LL1-05 digest — Ch5 Asbestos: from ‘magic’ to malevolent mineral#

Late lessons from early warnings (EEA 2001), pp. 52–63. David Gee and Morris Greenberg.

Gee: EEA staff and one of the report’s two EEA editors (p. 1), former NGO director and union health and safety adviser. Greenberg: former UK Medical Inspector of Factories who helped set up the first UK mesothelioma register (p. 196). Authors were disclosed as participants in the histories (p. 12). No panels.

Core story#

Chrysotile mining began in Canada in 1879. Within 20 years there were over 100 products from the “magic mineral” (pp. 52–53).

Warnings came early and from outside the expert core: - UK Women Factory Inspector Lucy Deane (1898) described “easily demonstrated danger” and jagged particles under the microscope, and proposed mortality statistics (p. 53). - A clinician’s case, reported in 1906, came with the patient’s claim that his co-workers had all died around age 30 (p. 53). - A French inspector reported about 50 deaths among female textile workers in 1906 (p. 53). - US and Canadian insurers refused cover in 1918, which the chapter calls “early precautionary action” that was “later forgotten” (p. 54). (Notes’ reading: it protected the insurers, not the workers.)

The authors say the Women Inspectors’ reports “were not refuted but simply ignored”, and the French report was “largely ignored” (p. 53). The Factory Department did press for exhaust ventilation after the 1910–11 evidence gave “reasonable grounds for suspicion”, but its inquiries in 1912 and 1917 found “insufficient evidence to justify further action” (p. 54). Deane herself judged the decisive mortality statistics “practically unattainable” with the resources of the day (p. 60). Whether pre-1930 action was warranted is disputed by the historian Bartrip (1998), whom the chapter nonetheless cites in support (p. 59).

The first health study (Merewether & Price 1930) found asbestosis in 66% of workers with 20+ years’ service, probably an underestimate since only current workers were examined. It led to the world’s first asbestos regulations (1931), which covered manufacturing only and were barely enforced: two prosecutions in 1931–68 (pp. 54, 56).

Cancer evidence accumulated in several stages: - Lung cancer case reports from the 1930s, and German recognition in 1938/1943 (p. 54). - Industry-held unpublished mouse data (p. 54). - Doll’s 1955 finding of a 10-fold lung cancer risk after 20+ years’ exposure, which Turner directors tried to suppress (p. 54). - Mesothelioma, identified by Wagner and Sleggs (published 1960). The exposure needed “seemed to be a matter of months only”, and cases included children playing on waste dumps in the mining areas (p. 55). - Causal consensus by 1964, from Selikoff’s union-records study (he had been refused company records) and Newhouse’s London hospital cases, which showed domestic and neighbourhood exposure (pp. 55–56).

Regulation lagged. The 1969 “hygiene standard” covered asbestosis only, ignoring users and cancers (pp. 56, 61). The chapter credits media, an Ombudsman complaint, a 1982 documentary and victims’ advocates with driving later tightening (pp. 56–57). The UK banned asbestos in 1999, the EU by 2005. The WTO upheld the French and EU bans against Canada, accepting qualitative evidence and minority scientific opinion and rejecting “controlled use” (Box 5.1, p. 57). These points check out against the Appellate Body report (WT/DS135/AB/R, paras 167, 174, 178). The Box is wrong, however, to say the US “cross-appealed”: the US was a third participant. And “found in favour of France” simplifies a mixed Panel ruling: the Decree was held to violate GATT Article III:4 but was justified under Article XX(b). The dispute concerned the French Decree, with the EC as respondent.

Key evidence#

• About 250,000 projected EU mesothelioma deaths over 35 years, and 250,000–400,000 total asbestos deaths (p. 52).
• UK mesothelioma deaths are projected to peak 50–60 years after peak imports (Fig. 5.1, p. 52). The UK had about 3,000 asbestos cancer deaths a year, “and rising” (Health and Safety Commission estimate, 1994–95; p. 58).
• Smoking and asbestos together raise lung cancer risk more than 50-fold, against 5-fold and 10-fold separately (p. 55).
• Costs: EUR 400bn at EUR 1m per life; a Dutch estimate that a 1965 rather than 1993 ban would have saved 34,000 victims and NLG 41bn; Turner’s 1994 arrangements to pay up to GBP 1bn; Manville’s bankruptcy; Lloyd’s near-collapse (pp. 56, 58).
• Substitutes existed “for most uses” “by the 1970s”, earlier for some, but were slowed by cartels and by prices that excluded health costs (p. 58).

Authors’ lessons (pp. 59–61)#

1. Heed “competent observers”.
2. Institutionalise long-term monitoring. (The claim that surveys “would have been possible at the time” sits uneasily with Deane’s own “practically unattainable”, p. 60.)
3. The 1931–32 laws were poorly implemented and sanctions “trivial” (the latter asserted without figures).
4. Earlier action would have avoided loss (well supported for the 1950s–60s; the “before 1930” window is contested, and one of the two sources cited for it, Bartrip, argues the opposite); higher prices would have spurred substitutes (asserted).
5. Internalise costs (polluter pays), against short-term horizons.
6. “Latency lacuna”: absent good evidence that “today’s exposures to carcinogens are safe”, assume them unsafe (said to be relevant to “all long-latent-period hazards”), with proportionality and rebalanced false positives/negatives (normative).
7. The healthy-survivor (“pensioners’ party”) fallacy.
8. Speedy, anticipatory compensation.
9. Curb the “ignorant expert” (the Lancet’s 1967 “irreplaceable”).
10. Beware substitutes sharing the fibre form; minimise all exposure.

Main mechanisms#

• Structural biases toward false reassurance (latency, short follow-up, survivor selection, producer-only studies, fine fibrils not accurately monitored by optical microscopy).
• Industry control of data; marginalised messengers.
• Externalised costs and cartels.
• Regulation scoped to the visible and measurable, and moved by media, politics and litigation.
• Smoking synergy obscuring attribution.
• Self-defeating industry short-termism: a former Johns Manville director argued (in the chapter’s paraphrase) that medical research, assiduous communication, insistent warnings and a rigorous dust reduction programme “could have saved lives and would probably have saved the stockholders, the industry and, for that matter, the product” (Sells 1994, p. 56).

Transferable insights (technology-neutral)#

• Proximate observers detect harm first, and their reports are set aside rather than tested (pp. 53–55, 59). Strong for the case from the 1920s on; tempered for 1898–1910, where whether the warnings were actionable is historically disputed. Moderate as a generalisation.
• Latency plus improving conditions produces a self-renewing “all clear” (pp. 55, 56, 59–60). Strong.
• Short-follow-up “negative” studies are uninformative about slow harms (p. 55). Strong.
• Survivor visibility creates false reassurance (pp. 54, 60). Strong.
• Studying the best-controlled segment misses the worst exposures (pp. 54–56). Strong.
• Paper rules without enforcement create the appearance of control (pp. 56, 59). Strong for non-enforcement (two prosecutions in 1931–68); “trivial” sanctions asserted.
• Whoever controls data shapes what evidence exists (pp. 54–55, 58–59). Moderate: the sources are secondary.
• Externalised costs and incumbent coordination slow alternatives (p. 58). Moderate: “the common cause” is asserted.
• Standards follow measurability and the first-recognised harm (pp. 56–57). Moderate.
• Harm peaks decades after use, and the legacy persists in infrastructure (pp. 52, 58). Strong for the mechanism; the magnitudes are projections.
• Delay costs can rebound on the delaying industry (pp. 56, 58). Moderate.
• Public stories and politics move regulators when accumulated evidence did not (pp. 56–57). Moderate: the causation is asserted.
• Messengers get marginalised (p. 55). Suggestive.
• Rebalancing error tolerance yields net gain (p. 60). Asserted.
• Anticipatory compensation improves incentives and records (pp. 60–61). Asserted.
• A legal forum accepted qualitative evidence, respected minority scientific opinion and doubts about “controlled use” as grounds for a ban (Box 5.1, p. 57). Strong (verified against the Appellate Body report); “a lower level of proof” is the Box’s gloss.
• Similar evidence was weighed differently across jurisdictions: Germany in 1938/1943 against the UK in the 1980s (p. 54). Suggestive to moderate, because the chapter notes that lower background lung cancer eased Germany’s attribution.

Main caveats#

• Advocacy by participants, relying on secondary histories, self-citation and a Greenpeace-commissioned cost study.
• Benefits and substitution limits are thinly treated; France’s own ban kept exemptions (p. 57).
• The chrysotile potency debate is not engaged.
• UK-centred: omits the US EPA’s 1989 ban and its 1991 court reversal (general knowledge).
• False-positive rebalancing is argued from one false-negative case. The editors say they sought false-positive cases but none robust emerged (p. 12).
• Some hindsight, though Legge’s 1934 lament and the 1911 “reasonable grounds for suspicion” show that contemporaries suspected the risk. The pre-1930 “missed opportunity” reading is contested: Bartrip (1998) argues there was “no compelling medical or scientific evidence … until the late twenties”. The chapter cites him (misdated “1931”) as if in support, and does not mention that co-author Greenberg had publicly replied to him (1999).
• Doll’s “unscientific and emotional” is the authors’ paraphrase, not a quotation (p. 57).
• Errors confirmed against primary sources:
• World production did not “grow to” 2 Mt in 1998; USGS puts the peak at almost 4.8 Mt around 1977, and output had already exceeded 2 Mt by 1965.
• The US was a WTO third participant, not a cross-appellant (Box 5.1).
• Tage’s forename is Reginald, not “Ronald” (p. 54): see the title of the cited Greenberg 1993 paper.
• Bartrip is 1998, not 1931, and argues against the view it is cited for (p. 59).
• Slips still to check: the “USD 2 billion” US settlements; Table and text date mismatches; the Knox citations.