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

EEA 2013, report pp. 279–307 (text pp. 279–297; Table 13.1 p. 298). The authors are Susan Jobling (Brunel; her research is much of the evidence) and Richard Owen (formerly head of the Environment Agency’s Environment and Human Health Programme, whose research supported policy on endocrine disrupters and the UK National Demonstration Programme; the programme itself was run by the water industry with the EA and government). Both are insiders to the story (pp. 692, 696). There are no panels.

Core story#

Ethinyloestradiol (EE2) was synthesised at Schering in 1938. It was valued because it is stable and works when taken by mouth, and it became an active ingredient of the contraceptive pill, which is used by more than 100 million women (pp. 279–281).

Most of an oral dose is excreted. Sewage works were “not designed to remove pharmaceuticals” (p. 281). So EE2 reaches rivers along with the natural oestrogens E1 and E2, and their effects add up.

Early warnings - 1978: a Thames Water biologist doing a routine check found intersex roach in the River Lea at more than 100 times the baseline rate. Its significance was recognised only a year later, when the baseline paper (Jafri & Ensor, published 1979) was read (p. 282). - Undated (early 1980s by inference): unpublished rat studies found river water oestrogenic but water taken part-way through drinking-water treatment not. A Department of Health committee, “after conducting further studies”, agreed there was enough information to “discount any possibility of risk” to consumers; the authors call this a rapid decision on “this small set of studies” (p. 283). - Mid-to-late 1980s: vitellogenin (a female yolk protein) in male fish, caged trout showing a 100,000-fold rise in effluent, and a national survey (1987–1990) found oestrogenic effluent widespread. A contract kept the survey results unpublished until 1994 (p. 284). - Early 1990s: suspicion moved to industrial chemicals (nonylphenol), which were then phased out voluntarily in the UK and restricted in the EU. The chapter calls this “a constructive application of the precautionary principle … which has not been extended to EE2” (pp. 288–289). - 1998: steroid oestrogens were identified as the main culprits, EE2 the most potent (p. 289).

Evidence of harm - Intersex at 86% of 51 English sites (p. 293). - Permanent feminisation of reproductive ducts (p. 290). - A life-cycle study by Schering: breeding inhibited above 2 ng/L; all-female population at 4 ng/L (p. 290). - Collapse of a whole-lake fish population at about 5–6 ng/L (Kidd 2007; pp. 285, 290).

Regulatory response - 2004: the UK accepted the risk, about 25 years after the first observation. It chose end-of-pipe treatment over substituting the drug, which put the costs on water companies and the public (pp. 290–291). - Demonstration Programme (GBP 25–40m): conventional treatment removed at best about 54% of EE2. Granular activated carbon worked but would cost an estimated EUR 32–37bn for England and Wales. Sand filtration was almost as effective against fish feminisation and two-thirds cheaper (pp. 291–293). - January 2012: the EU proposed EE2 as a priority substance, with a limit of 0.035 ng/L in rivers and lakes (p. 293). - July 2012: an amendment (a draft European Parliament report) proposed deferring the limit to 2016 and compliance to 2027. The authors say the decision “may be stalled” by representations from the water and pharmaceutical industries, whose content they do not report (p. 294).

Authors’ lessons#

“is the price of being precautionary simply too high?” Treatment and monitoring at part-per-trillion levels are costly and technically hard. Redesigning the drug is offered as “a constructive precautionary approach”, but only as a question (pp. 294–295).
The Rio definition’s “cost effective” wording is precaution’s “Achilles heel”. Defined that way, precaution “can also paradoxically become a perfect excuse for inaction”, and long delays have, “one might argue”, been “completely within the spirit” of the principle (p. 296). The authors allow that the cost “may be a price too high to pay” (p. 296).
Open public debate is needed. The public have been “silent witnesses” to a “closed process”. Without their views, deciding “based largely on costs alone … loads the dice before they are thrown” (p. 297). The authors concede that the public might accept the harm.

Main mechanisms#

Early signals were held back by confidentiality and scattered across institutions (pp. 283–284).
Drug regulators had “limited expertise in environmental issues” and no requirement to consider them, so early published reports of hormones in water went unrecognised (p. 284).
Standards of evidence were asymmetric: reassurance for drinking water came fast, action for wildlife came slowly (pp. 283–284).
Measurement limits shaped discovery, testing and enforceability (pp. 284, 292, 295).
As evidence grew, contestation moved from “is it happening?” towards “so what?” and cost (pp. 282, 285, 292, 295). The industry concession and the “so what?” challenge are both cited to one 2003 industry paper, so that step is not a documented sequence.
The science grew more complex (mixtures, anti-androgens, progestins). An early-1990s focus on industrial chemicals delayed singling out EE2, and multi-causality complicates substance-by-substance costing: the authors ask whether it is “scientifically incorrect to blame just EE2 for the costs” when treatment removes many contaminants (pp. 288–290, 293).
Cost estimates were contested and depended on context: EC EUR 11–18 per person against EUREAU 25–50% of sewerage charges; Switzerland EUR 5 per person (pp. 293, 295).
A high-benefit product pushed the response towards the end of the pipe (the authors’ hedged explanation: it “may have partly reflected” the pill’s benefits), while alkylphenols were restricted at source earlier (pp. 289–290).

Transferable insights (selected; full list in the notes)#

Where in the causal chain you intervene decides who pays (pp. 290–291, 296). Strong.
Contestation shifts from existence to significance to cost, so evidence alone does not trigger action (pp. 282, 285, 292–295). Moderate: the concession and the “so what?” come from the same 2003 paper.
Measurement capability bounds both knowledge and enforceability (pp. 284, 292, 295). Strong.
Cost estimates are contested evidence that varies with who produces them, legacy infrastructure and the compliance measure (pp. 293, 295). Strong on variability.
Precaution rules with cost-effectiveness clauses can legitimise delay (p. 296). Moderate: the counterfactual is untested.
Asymmetric evidentiary thresholds (pp. 283–284). Moderate.
Confidentiality and gaps between regulators’ remits delay recognition (pp. 283–284). Moderate.
Measuring outcomes rather than proxies can reveal cheaper fixes (p. 293). Moderate: one study, by the authors’ group.
High-benefit products get end-of-pipe fixes; contributors without comparable benefits get restricted at source (pp. 289–290). Moderate: the benefit contrast is partly the analyst’s.
Leaving out public values tilts decisions towards what is quantified (p. 297). Asserted.

Main caveats#

Participant authorship. The authors’ own studies and commentary supply much of the evidence and the cost figure.
UK worst case. Low dilution and two-stage treatment in the UK inflate costs and harms.
Population effects are the weak point. Population-level evidence comes from dosing studies (2–6 ng/L). The chapter calls 2 ng/L “environmentally relevant” but gives no typical river concentrations, and admits “doubt and amazement” that EE2 could act at “the very low concentrations at which it is present” (p. 290). It calls certainty “extremely high” yet concedes uncertainty about populations (p. 295). It mislabels the Canadian lake study as “Great Lakes” (p. 296).
Human-health material is contextual and partly advocacy. The authors call the oestrogen theory of male reproductive decline “not entirely convincing” (p. 287). Box 13.1 says WHO “redefine[d] the ill people as ‘normal’”.
Hindsight sharpening. “not changed since the late 1980s” (p. 285) conflicts with the chapter’s own account of the early-1990s shift to industrial chemicals (p. 288).
Many internal inconsistencies, especially in Table 13.1 (e.g. first observation 1982 v 1978). Use the text or primary sources.
The substitution idea is under-developed. Progestins also affect fish (p. 290). “Polluter pays” is never discussed.
Fair in both directions. Despite the above, the chapter credits the pill’s benefits and industry’s contributions to the evidence, and turns its critique on precaution itself.