertainty about population effects and very high costs; - by treating irreversibility at the individual level as sufficient seriousness; - by describing the resistance of industry and implementers mainly as “representations” and “so what?” rather than setting out their evidence. - Case selection. EE2 is in some ways a hard case for precaution (high benefit, diffuse source, costly fix). Its inclusion strengthens the report’s credibility, because it is not a simple villain story.

Dissent within panels. Not applicable (no panels).

Notable quotes#

1. “it is entirely reasonable to invoke the precautionary principle and introduce regulation to limit aquatic exposure to EE2. But this is no trivial matter” (p. 280)
2. “for many medical practitioners and scientists there seemed no limit to the extent to which artificial oestrogens could be put to good medical use” (p. 280)
3. “It is striking that such a rapid decision was made on the basis of this small set of studies.” (p. 283)
4. “This information was not widely circulated beyond government and industrial organisations.” (p. 284)
5. “The fundamental conclusion … has not changed since the late 1980s. Only the level of uncertainty has reduced.” (p. 285)
6. “Ironically … oestrogens and progesterones have yet to be tested in fish ‘in combination’, despite the fact that human females have been carrying out this ‘test’ for 50 years.” (p. 290)
7. “For a number of years the decision of whether or not to regulate EE2 appeared to stall in the still waters of cost‑benefit analysis” (p. 293)
8. “the water and chemical industries (and indeed some scientists) have increasingly asked the question, ‘so what?’” (p. 295)
9. “Defined in this way the precautionary principle may be logical and rational, but it can also paradoxically become a perfect excuse for inaction” (p. 296)
10. “without asking the public it will be far easier to come to a conclusion based largely on costs alone. This loads the dice before they are thrown.” (p. 297)

Open questions#

Within the chapter’s own terms - What exactly did the Department of Health standing committee review, and when? Its rapid reassurance (p. 283) is the chapter’s main example of asymmetric standards, but it is undated and unsourced. - What was the basis of the EUR 32–37 billion England-and-Wales estimate (capital against lifetime costs; which works), and how does it compare with the EC’s EU-wide figure (pp. 292–293)? - Do typical in-river EE2 concentrations, as opposed to effluent concentrations, reach levels at which population effects occur? How much of the observed feminisation is due to EE2, as against E1 and E2, anti-androgens and progestins (pp. 289–290)? - Was substituting EE2 ever seriously considered by regulators or industry, and what clinical or commercial barriers exist (p. 295)? - What did the water and pharmaceutical industries’ representations to the EC actually say (p. 294)? - Would a precaution rule without a cost-effectiveness clause have produced faster action, or were the practical barriers (monitoring at 0.035 ng/L, treatment technology) binding anyway (pp. 295–296)? - How would public deliberation have been designed, and whose views (pill users, anglers, bill-payers, conservation interests) would count (p. 297)?

Leads for the hindsight stage. These come from my background knowledge and are not verified in this pass. Check them against primary sources. - EU outcome of the 2012 proposal. My understanding is that Directive 2013/39/EU did not set EQS for EE2 and E2, and instead created a Watch List mechanism. The first Watch List (Decision (EU) 2015/495) included E1, E2 and EE2 (with diclofenac). The Commission’s 2022 proposal to revise the priority-substance lists reportedly proposed EQS for the oestrogens. Check its final status and EQS values. - Urban Waste Water Treatment Directive recast (Directive (EU) 2024/3019). Reportedly requires advanced (“quaternary”) treatment for micropollutants at larger plants, phased in towards 2045. It is funded mainly through extended producer responsibility, with the pharmaceutical and cosmetics sectors paying at least 80% of costs. This bears directly on the chapter’s “who pays” question. Reportedly contested by industry, including through legal challenges. - Switzerland. Reportedly amended its Water Protection Act (in force 2016) to upgrade roughly 100+ plants for micropollutant removal, financed by a national wastewater levy (reported ceiling of CHF 9 per inhabitant per year). A test of the chapter’s footnote-10 projection. - UK. Follow-up under the water industry’s Chemicals Investigation Programme (CIP), which reportedly included EE2 treatment trials. Whether any EE2 standard was adopted in the UK after Brexit. - Population-level science. Hamilton et al. 2014 (BMC Biology) reported that UK roach populations are self-sustaining despite widespread feminisation of males, which bears directly on the “so what?” debate. Blanchfield et al. 2015 reported recovery of the lake fish population after EE2 dosing stopped. Check both, and any later evidence on intersex trends after treatment upgrades. - Human sperm counts. Later meta-analyses (Levine et al. 2017; 2023 update) reported continuing declines. Causes remain unresolved; check the status of the oestrogen and anti-androgen hypotheses. - Substitution. Newer combined oral contraceptives using estetrol (E4), approved in about 2021, reportedly with claims of lower environmental impact. Check whether environmental considerations played any role in their development. - EMA environmental risk assessment. Check whether the 2006 guideline’s 0.01 µg/L action limit had an exception for endocrine-active substances, and whether the revised guideline (reportedly in force 2024) changed treatment of legacy products like EE2. - Bradford Hill and WHO 2002. Check the wording of the WHO/IPCS 2002 global assessment against the chapter’s characterisation (p. 287). - Competing no-effect values. My understanding (unverified) is that Caldwell et al. 2008, cited by the chapter (pp. 281–282) as supporting evidence, derived an EE2 PNEC of about 0.35 ng/L, higher than the EA’s 0.1 ng/L (Young et al. 2004) and the EC’s proposed EQS of 0.035 ng/L. The chapter does not compare these values. Check, since the choice of threshold drives both the exceedance map (Fig. 13.3) and compliance costs.

Audit log#

Independent audit against the full text extract (PDF pp. 281–309), the report bios (pp. 692, 696), the cross-citing pages (pp. 211, 644, 649, 660) and the embedded images for Figs 13.3 and 13.4. Changes made:

1. Owen bio: corrected “he oversaw the programme that produced the chapter’s cost data”. His EA programme included research supporting the NDP; the NDP was run by the water industry with the EA and government (pp. 291, 696). The “Standpoint” line was amended to match.
2. “Other notes”: added that “innovative” appears once in the body (p. 290).
3. §13.1 Thesis: added the chapter’s p. 280 paraphrase of Rio, and noted that it omits the “cost-effective” wording later called the “Achilles heel”.
4. §13.1.1: flagged the Dodds BPA date inconsistency (1936 on p. 280, 1938 in fn 6 p. 288; no “Dodds et al., 1936” in the references).
5. §13.1.2: relabelled “unusual feature for a Late Lessons case” as an analyst note.
6. §13.2.1: noted that Jafri & Ensor’s baseline was itself published in 1979, so the “year later” lag was partly the baseline not yet being in print; added a description of Figure 13.1.
7. §13.2.1: added the authors’ favourable framing of Aherne et al. 1985 (“had recognised that steroidal oestrogens could potentially have adverse effects”).
8. §13.2.1 fn 5 analyst note: added that the chapter does not say whether the EMA “no other environmental concerns” caveat would catch EE2.
9. Box 13.1: marked the WHO reference-limit critique as analyst background, and noted that the box cites studies disputing the decline.
10. §13.4 fn 7: corrected the misattribution. The footnote does not mention Jobling; her co-editorship of Weybridge+15 comes from the reference list (p. 300).
11. §13.5 Schering study: added that the chapter calls 2 ng/L “the environmentally relevant concentration” and gives no typical river concentrations. Marked the “doubt and amazement” gloss as analyst reading.
12. §13.6: marked the “tax paying public = bill-payers” gloss as an analyst note; quoted fn 8’s hedge (“may not be strict enough”) accurately.
13. Fig. 13.3: checked the image; added eastern England to the red areas, and noted that the caption says “UK” but the map shows only England and Wales.
14. §13.6 cost attribution: corrected a misreading. “Is it scientifically incorrect to blame just EE2 for the costs?” is the authors’ co-benefits point, not an argument by opponents of regulation.
15. Lesson 2: added the authors’ concession that the cost “may be a price too high to pay” (p. 296), and fn 13’s wording that economic CBA should be included “where this is appropriate and possible”.
16. Lesson 2 “so what?”: added that the concession (“undoubtedly occurring”) and the “so what?” challenge are both cited to the same paper (Webb et al. 2003), and that the water-industry attribution is not separately sourced.
17. Table 13.1: added four further discrepancies (FDA and NHS dated 1962–1969; nonylphenol in effluent 1985 against Giger 1984; intersex surveys 1995–1996 against 1995–2000; anti-androgen activity 2002–2003 against Jobling 2009).
18. References: clarified that the chapter’s PNEC (0.1 ng/L) is cited to Young et al. 2004, not to Caldwell et al. 2008.
19. Timeline: rat-study row marked undated (early 1980s by inference) and corrected to “part-way through drinking-water treatment”; the chemical-industry response was rewritten to reflect the single 2003 source.
20. “When action came”: removed an unsupported worldwide claim for 2012. The p. 291 “no other country” statement refers to 2007 and to EQS only.
21. “What was known when”: the first stage was redated to the early-to-mid 1980s (Aherne 1985; rat study undated).
22. Authors’ lessons item 7: removed “between regulators and regulated parties”. The chapter says “closed process that has little engaged the public” and documents no specific private negotiation. Clarified that the formal lessons are §13.7.1–13.7.3.
23. Recommendations: “Regulate EE2” reworded to “Regulation of EE2 is justified”, since the authors pose cost as an open question.
24. Mechanism 1: removed “chance reading” and added the 1979 publication date of the baseline.
25. Mechanism 3: marked “nobody’s remit covered the whole route” as the analyst’s framing; the chapter blames drug regulators’ expertise and remit.
26. Mechanism 4: added the DoH committee’s “further studies”; softened “quickly” to “sooner”; marked the substitutes and low-benefit explanation for alkylphenols as analyst inference.
27. Mechanism 5: corrected (c) (co-benefits, not an argument for resisting regulation); softened (b); marked the conclusion as analyst inference.
28. Mechanism 7: noted that the Webb 2003 concession and “so what?” come from the same source; dated the implementer concerns to after the NDP.
29. Mechanism 8: the EC is no longer called an “interested party”; noted the different units; marked “UK figures anchored the debate” as unshown.
30. Mechanism 10: replaced “lobbied … apparently contributing” with the chapter’s “representations” and “may”; noted the Seeber draft report and that the content of the representations is not given.
31. Mechanism 14: corrected the Aherne framing (the authors present it favourably, not neutrally); marked “wildlife problem lost urgency” as inference; corrected “hard and costly” to the source’s “potent and hard to get rid of”.
32. Mechanism 15: moved “something from nothing” from the authors’ rhetoric to a borrowed phrase (Silva et al. 2002).
33. Mechanism 16: corrected “EE2 … used by more than 100 million women”. The chapter says the pill has more than 100 million users and does not give EE2’s share.
34. Mechanism 17: removed the claim that the case “generated” green pharmacy (the term appears only in a reference title); reworded the innovation drivers.
35. Mechanism 18: corrected “the same national pill use produces … little [harm] in Japan”. It is a modelled counterfactual (Johnson et al.).
36. Insight 1: replaced “low-benefit alkylphenols” with the chapter-grounded wording, and noted the technical reasons also given.
37. Insight 2: strength caveat now notes that the DoH decision is undated and its further studies undescribed.
38. Insight 7: strength downgraded from strong to moderate (same-source concession and challenge).
39. Insight 8: rewritten to remove the co-benefits misreading. Strength is now moderate for delayed identification only.
40. Insight 10: the second clause was downgraded to suggestive (a single comparison, with different units).
41. Insight 14: “emptied” softened. Separated the documented deferral from the hedged link to industry.
42. Insight 15: added the 1979 baseline publication date.
43. Insight 19: heading softened; clarified that “so what?” is attributed to the water and chemical industries, not pharma. Strength downgraded to suggestive.
44. Bias check: population-evidence bullet rewritten (the “above typical river levels” claim is now marked as inference); “not in dispute” tension clarified as individual against population level; noted the wild-roach fertility studies cited on p. 282.
45. Bias check: the WHO critique is now marked as the analyst’s judgement.
46. Inconsistencies list: added the Dodds BPA date, Henley 2007 (prepubertal, not “young men”) and Ibaretta 2001/Ibarreta 2002.
47. Open questions: added a lead (unverified) on competing EE2 no-effect values (Caldwell 2008 against the EA and EC figures).
48. Digest: corrected Owen’s role; “the oestrogen in the pill” became “an active ingredient”; added the 1979 baseline, the DoH “further studies” and the undated rat studies; VTG redated to the mid-1980s; “only about 54%” became “at best”; the July 2012 deferral link was hedged; added “one might argue” and “price too high”; the remit-gap, contestation and complexity mechanisms were corrected; contestation insight downgraded to moderate; population caveat and alkylphenol wording fixed.