Late lessons from early warnings the precautionary principle 1896-2000

David Gee, Coordinator, Emerging Issues and Scientific Liaison, EEAEuropean Parliament Meeting, Dec 11th ,2003

EEAs main TasksTo provide the Community and member countries with information needed to: identify, frame, prepare and implement sound and effective environmental policy measures monitor, evaluate and disseminate actual and expected results of such measures to clients and the publicTo establish and coordinate the European environment information and observation network (EIONET), for the collection, assessment and sharing of data with European Commission services, EEA member countries and international organisations.

Wisdom is to know,

that you do not know

(Socrates)

An early lesson, lately forgotten?

Chemicals the dearth of dataToxicityOf 2,500 high production volume chemicals, only 14 % have sufficient data for a minimal OECD risk assessmentExposureThere is very little data on exposure levels for ecosystems and people, particularly sensitive groups (e.g. children)External environmental costsLittle data but the costs of ill health, environmental damage, fires/explosives, etc., are largeand can be reduced by getting market prices & the regulatory framework rightEEA draft

Measuring is Not Knowing:The Marine Environment and the Precautionary PrincipleThe enormous number of papers in the marine environment means that huge amounts of data are available, but we have reached a sort of plateau in the understanding of what the information is telling us . We seem not to be able to do very much about it or with it. This is what led to the precautionary principle, after all we do not know whether, in our studied ecosystem, a loss of diversity would matter, and it might. Marine Pollution Bulletin, Vol 34, No. 9, pp. 680-681, 1997Context 1

The precautionary Principle: The working definition used in the Late Lessons ReportA general rule of public policy action to be used in situations of potentially serious or irreversible threats to health or the environment, where there is a need to act to reduce potential hazards before there is strong proof of harm, taking into account the likely costs and benefits of action and inaction

(Source: Late Lessons Report, Introduction, p. 13)

The Precautionary Principle is:NOT a prediction- it is a process that may/may not lead to exposure reduction measures.NOT the same as prevention which is concerned with known risks. Precaution = Uncertain/unknown hazards/risks. Eg.bans on asbestos or smoking in 1950/60s would have involved both Precaution and Prevention but bans in 2003 are Prevention only because risks are well known. (See Late Lessons, P192)NOT based on zero risks but aims to achieve lower/more acceptable risks/hazards with lower overall costs, both quantifiable and non-quantifiable. NOT proof against mis-use or bad decision making (as with any other policy tool).NOT same as risk assessment: it is broader, deeper, supplementary to R.A.Towards a Common Understanding of the Precautionary Principles

6.NOT oblivious of costs of all kinds, and in both directions (i.e. of acting or not acting to reduce risks/hazards), including secondary costs/benefits.7.NOT one sided it applies to substitutes/alternatives too: and it promotes innovation and combats monopolies such as asbestos, CFCs, PCBs, antibiotics in animal feed, etc.8.NOT based on anxiety/emotion- but uses the best of the systems sciences of complex processes to make, hopefully,wiser decisions.9.NOT a guarantor of consistency or predictability between cases each case is different/ has different facts. (as with legal cases)10.NOT affected by peoples motives for promoting/opposing the PP but their interests(economic, political, scientific) in the issue do need declaring.

Late Lessons CaseStudies: Types of Issue1 Biodiversity (Fisheries)1 Pathogen (BSE)2 Physical Agents (Radiation, Asbestos)3 Pharmaceuticals (DES, Beef Hormones, Antibiotics in animal feed)7 Chemicals (Benzene, PCBs, Great Lakes Pollution. MTBE, CFCs, TBT, SO2)

Late Lessons: An Analysis of 14 False Negative Case StudiesFalse Negatives = Whoops! It really is dangerous after all!

False Positives = Whoops! It really isnt dangerous after all!

Some more False Negatives Not Studied(See Late Lessons Vol 2, 2005)

Lead Aral Sea Minnamata Bay DBCP Thalidomide Mule spinning oil BCME VCM The Drins Bladder cancer Overhead powerlines DDT

Why So Many False Negatives?Sound science generates more false negatives than false positives (= good science but poor public policy)

Short term economic/political interests can dominate longer term, total welfare interests.

Late Lessons is based on case studiesStructured around 4 questions:When were the first scientifically based early warnings ? When and what were the main actions, or inactions, by societys actors ? What were the costs and benefits (all kinds) of the actions/inactions; andWhat lessons can be drawn that may help improve decisionmaking and reduce overall costs ?

What about False Positives? Harder to prove

Failed to find authors for them

Secondary Benefits often make them worthwhile eg. Y2K bug

Vol 2 of Late Lessons will have a false Positive chapter

Antimicrobial Feed Additives (AFA) Chapter An Early Warning example: 1969 UK Medical Research Councils Swann Committee: Despite the gaps in our knowledge .. We believe on the basis of evidence presented to us, that this assessment is a sufficiently sound basis for action .. The cry for more research should not be allowed to hold up our recommendationsSales/use of AFA should be strictly controlled via tight criteria, despite not knowing mechanisms of action, nor foreseeing all effectsMore rewarding to improve animal husbandry than to feed diets containing AFASource: (HMSO, UK, Nov. 1969)

CFCs Chapter: Skin Cancer and Time Lags

Twelve Late Lessons fromEarly WarningsA Identify/Clarify the Framing and AssumptionsManage risk, uncertainty and ignoranceIdentify/reduce blind spotsAssess/account for all pros and consAnalyse/evaluate alternative optionsTake account of stakeholder valuesAvoid paralysis by analysis by acting to reduce hazards via the precautionary principle.

Towards A Clarification of Key TermsSource: Late Lessons, page 192

How Can We Respond toIgnorance (Surprises)?Use intrinsic parameters as proxies for unknown but possible impacts (e.g. persistence and/or bioaccumulation potential of chemical substances. (See Case Studies on PCBs, MTBE, CFCs, TBT) Reduce specific exposures to potentially harmful agents on the basis of credible Early Warnings of initial harmful impacts (thus limiting the size of any other surprise impacts from the same agent, e.g. the asbestos cancers that followed asbestosis; and PCB neurotoxicological effects that followed wildlife impacts). (See 5 other responses to ignorance in Late Lessons supplementary report, EEA 2003)

Promote a diversity of robust and adaptable technological and social options to meet needs (which limits technological monopolies such as asbestos, CFCs, PCBs etc., and therefore reduces the scale of any surprise). EEA draft

Including their distribution; and secondary benefits and costsInclude effects of innovation and technological change, as well as social impacts of technology choicesProduct prices to include full costs of production, use & disposal (the 'polluter pays principle')This maximises efficiency, stimulates innovation and minimises environmental and health burdensPrecautionary costs should not greatly outweigh the benefits; the proportionality principle Lesson 3: Assess, justify and account for all pros and cons

Avoid Substance monopolies that stifle innovation (see Asbestos, CFCs, PCBs, AFA)Stimulate available alternatives (see antimicrobials, asbestos, radiation, CFCs)(The principle of substitution, now part of OSPARs Strategy on Hazardous Substances requires a comparative assessment of alternative means of providing services; as does the EU Biocides Directive)But use precaution, eco-efficiency and diversity with substitutes, tooLesson 4: Evaluate alternative means of providing services

(CONT.)B Broaden Assessment InformationIdentify/reduce interdisciplinary obstacles to learningIdentify/reduce institutional obstacles to learningUse lay, local and specialist knowledgeIdentify/anticipate real world conditionsEnsure regulatory and informational independenceLong term monitoring/research

Misplaced certainty about the absence of harm played a key role in delaying preventive actions in most of the case studies

(Preface, Late lessons from Early Warnings: the Precautionary Principle 1896-2000)Less Hubris, more Humility?

ON BEING WRONG: Environmental and Health Sciences and Their Directions of Error1 Some features can go either way (e.g.inapproriate controls) but most of the features mainly err in the direction shown in the tableEEA draft

SCIENTIFIC STUDIESSOME METHODOLOGICAL FEATURESMAIN1 DIRECTIONS OF ERROR-INCREASES CHANCES OF DETECTING A:ExperimentalHigh dosesFalse positiveStudiesShort (in biological terms) range of dosesFalse negative(AnimalLow genetic variabilityFalse negativeLaboratory)Few exposures to mixturesFalse negativeFew Foetal-lifetime exposuresFalse negativeHigh fertility strainsFalse negative (Developmental/reproductive endpoints)

EEA draft

Observational ConfoundersFalse positiveStudiesInappropriate controlsFalse positive/negative(Wildlife &Non-differential exposure misclassificationFalse negativeHumans)Inadequate follow-upFalse negativeLost casesFalse negativeSimple models that do not reflect complexityFalse negative

BothPublication bias towards positivesFalse positiveExperimentalAndScientific cultural pressure to avoid false positivesFalse negativeObservationalStudiesLow statistical power (e.g. From small studies)False negativeUse of 5 % probability level to minimise chances of false positivesFalse negative

Levels of proof - some examples Beyond all reasonable doubt Reasonable certainty Balance of probabilities/evidence Strong possibility Scientific suspicion of risk Negligible/insignificantEEA draft

Endocrine Disrupting Substances (EDSs):Some difficult featuresTiming of the doseMixturesNatural and synthetic EDSsHormonal imprintingEndocrine systems cross-talkGenerational impactsImpacts/Adverse impacts Adverse impacts due to imbalance between opposing affects in complex systemsEEA draft

Its the Timing of the Dose that Makes the Poison for Developmental and Reproductive HarmThe time of life when exposures take place may be critical in defining dose-response relationships of EDSs for breast cancer as well as for other health effects (WHO/IPCS State of the Science of EDSs, 2002)DESDDTTBTThalidomideEtc.EEA draft

TBT Antifoulants: A Tale of Ships, Snails and Imposex (Chapter 13)An increased appreciation of scientific complexity and indeterminacy accompanied the unfolding of the TBT impacts story. Very low doses caused adverse impacts (i.e. in parts/trillion)High exposure concentrations were found in unexpected places e.g. the marine microlayer

EEA draft

(CONT.)Bioaccumulation in higher marine animals, including sea-food for human consumption, was greater than expected A highly specific, initially uncommon impact (imposex) was quickly linked to one chemical, TBT: this easily identified linkage is not likely for more common impacts on development and reproduction in humans e.g. sperm counts, breast cancerEarly restrictive actions in 1982-85 were based on a strength of evidence for the association only: causality and mechanisms of action knowledge came much laterEEA draft

Towards a Clarification of Some Key Terms: From Association to Causation and Mechanisms of ActionEEA draft

Situation

Association

Relevant Information

Some weak to strong evidence linking an environmental and/or health impact with a stressor

(We see that X may cause Y)Time to produce relevant information

Weeks to years

Years to decades

EEA draft

Situation

Causation

Mechanisms of ActionRelevant Information

Substantial evidence and knowledge linking impacts with stressor(We know that X causes Y)

Substantial evidence and knowledge and understanding of the reasons why there is a causal link between an impact and a stressor(We understand why X causes Y)Time to produce relevant informationYears to decades

Usually Decades

Applying the Precautionary Principle: Key DecisionsAssigning burdens of proofEvaluating the evidence (in a framework)Choosing appropriate level of proofEvaluating the justification / needAssessing alternativesAssessing appropriate policy measuresTaking transparent decisionsInvolving stakeholders in all stages

EEA draft

Man has lost the capacity to foresee and forestall....he will end up destroying the earth.

- Albert Schweitzer

Will this prediction come true ?Or could the precautionary principle help us to foresee and forestall Hazards using the best of systems science whilst stimulating innovation?EEA draft

Multi-causality and ComplexityHOST GENETICSHOST STATEEXPOSURES/DOSESEFFECTSGenotypes and phenotypes++==>Nutritional statusImmune statusHealth status (lifestyle)Age etc.Multiple exposures/dosesVia several exposure routes (skin, inhalation, ingestion)Environmental indoorEnvironmental outdoorOccupational Low doses and/or peaks24 hours continuousOr single dose at critical timesPre- and post-natal dosesLifetime and cumulative target organ doseBiologically effective doseOften unknown doses

InitiatingPromotingRetardingSuppressingCausingChanges in Cells, tissues,HormonesOrgansFunctions;normal distributions of bio-functions.InfectionsSkin CancerNeurotoxicityRespiratoryCirculatoryReproductiveOther adverse developmental impacts

HARM

Possible strategies for avoiding, reducing and compensating harmHOST GENETICSGenetic Engineering ?

Remove sensitiveGroups from Exposure ?EXPOSURES...but which strategies would be most effective?Improve Nutrition ?

Promote good health ?

Vaccinate ?Reduce ?

Eliminate ?Predict ?

Identify earlier ?

Stop/slowProgression to Harm ?Cure ?

Compensate ? Clean up ?

Learn from mistakes ?

Warn ?HOST STATEEFFECTSHARM

Illustrative Multi-causality: Childhood asthma illustration ++=++=25%25%e.g.Mites 10% Pets 5% Passivesmoking5% NO 5% Damp5%e.g. Pollen 10% Industrialpoon5%c50%30% indoor20% outdoor100%Xe.g.Allergic sensitivity

e.g.Nutritional and immune status

30% indoore.g.Mites10%Pets 5%Passive smoking 5%NOx 5%Damp 5%20% outdoorPollen 10 %Industrial pollution 5%Traffic pollution 5%

Asthma caused/ Aggravated in some children

HOST GENETICSHOST STATEEXPOSURESHARM

Reducing traffic pollution may be a very cost effective policy action on asthma; given Inter-linked causal chains Less dependence on individual behaviour change Cost effectiveness/distribution Large secondary benefits e.g. reduced traffic congestion, accidents; community fragmentation; CO2 etc. (multi-pollutant, multi-effects strategy)

The figure identifies and locates the 3 main clusters of causal factors usually involved in disease and ill health viz: The genetics of the individual The host state of the individual, which contains most of the factors usually described as lifestyle that affect nutritional and immune status. These are partly determined by factors that are largely outside the control of the individual, such as societal choices over energy, transport & food manufacturing and distribution systems. This cluster also includes income, employment and housing, as well as age, as sensitivities to stressors vary greatly across the age range of day one after conception to 90+years. Host state factors contribute to ill health both directly, by providing a specific link in the causal chain, such as extensive use of the contraceptive pill, obesity, lack of exercise, and late childbirth, and indirectly, by modifying the extent to which environmental stressors interfere with health. Environmental stressors. This cluster includes air, water, food and consumer product contamination, as well as climate, radiation, noise, etc. These stressors can vary from large and acute exposures, to small, yet continuous chronic exposures over many tears. Usually, exposures are mixtures of physical, biological and chemical agents that vary with the emissions and the concentrations of the agents. Their impacts are determined by their abilities to deliver biologically effective doses ,which of course vary with both the genetics and the host state of the people exposed. Knowledge about specific exposures are usually difficult to obtain but generic exposures can usually be inferred from the production of stressors, such as thousands of tons of pollutants emitted to the environments that people inhabit.Most diseases, such as cancers, asthma, and neurological diseases are the result of factors from all three clusters eg a genetic /environment interaction that is mediated by the state of the host body (eg foetus, child ,adult) before it receives the environmental stressors. Such multi-causality provides many obstacles to understanding of the mechanisms and factors in the causal chain but it also provides many opportunities for removing links in the chain and preventing harm. The simple model in Fig 4 breaks down on close analysis because the boundaries between the boxes are fuzzy: the environment can influence genes, and genes can influence lifestyle choices, which in turn affect both the host state and the stressors that it receives. In most cases neither Nature (genetics) nor Nurture (environment) is an independent cause of harm both are usually necessary[1] . In other words: the genes provide the bullets but the environment provides the triggers; and Biology is not necessarily destiny. Despite its simplicity, the clustering of many interacting causes into the three groups of Fig 4 can be a useful tool to help policymakers and the public respond to the complexities of multi-causal reality. Interactions between causal factors from all three clusters seem to be involved in most diseases, with particular combinations of factors being responsible both for individual cases of disease and for particular fractions of total disease, such as cancer or asthma. Before applying the approach to a specific disease such as childhood asthma we need to consider the question: where best could policymakers intervene in the causal chain?. [1] Ref the nature/nurture article,1934 cited in the kenneth rothman article-to follow.

Genetics. Genetic engineering; and the removal of the genetically susceptible from relevant environmental exposures (eg banning pregnant women from work involving lead;or advising asthmatics to avoid outdoor exercise on high pollution days) are two possible interventions. Both have serious practical and ethical limitations that reduce their feasibility and effectiveness.Host state. Public health campaigns aimed at behaviour change can be successful (eg on smoking,alcohol,exercise, vaccinations) but there are difficulties in reducing risks from often voluntary lifestyle choices that are located in socially determined contexts but which are often perceived as issues of personal freedom. Environmental Exposures. These can often be reduced or eliminated by public authorities, supported by a public that usually sees them as externally imposed, but complicated by the benefits associated with the activities that generate the exposures. (The distribution of the costs and benefits is also a critical issue, particularly for children, defined as from conception to 18, as they often bear the costs of economic activities that deliver most benefits to adults. Similarly, for globally distributed harm from chemicals such as POPs, or CO2, or CFCs, the benefits are mostly concentrated in the developed regions of the world whilst the costs, from bio-accumulations and other impacts, are mostly concentrated in developing regions). Effects The precursors to Harm, which provide opportunities for early warnings, precaution and prevention, arise in the changes in cells ,tissues, hormonal systems ,organs etc. that may progress to disease. However, at least in the West ,there has been a reluctance to base preventive public policy on effects only: evidence of actual or very probable harm has usually had to be generated, using high levels of proof, which in the cases of serious, irreversible, and long term harm, has often been very costly. The EU Treaty mandates both precaution and prevention and there is sometimes confusion, particularly in the health field, about the differences between the two, but the examples of smoking and asbestos can be used to illustrate them. It would have been both precautionary and preventive to have banned asbestos in the 1930s and smoking in the 1960s, but it was only preventive to have banned them in the 1990s, as the risks were by then well known. If the precautionary principle is to be applied then it generally needs to targeted at the effects stage in the framework rather than at the Harm stage.Harm. Treating ,curing, and compensating disease, and abating or cleaning up contamination, are interventions that can dominate the lifecyle of a public health or environmental hazard but the economic, ethical and political costs of intervening only at this point can be huge.The societal choice about where best to intervene along the causal chain can be difficult, and clearly involves more inputs to decision making than from science alone, including choosing an appropriate level of proof. However, a the focus on reducing /eliminating exposures is often the most cost effective point of policy intervention.

The figure provides an illustration of what is currently known about the many causes and aggravations of childhood asthma. The percentages of attributable risk are illustrative only, but are broadly supported by the available science. The 100% of the Harm represents both the asthma in any one case that arises from any combination of some of the risk factors, and the total of all childhood asthma in a population. The sum of the attributable risk factors in Fig 6 add to 100% in order to illustrate relative causality, but in practice they will add up to more than 100% because the risk factors are often inter-dependent, over-lapping, and competing with each other, a point that is often misunderstood.(See Box A) Box A. Inter-dependent,over-lapping and competing risk factors: an illustration from Traffic accidents.There are,say,100 fatal accidents on a road over a 10 year period. Studies show that compulsory headlights would prevent 20% of them, speed limits would prevent 30%, speed bumps in the road would prevent 70%, and putting a policeman in every car would prevent 98% of them. These sum to nearly 200% which is impossible as there is only 100% (100 deaths) to save. The risk factors clearly compete, inter-react, and are also focused on different parts of the causal chain. It also follows that those risk factors that are tackled first have the greater opportunity to save deaths, relative to their effectiveness, as there is more harm to save. So, headlights and speed limits would save more lives (50) if tackled before speed bumps (35 ie 70 % of the 50 lives left), but less lives if tackled after speed bumps. In reality, even relatively simple accident scenarios are changed by a change in any one of the main variables, such that the lives actually saved by taking action on the risk factors wouldnt be as simple as the studies estimated. With health hazards, reality is even more complex, but the general point about the significance of the order in which preventive measures are implemented still applies. (See ref 6, p. 577/8 from which this example is derived). This provides another reason why focusing on seemingly small environmental risk factors can be cost effective.Although much remains unknown about the causes of childhood asthma, there are many studies showing the different contributions that genetics, host state, and environmental stressors ,both indoor and outdoor, seem to play in the causation and aggravation of asthma. The attributable percentages are illustarttive only .but are broadly supported by the scientific evidence. With so many environmental stressors having been identified it follows that any one may only play a small part in overall disease causation but reducing them may have disproportionately large impacts but to inter-dependence ,where removing even a small link in a causal chain can break the chain of causation, thereby reducing the disease potential of several other risk factors. In choosing which risk factor to focus on first (which, as we saw above, will also maximise its disease reduction potential), other considerations such as feasibility, effectiveness, and secondary benefits need to be taken into account. From this analysis it can be seen that even small risk factors, such as traffic fumes, can de very cost effective environmental stressors to focus policy actions on.This type of approach to characterising the cause /effect framework can be applied to the other priority diseases such as childhood cancer, neurological diseases and breast cancer.