LL1-06 digest — Ch6 PCBs and the precautionary principle#

Late lessons from early warnings (EEA, 2001), pp. 64–75.

Authors. Janna G. Koppe, a neonatologist whose own PCB and dioxin research feeds the chapter, and Jane Keys, a freelance researcher who was also one of the report’s editors (biographies p. 197; editorial team PDF p. 1). Box 6.2 is by Søren Jensen. There are no panels and no industry responses; industry appears only as quoted via one web article, and a group critical of “over-regulation” (ACSH) is quoted once, for a concession (p. 70).

Core story#

• Why PCBs were used. They were mass-produced from 1929 because they were less flammable, more stable and more compact than what they replaced, and “thought to be safer” (p. 64).
• 1930s warnings. Halowax workers (exposed to PCBs plus chlorinated naphthalenes) developed chloracne; three died, and autopsies of two showed severe liver damage. Drinker’s 1937 rat study was shown to Monsanto, General Electric and public health officials, at a meeting that closed with a plea to avoid “mob hysteria” among workers. It was published but “did not gain the wider attention of policy-makers” (p. 64).
• Discovery in the environment. In 1966 Jensen, working on DDT, found unknown persistent compounds in Swedish sea eagles by chance; two more years of work identified them as PCBs, published in 1969. Baltic seal infertility (1960s–70s) and the Yusho poisoning in Japan (1968) came in the same period, and evidence of global spread built through the 1970s (pp. 64–66).
• Monsanto’s two positions. Publicly, Monsanto said the suggestion that PCBs were “highly toxic” was “simply not true”, that the “source” of the residue was “not yet known”, and asked for “extensive research, on a worldwide basis” (the chapter summarises this as denying the chemicals were PCBs, which goes slightly beyond the quoted words). Its 1969 internal plan acknowledged contamination on several continents and ruled out stopping production, which would cause “profits to cease and liability to soar” (p. 65). All of this comes via one secondary source, Francis 1998.
• Staged action.
• Monsanto’s 1971 reformulation rested on an oversimplified idea of toxicity (p. 66).
• Governments restricted open uses first (Sweden 1972, OECD 1973), then new closed uses and production (US TSCA 1976; UK and US production ended 1978–79), then all new uses (OECD 1987, recommending an end by 1989), then existing equipment (Sweden 1995, EU deadline 2010) (pp. 66–69, 72). Some eastern European production continued to the mid-1980s (p. 67).
• Science. Understanding congeners and how the environment changes the mixtures let “the apparently conflicting evidence” be resolved (pp. 66–67); recognising fetal vulnerability added “a new paradigm for toxicology” in which timing matters as well as dose (p. 67). Some of the science was “still being debated” in 2001 (p. 72).
• Ending. A 1999 Belgian feed contamination, “strongly suspected” but not proven to derive from illegally disposed-of old transformers (p. 71). The verdict: “At almost every stage government action was taken only when there was a high level of scientific proof” (p. 72).

Key evidence#

• Rice-oil poisonings (Box 6.1, pp. 66–67).
• Weathered residues apparently more toxic than the product. Mink fed Great Lakes fish showed liver and reproductive toxicity “comparable to” mink fed three times the quantity of the commercial mixture (p. 67). The chapter hedges (“appear to be more toxic”) and says this question is still debated (p. 72).
• Neurodevelopmental associations in the Lake Michigan and Dutch cohorts (pp. 69–70). The Dutch results are mixed: reduced attention at 42 months was linked to prenatal exposure, but hyperactivity was linked to current levels, an 18-month neurological effect was gone by 42 months, and no effect on attention or activity was found for dioxin-like congeners.
• Figure 6.1 (p. 65). PCBs in Baltic guillemot eggs peaked around 1975 and fell about 90% by the late 1990s, and eagle breeding success regained its pre-1950s level only around 2000. DDT fell in parallel, so the recovery is not attributable to PCB measures alone. The text is silent on these trends.

Authors’ lessons#

• 1930s. Evidence, “some at a low level of proof”, was “largely retained within the industry”. Precaution “would have prevented the toxic legacy” (p. 71).
• Late 1960s. Proof was high for human harm “in certain circumstances” (from PCBs “or their breakdown products”) and for bioaccumulation, and “lower, but still substantial” for reproductive effects in marine mammals. Acting below “beyond reasonable doubt” would have left “a more manageable, less costly problem” (p. 71).
• 1970s. Measures were a “half measure”, leaving existing uses and contaminated sites alone; “It is probable that the technical difficulties and costs” explain this, even though alternatives existed by the late 1970s (pp. 71–72).
• More science. Framed as two arguments: earlier study might have allowed “an earlier resolution”, but calls for more science can also delay “justifiable action” (p. 72).
• Legacy and advocacy. “A large percentage of historic PCB production has escaped beyond our control”. Body burdens must be reduced. Children’s behavioural and respiratory problems may be “in substantial part” due to PCBs. Who judges acceptable risk for the unborn? (p. 72)

Main mechanisms#

• Warnings contained within an industry and occupational-health circle (pp. 64, 71).
• Public denial alongside private acknowledgement, with liability deterring admission (p. 65).
• A heterogeneous agent treated as one, and transformed by the environment (pp. 66–67).
• Containment assumed, though “closed systems” leaked, aged and were dumped (pp. 66, 69–71).
• Action sequenced by cost (pp. 66, 72).
• Persistence, bringing transboundary, irreversible harm to the unborn, the Arctic and wildlife (pp. 70, 72).

Transferable insights (strength)#

1. The properties a product is valued for (stability) can be the source of lasting harm (pp. 64, 70, 72). Moderate.
2. Warnings can stay in a closed circle, and publication does not mean uptake (pp. 64, 71). Moderate.
3. Public calls for more research can coexist with private acknowledgement (pp. 65, 72). Moderate to strong for the divergence, via a single secondary source; moderate for reading the research call as delay, which the chapter makes only in general terms.
4. Liability deters admission and prompts partial fixes (pp. 65, 71). Moderate.
5. Fixes built on an incomplete causal model can be misdirected (pp. 66, 71). Moderate.
6. Lumping a heterogeneous class together breeds contradictory evidence (pp. 66–68). Moderate to strong (clear narrative, but the conflicting studies are uncited and the conclusion claims only “some discrepancies” explained, p. 72).
7. The tested product can differ from the real exposure after transformation (p. 67). Strong that it differs; moderate that the transformed form is more hazardous (hedged, and “still being debated”, p. 72).
8. Detection favours acute, visible events (pp. 65, 71). Moderate.
9. Timing of exposure matters as well as dose (pp. 67, 70). Moderate.
10. Single-mechanism metrics can create blind spots (Boxes 6.2–6.3; p. 71). Suggestive (my inference; weakened by Box 6.3’s TEQ including two di-ortho congeners that Box 6.2 says have zero TEF).
11. Controls on new use leave the installed stock as the long-term source (pp. 66, 69–72). Strong.
12. Regulation proceeds from cheap to costly measures (pp. 66, 72). Moderate.
13. Persistent agents escape control and shift harm to distant, future and non-human parties (pp. 66, 70, 72). Strong for persistent agents.
14. Reuse pathways can recirculate legacy hazards (pp. 66, 71). Moderate.
15. Grading evidence at each decision point shows action lagging proof (pp. 71–72). Moderate, with hindsight risk.
16. Protective advice carries health trade-offs (p. 69). Moderate.
17. Recovery after restriction can be substantial but slow, taking decades (Fig. 6.1). Moderate (correlational, and confounded by the parallel fall in DDT).

Caveats#

• Unshown counterfactuals. The 1930s warning was occupational and involved mixed exposures, and environmental accumulation was not recognised until 1966; the chapter never says what precaution in 1937 would have required.
• Overreach. The paediatric respiratory claim exceeds the evidence (it rests on high-dose poisonings), and the Belgian toxicologists are likened to a 1937 industrialist, on the uncited premise that Belgian background levels match Dutch ones.
• Sourcing. All the industry quotations come from one secondary article.
• Inflated framing. “100 years” presumably counts from a class-level 1899 warning; from the first PCB-specific evidence (1936–37) it is about 60 years.
• Internal slips. Box 6.3’s TEQ counts two di-ortho PCBs that Box 6.2 says have zero TEF; Table 6.1 dates Jensen’s identification to 1969, the text to about 1968, and the table’s 1937 row adds chloracne in Drinker’s rats, which the text does not report. The Belgian contamination levels are set beside the WHO intake limit without any intake estimate (p. 71).
• Omissions. No weighing of fire safety or alternatives, no cost figures, and the recovery in Figure 6.1 is ignored.
• Standpoint. The authors’ research, advocacy and editorial roles shape the chapter. Koppe’s three co-authored papers support Box 6.3’s supplementary links (thyroid effects, spatial ability, late haemorrhagic disease); she is not an author of the Dutch cohort studies at its core.
• Strength. The graded levels-of-proof analysis and the legacy-stock dynamics are the chapter’s strongest material.