LL2-08 — Ch8 Vinyl chloride: a saga of secrecy#

Late lessons from early warnings: science, precaution, innovation (EEA Report No 1/2013), Chapter 8. Report pages 179–202 (PDF pages 181–204). Main text and panels run pp. 179–196; references run pp. 196–202.

Read in full from the text extract (last marker seen: PDF 204 / report p. 202). I checked the following visually against the PDF: the chapter summary box (p. 179), Figure 8.1 (p. 180), Table 8.1 (p. 185), Box 8.1 (p. 187), Panel 8.1 (pp. 189–190), Box 8.2 and lessons 1–2 (p. 191), lessons 2–7 (p. 192), Table 8.2 and footnotes 9–10 (p. 193), and Panel 8.2 (pp. 194–196). The extract had misaligned Table 8.1: melanomas, acoustic duct tumours and lymphomas/leukaemias are hamster-only entries. The rest matched.

Material not in the chapter is marked [external — verify].

Authors and standpoint#

Chapter authors: Morando Soffritti, Jennifer Beth Sass, Barry Castleman and David Gee (p. 179). The chapter gives no affiliations. The report’s author-biography annex (same PDF) gives these:

• Morando Soffritti, MD: Scientific Director of the Ramazzini Institute (Cesare Maltoni Cancer Research Center, European Ramazzini Foundation, Castello di Bentivoglio). He has served on IARC working groups and is adjunct professor at Mount Sinai (annex, p. 698). His listed address is the Cesare Maltoni Cancer Research Center, so he writes from the institution that carries the name of Cesare Maltoni, the chapter’s scientific protagonist, and whose research much of the chapter rests on. (The annex does not say Maltoni founded it or that Soffritti succeeded him [external — verify].) Footnote 1 says many of the documents used are held in the Maltoni Center’s collection (p. 179).
• Jennifer Beth Sass: senior scientist at the Natural Resources Defense Council (NRDC). According to the annex she “advocates for health protective regulations” (annex, p. 697). She is first author of Sass, Castleman and Wallinga (2005), “Vinyl chloride: a case study of data suppression and misrepresentation”. That paper is a principal source for section 8.5 and lesson 3 (pp. 188, 192).
• Barry Castleman: environmental consultant (ScD, Johns Hopkins School of Public Health) who has worked “with public interest groups around the world” on asbestos and chemical hazards (annex, p. 688). He co-authored Castleman and Ziem (1988), “Corporate influence on threshold limit values”, which is the source for the ACGIH/TLV narrative (pp. 182, 184, 193).
• David Gee: has worked with the EEA since 1995 on science, policy and emerging issues. The annex calls him “the catalyst, EEA editor and a chapters author” of both Late Lessons volumes. From 1974 he worked for trade unions and NGOs, and he is a former Director of Friends of the Earth UK (annex, p. 689). He co-authored Fox, Gee et al. (1982, a GMB union publication), cited for the healthy-worker effect (p. 188), and Doyal … Gee (1983), cited in footnote 6 (p. 190).

Evident stance. The chapter is openly prosecutorial toward the chemical industry: the title is “a saga of secrecy” and section 8.1 is “A veil of secrecy”. It is strongly pro-precaution and pro-animal-bioassay, and it sides with labour. It is not one-sided, though. It credits industry with a “very prompt response” once the liver cancer evidence became public, including funding cancer testing and complying with a large cut in limits (p. 179). It also acknowledges “progressive business” (p. 193, fn 9). Four features shape the standpoint:

1. Heavy reliance on one secondary source, Markowitz and Rosner’s Deceit and Denial (2002), for the internal-document story.
2. Reliance on the authors’ own earlier work: Sass et al. 2005; Castleman and Ziem 1988; Soffritti et al. 1999 and 2002; Fox, Gee et al. 1982.
3. Close identification with the Maltoni/Ramazzini tradition.
4. No industry, regulator or dissenting panel. Some other 2013 chapters include industry or regulator responses; this one does not.

Panels: - Panel 8.1: “VCM: a personal perspective of a year in the United Kingdom after the Goodrich announcement of 1974”, by Charlie Clutterbuck (pp. 189–190). In 2013 he was a Research Fellow in Food Policy at City University. In the 1970s he worked for the British Society for Social Responsibility in Science (BSSRS), founded Hazards Bulletin, and spent 20 years in trade-union education (annex, p. 688). He calls himself one of the “‘radical’ scientists” (p. 190). The panel is a first-person activist memoir. It is broadly consistent with the chapter but adds its own analysis, notably why VC got unusual investment. It also gives some credit to industry. - Panel 8.2: “Value of animal testing for identifying carcinogens”, by James Huff (pp. 194–196). He was an NIEHS staff scientist, formerly Chief of the IARC Monographs Programme and coordinator of the NTP bioassay programme, and a Fellow of the Collegium Ramazzini. He received APHA’s David P. Rall Award “for science-based advocacy in public health” (annex, p. 692). The panel is a strong defence of long-term rodent bioassays. It is aligned with the chapter and with the Ramazzini tradition, and is not an independent or critical view.

Section-by-section notes#

Chapter summary box (p. 179)#

The box frames five themes:

1. Early warnings were hidden. Warnings in the 1950s–60s about harm to workers’ skin and bones and to animals’ livers were “initially hidden from other workers and regulators”. This happened “despite some early misgivings by company experts whose advice was initially ignored by their employers”. The pattern repeated with liver cancer, “revealed by long-term animal studies and by an attentive and concerned company physician”.
2. The contrast with most other histories: a “very prompt response from the global chemical industry” once the liver-cancer evidence was published. The industry funded cancer testing and complied with a large cut in limits. The box also claims “early evidence of reproductive effects”, which the body develops only briefly (pp. 185, 191).
3. Presaging later corporate behaviour: “greatly exaggerated estimates of the likely costs of complying”; “a frequent mismatch between the position of the trade association and that of many, more progressive companies”; “but also some relatively quick corporate responses”.
4. Two legal aspects, “more common in the US” but “also valuable for Europeans”: - The “potentially positive role” of judicial review of regulatory proposals in providing “a societal judgement about the behaviour of corporations”, covering “judgements about the state of the science and what society should do with it”. - Document discovery in compensation cases as the means of revealing “the real and until then secret activities of corporations”.
5. Recommendation: “Any proposals to promote justice for victims of environmental and health harms via no fault administrative arrangements need to be accompanied by other measures to extract information about corporate behaviour.”

Footnote 1: Rosner and Markowitz supplied original documents. Many documents are held in the Cesare Maltoni Cancer Research Center collection, and the Deceit and Denial industry documents are public at deceitanddenial.org.

8.1 A veil of secrecy (p. 180)#

• January 1973: NIOSH issued a mandatory “request for information” requiring all VC hazard information to be submitted.
• The 1972 secrecy agreement. This put industry “in a difficult position”. In 1972 Dow and other US manufacturers had signed a “secrecy agreement” with European companies, collected by ICI, to hold in confidence new Italian animal-testing research (Markowitz and Rosner 2002, pp. 182–183).
• Figure 8.1 (checked visually) is Dow’s signed “Secrecy Agreement, Re: Animal Testing Research – Vinyl Chloride Monomer”, dated 13 November 1972 and signed by Dow’s Director of Research. It refers to R. L. Lindsell’s letter of 16 August 1972 to the MCA and pledges “best efforts to hold such information strictly in confidence within Dow” until the “European sponsors” consent. It carves out “information already known to Dow” or “independently developed by Dow”. The agreement bound the companies, not Maltoni, who presented early results at a scientific symposium in April 1973 (p. 186).
• 14 November 1972 meeting at MCA headquarters, Washington. The findings came from Maltoni’s team at the Bentivoglio castle laboratory: “the first in the world” to see liver and kidney cancer in animals exposed to 250 ppm, half the permitted workplace level.
• “While the companies did not doubt the evidence from these studies, the Secrecy Agreement did not allow them disclose it”. This is an inference about corporate belief.
• ICI’s D. M. Elliott “insisted that the work tables be swept clear of paper for note taking before he would discuss anything … Such was done”.
• Framing. The MCA’s 500 ppm limit of 1954 “remained unchanged for two decades”. The industry had animal data in the 1950s showing the limit excessive, and “By 1961 company toxicologists were internally recommending a limit no higher than 50 ppm”. OSHA set 1 ppm only in 1974. “In the interim, workers were exposed to vinyl chloride concentrations that caused excruciatingly painful bone disease and cancer.”

8.2 1930–1999: rapid growth in PVC output (pp. 180–182)#

• Uses. More than 95% of VC goes to PVC. Construction is the main use; with plasticisers PVC is also used in medical devices (bypass, IV, blood-bag and dialysis tubing), consumer goods, vehicles and packaging (pp. 180–181).
• Output:

Forecast growth was ~4.7%/yr for 2010–15 and 4.2%/yr for 2015–20 (SRI Consulting 2011) (p. 181). - Geography. Asia overtook Western Europe and North America by 2005. Asia-Pacific holds ~54.8% of capacity and China is “the largest producer and exporter”; Europe and North America are “almost stagnant” (GBI Research 2010) (p. 181). - Exposure. Since closed-loop polymerisation arrived in the late 1970s, worker exposures in North America and Western Europe have “decreased substantially”; before that they had been “higher than 38 ppm”. Processing-plant exposures are usually much lower than those in VCM/PVC production, and current Western levels are “generally below 0.4 ppm” (the sentence leaves unclear whether this applies to processing only or to all plants). But “In low and medium-resource countries, older technologies have continued to be used and therefore high exposures probably occur” (no data given). Ambient air is normally <0.004 ppm but higher near plants (p. 181). - Consumer routes and early consumer-side actions: - VC was used as an aerosol propellant until the mid-1970s (including Clairol hair spray), causing “elevated exposures and eventually cancer in hairdressers” (Castleman 1981; Infante et al. 2009). - FDA warned about small-room aerosol use (1974). EPA made an emergency suspension of VC as a pesticide propellant (1974), and CPSC banned it in consumer aerosols (1978). - In May 1973 FDA, citing the Delaney amendment, suspended PVC bottles for alcoholic beverages. - VC was once detected in oils, vinegars, margarine and bottled water. It is no longer detected, since the US limits VC in PVC food packaging to ≤1 ppm (ATSDR 2006). - VC also forms in landfills from degrading chlorinated solvents and occurs in tobacco smoke (p. 181). - Residual monomer in finished PVC “can still pose dangers” (p. 182).

8.3 1930–1961: early warnings from animals and humans meet industry indifference (pp. 182–186)#

8.3.1 Earliest warnings (p. 182). - Acute effects. Narcosis in animals at 5,000–120,000 ppm, with organ congestion, pulmonary oedema and cardiac irregularities in dogs. In humans: giddiness, nausea, headaches, “almost immediate death” in an accidental poisoning (Danzinger 1960), and burns from spraying (Harris 1953). - First chronic signal: Soviet scientists in the late 1940s. Hepatitis-like liver inflammation in 15 of 48 workers (Tribukh et al. 1949), and angioneuropathy at 5–15 ppm (Filatova and Gronsberg 1957). USSR plants cut levels from 40 ppm to 4–15 ppm by 1966 (Filatova 1966). The chapter does not say whether Western actors knew of this work.

8.3.2 Ignoring evidence from animal studies (pp. 182–183). - The 500 ppm TLV (1954), unchanged for 20 years. Before OSHA (1971), TLVs were set by the ACGIH, “a volunteer organisation with no formal ties to the government” whose members included officials, academics and industry consultants. TLVs “represented what the industry felt was achievable but not necessarily health protective” (Markowitz and Rosner 2002; Castleman and Ziem 1988). The text variously says the MCA (or “CMA”) “established” or “in effect set” the TLV (pp. 182–183). - “A turning point in the history of VC should have occurred by 1961 at the latest.” Dow’s Torkelson, Oyen and Rowe (1961) reported liver abnormalities in rats at 500 ppm, increased liver weight at 100 ppm, and liver enlargement and degeneration in rabbits at 200 ppm (p. 182). - Known two years earlier. In November 1959 Union Carbide consultant Henry Smyth reported an “off-the-record phone call” from Dow’s V. K. Rowe. VC was “more toxic than has been believed”; “Even 100 ppm produced organ weight changes and gross pathology”. Dow was weighing the implications for “(food) packaging uses”. Smyth noted the 500 ppm TLV rested on a single 1930s guinea-pig study, and he asked that the information stay confidential until Dow published (p. 182). In May 1959 Rowe told B. F. Goodrich’s W. E. McCormick that 500 ppm would cause “appreciable injury” to full-time workers (pp. 182–183). - Limit history. The TLV went from 500 ppm (1954) to 200 ppm (1971) to 1 ppm (1974). Because OSHA adopted the 1968 TLVs as its first PELs, the enforceable US limit stayed at 500 ppm until 1974 (p. 183).

8.3.3 1963–1971: secrecy over evidence of bone disease in workers (pp. 183–184). - 1963, Suciu et al. (Romania). A study of 168 PVC workers first described VC-caused acroosteolysis (AOL), an “extremely rare bone disease characterised by excruciatingly painful bone reabsorption” with skin and vascular changes. - Solvay. “At around the same time” Solvay observed two AOL cases at a Belgian factory and “informed other VC/PVC manufacturers”. The same pathology appeared at a company factory in Romania, and “several cases” at other Solvay factories in 1962–65 (Maltoni 1974a). But “no reports were made public, or to the independent scientific community, governments or workers” (p. 183). This sits awkwardly with the same page’s account of a Solvay doctor publishing despite Goodrich’s efforts. The reference list includes Cordier et al. (1966), a French paper on AOL in two autoclave cleaners, which is not cited in the text (p. 197). It is probably that publication (my inference). - 1964, B. F. Goodrich. Dr John Creech, doing medical checks for Goodrich, followed up a worker’s complaint of “painfully tender fingers” and found four AOL cases in one plant area, “obvious evidence of a link” for him. Goodrich engaged Robert Kehoe of the Kettering Laboratory, who also figures in the lead-in-petrol chapter (fn 3). Kehoe called AOL an “entirely new” occupational disease (p. 183). - Documented secrecy (Markowitz and Rosner, citing MCA papers) (p. 183): - Goodrich’s Medical Director (1964): “determine as quietly as possible whether similar conditions existed … We do not wish to have this discussed at all”. - A Monsanto official (1966): “I am sure that Dr Nessel can prepare these people with an adequate story so that no problem will exist”. - Goodrich’s vice-president tried to “discourage or edit” a Solvay doctor’s publication, unsuccessfully. - June 1966 private meeting of European and US industry medical staff. They learned that 1% of PVC workers and 6% of VCM vat cleaners had AOL. Goodrich asked firms to “use discretion in making the problem public”. Industry feared the public would see PVC food packaging as hazardous, which “would have been very damaging for industry” (p. 183). - 1967 Goodrich article (31 AOL cases). Per Creech’s later legal testimony, the first draft named VC as the cause; the published version did not (p. 183). - 1969 MCA-funded University of Michigan report (presented confidentially). It found that AOL could affect tissue beyond the fingers, and that the odour threshold is 4,000 ppm, not 400 ppm “as previously thought”. It found that “it could not be assumed that workers would be safe from disease” at 500 ppm, and that ventilation should keep VC below 50 ppm. Because this implied causation, the MCA Occupational Health Committee voted 7–3 to refuse it. It was accepted unanimously only with: “Inasmuch as the etiologic agent of the disease is unknown, a level of vinyl chloride below 50 ppm should be used as an index of adequate ventilation” (p. 184). - Repeated 50 ppm recommendations: - Dow, 1961. - Robert Scala (Esso), 1965, leading to an ACGIH proposal. - In a 1966 meeting with ACGIH TLV chairman Stokinger, “50 member companies said 50 ppm was too low”, and the change “was put off” (Mellon Institute report, 1–2 Feb 1966; Castleman and Ziem 1988). - ACGIH moved only in 1971, to 200 ppm, on unpublished worker liver data. Dow claims an internal 50 ppm limit from 1961, but “it is known that this limit was exceeded” (p. 184). - Explanation (p. 184): “the toxicologists and company doctors do not run the companies; corporate toxicological policy decisions are made by corporate executives as business decisions.” No effort was made to limit residual monomer in resins. A 50 ppm limit could raise costs, and “Disclosure and recognition of the toxicity of VC could also endanger sensitive markets for PVC food packaging and medical applications”, as well as aerosols. - 1971 published Michigan report (Archives of Environmental Health). It dropped the non-protective-limit and odour findings and called the cause unknown. It recommended research, but “B.F. Goodrich executives explicitly decided not to accept any proposals for additional research … This enabled Goodrich and the rest of the industry to act as if the cause was ‘unknown’” (p. 184).

8.3.4 1970–1983: recognising VC as a carcinogen (pp. 184–186). - Viola. Solvay asked Prof. P. L. Viola, its Livorno factory doctor (Regina Elena Institute, Rome), to study AOL long-term. From 1967 he exposed rats to 30,000 ppm (4 h/day, 5 days/week, 12 months). The result was “a surprise”: 70% had malignant skin and lung tumours after 10 months. He reported this in Houston in May 1970 (p. 184). A study commissioned by a company produced the first cancer signal. - Maltoni. Maltoni (Director, Addarii Institute of Oncology, Bologna) contacted Viola at once. His worker surveillance was already finding abnormal cells in the saliva of VC-exposed workers (1970) (pp. 184–185). - Funding the large study. Montedison “expressed interest in supporting research”. ICI, Solvay and Rhône-Progil (text: “Rhône-Prugil”) joined to form the European Cooperative Group (p. 185). - Maltoni’s programme (July 1971 onward, Bentivoglio). Partly funded by the Bologna Hospital Administration. It used multiple routes, doses, durations, species, strains and ages; ran 10 years; used >7,000 animals observed to natural death; examined >200,000 slides; and cost >USD 2 million, “mainly from public institutions and also from the chemical industry” (p. 185). The terms of the industry co-funding, and their relation to the secrecy agreement, are not discussed. - Results (pp. 185–186): 1. Carcinogenic in rats, mice and hamsters: mammary, lung, Zymbal gland, skin, liver angiosarcoma and others. 2. Carcinogenic by inhalation, ingestion and prenatal exposure, and possibly by injection. Footnote 4 notes that foetal toxicity recurs across chapters. 3. Tumour incidence correlated with concentration, dose and duration. 4. Newborns were more susceptible (liver). 5. Multi-organ effects in rats. 6. Tumours down to 10 ppm inhaled and 0.3 mg/kg ingested, “Although these did not reach statistical significance, they were considered to be biologically relevant” (Maltoni et al. 1977). A few paragraphs later the same page, citing the same source, gives the lowest ingestion dose with tumours as 1 mg/kg. This is an unreconciled inconsistency (p. 186). - Table 8.1 (Maltoni et al. 1984; checked visually; + clear, (+) borderline): - Liver angiosarcomas etc.: + in all three species. - Angiosarcomas at other sites: + + (+). - Mammary malignant tumours: rat and mouse. - Zymbal gland, nephroblastoma and encephalic neuroblastoma: rat only. - Hepatomas: rat +, mouse (+). - Forestomach papillomas: + (+) +. - Lung adenomas: rat and mouse. - Cutaneous epithelial tumours: (+) in all three. - Melanomas: hamster (+). Acoustic-duct tumours: hamster +. Lymphomas and leukaemias: hamster (+) (p. 185). - April 1973. Maltoni presented early results at the Bologna symposium; “This presentation does not appear to have attracted the attention of government regulators or the general scientific community.” Manufacturers stayed under the secrecy agreement until 1974 (p. 186). - 17 July 1973 NIOSH meeting (Dow, Ethyl, Union Carbide, MCA). Viola’s 30,000 ppm cancers were mentioned; Maltoni’s 250 ppm cancers were not (Markowitz and Rosner, p. 189). Maltoni’s 250 ppm data were published only in 1974 (p. 186). - 22 January 1974. B. F. Goodrich alerted government, and the next day announced it was investigating three Louisville workers’ deaths from liver angiosarcoma. Creech had actually identified four (1968–73), all in PVC polymerisation. The cancer caused only about two dozen US deaths a year, “making four deaths among the several hundred plastics workers akin to an epidemic”. It was also appearing in Maltoni’s animals at half the workplace limit (p. 186). - Seven cases in all. Creech ultimately found seven cases (1964–74), all among “pot cleaners” lowered into 10-foot reactor tanks with no more than a six-foot opening, to chip out residue (p. 186).

8.4 1974: governments swift response (pp. 186–188)#

• 15 Feb 1974 OSHA hearing. Maltoni’s testimony was “the first time that his findings came officially to the attention of the public and government officials”. Selikoff (Mount Sinai) thanked him for “the beautiful piece of work you have done” (p. 186).
• Creech and Johnson (1974) published four deaths. An editor’s footnote credited Maltoni with first showing the link in mammals. The chapter infers Goodrich’s notification “was influenced by Maltoni’s experimental data”. Maltoni’s full results were published in 1981 and 1984 (p. 186).
• OSHA limits. On 5 April 1974 OSHA required that no employee be exposed above 50 ppm (cited in text as “Federal Register, 1974”). The chapter does not call this an emergency temporary standard [external — verify]. “A few months later the OSHA standard was lowered again to a ceiling of 1 ppm”, after Maltoni’s May 1974 New York Academy of Sciences report of cancer at 50 ppm (Maltoni and Lefemine 1975) (p. 186). Citation note: the reference list’s “Federal Register, 1974” entry is the 4 October 1974 final standard (p. 198). A separate “NARA, 1974” entry for the 5 April 1974 Federal Register (p. 200) is listed but never cited, so the in-text citation for the April action appears to point to the wrong entry. The October date for the 1 ppm standard comes only from the reference list. [external — verify:] the final standard is usually described as a 1 ppm 8-h TWA with a 5 ppm 15-minute ceiling. Panel 8.1 says “1 ppm, with a 15 minute excursion to 15 ppm” (p. 189); lesson 6 says “1–3 ppm” (p. 192).
• 1976 low-dose results. Tumours “considered biologically correlated (but not statistically significant)” down to 10 ppm inhaled and 1 mg/kg ingested. “None of the specifically VC-related tumours were found at doses of 1 ppm (by inhalation) and 0.03 mg/kg by ingestion” (Maltoni et al. 1977) (p. 186). Confirmatory bioassays followed (Keplinger 1975; Lee 1978; Feron and Kroes 1979) (p. 187).
• Industry appeal. “Despite this very strong evidence” the US plastics industry appealed the 1 ppm standard. It lost in January 1975 “and received a scathing condemnation from the judge” (p. 187).
• Box 8.1: “The judicial critique of industry’s lack of action on VCM” (p. 187; checked visually):
• Retired Justice Tom Clark, sitting on the 2nd Circuit, rejected all the basic industry arguments. The box says he was “Noting the industry policy of delay, feigned ignorance and irresponsibility”. That phrase is not in quotation marks in the source (checked visually). It is the box’s characterisation, drawn from Markowitz and Rosner, and not a quotation from the ruling. The words the box does quote as the judge’s are that “strong warning signals … of long before” had been ignored, and that despite research since 1949, including the AOS episode (sic, AOL), “nothing was done” (Markowitz and Rosner 2002, pp. 222–223).
• The proper standard was “in dispute” and “on the frontiers of scientific knowledge”, but “the factual finger” pointed to a low limit on the animal evidence. The standard took effect on 1 April 1975.
• Setback: the Supreme Court benzene decision (2 July 1980) required OSHA to show quantitatively a “significant risk of material health impairment”. OSHA lost 1 ppm for benzene. Marshall, dissenting: “the burden of medical uncertainty squarely on the shoulders of the American worker”.
• Authors’ gloss: “The precautionary principle is designed in part to reverse this burden of proof of harm so that risk makers, not risk takers, have to show, at least beyond reasonable doubt, that chemicals and other stressors are acceptably safe”.
• [external — verify:] the cases are Society of the Plastics Industry v. OSHA, 509 F.2d 1301 (2d Cir. 1975) and Industrial Union Dept. v. American Petroleum Institute, 448 U.S. 607 (1980).
• Compliance costs (p. 187). “Industry responded quickly and easily”. “An analysis in 2000” (unsourced) put compliance costs at “only USD 278 million”, against industry’s earlier estimate of “up to USD 90 billion and 2 million jobs”. The chapter quotes the NYT (Rattner, “Did Industry cry Wolf?”, 28 Dec 1975): “not one of the doomsday predictions (from industry) has proven accurate”. Supplies expanded, prices held and the industry grew. Lesson 3 (p. 192) instead gives “USD 60 million annually and 2 million jobs (Washington Post, 1974)”, which is inconsistent.
• Trade-association tactic (p. 187). “This public relations tactic of using an industry trade association to threaten and sue regulators and issue claims no member company would dare make in its own name would become more familiar after 1974” (cross-refs Chs 3 and 7). “Industry has generally exaggerated the projected costs of meeting new regulations” (cross-ref Ch 23).
• Rarity and registry data. Angiosarcoma is very rare, 20–30 US cases a year. In 1975–78 other cases were found in the US and elsewhere. The ICI-maintained registry recorded 103 cases among VC-exposed workers “in various countries”, 1974–1983 (Stafford 1983). Cases were also reported in Europe and Asia (p. 187).
• Industry-commissioned Tabershaw and Gaffey (1974). Overall mortality was 75% of expected and no cause showed a statistically significant excess. But digestive (mainly angiosarcoma), respiratory, brain and unknown-site cancers and lymphomas “occurred more often than expected” in the most exposed (pp. 187–188).
• Critique (p. 188):
• NIOSH: the study left out ~75% of the longest-exposed workers. Independent investigator Joseph Wagoner, adding them, found a 57% excess of cancer deaths.
• Wagoner told a Senate committee the study diluted risk with one- to two-year workers. The chapter calls it the “Tabershaw-Cooper study” here and Tabershaw and Gaffey elsewhere; these appear to be the same study. He restricted his analysis to workers exposed more than 15 years, “looking for the latent effects”.
• Healthy-worker effect. Expected cancer mortality in long-exposed workers, absent occupational risk, is ~75–80% of the general-population rate (Fox et al. 1982).
• Cross-refs to Ch 26 on methods that avoid false positives more than false negatives, and Ch 2 on false alarms.

8.5 1974–present: brain cancer, other cancers and reproductive effects (pp. 188–193)#

• Continued downplaying. Industry “was no longer able to deny” angiosarcoma, but Sass et al. (2005) “document continued efforts to suppress or play down the evidence of other cancer types, particularly brain cancers”. A 1976 Ethyl internal memo acknowledged liver angiosarcoma, brain and lung cancer risks (p. 188).
• IARC 1979. The chapter reports: Group 1, “associated with cancers of the liver, brain, lung and haemo-lymphopoietic system”, and “no threshold or safe level of exposure”. This was re-confirmed in 1987 and 2007/2008. The 2008 update notes greater early-life susceptibility in animals (p. 188). The attributions to IARC 1979 should be checked against the monograph.
• Wong et al. (1991), industry-funded, reported an “excess in cancer of the brain”. “However, at the request of the chemical industry, which had funded this study, two of the authors made a public retraction two years later”. They concluded the excess “was not likely related to the chemical” (Wong and Whorton 1993) (p. 188). The retraction’s title, “Diagnostic bias in occupational epidemiologic studies…” (p. 202), signals a methodological rationale that the chapter does not engage.
• 2010 Rohm and Haas brain-cancer suit (Morris, Center for Public Integrity). The suit concerned a brain-cancer cluster “surrounding a chemical plant”, a community rather than a workforce. Company experts called the link “tenuous at best”. Per “Morris’s report of the proceedings”, the industry-funded study the company relied on “failed to include as many as two dozen fatal cases of brain cancer” (p. 188). This is a claim made in litigation and reported by a journalist, not an adjudicated finding.
• Excess liver-carcinoma deaths reported in the US, France, Germany, the European four-country study, Taiwan, Italy and Japan (p. 188). The French (Saurin 1997) and Japanese (Makita 1997) citations are case reports, going by their titles in the reference list (“a report of two cases”; “A case of…”, pp. 199, 201). They document cases, not excess mortality.
• Associations beyond angiosarcoma (p. 191):
• brain cancer (nine citations, including IARC 2008)
• hepatocellular carcinoma
• haemolymphoreticular neoplasias
• lung cancer
• liver cirrhosis
• birth defects near PVC plants (Infante 1975)
• miscarriages among workers’ wives (Infante 1976a,b; NIOSH 1977; ATSDR 2006)

The chapter says only that VC “has been associated with” these outcomes. The human reproductive claim rests on these few sources: primary studies from 1975–77, plus the ATSDR 2006 review. The animal evidence is separate: carcinogenicity after prenatal exposure, and footnote 4 on foetal toxicity (p. 185). - Box 8.2: 2008 IARC Monograph (p. 191; checked visually): - VCM causes angiosarcoma and hepatocellular cancer; the evidence is sufficient for Class 1. - For other cancers, “the Working Group did not find strong epidemiological evidence for associations … with cancers of the brain or lymphatic and haematopoeitic tissue or melanoma”, since findings “were inconsistent between studies, no clear exposure–response relationships were found in the European multicentric study and … the numbers of observed and expected cases were small”. - Vinyl bromide was classed 2A by analogy (Bradford Hill), given “a consistently parallel response”. The box concludes: “Experience has shown that similar chemical structures commonly exhibit similar toxic effects. Acting on this knowledge enables us to best utilise the very limited data base in toxic substances control.” - This sits in tension with the chapter’s own summary of IARC (p. 188; Table 8.2 p. 193; bullet list p. 191) and with lesson 7.

Panel 8.1: Clutterbuck, UK perspective (pp. 189–190; checked visually)#

• UK response. “Very expensive and comprehensive measures”. EMAS, the Factory Inspectorate and industry studied PVC and VCM plants, fabrication, warehouses, transport, meat packers using PVC wrap, drinks in PVC bottles and emissions (Forman et al. 1985). “Later there would be research into the incidence of liver cancer in the vicinity of VCM plants, where some plants were clearly more dangerous than others” (Elliott and Kleinschmidt 1997) (p. 189). The panel does not report what that research found about community liver cancer.
• Monitoring evolved. It started irregular and accurate only to ~200 ppm, then reached 10 ppm, 1 ppm and 1 ppb, and within years was continuous with immediate readout to workers. Alarms followed, then ventilation, valve-leak reduction and high-pressure water cleaning of autoclaves (p. 189).
• UK costs. GBP 9 million on plant redesign in year one plus an estimated GBP 4 million in lost production. “There may have been the same exaggeration of costs beforehand … but it was all less transparent in the United Kingdom”. Much of the cost “would have been unnecessary if control measures had been built in at the design stage”, but “it must be granted that the industry did undertake expensive control procedures” (p. 189).
• Why such investment, when “Coal miners and coke oven workers never got that sort of investment”? Two reasons (p. 189): 1. The disease. Angiosarcoma is deadly and extremely rare. “The moral obligation was clear, especially considering the industry’s failure to spot it previously. The rarity of the disease also meant that the causal connection with vinyl chloride was undisputed. What if VCM had caused lung cancer? We may not know even today.” 2. The hazard was controllable. About six VC and six PVC plants; control costs “relatively low in such a capital-intensive industry”. Trichloroethylene, “found in every engineering shop in the country”, would have been “much harder to control”.
• “Clean” industry. Plastics were seen as “‘clean’”, but “the lack of evidence did not mean there was no hazard” (p. 189).
• US vs UK unions (pp. 189–190):
• US: OCAW, already striking over health and safety, “took up the campaign vigorously”. Its role was “critical”. In “transparent court hearings” unions sought “no detectable level”.
• UK: a tripartite TUC/CBI/government Code of Practice was “less legally binding” and recommended 25 ppm (p. 189). The panel does not date it, beyond covering the year after the 1974 announcement.
• Note that the panel says “the court decided the level should be 1 ppm”. Elsewhere the chapter says OSHA set the standard and the court upheld it on appeal (pp. 186–187).
• Local action varied. There were spasmodic strikes at ICI Runcorn for “danger money”. At Vinatex, Chesterfield, 20 men had AOL and the focus was compensation, “typical of UK trade unions at that time, where compensation was more important than prevention” (p. 190).
• None of the main unions (GMWU, TGWU, ASTMS) had health and safety officers; the first were appointed in 1977/78 under the 1974 Act (fn 5) (p. 190).
• TUC Medical Adviser Dr Robert Murray (Nature, 15 Feb 1974): tell workers “the risk is small; that apart from the risks involved when he drives a car and eats too much, and drinks too much, here is another risk which he has got to live with”. Footnote 6: he had also downplayed BCME in 1969, 1971 and 1974 despite union rep Andrew Tree’s evidence. BCME workers at Tree’s Welsh plant were told of the risks only in 1982, “some eight years after the US OSHA had labelled BCME as a human carcinogen” (Doyal et al. 1983, co-authored by Gee) (p. 190). The footnote does not say whether it is Clutterbuck’s or the editor’s.
• BP Baglan Bay. A manager: “Actually we are quite pleased about vinyl: it’s the only issue that we are in agreement on with the unions”. Clutterbuck supplied unions with US and scientific-press information and TV (World in Action). “The attitude changed dramatically”: management refused him entry, workers threatened a 24-hour strike, and they met outside the plant and on TV. ICI Runcorn reps said that, informed by outside organisations, “they could look after their members’ interests better” (p. 190).
• Containment of the issue. “The petrochemical industry was successful in removing the grosser hazards while keeping the issue confined to the risk of rare cancer”. Trichloroethylene was not addressed, other cancers were not accepted, and the focus stayed on fabrication units rather than PVC handlers. The consumer lobby later ensured residual VC was cleaned out and PVC replaced in food wrap (p. 190).
• Lesson for “radical” scientists. “Good scientific work could be left in academic filing cabinets, having no impact on people’s health. We needed to translate that work, without twisting it”. This led to BSSRS founding Hazards Bulletin, now Hazards (p. 190).

8.6 Some late lessons from vinyl chloride (pp. 191–193)#

• The eight numbered lessons are set out under “The authors’ own lessons and conclusions” below.
• Footnote 7 (p. 192): the Porter link between challenging regulation and innovation “seems to hold up in many but only specific circumstances (Porter, 2011)”.
• Footnote 8 (p. 192): “Extensive research efforts are underway to minimise the use of animals”.
• Table 8.2 “Early warnings and actions” (p. 193; checked visually), rows 1930s–1959 to 1982:
• The first row’s “Russia (where limit of 500 ppm exposure was established)” is ambiguous or garbled.
• The 1964 row says Creech’s four AOL cases were “clearly linked to workplace exposure to VC”. The body is more guarded: to Creech this “seemed to be obvious evidence of a link” (p. 183).
• The 1965 row says ACGIH “accepts industry claims that this is too low and costly”. The body does not mention cost.
• The 1966 row says the AOL results “were not shared with regulators or workers”.
• The 1967 row says the Goodrich article “Recommended that workplace exposure be reduced to 50 ppm”. The body attributes the 50 ppm recommendation to the 1969 Michigan report, and to Dow (1961) and Esso (1965), not to the 1967 article (pp. 183–184). This is an unreconciled inconsistency.
• The 1968–1970 Viola row differs from the body, which dates the start of his experiment to 1967 (p. 184).
• The 1979 row: IARC Category 1 “for liver, brain, lungs and blood-forming system and found no safe exposure level”.
• The 1982 row: IARC reconfirms.
• Footnote 10: the 500 ppm limit rested on one small 1930s Bureau of Mines guinea-pig study, “an example of the ‘unsubstantiated authoritative assertion’” (cross-ref Ch 3).
• After the lessons. Slow or hostile corporate responses are cross-referred to Ch 25 (“Why did bisiness [sic] not react with precaution…”). “Of course there are, and always have been cases of ‘progressive business’”. Footnote 9: Quakers and Josiah Wedgwood, who asked in 1833 for Factories Act rules to extend to his potteries (p. 193).

Panel 8.2: Huff, value of animal testing (pp. 194–196; checked visually)#

• Opening. It opens with David Rall: animal tests identify carcinogens “before human exposure … a foolproof way to prevent human cancer” (p. 194).
• Scale. Probably “less than 10–15 %” of agents are carcinogenic, but that is many given ~100,000 chemicals in commerce. The panel lists 11 categories of known human carcinogens, bolding those covered in Late Lessons (p. 194).
• Operational definition of a carcinogen, by six experimental criteria: 1. increased incidence 2. reduced latency 3. rare tumour types 4. multiplicity 5. increased total tumours 6. increased malignant/benign ratio

In humans usually only criteria 1 and 3 are used. VCM, asbestos and DES are examples of criterion 3 (pp. 194–195). - Methods listed include high-throughput screening “in development”. Tomatis: prevention was implemented on animal predictivity “independently of the extent of understanding of the underlying mechanisms”. The panel calls for replacement by “equally effective, non-animal methods”. But because human studies are “typically absent, … costly and time-consuming or of low power” and “alternatives to animals have so far proven unsuccessful”, long-term bioassays remain “the most useful time-proven method” (p. 195). - Concordance claim. The claim covers ~120 recognised human carcinogens and also those rated probable (IARC, 61) or “reasonably anticipated” (RoC, 186): “all that have been tested adequately are likewise carcinogenic in mammalian cancer bioassays”. Hence animal carcinogens “should be considered as being likely and anticipated to present cancer risks to humans” (p. 195). - Nine “facts” (p. 196). Among them: - Rodent–human similarities outweigh differences. - “Nearly one-third of human carcinogens were first discovered in animal bioassays”. Fact 4 qualifies this: some human carcinogens predate standard bioassays, and some are “undefined ‘exposure circumstances’” such as aluminium production or furniture-making, which are “not readily testable in animals”. - Independent bioassays are consistent, “albeit sometimes with additional or different target sites”. - Bioassays predicted VCM, 1,3-butadiene, TCE and TCDD, and confirmed arsenic and benzene. - Early-tested chemicals were pre-selected on suspicion, and random selections find fewer. - “Less than 10–15 %” of chemicals would test positive. - “No other in vitro assay or in vivo bioassay or combination of tests, or even epidemiology … can claim these collective facts and advantages”. - A hedge follows: “Of course not all animal (or human) carcinogens are equal, and one must combine the collective experimental findings with experience to best predict human cancer risks” (p. 196). - Vested interests. The panel frames opposition to bioassays as “vested interest opposition”. It quotes Tomatis (2006): primary prevention “has stumbled from the very beginning because of the interference of powerful economic interests”. Near the close: “The history of public health is characterised by persistent struggles between short-term economic interests and long-term public, environmental, and occupational health concerns”. The panel’s final sentence repeats that efforts to “minimise the use of animals” are under way (p. 196). - Counter-view only implied. The references include Huff and Melnick (2006), an “Opinionated Book review” of Gold, Ames et al.’s “Misconceptions about the Causes of Cancer”, and Tomatis et al. (2001), “‘Alleged misconceptions’ distort perceptions…” (pp. 198, 202). These show awareness of the high-dose-bioassay critique, which the panel does not present.

Case timeline#

Lag analysis. These are my calculations from the chapter’s dates. - From Rowe’s 1959 “appreciable injury” letter, or the 1961 publication and internal ≤50 ppm recommendation, to OSHA’s 1974 action: ~13–15 years. The US limit stayed at 500 ppm throughout. It was a voluntary ACGIH TLV until OSHA adopted the 1968 TLVs as enforceable limits, and although ACGIH lowered its TLV to 200 ppm in 1971, the OSHA limit remained 500 ppm until 1974 (p. 183). - From setting the 500 ppm limit on thin evidence (1954) to its replacement: ~20 years. - From the first public clinical description of AOL (1963) to meaningful exposure limits: ~11 years. - From Maltoni’s cancer finding at 250 ppm (revealed to firms on 14 Nov 1972) to the 1 ppm standard: ~2 years. For roughly 15 months (Nov 1972 to Jan–Feb 1974) the signatory firms did not disclose it to US regulators, although Maltoni presented early results publicly in April 1973 (pp. 180, 186). - From the Goodrich announcement (Jan 1974) to the final 1 ppm standard: ~9 months. That is very fast once the signature harm was public. - In the UK, the response after 1974 was a “less legally binding” tripartite code recommending 25 ppm (p. 189). - The chapter says nothing about the European Community or EU regulatory timeline. [external — verify] An EEC worker-protection directive of 1978 set a long-term limit of about 3 ppm; lesson 6’s “1–3 ppm” may allude to it. - Harms visible in the chapter: - AOL in ~1% of PVC workers and ~6% of vat cleaners (1966) - 31 AOL cases (Goodrich 1967) - 20 AOL cases at Vinatex, UK - 7 angiosarcoma cases at Goodrich Louisville - 103 registry cases “in various countries”, 1974–83 - cancers among hairdressers - contested excesses of other cancers - possible reproductive effects

The chapter gives no aggregate estimate of health burden or of the costs of inaction.

The authors’ own lessons and conclusions#

As stated in section 8.6 (pp. 191–193). I label each as (E) derived from the case evidence, (A) advocacy or recommendation, or a mix.

1. What changed between 1959 and 1974 (E, interpretive), p. 191. Four deaths from a rare liver cancer at one company “were publicised (unlike other earlier warnings)”. The evidence was “supported by strong animal evidence”. “Public awareness and concern about toxic chemicals” ensured “quick and radical regulatory and company action, at least in the US”. The new OSHA and EPA, “in a climate of union and environmental activism and media interest, helped to create this strong and prompt public health response.”
2. A simple engineering fix existed (E, with an A tail), pp. 191–192. “It turned out that a simple engineering solution had been available all along”. Steam-stripping PVC cut residual VC in resins by 99%, lowering exposures in polymerisation and fabrication. OSHA went from 500 ppm to 1 ppm, and EPA emission standards “greatly reduced the 4 % loss of VC to the environment”. The industry “was easily able to comply and keep growing to this day. It was finally a regulatory success story and another example of how clear and challenging regulations can stimulate innovation” (Ashford and Hall 2011, 2012). Footnote 7 hedges this: the Porter link holds “in many but only specific circumstances”. The chapter does not document when steam-stripping was known or available before 1974, which weakens “all along”. There is also a tension inside the lesson: if the fix had been “available all along”, regulation forced its adoption rather than stimulating its invention.
3. Cost forecasts were wrong (E, then generalised), p. 192. Industry threatened collapse, “losses of USD 60 million annually and 2 million jobs” (Washington Post 1974). The lesson: “technological innovation to reduce exposure levels can indeed be accomplished in a short time without catastrophic consequences on production or employment” (Sass et al. 2005).
4. Corporate conscience is insufficient (mixed E and A), p. 192. “Based on information available to major companies using VC in the 1950s, toxic exposures should have been lowered and applications restricted.” Industry resistance to discretionary cost increases “is rarely if ever overcome merely by corporate conscience (corporate social responsibility is the current term for this concept)”. For executives, “the immediate prospects of regulation, liability and market losses to an informed public are much more persuasive”. The “rarely if ever” generalisation goes beyond this one case. Footnote 9 on progressive business qualifies it (p. 193).
5. Long-term bioassays are valid (E for VC, generalised as A), p. 192. The VC story “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 affected” (Maltoni 1984; Soffritti 2002; Huff 2002). Cross-reference to Panel 8.2.
6. Exposure limits should keep falling (A), p. 192. The drop from 500 ppm to 1–3 ppm was “a great achievement”, but levels “did not continue to be lowered despite experimental evidence which demonstrated even then the risks of exposure at concentrations of 50 ppm and less”. “The approach to setting exposure levels should be a dynamic one, in which levels are constantly lowered throughout time in light of scientific evidence and technological achievements.” The chapter says the same lesson appears in the lead (Ch 3), mercury (Ch 5) and beryllium (Ch 6) chapters.
7. Epidemiology underestimated the risk (mixed E and A), pp. 192–193. “The epidemiological evaluation of the carcinogenic effects of VC has certainly underestimated the risk”. There were attempts from the beginning to “reduce the quantification of risk to a few dozen cases of liver angiosarcoma”, cases often found “thanks to observations made by workers and clinicians”. 1970s data linked VC with lung cancer and hepatocarcinomas, “Unfortunately, not enough weight was ever given to these studies”. Doll’s “widely accepted” 1988 review found no risk beyond angiosarcoma plus modest lung-cancer risk in the heavily exposed, and “dismissed” a brain-cancer excess as non-significant. The largest cohort’s average age was only 54 (US EPA 2000), while 80% of cancers are diagnosed after 55 (ACS 2005). So relying exclusively on epidemiology “would be a failure of public health, ignoring laboratory evidence while we count the bodies until they become statistically significant.”
8. Maltoni’s lesson (A), p. 193. Studies in environmental and occupational carcinogenesis “must … represent an important component of the decision-making processes which regulate the developmental trends of society” (Maltoni et al. 1984).

Further conclusions and recommendations elsewhere in the chapter: - (A) No-fault compensation schemes should be paired with measures to extract information about corporate behaviour (p. 179). - (A) Judicial review can deliver “a societal judgement” on corporate behaviour and on the state of the science (p. 179, Box 8.1 p. 187). - (A) The precautionary principle should reverse the burden of proof so that “risk makers, not risk takers” show acceptable safety, “at least beyond reasonable doubt” (p. 187). - (E, generalised) Industry generally exaggerates projected compliance costs (p. 187, citing Ch 23). - (E, generalised) The trade-association tactic of making claims no member company would make in its own name (p. 187). - (Box 8.2, A) Acting on structural analogy makes the best use of scarce data (p. 191).

Mechanisms and dynamics#

How warnings arose#

Warnings came from several independent channels: - Soviet occupational hygienists (p. 182) - in-house toxicologists at Dow and Union Carbide (pp. 182–183) - company physicians: Creech, who followed up a worker’s complaint; Solvay’s doctors; and Viola, Solvay’s factory doctor, whom Solvay commissioned to run an animal experiment on AOL (pp. 183–184) - an MCA-funded academic study (p. 184) - an independent, publicly and partly industry-funded experimental oncology programme (Maltoni) (p. 185) - workers’ own bodies and complaints: “painfully tender fingers” (p. 183) - clinicians and workers who found angiosarcoma cases (lesson 7, p. 192)

The detectors were mostly insiders or near-insiders. For the non-cancer hazards (liver damage, AOL), detection was not the bottleneck: the firms had credible signals from 1959 onwards, and the bottleneck was disclosure and action. For cancer, detection depended on long-term bioassays (1967–72). After that, disclosure again lagged, for about 15 months.

How warnings were contained#

The chapter documents a repertoire of containment moves, mostly via Markowitz and Rosner’s reading of MCA papers: - Confidentiality among firms but not beyond. Information moved freely between competitors through off-the-record calls, private medical meetings and trade-association briefings, but not to workers, regulators or independent scientists (pp. 182–183). The 1972 secrecy agreement formalised this across the Atlantic (p. 180). - Quiet case-finding. Physicians were told to “determine as quietly as possible” and to prepare workers “an adequate story” (p. 183). - Attempted suppression of third-party publication, e.g. the effort to “discourage or edit” Solvay’s paper (p. 183). - Editing the record. VC causation was removed from the 1967 article, and “etiologic agent … unknown” was inserted by committee vote in 1969. The final 1971 publication omitted the protective-limit and odour-threshold findings (pp. 183–184). - Strategic non-research. Goodrich “explicitly decided not to accept any proposals for additional research”, which let industry “act as if the cause was ‘unknown’” (p. 184). Here uncertainty was manufactured by deliberately not reducing it. - Selective disclosure. The high-dose Viola result was disclosed to NIOSH; the low-dose Maltoni result was not (p. 186). - Narrowing the harm after disclosure. Quantification of risk was confined to “a few dozen cases” of angiosarcoma (lesson 7, p. 192; the chapter does not say by whom). Concern was confined mainly to fabrication units and kept away from related chemicals such as trichloroethylene (Panel 8.1, p. 190). Evidence on other cancers was played down (section 8.5, p. 188). - Industry-controlled epidemiology. Cohort selection that excluded long-exposed workers, dilution with short-tenure workers, healthy-worker comparisons, omitted cases, and a retraction at the funder’s request (per the chapter) (pp. 187–188).

Why firms behaved this way (the chapter’s causal account)#

• Business-decision primacy. Toxicologists and physicians advised; executives decided “as business decisions” (p. 184).
• Market contagion fears. The central motive offered is that worker-hazard news would taint consumer markets: food packaging, medical uses, aerosols (pp. 183–184). This explains why a workplace issue was treated as a commercial secret.
• Costs. A 50 ppm limit “could increase costs” for resin makers and fabricators (p. 184).
• Collective action through the trade association. Firms could shelter behind the MCA and the “50 member companies” (pp. 184, 187). The chapter’s summary box also claims a “mismatch” between trade-association positions and more progressive members (p. 179), but the body gives almost no evidence of progressive members. Arguably Dow’s claimed internal 50 ppm limit (p. 184), Solvay’s informing peers (p. 183) and the funding of Maltoni (p. 185) are partial examples.

Standard-setting institutions and their culture#

• Before OSHA, the key limit was set by a voluntary body including industry consultants. It was based on what was “achievable but not necessarily health protective” and on a single guinea-pig study (pp. 182, 193 fn 10).
• The standard acquired authority out of proportion to its evidence: the “unsubstantiated authoritative assertion” (fn 10, p. 193).
• Objections from member companies were enough to defer a proposed revision (p. 184). The chapter does not describe the ACGIH’s formal decision rules.
• Institutional novelty mattered. OSHA and EPA, created “at the start of the 1970s” amid union and environmental activism, gave the 1974 evidence somewhere to go (lesson 1, p. 191).
• Information-forcing law had limits. The mandatory NIOSH request of January 1973 did not stop the omission of Maltoni’s results at the July 1973 meeting (pp. 180, 186). The chapter does not say whether any sanction followed.

Burden and standard of proof#

• Standards of proof were set by those with the strongest interest. Within the trade association’s occupational health committee, recommending a protective level was treated as tantamount to admitting causation. The chapter gives this as the reason for the 7–3 vote (p. 184).
• The courts and the burden of uncertainty. In court, Judge Clark accepted action “on the frontiers of scientific knowledge” when “the factual finger” pointed to it (p. 187). The 1980 benzene decision then required quantified “significant risk”, putting “the burden of medical uncertainty squarely on the shoulders of the American worker” (p. 187). The chapter uses this contrast to argue for reversing the burden of proof under the precautionary principle.
• Statistical conventions. Significance thresholds, short follow-up and healthy-worker comparisons all bias toward false negatives. The chapter links this to Ch 26 (p. 188; lesson 7 pp. 192–193).

Mental models and blind spots#

• Faith in a thin standard. The 500 ppm limit was treated as protective for 20 years although its evidential basis was known to be one 1930s study (p. 182).
• Odour as a safety proxy. It had been thought that VC could be smelled at 400 ppm. The Michigan study found the odour threshold was 4,000 ppm, so odour indicated “much more hazardous exposure than had been thought” (p. 184). The chapter does not say who relied on odour. The sensory warning was miscalibrated tenfold.
• Measurement limits. In the UK after 1974, monitoring was initially accurate only to ~200 ppm (p. 189). Hazards below what could be measured were in effect invisible.
• “Clean” industry and absence of evidence. Plastics were perceived as “clean” relative to mines and mills. The panel’s corrective: “the lack of evidence did not mean there was no hazard” (p. 189).
• Novelty framing. Kehoe called AOL an “entirely new” disease (p. 183). The chapter does not analyse this framing, but calling a disease “new” can sit alongside refusing to name its cause.
• Risk normalisation by trusted intermediaries. The TUC Medical Adviser likened the risk to driving and overeating (p. 190). Footnote 6 shows a pattern: he also downplayed BCME.
• Management complacency about agreement. A manager was “quite pleased” that vinyl was the one issue of agreement with the unions (p. 190). As Clutterbuck saw it, the consensus rested on the local unions lacking information (“something wrong with communications”). Once they were informed, the consensus collapsed.
• The scientist’s mental model. Panel 8.2 (Huff, quoting Tomatis) sets out a prevention-first view: act on animal predictivity “independently of the extent of understanding of the underlying mechanisms” (p. 195). Maltoni’s lesson 8 and lesson 5 are consistent with it (pp. 192–193). This is explicitly a different epistemic stance from demanding human proof.
• The judge’s mental model. Scientific dispute is not a bar to protective action (p. 187).
• Epidemiologists’ mental model. In Doll’s review, dismissing a non-significant excess treats absence of significance as absence of effect. The chapter criticises this (p. 192).

Framing and language#

Language that did work in this history: - Industry: “adequate story”, “as quietly as possible”, “use discretion”, “etiologic agent … unknown” (pp. 183–184). - Media rebuttal: “doomsday predictions”; “Did Industry cry Wolf?” (p. 187; refs p. 201). - The chapter’s own frames: “a saga of secrecy”, “veil of secrecy”, “count the bodies” (pp. 179–180, 193). - Box 8.1: “risk makers, not risk takers” (p. 187). - Footnote 10: “unsubstantiated authoritative assertion” (p. 193). - Panel 8.2: “foolproof”, “vested interest” (pp. 194, 196). - Panel 8.1: “clean” industry; “danger money” (UK strikes, p. 190). The latter is a framing of risk as something to be compensated rather than prevented.

Signal detectability and causal attribution#

Panel 8.1’s central insight (p. 189): causation was “undisputed” because angiosarcoma is so rare that a cluster is unmistakable. Had VC caused a common cancer such as lung cancer, “We may not know even today”. Lesson 1 (p. 191) makes rarity and publicity joint triggers. Panel 8.2’s criterion 3, “tumour types not typically seen”, ties the same point to the animal–human concordance (p. 195). The flip side is that VC’s other claimed harms, which add to common background cancers, remain contested (Box 8.2, p. 191).

Lock-in, substitution and innovation#

• Growth and embedding. PVC grew from 450 t (1933) to ~34 Mt (2010) and became embedded in construction, medicine, vehicles and packaging (pp. 180–181). The regulatory response took the form of exposure control (steam stripping, closed-loop reactors, monitoring, ventilation), not substitution or phase-out (pp. 181, 189, 192).
• Limited substitution. It occurred only at the margins: aerosol propellant bans (1974, 1978), the FDA’s action on PVC liquor bottles, and “replacement of PVC in food wrapping” driven by the consumer lobby (pp. 181, 190).
• Lock-in unquestioned. The chapter does not question PVC’s continued expansion. It presents continued growth as evidence that regulation did no harm (pp. 187, 192).
• Innovation came quickly under pressure (lesson 3, p. 192). The chapter attributes it to “clear and challenging regulations” (p. 192). Panel 8.1 adds that retrofitting cost more than designing controls in from the start would have (p. 189).

Time lags and irreversibility#

• Latency. Long latency means harms keep appearing for decades after exposure. There were 103 angiosarcoma cases in 1974–83 (p. 187). The cohort average age of 54 means much of the cancer burden may not yet be observed (p. 192).
• Irreversibility. AOL is “excruciatingly painful” (p. 183). Angiosarcoma is “one of the deadliest cancers” (p. 189).
• The cost of delay. The chapter’s implied counterfactual is that roughly 15 years of 500 ppm exposure after credible warnings could have been avoided. The panel’s point that design-stage controls would have been cheaper (p. 189) suggests delay also raised costs.

Distribution of costs, benefits and risks#

• Who bore the harm:
• The most exposed workers, especially “pot cleaners” (p. 186) and vat cleaners with a 6% AOL rate (p. 183).
• Hairdressers exposed through consumer aerosols (p. 181).
• Communities near plants: liver-cancer incidence was later researched, with results not reported (p. 189); birth defects are cited as an association (p. 191); and there was the brain-cancer cluster litigation (p. 188).
• Consumers exposed through foods contaminated from packaging (p. 181). This was exposure; the chapter does not document resulting harm.
• In future, workers in low- and medium-resource countries where “older technologies” persist (p. 181).
• Who gained from delay: resin makers, fabricators and markets for PVC food packaging, medical uses and aerosols (p. 184).
• Who paid for compliance: industry, at modest ex post cost (USD 278 million, p. 187; GBP 9m + 4m, p. 189).
• Who bears uncertainty depends on legal standards (the benzene decision, p. 187).
• Asymmetry in investment. The panel observes that coal miners and coke-oven workers never received comparable investment despite their diseases (p. 189). Protective investment tracks the detectability and controllability of the harm, not only its size.

Role of labour, publics and media#

• Labour. In the US, OCAW’s prior mobilisation mattered for the standard (p. 189). In the UK, unions without health and safety capacity and focused on compensation responded weakly until outside information arrived (p. 190).
• Media. Media interest mattered in the US (lesson 1, p. 191) and in the UK (World in Action, p. 190). The 1975 NYT piece was an ex post rebuttal of cost claims, not a driver of action (p. 187).
• Consumer lobby. It later drove cleanup of residual VC and substitution in food wrap (p. 190).
• Independent translators of science. BSSRS/Hazards were created in response (p. 190).

Law as an information and judgement system#

• Discovery. Document discovery in compensation litigation is presented as the way the “real and until then secret” history came to light (p. 179). Part of the chapter’s documentary base comes from litigation: Creech’s “later legal testimony” (p. 183). The litigation origin of the MCA papers cited via Markowitz and Rosner is implied but never stated [external — verify].
• Why no-fault schemes need a counterpart. No-fault compensation would remove this information channel unless replaced (p. 179).
• Judicial review as societal adjudication of science and conduct (p. 187).

Research funding and the independence of science#

• Industry-funded research cut both ways:
• Solvay commissioned Viola’s study, which found the first cancers (p. 184).
• European producers co-funded Maltoni’s programme (p. 185).
• The MCA funded the Michigan study (p. 184).
• Industry commissioned Tabershaw–Gaffey (p. 187) and funded Wong et al. (p. 188).

Funders exercised control after the fact: secrecy agreements, committee rewording, a refusal to fund further research and the retraction request. The chapter’s own account shows industry funding as both a precondition of discovery and an instrument of control. It does not analyse this ambivalence. - Public funding. Maltoni’s programme was funded “mainly from public institutions”, including the Bologna Hospital Administration (p. 185). The chapter does not discuss how this mixed funding related to the sponsors’ secrecy agreement. Nor does it discuss how it related to Maltoni’s own freedom to present results (Bologna symposium April 1973; the OSHA hearing, where he was “invited”, February 1974) (pp. 185–186).

Geography and systemic spread#

• Western action versus migrating production. Western controls coexist with rapid growth in Asia and likely high exposures where older technologies persist (p. 181). Hazard reduction in one region may be offset by expansion elsewhere.
• Unintended formation. VC forms in landfills from chlorinated solvents (p. 181). This systemic pathway is mentioned but not analysed.

Transferable insights (technology-neutral)#

1. Producers can know about hazards long before anyone else, and that gap can be a main source of delay. Those who make and use a product can accumulate credible internal evidence years before regulators, workers or independent scientists. That asymmetry, not missing science, can drive the delay. Here it did so for the liver and bone hazards; for cancer, the science itself arrived only in 1970–72. Evidence: 1959 Rowe and Smyth letters; 1961 internal ≤50 ppm recommendation; 1966 private industry meeting (1%/6% AOL); 1969 Michigan report; 1972 secrecy agreement (pp. 180, 182–184). Strength: strong. Contemporaneous internal documents, though accessed through a secondary synthesis (Markowitz and Rosner).

2. Warnings can circulate freely among competitors while being withheld from those who bear the risk. The closed circle becomes an information cartel that can co-ordinate non-disclosure. Evidence: off-the-record inter-firm calls; Solvay informing other manufacturers; the transatlantic secrecy agreement collected by one firm; “use discretion” (pp. 180, 182–183). Strength: strong for this case (documented); moderate as a general pattern.

3. In-house technical experts who recognise a hazard are not enough to trigger action. Commercial decision-makers filter expert advice, and experts may themselves observe confidentiality norms. Evidence: “toxicologists and company doctors do not run the companies” (p. 184); Dow 1961 advice ignored; Smyth asking confidentiality until publication (p. 182); causal language removed from the draft of a company research article (p. 183). Strength: strong within the case (repeated instances); moderate as a generalisation.

4. Uncertainty can be maintained on purpose. Declining to fund research that would resolve causation, and rewording findings to call the cause “unknown”, keeps the uncertainty alive, and that uncertainty is then used to justify inaction. Evidence: the 7–3 committee vote and inserted wording; the published report omitting findings; the refusal to accept further research proposals (p. 184). Strength: strong (specific documented acts), with the caveat that it relies on a single historical source.

5. Standards resting on thin early evidence can acquire authority and persist for decades. Once institutionalised, a weakly grounded limit shifts the burden onto those seeking to change it. Evidence: the 500 ppm limit based on one 1930s guinea-pig study, unchanged 1954–1974 (pp. 180, 182, 193 fn 10). Strength: strong.

6. Standard-setting bodies dominated by regulated interests tend to set “achievable” rather than protective levels, and collective industry bodies can veto change. Trade associations can also voice positions no single member would voice openly. Evidence: ACGIH composition; “achievable but not necessarily health protective”; 50 member companies deferring the 50 ppm proposal; the trade-association tactic (pp. 182, 184, 187). Strength: moderate to strong (documented for this case; the chapter asserts it generalises and cites other chapters).

7. Proxy indicators of safety, such as senses, crude monitoring and absence of complaints, can be badly miscalibrated. They give false reassurance until independent measurement improves. Evidence: odour threshold 4,000 ppm, not 400 (p. 184); monitoring initially accurate only to ~200 ppm (p. 189); the “clean” industry perception (p. 189). Strength: moderate to strong (specific data points, but few).

8. Whether a harm gets recognised depends heavily on how distinctive it is. Rare, distinctive harms get attributed and acted on quickly. Harms that add to common background conditions may remain disputed indefinitely, so the most detectable harm can end up defining the whole issue. Evidence: Panel 8.1’s “What if VCM had caused lung cancer?” (p. 189); four deaths “akin to an epidemic” (p. 186); lesson 1 (p. 191); IARC 2008’s inconclusive findings on non-signature cancers (p. 191). Strength: strong as epidemiological logic; the counterfactual itself is the panel author’s reasoning.

9. Effective action followed a conjunction of factors. These were a publicised, unmistakable harm; independent experimental corroboration; newly created regulators with a mandate; and mobilised labour, media and publics. Evidence alone had not sufficed for 15 years. Evidence: lesson 1 (p. 191); Maltoni’s hearing testimony (p. 186); the OCAW role and World in Action (pp. 189–190). Strength: moderate. The authors’ interpretive synthesis is plausible but drawn from a single case, without a systematic comparison.

10. Once firm rules exist, the regulated industry can adapt quickly and cheaply. Ex ante cost and job-loss forecasts from regulated parties can be greatly overstated, and innovation under a clear, stringent requirement may reduce compliance costs. Evidence: USD 278 million ex post versus forecasts of up to USD 90 billion and 2 million jobs; NYT “doomsday predictions”; continued growth; steam-stripping (pp. 179, 187, 192); UK costs (p. 189). Strength: moderate. The case outcome is well known, but the chapter’s figures are internally inconsistent (USD 90bn vs USD 60m/yr), the ex post figure is unsourced, and footnote 7 limits Porter-type effects to “specific circumstances”. VC also had an unusually cheap fix.

11. Protective technical options may exist long before they are adopted. They go unused until external requirements make them necessary, so “no feasible alternative” claims made before regulation deserve scrutiny. Evidence: “a simple engineering solution had been available all along” (pp. 191–192); USSR reductions to 4–15 ppm by 1966 (p. 182); controls cheaper at design stage (p. 189). Strength: moderate. Asserted, with the USSR data as partial support. The chapter does not document when steam-stripping was known.

12. Voluntary responsibility rarely outweighs cost and market incentives. Regulation, liability exposure and informed markets are the levers that change organisational behaviour. Evidence: lesson 4 (p. 192); the rapid response after 1974 regulation and publicity (pp. 186–187); footnote 9 acknowledges exceptions (p. 193). Strength: moderate for this case; asserted as the general “rarely if ever” claim.

13. Studies controlled by interested parties can be structured so that they miss the effects they are meant to detect. Common ways are excluding the longest-exposed, diluting with short-exposed participants, using inappropriate comparison populations, having too little follow-up time for effects with long latency, and omitting cases. Relying only on such studies biases decisions toward false negatives. Evidence: Tabershaw–Gaffey versus Wagoner (57% excess), the healthy-worker effect, the Wong retraction, the Rohm and Haas study omitting cases, cohort average age 54 (pp. 187–188, 192–193). Strength: strong for the methodological mechanisms (standard epidemiology); moderate for the attributions of motive (the retraction’s own rationale is not engaged).

14. Independent, long-duration experimental evidence can anticipate harms before human evidence accumulates, even down to specific organs. It is most persuasive when the predicted harm is distinctive. Evidence: Maltoni’s angiosarcomas in animals preceding and matching human cases; Table 8.1 (pp. 185–186); lesson 5 (p. 192); Panel 8.2 (pp. 194–196). Strength: strong for this case (same rare tumour in both); moderate as a general claim. Panel 8.2 is advocacy by a proponent, and its concordance claims are contested in the toxicology literature (not presented).

15. Publishing a finding is not the same as it reaching regulators. Scientific disclosure may not reach decision-makers without a formal channel, a hearing or a triggering public event. Evidence: Maltoni’s April 1973 presentation “does not appear to have attracted the attention” of regulators; the first official airing came at the Feb 1974 hearing (p. 186); Panel 8.1: “Good scientific work could be left in academic filing cabinets” (p. 190). Strength: moderate.

16. Where the burden of proof falls decides who bears the uncertainty. Courts and legal standards can accept action under scientific dispute or demand quantified proof. Legal processes such as discovery can be a major route by which hidden organisational knowledge becomes public. Evidence: Box 8.1 (Clark’s “frontiers of scientific knowledge” versus the 1980 benzene decision and Marshall’s dissent) (p. 187); the document-discovery point and the no-fault caveat (p. 179). Strength: moderate to strong on the facts of the rulings; the recommendation on no-fault schemes is asserted.

17. How controllable a hazard is depends on how concentrated its use is. Few, capital-intensive sites make control feasible and affordable; widely dispersed uses are much harder to control. The same underlying risk can therefore yield very different protective outcomes. Evidence: Panel 8.1: about six VC and six PVC plants versus trichloroethylene “in every engineering shop”; coal miners and coke-oven workers never got comparable investment (p. 189). Strength: suggestive to moderate. This is the panel author’s reasoning, plausible but not tested.

18. Once a hazard is admitted, the issue tends to be narrowed to its most undeniable form. Related harms, related agents and downstream exposures then go unaddressed. Evidence: Panel 8.1: “keeping the issue confined to the risk of rare cancer”, trichloroethylene not addressed, finished-product handlers less considered (p. 190); lesson 7 (p. 192); section 8.5 (p. 188). Strength: moderate. It is partly undercut by IARC 2008 (Box 8.2), which treats evidence for some broader harms as weak, so narrowing may partly reflect real evidential limits.

19. Standards set at a moment of crisis tend to freeze there. The authors argue that limits should be revised continually as evidence and technology improve. Evidence: no further lowering after 1974 despite low-dose animal findings (pp. 186, 192), and parallels claimed with Chs 3, 5 and 6. Strength: suggestive/advocacy. This is normative, and the chapter’s own data show no VC-specific tumours at 1 ppm in Maltoni’s studies (p. 186).

20. Access to independent information changes how those at risk perceive and negotiate it. Trusted intermediaries, including advisers within protective institutions, can normalise risk. Evidence: the change of attitude at Baglan Bay and ICI Runcorn after outside information arrived; the TUC Medical Adviser’s “risk … he has got to live with” and his BCME record (p. 190, fn 6). Strength: moderate (first-person testimony with a documented footnote case).

21. Fear that a production-side hazard would damage product markets motivated concealment. Where occupational and consumer exposures share a substance, reputational spillover creates incentives to suppress workplace findings. Evidence: “would have been very damaging for industry”; “sensitive markets for PVC food packaging and medical applications” (pp. 183–184). Strength: moderate (documented quotes, but motive inferred).

22. Research funded by interested parties can both enable discovery and control what happens to it. Funding can generate the key evidence while the funders retain the ability to delay, edit or narrow its use. Evidence: Solvay commissioning Viola; producers co-funding Maltoni and then signing the secrecy agreement; the MCA-funded Michigan report reworded; Wong et al. retraction (pp. 180, 184–185, 188). Strength: moderate. The facts are documented, but the chapter does not itself draw out this dual role.

23. As production moves geographically, older, higher-exposure practices can persist or re-emerge where controls are weaker. Reducing a hazard in one region does not guarantee global reduction. Evidence: Asian growth and “older technologies … high exposures probably occur” (p. 181). Strength: suggestive (no exposure data given).

24. Structurally similar agents can justify provisional action when direct data are scarce. Evidence: Box 8.2 on vinyl bromide classed 2A by analogy (p. 191). Strength: moderate (an established IARC practice; the general claim “similar structures commonly exhibit similar toxic effects” is asserted).

Limitations, contestation and bias check#

Advocacy versus analysis - The chapter is written by advocates and allies: Maltoni’s institutional successor, an NRDC scientist, a public-interest consultant, and the EEA project originator with a trade-union and FoE background. Both panels come from allied positions: an activist-scientist memoir and a bioassay proponent. - There is no industry, regulator or independent-historian response. The reader never hears how the companies or their scientists explained their conduct, e.g. whether some believed AOL’s cause was genuinely uncertain in 1969, or what Wong and Whorton’s “diagnostic bias” rationale was. - The narrative is structured as a morality tale: “saga”, “veil”, and Box 8.1’s “delay, feigned ignorance and irresponsibility”. That last phrase is the box’s own characterisation of the ruling, not a quotation from the court (checked visually, p. 187).

Dependence on a single secondary source - Much of the pre-1974 documentary history is cited through Markowitz and Rosner (2002), including primary MCA papers cited via that book’s page numbers. The chapter did obtain original copies (fn 1, p. 179), but it does not show independent verification. - [external — verify] Markowitz and Rosner’s documentary work derived from litigation discovery, in which they served as experts for plaintiffs, and industry parties challenged their book. This does not make the documents unreliable, but the provenance should be noted when the story is used as a lens.

Internal inconsistencies and errors - Industry cost forecasts: USD 90 billion and 2 million jobs (p. 187) versus USD 60 million annually and 2 million jobs (p. 192). - “An analysis in 2000” giving USD 278 million is unsourced (p. 187). - IARC on non-liver cancers. Table 8.2 and p. 188 say IARC 1979 associated VC with brain, lung and haemo-lymphopoietic cancers and found no safe level. The bullet list on p. 191 cites IARC 2008 for brain cancer. But Box 8.2 says IARC 2008 found no strong evidence for brain or lymphohaematopoietic cancer or melanoma. Lesson 7’s “certainly underestimated” is not reconciled with this. - OSHA standard description. “Ceiling of 1 ppm” (p. 186), “1 ppm with a 15 minute excursion to 15 ppm” and “the court decided the level” (Panel 8.1, p. 189), and “1–3 ppm” (lesson 6, p. 192) are inconsistent. [external — verify] The standard is usually described as a 1 ppm TWA with a 5 ppm 15-minute ceiling, set by OSHA and upheld by the court. - Who set the 500 ppm limit. MCA, “CMA” and ACGIH are conflated (pp. 180, 182–183). Table 8.2’s first row attributes the 500 ppm limit to “Russia” ambiguously (p. 193). - Number of angiosarcoma deaths at Goodrich: three (press release), four (Creech 1968–73; Table 8.2) and seven (1964–74) (pp. 186, 193). These are explained in the text but could confuse. - Maltoni’s low-dose ingestion figures: tumours down to 0.3 mg/kg (result 6) versus 1 mg/kg (next paragraph), both on p. 186 and citing Maltoni et al. 1977. - Table 8.2’s 1967 row credits the Goodrich article with a 50 ppm recommendation that the body gives to the 1969 Michigan report (pp. 184, 193). Its 1964 row (“clearly linked”) is firmer than the body (p. 183). It dates Viola to 1968–70, where the body says the experiment started in 1967. - Citation mismatch: “(Federal Register, 1974)” for the 5 April 1974 50 ppm limit resolves to the 4 October 1974 entry. The 5 April “NARA, 1974” entry is listed but uncited (pp. 186, 198, 200). - Solvay publicity: “no reports were made public” (p. 183) versus the same page’s account of a Solvay doctor publishing despite Goodrich (see also the uncited Cordier et al. 1966, p. 197). - Minor errors: “Rhône-Prugil” (Rhône-Progil); “AOS episode” (AOL); “heptacellular”; “bisiness” (Ch 25 title); “Tabershaw-Cooper” versus Tabershaw and Gaffey (pp. 187–188); and IARC reconfirmation dated 1982 in Table 8.2 but 1987 in the text. Both dates may be correct, as separate IARC supplements [external — verify].

Thin evidence for some claims in the summary box - Reproductive effects (p. 179) rest on a few human citations in a bullet list (Infante 1975, 1976a,b; NIOSH 1977; the ATSDR 2006 review) (p. 191), plus animal evidence of prenatal carcinogenicity and footnote 4 on foetal toxicity (p. 185). - The “mismatch” between the trade association and more progressive companies (p. 179) is barely evidenced in the body. - Document discovery (p. 179) is asserted as a lesson but its role in this case is not explained.

Motive attribution - The Wong retraction is presented as made “at the request of the chemical industry” (p. 188), citing Sass et al. 2005 (co-authored by two of the chapter’s authors) and Markowitz and Rosner. The retraction’s own stated scientific basis, diagnostic bias, is not examined (title at p. 202). Readers cannot judge whether the retraction had merit. - “While the companies did not doubt the evidence” (p. 180) is an inference about belief.

Hindsight bias - It is largely mitigated for the non-cancer hazards (liver and AOL). Here the chapter shows that contemporaneous actors inside the firms judged 500 ppm unsafe and recommended ≤50 ppm (1959, 1961, 1965, 1969). This is contemporaneous knowledge, not retrospective judgement. - For cancer, the first signal (1970) came at 30,000 ppm and was reported publicly. Evidence at workplace-relevant doses dates from late 1972. The period in which industry did not disclose that cancer evidence to regulators was ~15 months (Nov 1972–Jan/Feb 1974), not decades. This is my calculation; the chapter does not frame it this way. - Lesson 4 says “toxic exposures should have been lowered and applications restricted” in the 1950s. For lowering exposures this is defensible from internal documents. For restricting applications, the chapter gives less basis in 1950s evidence (the food-packaging concern of 1959 is the main hint, p. 182).

Case selection and generalisability - VC is an unusually favourable case for precaution: - a signature disease - striking animal–human concordance (the same rare tumour) - a concentrated, capital-intensive industry - a cheap and available engineering fix - a political moment of new agencies and activism

Panel 8.1 explicitly recognises the first and third points (p. 189). - Lessons 2 and 3 (compliance was easy and innovation was stimulated) may generalise poorly to hazards with dispersed use, common-disease outcomes or no ready fix. Footnote 7’s Porter hedge acknowledges this, but the lesson text does not. - The chapter’s framing draws “late lessons” from a case that ended, in the US, in a fast response. It is partly a success story, which the authors acknowledge (pp. 179, 192).

The pro-precaution framing and scientific judgement calls - Lesson 6 calls for continual lowering of limits. The chapter’s own data say none of the specifically VC-related tumours appeared at 1 ppm inhaled or 0.03 mg/kg ingested (p. 186), and low-dose findings at 10 ppm were not statistically significant but judged “biologically relevant” (p. 186). The case for going below 1 ppm rests on a no-threshold presumption (attributed to IARC 1979, p. 188) rather than on direct evidence. That is defensible as precaution but should be labelled as such. - Lesson 7 (“certainly underestimated”) rests partly on sound methodological points (latency, cohort age, healthy-worker effect) and partly on a disagreement with Doll’s 1988 review and IARC 2008. The chapter presents the IARC 2008 view in Box 8.2 but does not reconcile it. - Box 8.1’s demand that risk makers show safety “at least beyond reasonable doubt” (p. 187) is a very high bar, stated without discussing its feasibility or its costs, including the risk of false positives. The chapter points to Ch 2 on false alarms but does not engage with it here. - Panel 8.2’s claims have long been contested in toxicology [external — verify]. These include that all adequately tested human carcinogens are positive in animals, that bioassays are the “most predictive” method, that under 10–15% of chemicals would be positive, and Rall’s “foolproof” quote. Points of contest include high-dose testing, species-specific mechanisms and false-positive rates. The panel shows awareness of the Ames/Gold critique (Huff and Melnick 2006 in the references) but does not present it. The panel’s call to reduce and replace animal testing sits in some tension with its insistence that nothing else matches bioassays; it acknowledges this as ongoing work. In fairness, the panel does include some hedges. Not all carcinogens “are equal” and findings must be combined “with experience”. Some human carcinogens are “exposure circumstances” not readily testable in animals. And independent bioassays sometimes show “additional or different target sites” (p. 196). - [external — verify] The Ramazzini Institute’s later lifetime-bioassay findings on some other substances were contested by regulatory bodies around 2006–2011, including questions about pathology diagnoses in aged animals. This bears on how far “the Maltoni model” can be generalised. It does not affect the VC findings, which were independently replicated (Keplinger 1975; Lee 1978; Feron and Kroes 1979, cited p. 187).

What is missing - European Community and EU regulatory history. For an EEA report the chapter covers only the UK (in a panel), with no account of other member states or Community directives. - Aggregate burden and costs of inaction. There are no estimates of total AOL, angiosarcoma or other disease burden, and no costs of inaction, although the report has a whole chapter on the latter (Ch 23). - Soviet warnings. How were Soviet findings received in the West? Were they known and discounted? - What regulators knew and did before 1970, beyond ACGIH, including any US Public Health Service involvement. - Industry’s side. No industry-side explanation or internal dissent is presented beyond Dow’s claim of an internal 50 ppm limit. - PVC’s wider lifecycle issues are out of scope, deliberately and reasonably. - Exposures in the fast-growing Asian industry. There is no data despite the claim that “high exposures probably occur” (p. 181).

Fairness in the other direction. The chapter does credit industry: - a “very prompt response” after 1974 (p. 179) - co-funding of the key research (p. 185) - Solvay informing other manufacturers (p. 183) - the failed attempt to stop Solvay’s publication, which implies Solvay’s doctors published anyway (p. 183; cf. the uncited Cordier et al. 1966, p. 197) - Dow’s claimed internal limit (p. 184) - expensive UK controls (Panel 8.1, p. 189) - “progressive business” (p. 193, fn 9)

Panel 8.1 also criticises labour’s own institutions (the TUC Medical Adviser, national unions) (p. 190). And the chapter includes the IARC 2008 caution (Box 8.2), which cuts against its own lesson 7.

Dissent within the panels. There is no dissent. Panel 8.1 adds nuance (costs, reasons for exceptional investment, union weaknesses), but it does not challenge the chapter’s thesis.

Notable quotes#

1. “That is because the toxicologists and company doctors do not run the companies; corporate toxicological policy decisions are made by corporate executives as business decisions.” (p. 184)
2. “Inasmuch as the etiologic agent of the disease is unknown, a level of vinyl chloride below 50 ppm should be used as an index of adequate ventilation” (the wording the MCA Occupational Health Committee required before accepting the report, p. 184)
3. “This enabled Goodrich and the rest of the industry to act as if the cause was ‘unknown’” (p. 184)
4. “I am sure that Dr Nessel can prepare these people with an adequate story so that no problem will exist” (Monsanto official, 1966, p. 183)
5. “What if VCM had caused lung cancer? We may not know even today. And it would have been disputed for many years.” (Clutterbuck, Panel 8.1, p. 189)
6. “not one of the doomsday predictions (from industry) has proven accurate” (New York Times, 1975, quoted p. 187)
7. “the burden of medical uncertainty squarely on the shoulders of the American worker” (Justice Marshall, benzene dissent, Box 8.1, p. 187)
8. “so that risk makers, not risk takers, have to show, at least beyond reasonable doubt, that chemicals and other stressors are acceptably safe” (Box 8.1, p. 187)
9. “ignoring laboratory evidence while we count the bodies until they become statistically significant” (lesson 7, pp. 192–193)
10. “Good scientific work could be left in academic filing cabinets, having no impact on people’s health.” (Clutterbuck, Panel 8.1, p. 190)

Open questions#

1. What is the source of the “analysis in 2000” giving USD 278 million in compliance costs (p. 187)? Which industry forecast is correct: USD 90 billion (p. 187) or USD 60 million a year (p. 192)? What were the forecast’s origin and assumptions? [external lead: a 1974 consultancy forecast for the plastics industry and a 1995 US Office of Technology Assessment retrospective; verify]
2. When were steam-stripping and closed-loop polymerisation known or available? Was the fix really “available all along” (pp. 191–192), or was it developed under regulatory pressure? This matters for the innovation-forcing lesson.
3. What exactly did the 1974 OSHA standard specify (TWA, ceiling, excursion), and how does it relate to the “1–3 ppm” in lesson 6 and the UK’s 25 ppm code?
4. What did European Community and national regulators do in 1974–78 and after? Have occupational limits been lowered since 2013, as lesson 6 urges? [external lead: EU binding limit reportedly set at 1 ppm in the 2017 carcinogens directive amendment; verify]
5. What does post-2013 evidence say about non-liver cancers (brain, lung, lymphohaematopoietic), hepatocellular carcinoma and reproductive effects? Has lesson 7’s claim of underestimation held up, or has the more cautious IARC 2008 view prevailed? [check: IARC Monographs vol. 100F (2012) and later cohort updates]
6. What was the stated scientific basis of the Wong and Whorton (1993) retraction (“diagnostic bias”), and how have independent reviewers judged it?
7. Were there conflicts of interest in the 1988 Doll review the chapter cites as “widely accepted” (p. 192)? The chapter does not raise the question. [external — verify: later reporting (2006) described a paid consultancy between Doll and a VC-producing company during the relevant period]
8. What happened to exposures and health outcomes as production shifted to Asia (p. 181)? Are “older technologies” still in use and at what exposures? [possible link to the mercury-catalyst production route and the Minamata chapter; verify]
9. Did the 1972–73 secrecy agreement or the omission at the July 1973 NIOSH meeting have legal consequences, given the mandatory January 1973 request for information?
10. How did the Soviet findings of 1949–1966 circulate in the West, and why were they not acted on?
11. Counterfactual: without the rare-tumour cluster at one plant and Creech’s persistence, when would action have come? Is there a comparable case where the harm was a common disease?
12. How reliable is the documentary base, and what is its provenance (litigation discovery, Markowitz and Rosner)? Is there any published industry-side or independent historical counter-account?
13. How should the pre-1974 industry funding of Maltoni’s research be characterised? What were the terms? Was the secrecy agreement a condition of funding?
14. Did the “trade association versus progressive companies” mismatch (p. 179) exist in the VC case, and which firms were on which side?

Audit log#

Independent audit on 2026-09-25 against the full text extract (PDF pp. 181–204). I re-checked Figure 8.1, Table 8.1, Box 8.1 and Table 8.2 visually in the PDF, and checked the author entries in the annex (report pp. 688, 689, 692, 697, 698). I also confirmed that Ch 16 carries an industry view and a reply, which supports the “other chapters include industry responses” remark. No mentions of contemporary technologies were found.

Box 8.1 attribution - Box 8.1: “delay, feigned ignorance and irresponsibility” had been presented as a quotation from Justice Clark. It is unquoted wording in the box. I corrected this in the Box 8.1 notes, the limitations section and the digest’s core story.

Authors - Soffritti: removed the unsupported “leads the institution founded by Maltoni” and marked the Maltoni link as external. In the digest, “Maltoni’s successor” is now his annex title. - Gee: “EEA’s science, policy and emerging-issues lead / originator” replaced with the annex wording (“worked with the EEA since 1995…”; “catalyst, EEA editor and a chapters author”).

Summary box and Figure 8.1 - Summary box: restored the “potentially positive” hedge and the “more common in the US” qualifier on the legal aspects. - Figure 8.1: added the agreement’s carve-out for information already known to or independently developed by Dow. Also noted that the agreement bound firms, not Maltoni.

Sections 8.2–8.4 - 8.2 exposure: corrected the conflated “>38 ppm to <0.4 ppm in processing” to the chapter’s actual (ambiguous) statements. - 8.3.3 Solvay: corrected the compression of the Belgian, Romanian and 1962–65 cases. Flagged the tension between “no reports were made public” and the Solvay doctor’s publication. Noted the uncited Cordier et al. 1966 reference. - 8.3.3 Michigan report: “500 ppm is not safe” softened to the source’s “could not be assumed that workers would be safe”. - 8.3.4 result 6: flagged the 0.3 vs 1 mg/kg inconsistency in the lowest tumour-producing ingestion dose (both p. 186). - 8.4 OSHA limits: corrected citation placement. “Federal Register 1974” is cited for the April 50 ppm limit but resolves to the October entry, and the April NARA entry is uncited. Restored the Maltoni and Lefemine citation for the 1 ppm cut. - 8.4 registry: “103 cases worldwide” corrected to “in various countries”. Added the 1975–78 case finding. - 8.4 epidemiology: noted the “Tabershaw-Cooper” vs Tabershaw and Gaffey naming inconsistency.

Section 8.5 - Rohm and Haas: clarified that it concerned a community cluster and a litigation claim reported by a journalist, not an adjudicated finding. - Liver carcinoma: noted that the French and Japanese citations are case reports, not excess-mortality studies. - Reproductive effects: added the chapter’s hedged “associated with” and the ATSDR 2006 review. Separated out the animal (prenatal) evidence.

Panel 8.1 - Elliott and Kleinschmidt: “later work found liver cancer near VCM plants” corrected to “research into” incidence, with results not reported by the panel. Fixed the parallel claim under distribution of harms. - UK code: noted it is undated. Noted the panel’s “the court decided the level” versus the chapter’s OSHA-set, court-upheld standard. - Footnote 6: “UK BCME workers” narrowed to workers at Tree’s Welsh plant. Noted the footnote’s ambiguous authorship and Gee’s co-authorship of the source.

Table 8.2 - Added the unreconciled 1967-row 50 ppm recommendation, the firmer 1964-row wording (“clearly linked”), the cost claim in the 1965 row, and the Viola date discrepancy.

Panel 8.2 - Added the panel’s own hedges: fact 4 on untestable exposure circumstances, fact 6’s “additional or different target sites”, and “not all … carcinogens are equal”. - Corrected “closing line”. Noted that the concordance claim extends to probable carcinogens and that the panel frames critics as “vested interest opposition”. Marked “long contested in toxicology” as external.

Timeline and lag analysis - Softened “captured by producers”, “unnoticed” and “emergency action”. - UK row: “voluntary European” changed to “less legally binding UK”. Also redated it to 1974–75. - Steam stripping marked as undated in the chapter rather than “late 1970s”. Added the Table 8.2 note to the 1967 row. - Lag analysis: “legal 500 ppm throughout” corrected to reflect the TLV-to-PEL history. “~18 months held under secrecy” corrected to ~15 months (Nov 1972–Jan/Feb 1974), with Maltoni’s April 1973 public presentation noted. The same fix applies to “culpable delay” in the hindsight section, now de-editorialised.

Lessons, mechanisms and insights - Lesson 2: added the internal tension between a fix “available all along” and regulation that “stimulate[s] innovation”. - How warnings arose: Viola was commissioned to run an experiment, not responding to worker complaints. “Detection was never the bottleneck” is now limited to the non-cancer hazards. - Mechanisms: - The “narrowing” sequence no longer imposes an order the source doesn’t give. - “Revision required consensus / veto” softened to what the source supports. - The 7–3 vote is located in the trade-association committee, not “within firms”. - The odour proxy no longer attributes assumptions to “workers and managers”. - Monitoring is placed in the UK after 1974. - “Consensus rested on shared ignorance” is now attributed to Clutterbuck’s view. - The Tomatis quote is no longer attributed to Maltoni. - Media: the 1975 NYT piece is noted as an ex post rebuttal, not a driver of action. - Law: the litigation provenance of the MCA papers is marked external. - Research funding: removed the unsupported claim that the programme’s independence “let the findings reach an OSHA hearing”. - Insight 1: “the main source of delay” softened, and limited to the liver and bone hazards. Insight 3’s evidence corrected (not necessarily “a company physician’s” draft).

Limitations - Internal-inconsistency list expanded (ingestion dose, Table 8.2 rows, citation mismatch, Solvay publicity, Tabershaw naming). The IARC 1982/1987 explanation is marked external.

Digest - 7–3 vote wording corrected. “Fifty companies blocked” changed to “objected, and the ACGIH put it off”. Added Suciu 1963 as public evidence. - “Cancer kept secret (1970–73)” corrected: Viola’s 1970 results were public, and the Maltoni finding was revealed to firms in November 1972. - Goodrich announcement corrected (three announced, four and then seven found by Creech). The USD 278 million figure is flagged as unsourced. The discovery rationale is reworded to the source’s claim. - Litigation provenance marked external. Insight 1 scoped. Two inconsistencies added to the caveats.