Setting and Using Environmental Standards

Highlights of SETAC workshopFaringdon, October 2006

Paul Whitehouse

Chemicals Science

Environment Agency

SETAC workshopAn opportunity to ‘take stock’

• of technical developments

• wider aspects e.g. role of stakeholders in regulatory decision-making

Scope

• chemicals

• environmental receptors

• human health ()

• microbes, radionuclides

SETAC workshop - Working Groups

Aquatic effects assessment

Implementation

Socio-economic issues

Terrestrial effects assessment

Selected highlights

Types of standard A process for delivering standards Effects assessment - data and extrapolation Implementing standards Incorporating an economic dimension

Types of standard

Statutory threshold - must not be exceeded

Threshold that prompts action - not usually statutory

An aspiration - not statutory

‘benchmark’‘trigger value’

‘screening value’

‘standard’‘limit value’

‘goal’‘guideline’

Usually, but not always numerical

always numerical, usually with accompanying conditions e.g. duration, confidence of failure, return period

• Failure can have serious implications (legal, financial)

• Costs and benefits important

• Implications of failure less serious

• Conservativism is appropriate

A process for developing standards

PROBLEM FORMULATION

DEVELOP SPECIFICATION

DERIVE STANDARD

IMPLEMENT STANDARD

What is the standard intended to achieve?To what extent should economic factors influence the outcome?Who will be affected?

How should the standard be expressed?Methodology - constraints?Who needs to be involved?

Are the data adequate?Account for uncertaintiesIncorporate socio-economic factors

Where is to be applied?

How confident do we need to be before we take action?

What will we do in the event of failure?

• Consistency across regulatory regimes

• Need for transparency - report

assumptions, decisions, uncertainties

• Involve stakeholders

• Value of peer review

Effects assessment - overview

Step 4Predicted no effect

concentration determination

Step 4Predicted no effect

concentration determination

Data extrapolation

Step 1 Data gathering Step 1 Data gathering

Step 2 Data selection

Step 3

Data Quality assessment of data important e.g. Klimitsch codes

acceptable - supporting - unacceptable Relevance

demographic endpoints (survival, reproduction, development) magnitude of effect (LOEC)

Reliability test conditions stated QA regime e.g. GLP dose-response, taking account of limit of solubility measured exposures NOECS are bounded (I.e. there is an effect conc)

Data not generated to standard guidelines are acceptable

How to use field and mesocosm data in deriving thresholds? Sometimes, we have substantial quantities of field data or data from

mesocosm studies

But …

can’t always eliminate other stressors

goals of study may not always be consistent with those of standard (e.g. ‘soil fertility’ vs ‘protection of ecoreceptors’)

Use as critical data to derive a standard or to corroborate one based on lab data (i.e. adjust AF)?

Support for use as ‘driving’ data as long as study goals are consistent with those identified at Problem Formulation and other quality criteria are met

What level of protection?

Threshold can protect against ‘no effects’ or

A more conservative than B

Screening values might be type A and mandatory standards more like type B

‘Warning’ and ‘Action’ limits

May be useful as a way of setting bounds within which economic or policy factors can operate

‘Horses for courses’

Soil use (decreasing sensitivity)

NATURERESERVE

INDUSTRIALSITE

RESIDENTIALWITH GARDEN

Protectionlevel

Minimumprotectionlevel

• Selection of data: magnitude of effecte.g. EC10 vs EC50

• Extrapolation: assessment factor or percentileat risk (e.g. HC5 vs HC10)

• Burden of proof before we will take action

Extrapolation methods - workshop view on reliability

Reliability Derivation methodModel ecosystem data with small assessmentfactorSSDs based on chronic NOECs (or ECx) withminimum data set or greater, plus small or noassessment factor

High

Medium assessment factor applied to lowestchronic NOEC from data set of 5 or morespeciesSSDs based on acute data, with largeassessment factor

Medium

Lowest acute LC(EC)50 data for 5 or morespecies, with large assessment factor

Low Small acute dataset with very large assessmentfactor;Small chronic dataset with large assessmentfactor

Dealing with background Principle of allowing for background accepted for naturally

occurring metals and some organics e.g. PAHs Assumes adaptation to background and hence need to manage

only the anthropogenic fraction ‘Added Risk’ currently the only feasible approach:

Threshold = Background + Maximum Permissible Addition* Can we reliably estimate a background?

should it include anthropogenic inputs (mining from 2000 years ago)?

distribution of backgrounds may have large variance - where to set the background?

What scale? Site-specific? Geotype? National?

* PNEC derived from ecotoxicity testing

Dealing with background

F

F

Background conc

express as ‘total risk’

? express as ‘added risk’

threshold = background?

The ‘Added Risk’ approach

Env conc < threshold? NFW

Determine background

NFWEnv conc < threshold + background?

Take action

Y

Y

N

N

(Bio)availability

Metals can exist in different chemical forms - largely influenced by prevailing environmental conditions

Only a small proportion of total metal may be in a form that can be taken up or exert biological effects

Availability can now be predicted for a number of metals (e.g. ‘WHAM’, BLMs)

Accounting for speciation and availability can remove much of the scatter in conc-effect relationships

[Total] [Dissolved] [Speciation-based]

implementation costs increasing

risk of false +/- increasing

Implementing a standard

numeric value - only one part of a standard, especially if measurement required to determine pass/fail

design risk - how often is it acceptable to fail? e.g. “1 in 20 years”

period of time over which this statistic applies e.g. a year

How often the limit may be exceeded (e.g. 5% of the time) - express standard as mean or percentile

statistical confidence with which failure must be demonstrated before action taken (“burden of proof”)

“Burden of proof”

threshold

Concentration (or dose)

F

Do we give benefit of doubt to the environment … or polluter … or face value?

If we give benefit of doubt to the ‘polluter’ then we require a higher level of confidence before taking action - effectively raising the standard (or increase sampling frequency)

Depends on seriousness of failure?

Standards - social and economic aspects

Costs of monitoring (regulators, industry) of compliance e.g. limiting emissions so that

the standard can be met

Socio-economic analysis when significant risk of failure, investment implications, risks to certain sectors

Openness and consultation are now important ways of working Regulators are required to address costs Social and economic aspects of standards are dealt with through

Regulatory Impact Assessments derogations because of disproportionate cost non-implementation of standards

MCDA - options appraisal Multi Criteria Decision Analysis is a technique for ‘balancing’

conflicting risks Formal approach that involves identifying criteria against which

we will make a decision, measuring preferences and finding an option that provides the best overall balance for a standard

Recently used to assess options for sheep dip chemicals, taking account of concerns about environmental protection, animal welfare, farmer livelihoods etc

Can include a ‘do nothing’ option Robust scientific analysis is a key element - but other elements

will also influence the standard Participative and transparent - opportunity to involve stakeholders

Key points

Stepwise process with clear roles for policy, science and stakeholders

Technical groups largely ‘consolidated’ conventional practice Methods emerging for dealing with backgrounds and

bioavailability - ‘research to regulation’ Flexibility recognised in standards for different purposes - in

the way standards are set and the way they are used 5 requirements of an ‘ideal’ standard Recognise socio-economic realities - robust scientific analysis

could be just one of a number of inputs to standard setting