LL2-08 hindsight check: Vinyl chloride: a saga of secrecy (Soffritti, Sass, Castleman and Gee), Late lessons from early warnings: science, precaution, innovation (EEA Report No 1/2013), Ch. 8, pp. 179–202#

Checked 25 September 2026. The check covers what happened between publication (January 2013) and late September 2026 that bears on the chapter’s claims, evidence, predictions and recommendations. Page numbers are report pages (PDF page minus 2).

The authors are allied advocates: - Soffritti directs research at the Ramazzini Institute, as Maltoni’s successor. - Sass is a scientist at NRDC. - Castleman is a public-interest consultant. - Panel 8.2, on animal bioassays, is by James Huff of NIEHS.

No industry or regulator voice appears. That matters most for the contested epidemiology (Claims 4 and 10), where the chapter presents one side of a live dispute.

Annex 3 note. Not applicable. Vinyl chloride (VC) was not one of the 2001 report’s case studies, so no Annex 3 update exists. [This rests on the 2001 report’s case list; I did not re-open the 2001 report for this check.]

Access note. The web-search quota was used up before this check began. I verified points by fetching primary sources directly: - IARC Monograph Vol. 100F (2012) and the 2019 Preamble; - the ATSDR Toxicological Profile for Vinyl Chloride (January 2024); - EU legal texts, through the EU Publications Office; - the US Federal Register API and eCFR; - PubMed, Europe PMC, OpenAlex and Crossref, for the literature.

Some sources returned 403 errors or bot challenges, which I did not try to get round: - ECHA’s 2023 investigation report on PVC and PVC additives; - the 1995 Office of Technology Assessment (OTA) report on OSHA’s regulatory analyses; - Doniger (1978); - the Minamata Convention secretariat site; - Germany’s TRGS 910; - NCBI Bookshelf.

Where a point rests on a secondary account or was not re-verified, it is flagged in square brackets.

Overview#

Nothing I found challenges the chapter’s core history, which runs in three steps: - company insiders knew about liver and bone harm years before regulators or workers did; - that knowledge was actively suppressed, including through the 1972 secrecy agreement; - once the signature cancer became public in 1974, control came quickly and at bearable cost.

The documentary base for that history has since grown. Toxic Docs (2018) puts millions of pages of once-secret industry documents online, including on PVC. The record on the chapter’s wider scientific, economic and policy claims divides much more sharply.

Vindicated or strengthened - Liver cancer and cirrhosis. Longer follow-up confirmed exposure–response for hepatocellular carcinoma (HCC) in two cohorts: - US industry-wide cohort, followed to 2013 (Mundt et al., 2017): HR 5.3 in the top exposure category. - Porto Marghera cohort (Fedeli et al., 2019): HCC RR 5.52; cirrhosis RR 2.60.

Latencies of 24–56 years for angiosarcoma of the liver (ASL) and 32–67 years for HCC (ATSDR, 2024) bear out the chapter’s warning that short follow-up understates risk (lesson 7, p. 192). - Exposure in lower-resource settings (Claim 2). The chapter offered no data; there now are some. - Chinese studies since 2013 report VC levels above China’s own limit of 10 mg/m³ (about 3.9 ppm, nearly four times the EU and US limits) before plant upgrades in 2013–16. - A systematic review reports mean exposures of about 50 ppm in South Indian and former-Yugoslav PVC plants. - The older acetylene (“carbide”) production route persists: about 35% of global capacity, mainly in China. Its workers show poorer liver indices, and its mercury use is rising. - Concern about low doses (Claim 5, in part). - Mouse studies at under 1 ppm show VC aggravating diet-induced fatty liver disease and increasing liver tumours. - Genotoxicity benchmark doses in exposed workers fall far below 1 ppm. - The EU cut its binding workplace limit from 3 ppm to 1 ppm (Directive (EU) 2017/2398). - Extracting corporate information (Claim 9). - The EU’s new Product Liability Directive (2024/2853) pairs compensation schemes with court-ordered disclosure of evidence. - The US EPA ordered VC manufacturers to submit unpublished health and safety studies (TSCA §8(d), December 2024). The deadline has since slipped to May 2027.

Weakened, contested or wrong - The multi-site cancer list (Claim 4). - IARC’s 2012 re-evaluation (Vol. 100F) found “inconsistent or scanty” evidence for lung, brain and lymphohaematopoietic cancers. The chapter does not cite it, although it was published a year earlier. - The 2017 US cohort update found no excess and no exposure–response for brain cancer, lung cancer or non-Hodgkin lymphoma. - ATSDR (2024) reports that most studies fail to link VC with birth defects or fetal loss. - The chapter’s own Box 8.2 already conflicted with its list (p. 191). - The Wong retraction (Claim 10). Press reporting of internal documents describes pressure from the study’s industry sponsor. But no later assessment has supported a causal link between VC and brain cancer, which is consistent with the retraction’s scientific conclusion. In the 2017 update, the small excess was confined to early years and showed no exposure–response. - The cost figures (Claim 3). - The “USD 278 million” looks like the top of a USD 228–278 million range from a later survey. OTA reported it in 1995; McGarity and Ruttenberg (2002) relayed it. - The same account puts the ex ante estimate at “over $1 billion”. That is roughly a four-fold overestimate, not the 300-fold gap implied by setting USD 278 million against USD 90 billion. - Lesson 3’s “USD 60 million annually” contradicts the p. 187 figure, and I could not trace it. - Bioassays predicting target organs “without a doubt” (Claim 6). - For VC, ATSDR (2024) finds “little consistency in tumor types across species” outside liver angiosarcoma. - Pharmaceutical regulators now waive some two-year rat studies (ICH S1B(R1), 2022). IARC’s 2019 Preamble gives mechanistic evidence more weight. The EU adopted a roadmap to phase out animal testing in chemical safety assessment (June 2026). - Even so, no validated non-animal replacement for carcinogenicity testing exists yet. - Reversing the burden of proof (Claim 8). - Some moves went the chapter’s way: TSCA’s 2016 reform requires an affirmative EPA finding before a new chemical enters the market. - In 2025–26 the direction reversed in part. Presidential proclamations granted two-year exemptions from EPA’s 2024 air-toxics rule for chemical plants (the “HON rule”), which requires fenceline monitoring for VC. The EU turned towards “simplification”. - VC itself entered a TSCA risk evaluation only in December 2024, fifty years after OSHA’s 1 ppm standard.

What the chapter could not see - Market geography. EU PVC went from stagnation to decline: output fell 22% and capacity use dropped to 69% by 2023. The industry has turned to anti-dumping defence: - against US and Egyptian imports (2024–25); - from 23 September 2026, against imports from China, Korea, Mexico and Taiwan.

The US became a large exporter to Europe. - Mercury. VC production using mercury catalysts became a treaty issue (Minamata Convention, in force since 2017), and the EU banned the practice from 2022. Yet mercury use in China’s PVC sector reportedly doubled between 2004 and 2023. - Infeasibility claims recur. The 2025 and 2026 proclamations exempting plants from the HON rule assert that the required technology “is not available”. That is the kind of claim the chapter’s history shows to have been wrong in 1974. Whether it is wrong this time cannot yet be judged.

Weight for the lens.

What the chapter is strong evidence for: - the documentary history of concealment; - the mechanism by which insider knowledge outruns public knowledge; - the finding that, in this case, control came fast and was affordable once required.

Moderate evidence: that regulated parties overstate compliance costs. Recalibrate this to roughly four-fold rather than 300-fold, and remember that VC had an unusually cheap fix.

Weak evidence: - the multi-site cancer and reproductive claims; - the portrayal of the Wong retraction; - the absolutist defence of bioassays.

Advocacy that policy has taken up only partly and reversibly: the ratchet (lesson 6) and the burden-of-proof prescription (Box 8.1).

Most durable lessons: - read the internal record; - long latency makes a missing epidemiological signal weak evidence of safety; - hazards move with older technologies when production relocates; - infeasibility claims recur and should be checked against ex post outcomes.

Claim 1. Global VC demand was forecast to grow about 4.7%/yr (2010–15) and 4.2%/yr (2015–20), with Asia and China dominant and European and North American capacity “almost stagnant” (p. 181)#

Original claim. Global VC production and consumption was about 34 million tonnes (Mt) in 2010. VC output roughly equals PVC output. Demand was expected to grow “around 4.7% per year from 2010 to 2015, and 4.2% from 2015 to 2020” (SRI Consulting, 2011). Asia-Pacific held 54.8% of installed capacity. China had become “the largest producer and exporter”. European and North American capacity was expected to be “almost stagnant” (GBI Research, 2010) (p. 181).

What happened since. - Global growth ran below forecast but was still substantial. - A 2023 techno-economic study puts global PVC manufacturing at 45 Mt in 2018, forecast to grow by about 2 Mt a year (Medrano-García et al., ACS Sustainable Chem. Eng., Aug 2023, citing industry sources). From 34 Mt in 2010, that implies about 3.6% a year, below the 4.7% forecast. - A December 2025 Nature Communications paper gives “an annual production of 61 million tons” (He et al., 2025) [probably a capacity-type figure; industry-derived, not independently checked]. - The EU’s 2024 anti-dumping regulation records that global PVC capacity “has systematically exceeded global demand”, and calls this overcapacity structural (Reg. (EU) 2024/1896, recitals 185–186). - Asian and Chinese dominance held. - About 35% of global PVC capacity uses the acetylene (coal and calcium-carbide) route, “mainly in coal-rich countries like China”. The ethylene route (65%) is concentrated in the USA and the Middle East (He et al., 2025). - On 23 September 2026 the Commission opened an anti-dumping investigation into PVC from China, Korea, Mexico and Taiwan. The complaint alleges significant distortions in China’s calcium-carbide inputs, plus “freely disposable capacity” and an “imminent and substantial increase” in capacity in the countries concerned (Notice 2026/C 4912). - [I could not verify China’s exact production or export share from a primary source.] - Europe did worse than stagnate. For the EU PVC industry, from 2020 to the investigation period ending September 2023: - capacity was flat at 5.46–5.56 Mt; - production fell from 4.95 Mt to 3.84 Mt (−22%); - capacity use fell from 91% to 69%; - EU consumption fell from 3.95 Mt to 3.22 Mt; - the industry became loss-making (Reg. (EU) 2024/1896, Tables 1 and 4, recitals 155–156).

The Commission’s Chemicals Industry Action Plan (8 July 2025) reports that more than 20 major EU production sites announced closures in two years, and that 8–10% of EU cracking capacity has been lost. - North America was not simply stagnant; it became a major exporter. US PVC imports into the EU rose from 33,737 t in 2020 to 267,390 t in the investigation period. Their market share went from 0.85% to 8.30% (Reg. (EU) 2024/1896, Table 2). The EU imposed definitive anti-dumping duties on US and Egyptian PVC (Reg. (EU) 2025/36, 9 Jan 2025). [I did not verify North American capacity figures.] - A hazard dimension the chapter did not mention: mercury. - The acetylene route uses mercury catalysts. Mercury consumption in China’s PVC industry reportedly rose from over 600 t in 2004 to more than 1,200 t in 2023 (He et al., 2025). Tsinghua researchers call the process “one of the most important sources” of anthropogenic mercury in China (Ren et al., Environ. Sci. Technol., 2014). - The Minamata Convention (in force 16 August 2017) treats VC production as a mercury-using process. It requires a 50% cut in mercury use per unit of production by 2020 against 2010. It bans mercury only five years after the Conference of the Parties finds mercury-free catalysts “technically and economically feasible” (Annex B, Part II). - The EU prohibited mercury use in VC production from 1 January 2022 (Reg. (EU) 2017/852, Annex III).

Verdict: partly held up. - Held up: the direction, with production and growth in Asia (above all China), and Europe stagnant or declining. - Too high: the growth rates. - Wrong: “almost stagnant” North America, at least as a description of trade.

Weight. Descriptive context, not a lesson. It matters for the lens because it supports the chapter’s point, made without data, that production and exposure move to places with older technologies and weaker controls (Claim 2). It also shows the hazard profile changing with the production route (mercury).

Claim 2. In low- and medium-resource countries “older technologies have continued to be used and therefore high exposures probably occur” (p. 181; no data given)#

Original claim. Exposures in North America and Western Europe fell sharply after closed-loop polymerisation came in during the late 1970s. The chapter says they are now generally below 0.4 ppm in PVC processing. Elsewhere, it infers, high exposures “probably occur”. It cites no measurements (p. 181).

What happened since. Occupational-health literature since 2013, mostly Chinese, partly fills the gap. - Exposure above China’s own limit. China’s occupational limit is 10 mg/m³ (about 3.9 ppm). At a Tianjin PVC plant using the calcium-carbide process (227 exposed workers, 179 controls): - before protective upgrades in 2013 and 2016, 8-hour VC levels exceeded that limit at all eight job positions, reported as “1 to 2 folds” of the limit; - after the upgrades, levels met the limit and liver-function indices improved; - the indices stayed above those of controls. The authors conclude workers “might still be at risk of liver injury” at the lower levels (Dong et al., Wei Sheng Yan Jiu, Jan 2023). - Older route, worse liver indices. In a Tianjin petrochemical complex, workers on the acetylene route had worse liver indices and ultrasound findings than ethylene-route workers in the same enterprise (Dong et al., 2024). A companion study found that acetylene-route job positions (ore breaking, acetylene generation, steam stripping and others) failed exposure limits in 2020. They met the limits in 2021, after ventilation and sealing were improved (Dong et al., Front. Public Health, Nov 2022). - Firm type matters. Among 478 VC workers in Tianjin, abnormal health-examination findings were more common in domestically owned than in foreign-invested firms (OR 3.56) (Li et al., 2022). - High exposures elsewhere. A systematic review of micronucleus studies found 13 of 15 Chinese studies showing increased chromosome damage. It also reports PVC workers “exposed to a mean of 50 ppm” in the former Yugoslavia and South India (Bolognesi et al., Mutat. Res. Rev., Oct 2017). It warns that PVC products from China may contain high residual VC. - Structural driver. The acetylene route’s persistence is economic: ethylene-based PVC costs 7–17% more than acetylene-based PVC in China (He et al., 2025).

Verdict: strengthened. The inference was right, and there are now data behind it, at least for China and India. Two limits remain: nothing I found estimates the number of highly exposed workers globally, and none of the newer studies has cancer outcomes.

Weight. Supports a technology-neutral lesson: when production relocates, hazards controlled in one place can reappear elsewhere with older processes, lower limits and weaker enforcement. It also shows that the “quick fix” of 1974 (Claim 7) did not spread on its own.

Claim 3. Compliance with the 1 ppm standard cost only USD 278 million (“an analysis in 2000”), against industry forecasts of up to USD 90 billion and 2 million jobs (p. 187); lesson 3 cites “USD 60 million annually and 2 million jobs” (Washington Post, 1974) (p. 192)#

Original claim. The chapter says “Industry responded quickly and easily”. It gives the USD 278 million figure without a source. Industry had “earlier estimated that it would cost them up to USD 90 billion and 2 million jobs”. The New York Times (Rattner, 1975) noted that “not one of the doomsday predictions” had come true (p. 187). Lesson 3 instead gives “USD 60 million annually and 2 million jobs” (p. 192). Panel 8.1 reports UK first-year costs of GBP 9 million, plus GBP 4 million in lost production (p. 189).

What happened since: provenance. - The USD 278 million. - Wagner (Duke Law Journal, 2004, n. 225) summarises McGarity and Ruttenberg (Texas Law Review 80:1997, 2002, at 2031). Their account runs as follows: the 1995 OTA study found ex ante cost estimates for OSHA’s 1974 VC standard “exceeded $1 billion”, while a later survey found compliance costs “in the $228–278 million range”. - The chapter’s “analysis in 2000” is therefore probably this legal-scholarship chain, which rests on OTA (1995). [I could not access OTA (1995) or McGarity and Ruttenberg directly.] - The chapter quotes only the top of a range. - The like-for-like comparison (estimate for the standard versus measured cost) is roughly four-fold, not about 300-fold. - The USD 90 billion and 2 million jobs. - These figures match the widely cited Foster D. Snell study for the Society of the Plastics Industry (USD 65–90 billion; 1.7–2.2 million jobs). [Not verified from a primary copy.] - That study is usually described as costing OSHA’s May 1974 proposal of “no detectable level” and assuming plant closures, not the 1 ppm standard adopted. [Also not verified here.] - If so, setting it against the actual cost of the adopted standard compares different things. - The USD 60 million a year. I could not trace this figure (Washington Post, 20 May 1974). It is inconsistent with p. 187 by three orders of magnitude, and one of the two must be wrong or refer to something else. - Lesson 3’s citation. The lesson cites Sass et al. (2005). That paper does not give cost figures. It says industry predicted “job loss and plant closures”. It says virtually all US plants met the standard within two years “while still maintaining rapid growth of sales volume”, citing OSHA (1975). That qualitative account supports the lesson.

What happened since: the general question. Later research on how accurate ex ante cost estimates are gives a more guarded picture: - Simpson (J. Benefit-Cost Analysis, 2014). Overestimates are more common, and the average ratio of ex ante to ex post estimates exceeds one. Simpson argues this does not prove bias, because cost distributions are skewed. His regression test “cannot reject the hypothesis that estimates are unbiased”, and he calls for better data. - Morgenstern (J. Benefit-Cost Analysis, 2018; 34 comparisons across nine rules). He finds “a slight tendency to overestimate both costs and benefits”.

Verdict: partly held up. - Held up: the direction. The worst forecasts were far off, compliance was fast, and the industry grew. - Not reliable: the magnitudes. - The two figures conflict. - The USD 278 million is unsourced in the chapter and is the upper end of a range. - The headline contrast probably sets a worst-case estimate for a stricter proposal against measured costs of a less strict final rule.

Weight. Moderate support for the lesson that regulated parties’ forecasts of compliance costs can be greatly overstated. Anyone using this case should quote the four-fold comparison (over USD 1 billion forecast against USD 228–278 million actual), not USD 90 billion against USD 278 million. The case also had an unusually cheap technical fix (steam stripping, closed reactors).

Claim 4. Epidemiology “certainly underestimated” VC’s cancer risk; VC is associated with brain cancer, HCC, haemolymphoreticular neoplasms, lung cancer, cirrhosis, birth defects and miscarriages (pp. 188, 191–193)#

Original claim. The chapter lists these outcomes (p. 191). It states that IARC in 1979 linked VC to liver, brain, lung and haemo-lymphopoietic cancers, and that the 1987 and 2007 reviews “re-confirmed the previous evaluations” (p. 188). It argues that epidemiology underestimated risk because early studies were short, cohorts young (average age 54) and industry-controlled, and the healthy-worker effect masked excesses (pp. 188, 192–193).

Box 8.2 quotes the 2008 IARC Monograph finding no strong evidence for brain or lymphatic and haematopoietic cancers (p. 191). The chapter does not reconcile this with its list.

What happened since, outcome by outcome.

• “Certainly underestimated.”
• Partly vindicated. Follow-up to 2013–17 revealed HCC and cirrhosis risks that short studies missed. The long latencies (HCC 32–67 years) confirm the chapter’s point about the young 1990s cohort.
• Not vindicated for the other sites the chapter named. Longer follow-up tended to attenuate them.
• Worth noting: the 2017 update, which found the HCC association, was funded by the American Chemistry Council. Industry-funded work does not uniformly minimise risk.
• Low-weight counter-evidence: an ecological study by consultants found no county-level association between ambient VC and liver cancer in Texas (Towle et al., 2021).
• The chapter’s account of IARC. The chapter says the later IARC reviews “re-confirmed” the 1979 evaluation, including brain, lung and blood-forming system (p. 188; Table 8.2, p. 193). That is inaccurate. IARC’s 2008 and 2012 evaluations confirmed carcinogenicity overall but restricted the established sites to ASL and HCC.

Verdict: partly held up. - Strengthened: the liver claims (HCC, cirrhosis) and the underlying warning about latency. - Weakened: the multi-site list (brain, lymphohaematopoietic), the reproductive claims and the characterisation of IARC. These have moved away from the mainstream assessment. - Contested: lung, possibly through PVC dust.

Weight. The latency lesson is strong: a missing signal in short or young cohorts is weak evidence of safety. The multi-site claims should not be cited as established. Doing so repeats an error the chapter criticises in others, reading one’s preferred conclusion into inconsistent data.

Claim 5. There is no safe threshold (attributed to IARC 1979); occupational limits should keep falling below the 1974 level of 1–3 ppm as evidence and technology advance (p. 188; lesson 6, p. 192)#

Original claim. IARC (1979) found “no threshold or safe level of exposure” (p. 188). Table 8.2 says IARC “found no safe exposure level” (p. 193). Lesson 6 says limits “did not continue to be lowered” and that limit-setting should be “a dynamic one, in which levels are constantly lowered” (p. 192).

The chapter itself reports Maltoni’s finding that no VC-specific tumours occurred at 1 ppm by inhalation (p. 186).

What happened since. - The IARC 1979 wording. As quoted by Sass et al. (2005), IARC said there was “no evidence that there is an exposure level below which no increased risk of cancer would occur”. That is absence of evidence for a threshold, not a finding that none exists. The chapter turns it into a positive finding. [IARC 1979 text as quoted by Sass et al.; not retrieved directly.] - Mechanism supports the no-threshold default. IARC (2012) found “strong evidence” of a genotoxic mechanism: metabolic activation, DNA adducts and mutations, with many key events also shown in humans. That supports linear low-dose extrapolation as the regulatory default. A minority literature argues for carcinogen thresholds in general (e.g. Calabrese et al., Chem.-Biol. Interact., 2021), but it is not VC-specific and not mainstream. - Limits since 2013. - EU. Directive (EU) 2017/2398 (12 Dec 2017) cut the binding limit for VC from 7.77 mg/m³ (3 ppm), the value in Directive 2004/37/EC, to 2.6 mg/m³ (1 ppm), with transposition due within two years. Recital 13 says the limit “should be revised in the light of more recent scientific and technical data”. This one step down is to the US 1974 level. - US. The OSHA limit remains 1 ppm (8-hour average) and 5 ppm (15-minute) (29 CFR 1910.1017(c)). I found no rulemaking to revise it. That makes 1 ppm a 50-year-old standard. - China. The limit is 10 mg/m³, about 3.9 ppm (Dong et al., 2023). - Beyond the workplace. EPA began a TSCA risk evaluation of VC in December 2024 (Claim 8). - Low-dose evidence since 2013. - In favour of lower limits: - Mice exposed to VC below 1 ppm showed more liver injury and steatosis on a high-fat diet (Lang et al., Hepatology Communications, 2018). - Twelve weeks of VC below 1 ppm, followed by a Western diet, increased liver tumours, including HCC (Liu et al., Toxicol. Appl. Pharmacol., 2023). Both groups conclude that current limits “may be insufficient”. - Benchmark-dose analyses of chromosome damage in exposed workers give lower bounds of 0.23–0.54 mg/m³ (Wang et al., 2013). The meta-analytic bound is 0.03–0.07 mg/m³, i.e. hundredths of a ppm (Bolognesi et al., 2017). These are biomarkers, not disease. - Chinese workers showed liver changes at exposures meeting China’s limit (Dong et al., 2023). - On the other side: - No ASL deaths occurred among North American workers exposed after 1974 (Collins et al., 2014). - No excess mortality occurred among Italian workers hired after 1971 (Scarnato et al., 2017). - In the US cohort, ASL and HCC risks were elevated only above about 1,000 ppm-years (Mundt et al., 2017).

These studies have low power and are still inside the latency window, so they cannot show that 1 ppm is safe. They do show that the 1974 cut largely stopped the signature disease.

Verdict: partly held up. - Held up: the no-threshold presumption is consistent with the mechanistic consensus, and new low-dose animal and biomarker evidence points below 1 ppm. - Not borne out: the “constantly lowered” prescription. Limits fell once, in the EU and only to 1 ppm; in the US they did not move. - Overstated: the attribution to IARC 1979.

Weight. Treat lesson 6 as advocacy grounded in a defensible default, not as an empirical finding. The durable, technology-neutral point is narrower: a standard set in one era tends to persist as a de facto “safe” level long after the evidence that justified it has moved on.

Claim 6. Long-term rodent bioassays have been shown “without a doubt” to predict human carcinogenicity and even target organs (lesson 5, p. 192); no other test, “or even epidemiology”, matches them; non-animal alternatives “have so far proven unsuccessful” (Panel 8.2, pp. 194–196)#

Original claim. Lesson 5 says VC “showed without a doubt the validity of long-term bioassays in predicting not only the general carcinogenicity of industrial agents, but even specific target organs and tissues” (p. 192). Panel 8.2 argues: - all adequately tested human carcinogens are also animal carcinogens; - about a third were first found in animals; - no other method “can claim these collective facts and advantages”; - alternatives “have so far proven unsuccessful” (pp. 195–196).

A footnote notes efforts to reduce animal use (p. 192, fn 8).

What happened since. - The VC case itself. - ATSDR (2024): “With the exception of liver angiosarcomas, which have been observed in all species (including humans), there is little consistency in tumor types across species.” - Maltoni’s rats developed Zymbal gland, mammary and kidney (nephroblastoma) tumours that have no human counterpart. Viola’s early positive study (1970) found skin and lung tumours. - The bioassays did predict carcinogenicity and the signature human tumour. They did not predict the human target-organ profile in general. - The broad concordance evidence. An IARC-linked analysis of 111 Group 1 carcinogens found sufficient or limited animal evidence for 91. Of 60 agents with tumour sites in both humans and animals, 52 (87%) shared at least one organ or tissue system: “good concordance … for many, but not all, tumor sites” (Krewski et al., J. Toxicol. Environ. Health B, 2019). That supports Panel 8.2’s general case at the level of organ systems, not specific organs. - Scientific critique grew louder, on both sides. - Goodman (Toxicol. Res., 2018) argued it is “time to say goodbye” to the two-year bioassay. He notes that larger dose groups would make almost every tested chemical positive. - A 2021 review found the bioassay sensitive but significantly non-specific, largely because of rodent-specific mechanisms (Suarez-Torres et al.). - European researchers call its human relevance “questionable”, especially for non-genotoxic carcinogens (Audebert et al., PARC, 2023). - Regulators moved to reduce reliance on it, without replacing it. - Pharmaceuticals. ICH S1B(R1) (August 2022) allows a weight-of-evidence case in place of a two-year rat study. In the prospective evaluation, about 27% of such studies could have been avoided (Bourcier et al., 2024). - IARC’s 2019 Preamble. “Sufficient evidence” in animals alone now supports only Group 2B (“possibly carcinogenic”). Animal evidence combined with strong mechanistic evidence in exposed humans can support Group 1. - FDA announced a plan (10 April 2025) to phase out animal-testing requirements for monoclonal antibodies and other drugs. - EU. The Commission’s Roadmap towards phasing out animal testing for chemical safety assessments (3 June 2026) lists carcinogenicity among the endpoints where a reliable non-animal battery is still a mid-term goal.

So Panel 8.2’s statement that alternatives had “so far proven unsuccessful” for carcinogenicity remains broadly true in 2026. EU hazard classification still treats positive animal studies as grounds for “presumed” human carcinogen status.

Verdict: contested. - Broadly held up: the claim that bioassays identify likely human carcinogens, especially genotoxic ones. - Overstated: “without a doubt” and “even specific target organs”, even for VC. - Weakened as a policy stance: “no other method … or even epidemiology” can match them. Regulatory trends since 2013 favour weight-of-evidence and mechanistic integration and reduced use of the two-year bioassay. A full replacement has not yet arrived.

Weight. The narrow lesson is durable and important: laboratory evidence can and should trigger protective action before human cases accumulate, and VC is the textbook example. Cite the panel as advocacy by a leading proponent, not as a settled consensus. The later literature splits into three positions: bioassays are indispensable; bioassays are obsolete; weight-of-evidence approaches should replace routine bioassays.

Claim 7. Clear, challenging regulation stimulates innovation; exposure cuts can be achieved quickly without catastrophic effects on production or jobs (lessons 2–3, p. 192), hedged in fn 7 as holding “in many but only specific circumstances”#

Original claim. Steam-stripping cut residual VC in PVC by 99%. The industry “was easily able to comply and keep growing”, “another example of how clear and challenging regulations can stimulate innovation” (lesson 2, citing Ashford and Hall). Rapid exposure reduction came “without catastrophic consequences” (lesson 3). Footnote 7 concedes the Porter link holds “in many but only specific circumstances” (p. 192).

What happened since. - The VC case holds. - Nothing contradicts the account of rapid compliance and continued growth. Sass et al. (2005): “in < 2 years virtually all U.S. manufacturing plants” complied. - Global PVC output grew from about 7 Mt (1971) to 26 Mt (1999; p. 181) and about 45 Mt (2018). - The absence of ASL deaths among post-1974-exposed workers (Collins et al., 2014) shows the fix worked for the signature disease. - The general evidence is mixed. - Cohen and Tubb (meta-analysis of 103 studies, JAERE, 2018): “considerable heterogeneity”. Positive effects on competitiveness are more likely at country than at firm level, but “the most likely scenario is statistical insignificance”. - Dechezleprêtre and Sato (REEP, 2017): regulation induces clean-technology innovation, but “the resulting benefits do not appear to be large enough to outweigh the costs” for regulated firms. Short-run adverse effects on trade, jobs and productivity are “small relative to general trends”. - Zhang et al. (2024 meta-analysis of 58 papers): finds a positive effect of regulation on green innovation. - Ambec et al. (REEP, Jan 2013): the evidence remained “conflicting” twenty years on.

Taken together, the literature supports induced innovation (the “weak” Porter hypothesis) better than net gains for the firms themselves (the “strong” version). That matches footnote 7 more than the unhedged lessons. - Two later VC-sector episodes complicate the lesson. - Mercury in China. Minamata’s soft, per-unit target for VC production triggered a large body of research on mercury-free catalysts (He et al., 2025; Wang et al., Chem. Commun., 2025). Yet mercury use in China’s PVC sector reportedly doubled between 2004 and 2023, because the incumbent coal route is 7–17% cheaper (He et al., 2025). Research stimulated by a non-binding target has not displaced a cheaper incumbent. That is consistent with the chapter’s emphasis on clear and challenging rules. - The US HON rule. EPA’s 2024 rule for chemical plants requires fenceline monitoring for benzene, 1,3-butadiene, chloroprene, ethylene oxide, ethylene dichloride and VC (89 FR 42932, 16 May 2024). - Proclamation 10957 (17 July 2025) granted listed facilities a two-year exemption. It determined that “the technology to implement the HON Rule is not available” and that compliance timelines “would require shutdowns or massive capital investments”. - Proclamation 11041 (9 July 2026) repeated this. - The argument is structurally the 1974 one. This time it prevailed, through executive action. [Annex I, the facility list, is published as an image, so I did not check whether it includes VC or PVC plants.]

Verdict: partly held up. - Held up: the VC-specific lessons. - Contested: the general claim that clear, challenging regulation stimulates innovation without net harm. The evidence supports induced innovation more than net benefit to firms, as footnote 7 already concedes.

Weight. Moderate. VC had a cheap, available engineering fix, a concentrated industry and a signature disease; generalising lessons 2–3 needs those conditions to be stated. The 2025–26 US episode adds a technology-neutral lesson: claims that compliance technology “does not exist” recur and can win. Such claims should be tested against ex post outcomes, as the chapter did for 1974, rather than accepted or dismissed by analogy.

Claim 8. The precautionary principle should reverse the burden of proof so that “risk makers, not risk takers” show “at least beyond reasonable doubt” that chemicals are acceptably safe (Box 8.1, p. 187)#

Original claim. After the 1980 US Supreme Court benzene decision, OSHA had to show “significant risk” quantitatively; Justice Marshall’s dissent said this put “the burden of medical uncertainty squarely on the shoulders of the American worker”. The chapter says precaution “is designed in part to reverse this burden of proof” (Box 8.1, p. 187).

What happened since. - Partial moves in the chapter’s direction, 2016–24. - US. The Lautenberg Act (22 June 2016) amended TSCA. EPA “must make an affirmative finding on the safety of a new chemical … before it is allowed into the marketplace”. Existing-chemical risk evaluations exclude costs and must consider susceptible and highly exposed groups (EPA summary). That is a genuine burden shift for new chemicals. For existing chemicals like VC, the burden stays with EPA. - VC under TSCA. EPA designated VC a High-Priority Substance on 18 December 2024. More than 600 commenters asked for a ban, many citing the 3 February 2023 derailment and VC release at East Palestine, Ohio (89 FR 102900 ff.). A draft scope followed in January 2025; the risk evaluation is pending. - Related findings. EPA found “unreasonable risk” to workers from 1,2-dichloroethane, the VC feedstock, under 15 conditions of use (final risk evaluation, 5 May 2026). - Data call-in. EPA ordered manufacturers of 16 chemicals, including VC, to submit unpublished health and safety studies (TSCA §8(d), 13 Dec 2024). The reporting deadline has been extended to 21 May 2027. - Reversal and retrenchment, 2025–26. - US. - The HON exemptions (Claim 7). - A proposed rule (23 Sept 2025) to rescind or revise the 2024 amendments to the TSCA risk-evaluation procedures. - Executive Order 14303, “Restoring Gold Standard Science” (May 2025), invoked in later risk notices. - EU. - The Chemicals Industry Action Plan (8 July 2025) put the emphasis on competitiveness. It committed to “a targeted revision of REACH to simplify the rules and speed up the procedures for industry” (planned Q4 2025) and to a chemicals simplification omnibus (COM(2025) 531). [I did not find an adopted REACH revision proposal in the Publications Office index as of September 2026.] - The existing “no data, no market” rule in REACH (2006) is the EU’s main burden-shifting tool, and it predates the chapter. - No jurisdiction I checked has adopted anything like a “beyond reasonable doubt” safety standard for existing chemicals.

Verdict: contested. - Moved the chapter’s way: the burden shifted in part for new chemicals (US 2016) and for data generation (TSCA §8(d)). - Moved the other way: retrenchment followed in 2025–26. - Not adopted anywhere: the “beyond reasonable doubt” formulation.

Weight. A normative recommendation, still contested. The hindsight record supports the underlying diagnosis rather than the remedy: who bears the burden of uncertainty decides outcomes, and that allocation can be reversed by later political choices.

Claim 9. No-fault compensation schemes must be paired with other measures to extract information about corporate behaviour, because litigation document discovery is what revealed the secret history (p. 179)#

Original claim. Document discovery in compensation cases revealed “the real and until then secret activities of corporations”. Any move to no-fault administrative compensation “need[s] to be accompanied by other measures to extract information about corporate behaviour” (p. 179). The chapter rests heavily on Markowitz and Rosner (2002), whose documents came from litigation.

What happened since. - The discovery archive grew and went public. Toxic Docs (Columbia University and CUNY, version 1.0, 2018; Chowkwanyun, Markowitz, Rosner) hosts “millions of pages of once-secret corporate documents” on asbestos, PVC, benzene, silica and lead. - EU law paired compensation with disclosure. Directive (EU) 2024/2853 on liability for defective products (23 October 2024; transposition by 9 December 2026) does both: - it lets Member States use or create sectoral compensation schemes, “preferably not funded by public revenue”, where no liable person can be found (Art. 8(5)); - it requires courts to be able to order defendants to “disclose relevant evidence that is at the defendant’s disposal” once a claim is plausible, with trade-secret safeguards (Art. 9).

This is close to the chapter’s recommendation, but it covers defective products, not workplace or environmental exposures. - Regulatory information-forcing. The US TSCA §8(d) rule (Dec 2024) compels submission of unpublished VC health and safety studies. The deadline has been pushed to May 2027 (Claim 8), a reminder that information-forcing rules can themselves be delayed. - No test of the premise. I found no study showing that no-fault schemes suppress information in practice.

Verdict: held up as reasoning. Institutional developments since 2013 run in the direction the chapter urged. The empirical premise, that no-fault compensation reduces disclosure, remains untested.

Weight. Strong for the technology-neutral insight that independent access to internal records is often what closes the gap between what producers know and what others know. That is the chapter’s central mechanism (digest insight 1). Moderate for the specific institutional prescription.

Claim 10. A 1991 industry-funded study’s brain-cancer excess was publicly retracted by two authors in 1993 “at the request of the chemical industry, which had funded this study” (p. 188)#

Original claim. Wong et al. (1991) reported an “excess in cancer of the brain”. Two of the authors retracted it “at the request of the chemical industry”, concluding that the excess “was not likely related to the chemical” (Wong and Whorton, 1993) (p. 188). The chapter does not engage the retraction’s title or stated rationale, “diagnostic bias” (reference list, p. 202).

What happened since, and what the sources say. - The process. Sass et al. (2005), a co-author of this chapter, rely on a 1998 Houston Chronicle investigation (Morris, 1998). Its account, which Sass et al. quote: - Wong had published without the sponsor’s permission; - the Chemical Manufacturers Association’s Vinyl Chloride Panel was provoked; - documents show “a months-long effort to persuade Wong to recant”; - Wong denied being pressured.

A companion piece in the same journal issue was by H.C. Shah (Shah, Am. J. Ind. Med. 24:249, 1993). Sass et al. identify him as the manager of the industry panel.

So there is documented industry effort to obtain a retraction. “At the request of” compresses that into a finding of cause. [Wong and Whorton (1993) and the Houston Chronicle article were not accessible; both are relied on here as described by Sass et al.] - The rationale. The retraction argued that better diagnosis and reporting of brain tumours among monitored workers than in the general population could produce a spurious excess. - The science since. No later assessment has supported a causal link, which is consistent with the retraction’s substantive conclusion: - A 2003 meta-analysis gave a meta-SMR for brain cancer of 1.26 (0.98–1.62). Its authors (Boffetta, then at IARC, with industry-consultant co-authors) said an increase “cannot be excluded”. - IARC (2008 and 2012) found no strong evidence, citing inconsistency and no exposure–response. - The US cohort shows attenuation over successive updates. SMR 1.80 (to 1982, the Wong analysis), 1.42 (to 1995), 1.27 (0.95–1.66, to 2013), with 16 new deaths observed against 16 expected in the latest period. There was no exposure–response. The authors conclude that any excess “was limited to early years and unlikely related to VC exposure” (Mundt et al., 2017; ACC-funded). - The European cohort found no excess (SMR 0.93).

An early-period excess without exposure–response fits the diagnostic-bias explanation, or chance, better than causation. It does not prove either. - Contemporary rebuttal to the chapter’s source. An American Chemistry Council scientist replied to Sass et al. (Price, EHP, Oct 2005). The letter argued that the later European and US cohort updates, by NIOSH-, IARC- and NCI-affiliated scientists, “reach the opposite conclusion”. Subsequent evidence has broadly supported that reading on brain cancer.

Verdict: partly held up. - Supported: industry pressure for a retraction, by press reporting of internal documents. - Weakened: the chapter’s implicit suggestion that the retraction suppressed a real brain-cancer risk. - Unreliable as presented: the episode as evidence of suppressed science.

Weight. Low as an illustration of “manufactured doubt”. The episode is better used to show that motive and truth are separate questions. Documented pressure does not make the retracted finding correct, and industry funding does not make a conclusion wrong. The chapter’s wider, well-documented concealment story (1959–74) does not depend on this episode.

Minor factual checks#

• The OSHA standard. The chapter describes it three ways:
• “a ceiling of 1 ppm” (p. 186);
• “1 ppm, with a 15 minute excursion to 15 ppm” (Panel 8.1, p. 189);
• a reduction to “1 ppm in the workplace air” (p. 192).

The codified standard is 1 ppm averaged over 8 hours and 5 ppm over any 15 minutes (29 CFR 1910.1017(c)). The p. 186 and p. 189 descriptions are inaccurate. - IARC reviews “re-confirmed” the 1979 evaluation (p. 188; Table 8.2, p. 193). As explained under Claim 4, this is inaccurate for the non-liver sites. The chapter also does not cite IARC Vol. 100F (2012), which appeared before the report and confirmed only ASL and HCC. - “No safe exposure level” attributed to IARC 1979 (pp. 188, 193). IARC said there was no evidence of a threshold, not that none exists (Claim 5). - Lesson 3’s source. Sass et al. (2005) support the qualitative point about rapid compliance but do not contain the USD figures. - The young cohort (lesson 7, p. 192; average age 54, US EPA 2000). This concern has since been addressed: the US cohort (hired 1942–72) was followed to 2013 (Mundt et al., 2017). - “1–3 ppm” in 1974 (lesson 6). The US set 1 ppm in 1974. The EU value of 3 ppm (7.77 mg/m³) appears in Directive 2004/37/EC. [Its 1978 origin, Directive 78/610/EEC, was not re-fetched.]

Implications for the section’s transferable insights#

Grades refer to the LL2-08 digest’s “Transferable insights”.

1. Internal knowledge often runs years ahead of external knowledge, and the gap drives delay (pp. 180–186). Strong, reinforced. Toxic Docs has made more of the record public. Newer information-forcing tools (EU PLD Art. 9, TSCA §8(d)) show institutions acting on this mechanism, if slowly.
2. Uncertainty can be deliberately maintained by not doing, or by rewording, the research that would resolve it (p. 184). Strong for the 1960s acroosteolysis episode, which is documented. The Wong retraction (Claim 10) is a poor example of this mechanism, because later evidence supports the retracted conclusion.
3. Thinly based standards acquire authority and persist (pp. 182, 193). Strong. Hindsight adds a twist: the 1974 US limit of 1 ppm, a great advance at the time, has itself persisted unchanged for 50 years while low-dose evidence accumulated (Claim 5).
4. Distinctive harms get attributed and acted on; harms that add to common diseases may never be settled (p. 189). Strong, reinforced. ASL was settled at once. HCC took about 35 years and cirrhosis longer. Lung and brain cancers remain unresolved 50 years on, just as Panel 8.1 predicted.
5. Studies controlled by interested parties can miss effects through design (pp. 187–188). Moderate, and needs balance. The design critiques (cohort exclusions, healthy-worker effect, short follow-up) remain valid. But the industry-funded 2017 update reported the HCC association, and industry-linked work on brain cancer has been borne out. Funding source predicts little on its own; design and follow-up predict more.
6. Ex ante cost forecasts from regulated parties can be greatly overstated (pp. 187, 192). Moderate. The like-for-like overestimate is about four-fold, not 300-fold, and the general literature finds overestimates common but not demonstrably biased.
7. Action needs a conjunction of signature evidence, independent corroboration, institutions with a mandate and mobilised publics (p. 191). Moderate, reinforced. The pattern recurs in the 2024 TSCA prioritisation of VC. Mass comments followed a vivid public event (East Palestine, 2023), and EPA’s action came through an institution created for the purpose.
8. Controllability depends on how concentrated use is (p. 189). Moderate. Relocation of production to many carbide-route plants under weaker limits (Claim 2) shows control eroding as geography and technology change.
9. Standards should ratchet down over time (p. 192). Advocacy; partly enacted, once, in the EU.

Two lessons the chapter points towards but does not state: - Hazards travel with technologies, not just with substances. The same product made by an older process in a different regulatory setting brings back old exposures and adds new hazards (mercury). - Winning once does not settle the argument. The 1974 infeasibility argument lost. Its structural twin won in 2025–26. Whether ex ante claims were right can only be judged ex post, so retrospective evaluation needs to be routine rather than anecdotal.

Method and access notes#

• The web-search quota was exhausted before the check started. Sources were verified by direct retrieval:
• EU legal texts, from the EU Publications Office (cellar), with titles found through its SPARQL endpoint;
• US regulatory texts, from the Federal Register API and eCFR;
• IARC, from publications.iarc.who.int and monographs.iarc.who.int;
• ATSDR, from CDC servers (Chapters 1 and 7 of the 2024 profile);
• literature, from PubMed E-utilities, Europe PMC (full text where open access), OpenAlex and Crossref.
• Items marked in square brackets rest on secondary accounts or were not re-fetched. Three matter most:
• the provenance of the USD 90 billion and USD 278 million figures (Claim 3);
• the text of Wong and Whorton (1993) and the Houston Chronicle investigation (Claim 10);
• the IARC 1979 wording (Claim 5).
• ECHA’s November 2023 PVC investigation report is relevant to Claims 1, 5 and 8 but could not be retrieved (HTTP 403), so it is not used.
• Market statistics (Claim 1) come from EU trade-defence findings and scientific papers citing industry data, not from industry databases. They are indicative only.

Sources#

Claim 1: market and production route - Medrano-García, J.D. et al. (23 Aug 2023). Economic and environmental competitiveness of ethane-based technologies for vinyl chloride synthesis. ACS Sustainable Chemistry & Engineering. https://doi.org/10.1021/acssuschemeng.3c03006 (full text via Europe PMC PMC10481392) - He, D. et al. (16 Dec 2025). Hg-free electro-thermal cascade catalysis for acetylene upgrading into polyvinyl chloride precursor. Nature Communications. https://doi.org/10.1038/s41467-025-67389-x (full text via Europe PMC PMC12816569) - Ren, W. et al. (Feb 2014). Mercury transformation and distribution across a polyvinyl chloride (PVC) production line in China. Environ. Sci. Technol. 48:2321–2327. https://doi.org/10.1021/es404147c - Wang, H. et al. (May 2025). Current status and perspective of metal-free materials as catalysts in acetylene hydrochlorination. Chem. Commun. 61:7368–7383. https://doi.org/10.1039/d4cc06830k - Commission Implementing Regulation (EU) 2024/1896 of 11 July 2024 (provisional anti-dumping duty, PVC from Egypt and the USA). https://eur-lex.europa.eu/eli/reg_impl/2024/1896/oj (text retrieved via http://publications.europa.eu/resource/celex/32024R1896) - Commission Implementing Regulation (EU) 2025/36 of 9 January 2025 (definitive duty) [title and date only]. https://eur-lex.europa.eu/eli/reg_impl/2025/36/oj - Notice of initiation of an anti-dumping proceeding concerning imports of certain PVC originating in China, Korea, Mexico and Taiwan (OJ C, 23 Sept 2026; CELEX 52026XC04912). https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:52026XC04912 - European Commission (8 July 2025). A European Chemicals Industry Action Plan, COM(2025) 530. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:52025DC0530 - Minamata Convention on Mercury, Annex B Part II (text as published with Council Decision (EU) 2017/939 of 11 May 2017). https://eur-lex.europa.eu/eli/dec/2017/939/oj - Regulation (EU) 2017/852 of 17 May 2017 on mercury, Annex III. https://eur-lex.europa.eu/eli/reg/2017/852/oj

Claim 2: exposure in lower-resource settings - Dong, Y. et al. (Jan 2023). [Improvement of protective facilities in vinyl chloride units on liver injury status of occupational exposed group]. Wei Sheng Yan Jiu 52:100–114. https://doi.org/10.19813/j.cnki.weishengyanjiu.2023.01.017 - Dong, Y. et al. (22 Nov 2022). Improvements in protective measures in factories with acetylene hydrochlorination and ethylene oxychlorination techniques… Frontiers in Public Health 10:1053300. https://doi.org/10.3389/fpubh.2022.1053300 - Dong, Y. et al. (Mar 2024). [Liver impairment status toward workers exposed to vinyl chloride under different techniques]. Wei Sheng Yan Jiu 53:257–266. https://doi.org/10.19813/j.cnki.weishengyanjiu.2024.02.013 - Li, B.S. et al. (20 Nov 2022). [Analysis of occupational health examination status of people exposed to vinyl chloride]. Zhonghua Lao Dong Wei Sheng Zhi Ye Bing Za Zhi 40:850–853. https://doi.org/10.3760/cma.j.cn121094-20210625-00305 - Bolognesi, C. et al. (Oct 2017). The lymphocyte cytokinesis block micronucleus test in human populations occupationally exposed to vinyl chloride: a systematic review and meta-analysis. Mutat. Res. Rev. Mutat. Res. 774:1–11. https://doi.org/10.1016/j.mrrev.2017.07.003

Claim 3: costs - Wagner, W.E. (April 2004). Commons ignorance: the failure of environmental law to produce needed information on health and the environment. Duke Law Journal 53:1619 (n. 225, summarising McGarity and Ruttenberg, 80 Tex. L. Rev. 1997, 2031 (2002), and OTA 1995). http://hdl.handle.net/10535/2725 - Sass, J.B., Castleman, B. and Wallinga, D. (July 2005). Vinyl chloride: a case study of data suppression and misrepresentation. Environ. Health Perspect. 113:809–812. https://doi.org/10.1289/ehp.7716 (PMC1257639) - Simpson, R.D. (June 2014). Do regulators overestimate the costs of regulation? J. Benefit-Cost Analysis 5(2):315–332. https://doi.org/10.1515/jbca-2014-0027 - Morgenstern, R.D. (2018; online 10 Nov 2017). Retrospective analysis of U.S. federal environmental regulation. J. Benefit-Cost Analysis 9(2):285–304. https://doi.org/10.1017/bca.2017.17 - [Not accessed: US Office of Technology Assessment (Sept 1995), Gauging Control Technology and Regulatory Impacts in Occupational Safety and Health, OTA-ENV-635; McGarity, T.O. and Ruttenberg, R. (2002), Counting the cost of health, safety, and environmental regulation, Texas Law Review 80:1997.]

Claim 4: epidemiology - IARC (2012). Vinyl chloride. IARC Monographs Vol. 100F, pp. 451–478. https://publications.iarc.who.int/123 (chapter PDF: https://publications.iarc.who.int/download/mono100F-31_new.pdf) - ATSDR (January 2024). Toxicological Profile for Vinyl Chloride. https://wwwn.cdc.gov/TSP/ToxProfiles/ToxProfiles.aspx?id=282&tid=51 (Ch. 1: https://www.atsdr.cdc.gov/ToxProfiles/tp20-c1.pdf) - Mundt, K.A. et al. (Oct 2017). Quantitative estimated exposure to vinyl chloride and risk of angiosarcoma of the liver and hepatocellular cancer in the US industry-wide vinyl chloride cohort: mortality update through 2013. Occup. Environ. Med. 74:709–716. https://doi.org/10.1136/oemed-2016-104051 (PMC5629943; funded by the American Chemistry Council) - Fedeli, U. et al. (Jan 2019; online Nov 2018). Mortality from liver angiosarcoma, hepatocellular carcinoma, and cirrhosis among vinyl chloride workers. Am. J. Ind. Med. 62:14–20. https://doi.org/10.1002/ajim.22922 - Fedeli, U., Girardi, P. and Mastrangelo, G. (7 Sept 2019). Occupational exposure to vinyl chloride and liver diseases. World J. Gastroenterol. 25:4885–4891. https://doi.org/10.3748/wjg.v25.i33.4885 - Lotti, M. (May 2017). Do occupational exposures to vinyl chloride cause hepatocellular carcinoma and cirrhosis? Liver International 37:630–633. https://doi.org/10.1111/liv.13326 - Collins, J.J. et al. (Nov 2014). Surveillance for angiosarcoma of the liver among vinyl chloride workers. J. Occup. Environ. Med. 56:1207–1209. https://doi.org/10.1097/JOM.0000000000000247 (authors from The Dow Chemical Company) - Scarnato, C. et al. (2017). [Mortality study update of workers exposed to vinyl chloride in plants located in Ferrara and Ravenna]. Epidemiol. Prev. 41:271–278. https://doi.org/10.19191/EP17.5-6.P271.088 - Girardi, P. et al. (20 May 2022). Mortality for lung cancer among PVC baggers employed in the vinyl chloride industry. Int. J. Environ. Res. Public Health 19:6246. https://doi.org/10.3390/ijerph19106246 - Boffetta, P. et al. (June 2003). Meta-analysis of studies of occupational exposure to vinyl chloride in relation to cancer mortality. Scand. J. Work Environ. Health 29:220–229. https://doi.org/10.5271/sjweh.725 - Towle, K.M. et al. (Feb 2021). An ecological evaluation of vinyl chloride exposure and liver cancer incidence and mortality in Texas. J. Clin. Transl. Hepatol. 9:99–105. https://doi.org/10.14218/JCTH.2020.00073 (authors from Cardno ChemRisk)

Claim 5: thresholds and limits - Directive 2004/37/EC, Annex III (VCM 7.77 mg/m³, 3 ppm). https://eur-lex.europa.eu/eli/dir/2004/37/oj - Directive (EU) 2017/2398 of 12 December 2017, recital 13 and Annex (VCM 2.6 mg/m³, 1 ppm). https://eur-lex.europa.eu/eli/dir/2017/2398/oj - 29 CFR 1910.1017(c) (OSHA vinyl chloride standard; eCFR version history checked back to 2016). https://www.ecfr.gov/current/title-29/subtitle-B/chapter-XVII/part-1910/subpart-Z/section-1910.1017 - ATSDR (2024), Ch. 7, Regulations and guidelines. https://www.atsdr.cdc.gov/ToxProfiles/tp20-c7.pdf - Lang, A.L. et al. (March 2018). Vinyl chloride dysregulates metabolic homeostasis and enhances diet-induced liver injury in mice. Hepatology Communications 2:270–284. https://doi.org/10.1002/hep4.1151 - Liu, S. et al. (1 June 2023). Western diet unmasks transient low-level vinyl chloride-induced tumorigenesis… Toxicol. Appl. Pharmacol. 468:116514. https://doi.org/10.1016/j.taap.2023.116514 - Wang, Q. et al. (Jan 2013). Estimation of benchmark dose for micronucleus occurrence in Chinese vinyl chloride-exposed workers. Int. J. Hyg. Environ. Health 216:76–81. https://doi.org/10.1016/j.ijheh.2012.02.008 - Calabrese, E.J., Priest, N.D. and Kozumbo, W.J. (May 2021). Thresholds for carcinogens. Chem.-Biol. Interact. 341:109464. https://doi.org/10.1016/j.cbi.2021.109464 - (IARC 1979 wording as quoted in Sass et al. 2005, above.)

Claim 6: bioassays and alternatives - IARC (2019). Preamble to the IARC Monographs (amended January 2019), Part B, section 6(d). https://monographs.iarc.who.int/wp-content/uploads/2019/07/Preamble-2019.pdf - Krewski, D. et al. (2019). Concordance between sites of tumor development in humans and in experimental animals for 111 agents that are carcinogenic to humans. J. Toxicol. Environ. Health B 22:203–236. https://doi.org/10.1080/10937404.2019.1642586 - Goodman, J.I. (2018). Goodbye to the bioassay. Toxicology Research 7:558–564. https://doi.org/10.1039/c8tx00004b - Suarez-Torres, J.D., Orozco, C.A. and Ciangherotti, C.E. (2021). The 2-year rodent bioassay in drug and chemical carcinogenicity testing: performance, utility, and configuration for cancer hazard identification. J. Pharmacol. Toxicol. Methods 110:107070. https://doi.org/10.1016/j.vascn.2021.107070 - Audebert, M. et al. (10 July 2023). New approach methodologies to facilitate and improve the hazard assessment of non-genotoxic carcinogens: a PARC project. Frontiers in Toxicology 5:1220998. https://doi.org/10.3389/ftox.2023.1220998 - Bourcier, T. et al. (11 April 2024). ICH S1 prospective evaluation study: weight of evidence approach to predict outcome and value of 2-year rat carcinogenicity studies. Frontiers in Toxicology 6:1353783. https://doi.org/10.3389/ftox.2024.1353783 - US FDA (10 April 2025). FDA announces plan to phase out animal testing requirement for monoclonal antibodies and other drugs. https://www.fda.gov/news-events/press-announcements/fda-announces-plan-phase-out-animal-testing-requirement-monoclonal-antibodies-and-other-drugs - European Commission (3 June 2026). Communication: Roadmap towards phasing out animal testing for chemical safety assessments (CELEX 52026XC03087; with SWD(2026) 144). https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:52026XC03087

Claim 7: innovation and costs - Ambec, S., Cohen, M.A., Elgie, S. and Lanoie, P. (Jan 2013). The Porter hypothesis at 20. Rev. Environ. Econ. Policy 7:2–22. https://doi.org/10.1093/reep/res016 - Cohen, M.A. and Tubb, A. (2018; online 10 Nov 2017). The impact of environmental regulation on firm and country competitiveness: a meta-analysis of the Porter hypothesis. JAERE 5(2):371–399. https://doi.org/10.1086/695613 - Dechezleprêtre, A. and Sato, M. (2017; online 16 May 2017). The impacts of environmental regulations on competitiveness. Rev. Environ. Econ. Policy 11:183–206. https://doi.org/10.1093/reep/rex013 - Zhang, W.X. et al. (8 Feb 2024). Revisiting the Porter hypothesis: a multi-country meta-analysis… Humanities and Social Sciences Communications 11:232. https://doi.org/10.1057/s41599-024-02671-9 - US EPA (16 May 2024). NSPS and NESHAP for the Synthetic Organic Chemical Manufacturing Industry and Group I & II Polymers and Resins Industry (HON rule), 89 FR 42932. https://www.federalregister.gov/documents/2024/05/16/2024-07002 (full text: https://www.federalregister.gov/documents/full_text/text/2024/05/16/2024-07002.txt) - Proclamation 10957 of 17 July 2025, Regulatory relief for certain stationary sources to promote American chemical manufacturing security, 90 FR 34587 (23 July 2025). https://www.federalregister.gov/documents/2025/07/23/2025-13890 - Proclamation 11041 of 9 July 2026 (same title), 91 FR 44719 (16 July 2026). https://www.federalregister.gov/documents/2026/07/16/2026-14452

Claim 8: burden of proof - US EPA. Highlights of key provisions in the Frank R. Lautenberg Chemical Safety for the 21st Century Act (Pub. L. 114-182, 22 June 2016). https://www.epa.gov/assessing-and-managing-chemicals-under-tsca/highlights-key-provisions-frank-r-lautenberg-chemical - US EPA (18 Dec 2024). High-priority substance designations under TSCA and initiation of risk evaluation (includes vinyl chloride), 89 FR 102900 ff. https://www.federalregister.gov/documents/2024/12/18/2024-29830 - US EPA (16 Jan 2025). Vinyl chloride; draft scope of the risk evaluation under TSCA. https://www.federalregister.gov/documents/2025/01/16/2025-00948 - US EPA (5 May 2026). 1,2-Dichloroethane; final risk evaluation under TSCA. https://www.federalregister.gov/documents/2026/05/05/2026-08682 - US EPA (23 Sept 2025). Procedures for chemical risk evaluation under TSCA, proposed rule. https://www.federalregister.gov/documents/2025/09/23/2025-18431 - European Commission (8 July 2025). Chemicals Industry Action Plan, COM(2025) 530 (above); proposal amending CLP and other regulations for simplification, COM(2025) 531. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:52025PC0531

Claim 9: information extraction - Toxic Docs (version 1.0, 2018; Columbia University and City University of New York). About. https://www.toxicdocs.org/about - Directive (EU) 2024/2853 of 23 October 2024 on liability for defective products, Arts. 8(5), 9, 22. https://eur-lex.europa.eu/eli/dir/2024/2853/oj - US EPA (13 Dec 2024). Certain existing chemicals; request to submit unpublished health and safety data under TSCA (16 chemicals incl. vinyl chloride). https://www.federalregister.gov/documents/2024/12/13/2024-29406 - US EPA (22 May 2026). Reporting deadline extension for the health and safety data reporting rule under TSCA §8(d) (to 21 May 2027). https://www.federalregister.gov/documents/2026/05/22/2026-10263

Claim 10: the Wong retraction - Wong, O., Whorton, M.D., Foliart, D.E. and Ragland, D. (1991). An industry-wide epidemiologic study of vinyl chloride workers, 1942–1982. Am. J. Ind. Med. 20:317–334. https://doi.org/10.1002/ajim.4700200305 - Wong, O. and Whorton, M.D. (Aug 1993). Diagnostic bias in occupational epidemiologic studies: an example based on the vinyl chloride literature. Am. J. Ind. Med. 24:251–256 [not accessible; title and bibliographic record via Crossref]. https://doi.org/10.1002/ajim.4700240215 - Shah, H.C. (Aug 1993). Diagnostic bias in occupational epidemiologic studies [comment on Wong et al. 1991]. Am. J. Ind. Med. 24:249–250. https://doi.org/10.1002/ajim.4700240214 - Sass, J.B. et al. (2005), above (quoting Morris, J., Houston Chronicle, 1998). - Price, C.M. (Oct 2005). Vinyl chloride and U.S. EPA research [letter; author employed by the American Chemistry Council]. Environ. Health Perspect. 113:A653–A654. https://doi.org/10.1289/ehp.113-a653 - Mundt et al. (2017); Boffetta et al. (2003); IARC (2012), above.