CONTRACT RESEARCH REPORT 295/2000

Policy, risk and science:Securing and using scientific advice

Prepared by OXERA(Oxford Economic Research Associates Ltd)

CONTRACT RESEARCH REPORT

295/2000

Policy, risk and science:Securing and using scientific advice

OXERA(Oxford Economic Research Associates Ltd)

Blue Boar CourtAlfred Street

OxfordOX1 4EH

The primary aim of this study is to offer recommendations for improving the quality of scientific advicereceived by government and used in policy development. Scientific methods are well established, andmany scientific theories are highly successful, and can therefore inform analysis leading to policydecisions. The area of most interest in this study, however, lies outside the secure region in whichscientific understanding is complete and undisputed. It is concerned with weak evidence, novel andincompletely known risks, hypotheses and gaps in data.

Information was gathered by means of a literature review, case studies, and informal interviews with awide range of interested parties (including non-governmental organisations (NGOs), scientists, policy-makers and ministers).

The conclusions of this study are a set of recommendations expressed in the form of: principles, whichare fundamental and comprehensive; a model process, which is the clearest way to secure scientificadvice that is fully compatible with the principles; and supplementary notes, which include detailedrecommendations for some aspects of the process.

The suggestions made in this report are radical because they call for significant changes in theoperation of existing mechanisms. If implemented, they should produce advice that is more robustbecause it will have more developed ways of embracing uncertainty, recording rational justification andmanaging bias. This should lead to policy decisions that are better informed and should help to providea system of policy development that achieves a higher level of public confidence and the support of thescientific community.

This report and the work it describes were funded jointly by the Health and Safety Executive; theCabinet Office (Regulatory Impact Unit); the Department of the Environment, Transport and theRegions; the Department of Health; the Environment Agency; the Food Standards Agency; the Officeof Science and Technology; and the Scottish Executive. Its contents, including any opinions and/orconclusions expressed, are those of the authors alone and do not necessarily reflect the policies of thesponsors or the government.

This report is equivalent to Environment Agency R&D Technical Report P414.

© Crown copyright 2000Applications for reproduction should be made in writing to:Copyright Unit, Her Majesty’s Stationery Office,St Clements House, 2-16 Colegate, Norwich NR3 1BQ

First published 2000

ISBN 0 7176 1852 8

All rights reserved. No part of this publication may bereproduced, stored in a retrieval system, or transmittedin any form or by any means (electronic, mechanical,photocopying, recording or otherwise) without the priorwritten permission of the copyright owner.

|O|X|E|R|A| Executive Summary

i

EXECUTIVE SUMMARY

Introduction

This report has been prepared primarily for the eight government bodies (the Health and SafetyExecutive; the Cabinet Office (Regulatory Impact Unit); the Department of the Environment, Transportand the Regions; the Department of Health; the Office of Science and Technology; the EnvironmentAgency; the Ministry of Agriculture, Fisheries and Food and the Food Standards Agency; and theScottish Executive) that co-sponsored this study. It may also be of interest to other governmentdepartments, executive agencies, and non-departmental public bodies that use scientific advice indeveloping policy, and to individuals and organisations that participate in giving scientific advice or areaffected by related policy decisions.

This report is the culmination of work carried out by OXERA (Oxford Economic Research AssociatesLtd) between April 1999 and August 2000. OXERA gratefully acknowledges the valuable contributionsof the Steering Group, OXERA’s internal Advisory Board, and others who participated in the study.OXERA takes sole responsibility for the content of this report, which presents the study’s findings thathave not been determined nor endorsed by the Steering Group or the sponsoring departments.

The study took part in three phases. Evidence was collected in the first phase; in the second phase itwas analysed, and solutions were proposed and tested; and, in the final phase, a consultation washeld on the findings, and the emerging recommendations were refined.

This evidence was collected in:

• eight case studies;• a literature review; and• more than 65 interviews with policy-makers, government Ministers, scientific advisers and

others—some interviews covered particular case studies and others were of general relevance.

The primary aim of this study is to offer recommendations for improving the quality of scientific advicereceived by government and the way that it is used in policy development. The study is concernedwith the way in which government takes decisions that rely in part on scientific advice to characterisethe risks that might result. The government has sought external scientific advice on many of therecent policy issues that have attracted greatest public concern. In addition, it employs manyscientists who provide internal scientific advice. The recommendations in this report could be appliedto both external and internal advisers. This report recognises and draws upon the important work thathas taken place within government to develop guidance on securing and using scientific advice forpolicy, including, for example, Guidelines 2000—Scientific Advice and Policy-Making. It is not the roleof this report to offer formal guidance, but to recommend good practice and model processes forconsideration in the further development of official guidelines in this area, such as the forthcomingCode of Practice for Scientific Advisory Committees.

The report addresses all stages in the resolution of policy questions in which science plays a part:

• identifying problems for which scientific advice is needed;• seeking and obtaining scientific advice; and• building that advice into policy.

The recommendations in this report have been designed to be widely applicable and to allowsufficient flexibility to accommodate a wide range of types of problem and solution. The reportcontains recommendations in the form of:

• principles, which are intended to be fundamental and comprehensive, and pervade theserecommendations. These principles form an internally consistent set; it is suggested that anyprocess for obtaining scientific advice should be consistent with all of the principles;

• a model process, which is a way to secure scientific advice that is fully compatible with theprinciples. It is not presented as the only acceptable process, but it is recommended thatwhatever process is used, it should be compatible with the principles.


ii

• supplementary notes, which discuss in more detail some of the themes that emerged duringthe study, and include detailed recommendations for some aspects of the process, notably goodpractice in taking decisions under risk and uncertainty, in securing scientific advice inemergencies, and in selecting advisers.

These three components constitute a possible foundation for a system for securing and usingscientific advice. Upon this foundation, more detailed structures and procedures could be built. In thecourse of the study, no reason was found to discard the existing advisory mechanisms, but manyimportant themes emerged that would have a strong influence on their effectiveness. These themesare described below. In addition, within the main text, the subtleties and complexities of the practicalimplementation of the suggestions are discussed and as far as possible resolved. These subtletiesare extremely important to the successful operation of the system. The foundations are broadlycompatible with the wide range of advisory structures currently operating within government, but ifimplemented, they would require substantial changes to current practice.

Four functions of individuals can be distinguished in the scientific advisory process:

• decision-taker—a person with the authority to take a policy decision. This may be agovernment Minister, or a person or body with the delegated authority to take a decision in thename of a Minister;

• policy-maker—a person or organisation charged with assisting a decision-taker in reaching adecision by providing policy analysis, generating policy options, or by conducting riskassessment (policy has been interpreted to include regulation);

• scientific adviser—a person or organisation responsible for providing scientific input to policy-making or decision-taking. This includes both scientists expert in narrow disciplines relevant tothe problem in question, and more broadly-based scientists able to integrate several disciplines,and those within and outside the civil service;

• stakeholder representative—a person or organisation representing the interests and opinionsof a group with an interest in the outcome of a particular policy decision.

EMERGENT THEMES

Rights of those at risk

The study observes that some stakeholders have been well represented in the policy-makingprocesses but that in some cases other stakeholders have not been so effectively represented. Therecommendations in this report to improve access for all stakeholders to the policy-making processalign with, and support, the current movement to broaden stakeholder inclusion. However, ifstakeholders are asked to produce the scientific advice itself, then the advice could be biased—without this being apparent.

The report recommends the full inclusion of stakeholders in other parts of the process, notably:problem identification; defining policy issues and possible options; framing of scientific questions;selecting the advisory mechanism and adviser(s); and assessing the findings. It is vital thatstakeholders are encouraged to participate in these stages of the process if the overall success of theprocess is to improve. In addition, all stages should be transparent, to avoid stakeholders beingexcluded through lack of information.

This transparency demands a presumption that the scientific advice will be published. However, itdoes not demand the publication of incomplete scientific analysis. It will usually be necessary toprepare a full draft of the scientific advice before it is published, because it could otherwise bemisleading. The presumption should always be that the scientific advice will ultimately be published.

Needs of decision-takers

On most policy issues, scientific advice will be one of many other inputs to policy-making, such as, forexample, advice on economic impacts, and equity considerations. The decision-taker will wish to takeaccount of all of these factors. If the recommendations in this report are adopted, the decision-taker


iii

will be obliged, and assisted, to consider the scientific and other uncertainties, and their potentialimplications.

The questions put to scientific advisers will often require them to consider the risks involved if differentpolicy options were to be followed, and to communicate any additional options that they identify.Decision-takers should be presented with a choice of policy options, each of which is accompanied byan analysis of its risks, benefits, and uncertainties, and the scientific and other inputs which underliethis analysis. They should be able to rely on policy-makers to gather and synthesise these inputs tothe analysis of policy options.

A rational explanation for the decision could be given, and this could include an explanation of whythe options that were not chosen were rejected. This offers decision-takers a basis for responding tocritics, and an easier way of revisiting decisions in the light of new scientific information.

Behaviour of the participants in the advisory process

It is concluded in the study that the functions of scientific adviser, policy-maker, stakeholder anddecision-taker should be clearly distinguished, so that the conflicting interests of different stakeholderscan be expressed and addressed in their proper place, and the biases of scientific advisers can bemanaged. If one individual (or group) is carrying out more than one of these functions, they should actin only one capacity at any given time. By introducing a clear structure into the relationship betweenthese functions, a more transparent, reliable and trustworthy process should emerge.

There is a conflict between independence and expertise. It may not be possible to acquire expertiserelevant to a policy issue without also acquiring interests in the issue. The independent scientist is anideal: individuals tend to have biases and personal motives. In practice it is better to identify andmanage the biases of expert advisers, than to require their total independence.

Competence as a scientific adviser means not only proficiency in the relevant scientific disciplines, butalso familiarity with the range of views of others within the field.

The responsibility to be placed upon scientific advisers is clear. They should be asked to act purely intheir capacity as professional scientists, excluding all partisan interests. When in doubt about wheretheir primary duty lies, they should always remember that when giving advice to government, theirduty is to the public interest—and that this duty over-rides any personal or professional interests. Thisprovision would empower and require advisers to resist external pressures.

As well as duties, scientific advisers could also be given rights. Scientific advisers could be givenrights to adequate support from a secretariat, independence from that secretariat, and someprotection from civil liability, so long as they conduct themselves honestly and competently. Thisprovision would also help then to resist external influences.

PRINCIPLES

These principles express an underlying philosophy which appears to be applicable to all scientificadvisory mechanisms, whatever the policy issue may be. In them a collection of ideas is distilled fromthe evidence-gathering phase of this study. The need for a set of principles applicable to scientificadvice emerged from the case studies and initial debates within the study team, when it becameapparent that points of principle were emerging that called for resolution.

The dominant consideration in settling the final form of the principles was the need for a disciplinedframework in which the participants in the scientific advisory processes can operate. Other schemeswere considered, in which there would have been more reliance on subjective judgement, but thesewere judged to carry greater risks of abuse and error. Some of the principles may seem self-evidentor trivial, but in practice all make an important contribution to the definition and implementation of avalid scientific advisory process. Each principle addresses (as far as possible) a distinct issue. Theprinciples are mutually compatible and not overlapping. It is suggested that these principles could beused to guide working practices and could be made available as a reference if the conduct of thescientific advisory process were to be challenged.


iv

The principles were conceived for application to the process of seeking and using advice from expertsin the natural sciences; their relevance to the social sciences has not been explored and remainsuntested.

Scientific advice should be founded upon observation and theory, and should describe boththe scientific conclusions and their uncertainty; these should be deduced from the evidenceby reasoned argument.

Scientists use a core of firmly established information. Outside this core, there is a large body ofvaluable indicative information that is less firmly established, and, beyond it, on the boundaries ofcurrent research, there is significant uncertainty. In these less secure regions there is still much thatcan be said to characterise a range of possible outcomes, by stating the risks and uncertainties. It issuggested that a disciplined approach be taken to the handling of evidence, and that the presentationof this evidence be made by deductive reasoning from what is known and from the incompleteevidence that is available.

The functions of scientific adviser, policy-maker, decision-taker and stakeholderrepresentative should be distinguished.

Scientific advice is but one contribution to a policy decision. In formulating policy, it should be takeninto account, together with economic and social considerations, political constraints, values, and theviews of stakeholders. Scientific advisers are not necessarily also qualified or authorised to makepolicy decisions, nor to provide advice involving judgements of social values; equally, decision-takersare usually not competent to assess scientific issues.

Policy-makers may wish to elicit the views of stakeholders during the policy-making process.However, the contribution that stakeholders can make to the scientific advice does not includenegotiating on behalf of their constituencies as a representative.

Full information about the process of seeking and using scientific advice should be madepublic, and stakeholders should be encouraged to comment. Scientific advice itself should bepublished promptly in a transparent manner, once it is complete.

There are many individuals and groups with a stake in policy, including the public (as the main riskbearers and potential beneficiaries), industry (as providers of benefits and generators of risks), andpressure groups. These stakeholders demand access to the decision-taking process, including thescientific advisory process. Stakeholders have an important part to play in ensuring that the scientificquestions being asked are pertinent to the policy issues that are of concern to them.

Decision-takers may find that it is useful to explain how scientific advice was taken into account in apolicy decision. This would confirm that the scientific advice has been correctly understood, and that itprovides assurance to stakeholders that the issues of concern to them have been addressed. It wouldprovide valuable feedback to the scientific advisers, and, hence, contribute to the continualimprovement of the scientific advisory system.

An audit trail of the scientific advisory process could be made available to policy-makers, decision-takers and, ultimately, to stakeholders. This audit trail might cover the generation of the scientificquestions; the scientific arguments, calculations and analyses; alternative scientific opinions anduncertainties; and the scientific appraisal of policy options.

Care should be taken to avoid a form of ‘openness’ that actually obscures key messages throughobscure terminology or dilution in a large volume of detail. Detail should be available for the expertreviewer, but the aim of this principle is effective communication with the public. This requirestransparent communication of the key findings, including uncertainty.

Scientific advisers should be selected for their competence. It is not necessary for advisers tobe independent of all interests in the policy question; however, the advisory process shouldensure that interests are declared and any resulting biases are balanced or taken into account.

Scientific advisers should be selected for their competence, and their field of expertise should beappropriate to the scientific question. They must possess not only personal proficiency but also


v

awareness of the wider state of knowledge in the field, and an aptitude for communication. Scientificadvisers should show willingness to consider scientific uncertainties and opinions that are at variancewith their own, and to discuss these in their advice. They should be able to analyse the scientificconsequences of a range of policy options, and assess the probability of various outcomes.

Some scientists who have the potential to contribute useful information to the advisory process—owing to their relevant knowledge and experience—may be employed by organisations, such asindustries or pressure groups, that have an interest in the policy issues. Such scientists should not beexcluded from the process, but they should advise as experts rather than negotiators. Scientificadvisory committees may achieve overall freedom from bias by appointing a balanced group ofmembers with different affiliations. This will help to define the range of scientific opinion on a particularissue and hence reveal the existence of uncertainties.

The declaration of interests by scientific advisers is necessary whether they are academics, employedby industry, or employed by pressure groups. Possession of interests should not be presumed tocompromise an individual’s integrity. It may not be possible to apply to scientific advisers the Nolanprinciple that appointees should not ‘place themselves under any financial or other obligation tooutside individuals or organisations that might influence them in the performance of their publicduties’; nevertheless its spirit can be recognised in other ways.

The scientific adviser’s over-riding duty is to the public interest. Advisers should not seek topromote any special interests in preparing their advice.

Scientific advisers may face a conflict of interest because they have other occupations and interestsoutside their role as advisers. This professional experience of experts contributes to their ability toprovide advice. Experts who are employed in industry or academia may possess expertise notavailable within government.

Some of these scientific advisers may feel under pressure to bias their advice, for example, to suit theconstituency that nominated them as advisers, to be consistent with established scientific consensus,to lead to a particular policy outcome, or to suit their employers. Individual advisers should beempowered, and required, to resist such influences.

It is difficult to define the public interest, but it is suggested that it would be sufficient for scientificadvisers to be required not to promote any special interest in preparing their advice. The aims of thisrequirement are to ensure that scientific advisers recognise that they are forbidden from placing aconscious bias on their advice, and to empower advisers to use this requirement to resist anypressure to do so. This requirement addresses only conscious bias, and does not guarantee thatadvisers will be free from unconscious bias, nor remove from policy-makers the duty to balance, orotherwise take into account, all sources of bias in the advice they receive.

Scientific advisers and policy-makers should be candid about the limitations of scientificknowledge and should always assess the uncertainty in scientific advice and the risksassociated with each policy option; this information should be taken into account by decision-takers.

There are limits to the extent of scientific knowledge, particularly concerning new technologies and theassessment of risks. Scientific advice in some areas is therefore characterised by uncertainty anddifferences of scientific opinion. For good decision-taking, it is essential that these uncertainties betaken into account.

Scientific advisers should seek out the sources of uncertainty in their advice, and be candid aboutthese uncertainties and any differences of opinion among their peers. They should be ready to explainhow the uncertainties arise and what assumptions have been made, and to provide an estimate of theoverall uncertainty. In particular, they should avoid the temptation to seek closure on scientific issues,where an error or misjudgement could have adverse consequences.

Scientific advisers are not duty-bound to provide a subjective judgement where the evidence cannotresolve a question, nor should they be (or feel they are) subject to any pressure to do so. Decision-takers need to accept and understand the uncertainty in the advice they receive, and to consider it inmaking their decision.


vi

The effort expended on securing scientific advice should be proportionate to the importanceof the policy issue and the difficulty of the scientific investigations required.

The choice of mechanism for generating scientific advice is likely to involve a trade-off, for examplebetween the cost of generating the advice and the degree of certainty achievable. Advice can be mostcost-effective when the effort expended reflects the potential magnitude of the policy consequencesand the value of reducing uncertainty.

MODEL PROCESS

The model process gives effect to the themes and principles identified earlier. It helps to demonstratethat the principles could be applied in practice and shows how this might be done. It is a benchmarkor ideal process, against which the practicalities of day to day pressures need to be set. The model iscompliant with the principles set out earlier. There may be other models that have equal validity.

The model described in the report is divided into a set of discrete and bounded steps. In practice,these may overlap or even merge, and may not always be clearly distinguishable. It is presented as asequence of steps with some explicit feedback loops. In practice, the steps may be iterated severaltimes.

The model process proposed here is not prescriptive and could be adapted, in which case:

• it is suggested that the adapted process should remain consistent with the principles;• the justification for omitting or adapting parts of the model process, and for the alternative

adopted, could be recorded so that the operation of the process can still be audited; and• it is recommended that efforts should be made to conform to the spirit of the model.

The model process allows functions to be clearly distinguished without compromising the quality ofthe scientific advice or the decisions based on it. It places demands on all the participants torecognise and fulfil their functions. It is suggested that stakeholders should be entitled and invited tocontribute to most stages of the process. Stakeholders have a valuable role to play in challengingboth the scientific advice and the actions of the policy-maker.

In distinguishing the functions, different people would normally perform each function. However, ifthere is a good reason for one individual (or group) to enact more than one function (perhaps becausethe problem is small, or specialised, or requires urgent action), the process requires the person to besure which function they are fulfilling at each point in time, and to act accordingly.

In practice, a written record of the logical argument of the case would provide an audit trail andjustification if subsequently challenged.

The steps in the model process

Detect the issue

The early detection of a policy issue is desirable because it maximises the range of policy optionsavailable and the time available to analyse them. Systematic procedures could be employed tomaximise the chance of detecting an issue promptly. Such procedures are discussed in the report.

Establish the policy context and policy options

Before seeking scientific advice, policy-makers identify the policy issues on which the science bears,the policy options that are available, and the scope of advice that is needed. It is recommended that,as far as possible, policy options should be kept open until scientific advice has been taken.

Scientific considerations may help to identify policy options that were not apparent before. Policy-makers are therefore encouraged in this report to reconsider the range of policies that are available inthe light of emerging scientific evidence.


vii

Define what the policy-maker needs to know

The policy-maker needs to be competent to specify scientific questions. The questions can be refinedthrough discussion between scientific advisers and policy-makers, and may be further refined bychallenge from stakeholders. In this model, the questions put to scientific advisers must be restrictedto scientific issues, and must admit a purely scientific answer, in order to maintain the distinction offunctions.

In order to comply with the principle that the functions of scientific adviser and decision-taker bedisinguished, the policy decision must not be delegated to scientific advisers by posing the question insuch a way that the scientific advice determines the policy option to be adopted.

Choose the scientific advisory mechanism

The report discusses the relative merits of four common mechanisms. The discussion is intended tohelp policy-makers make a reasoned choice of advisory mechanism, which should be proportionate tothe scale, complexity and importance of the policy issue and to the scientific challenge.

Choose the adviser(s)

The selection of scientific advisers is dominated by two drivers:

• the need to obtain a high level of expertise;• the need to identify and manage bias and conflicts of interest.

Taking into account the effect of the principle of over-riding duty to the public interest, the balancing ofcommittees and the valuable input that may not be available elsewhere, advisers affiliated tostakeholders are acceptable, and may be desirable, so that all schools of thought are included.

Where a single expert is used, vested interests or bias can be critically important. Also, the individualmust possess a suitably broad range of knowledge and expertise. However, the members of acommittee do not need to have all these qualities individually—collectively the committee should bebalanced.

Stakeholders could be invited to make recommendations for, or comment on, the choice of advisers.This may help the policy-maker to identify candidate advisers and may help to assess potential bias inthe advice that might be provided.

Agree and confirm the brief

The evidence collected in the study showed that it is essential that the scientific adviser and thepolicy-maker have the same understanding of the questions that are to be addressed. This can betested if scientific advisers replay the questions in their own terms. In order to avoidmisunderstanding, the questions should be stated explicitly and agreed with advisers. In order tofurther clarify the need for advice, the scientific adviser could be told the policy options and given anexplanation of the policy background. The scientific adviser could be asked to evaluate the scientificconsequences and likelihood of a range of outcomes of each policy option that fully represents therisks involved.

The distinction between analysing the scientific issues and making value judgements must be madeclear to the scientific adviser, in order that the adviser can offer advice which is not based on valuejudgements, and does not usurp the legitimate function of the decision-taker. At the same time, thepolicy-maker could make clear to the adviser any constraints on the way in which they should work,for example:

• the procedures to be followed to maintain an audit trail of the advice;• how uncertainty and risks should be presented;• whether meetings are to be open or closed;• when reports are to be prepared; and• whether any draft scientific advice should be published, or if there is to be any communication

with the press or general public while the work is in progress.


viii

This report makes suggestions about how some of these constraints might be framed by the policy-maker and met by the scientific adviser.

Prepare the scientific advice

The report reaches a firm conclusion that the preparation of scientific advice is not the same as theconduct of scientific research. The former involves the synthesis of known science with estimates ofwhat is unknown or uncertain, directed towards the resolution of a policy issue. The latter is notnecessarily policy-focused and customarily applies very strong tests of evidence before a positivefinding is accepted. Furthermore, scientific advisers are required to appreciate the information needsof policy makers, and to reflect candidly the full scope of scientific views, and not just their ownscientific position. The advisers may recommend, or even conduct, further research in the course ofpreparing the advice, but the two activities are clearly different.

In order to communicate the range of possible scientific outcomes identified in their advice, scientificadvisers will need a systematic procedure for identifying and handling uncertainty. They will need toconsider all types of uncertainty (for example, inadequate data, incomplete theory, conflicting theory,limitations of theory) and will need to characterise their advice accordingly. Where evidence isinconclusive by the conventional standards of scientific research, they should evaluate what risks areimplied by the evidence as it stands. Sensitivity analysis could be used to test how the advice mightneed to be changed if new evidence were to emerge, or if a theory that has been used were found tobe incorrect. A statement could be made of what new evidence, if it were to emerge, would require theadvice to be changed. The study concludes that it is not the function of scientific advisers to apply theprecautionary principle or any hidden safety margin. The application of the precautionary principle isimportant but it should only be applied at the point where a final decision is taken, or there will beexcessive caution.

If dissenting advisers are encouraged to present their views constructively then it will be easier toobserve bias and to avoid overlooking less likely outcomes or outcomes based on unconventionaltheories. Committee chairpersons could be asked to ensure that those views are considered,accurately reported, and given appropriate weight so that the rights of advisers to submit their findingsare protected.

Communicate the advice

The communication of the scientific advice is an important step, and it is important that the full contentof the advice is preserved in the policy process. The report suggests ways in which scientific adviserscan present their advice in such a way that it will be fully understood by the policy-maker anddecision-taker. If asked to present and explain scientific advice to the general public, it has beenobserved that scientific advisers may have to exercise great care in order to restrict their comments tothe science, without discussing the wider policy issues.

Prepare advice on policy options

The policy-maker is responsible for managing the process by which scientific advice informs policy.The policy-maker must therefore be able to make use of advice that carries explicit scientificuncertainty, because most scientific advice will carry some uncertainty.

When presenting policy options to the decision-taker, the policy-maker will need to make clear theimplications and robustness of the scientific analysis if the decision-taker is to be able to take thedecision based on all the evidence available. The report suggests how this might be achieved, bycharacterising the policy options according to:

• expected outcome and possible worst-case outcome;• the degree of reversibility of the option;• sustainability—whether the option can be sustained in the long term;• any precautionary arguments that may be relevant;• reliability of the key assumptions and evidence; and• the source of the advice, and the weight that should be attributed to that source.


ix

Take the decision

The report argues, from constitutional principles and existing guidance to Ministers, that it is thelegitimate function of the decision-taker to take the decision, and not to allow it to be taken by thescientific advisers. The decision-taker has a duty to expressly consider the costs, risks and benefits ofeach policy option.

Since the decision-taker is accountable for the decision, they must be satisfied that the scientificadvice is sound, must understand it, and must be able to explain, and defend, a reasonable decision.The decision-taker could consider stating the advice and explaining the reasoning that led to thedecision: this would have the advantage that a decision can more easily be reversed in the light ofnew evidence; both accountability and general understanding of policy would be improved.

Audit and maintain the process

The study considered the maintenance of scientific advisory mechanisms and concluded that qualityassurance and audit functions have a valuable role to play, in:

• auditing the operation of the scientific advisory systems to ensure that they are compliant withagreed guidelines;

• training those who provide and those who use scientific advice to ensure that they understandand fulfil their duties. All participants in the system, particularly advisory committee chairpersonsand policy-makers, could be trained in techniques for posing questions and operating thevarious advisory mechanisms;

• reviewing, adapting and maintaining the advisory system to ensure that it remains fit forpurpose.

There are four checks on the quality of scientific advice: quality assurance, peer review, audit, andParliamentary scrutiny. Quality assurance could be provided by the policy-maker who would checkthat the scientific advisers are following good practice, consistent with the recommendations in thisstudy. To ensure the system, including the communication of advice, remains fit for purpose, it isrecommended that the departmental chief scientists should arrange for it to be periodically andindependently audited and reviewed. The results of these audits and reviews, and any changes to thesystem, could be reported to the department’s management board and to OST.

The study found strong support for the production by Ministers of a ‘reasoned opinion’ that explainsthe links between the policy options and evidence collected in the policy-making and advisoryprocess, as described above.

RIGHTS AND DUTIES

The report argues that scientific advisers should be given clearly defined rights and duties that reflectthe principles. The key rights are:

• a brief which sets a question that is amenable to a scientific answer and that does not demandbroader policy questions to be addressed by the adviser;

• sufficient resources;• protection as far as possible from any civil action arising from their work as advisers, whether

brought by a third party or by the government department which appointed them; and• to record a personal statement which sets in context their declarations of interest, and their

understanding and acceptance of their duties.

Scientific advisers would have duties to:

• disclose any interest which might be seen to have the potential to distort their judgement; and• keep confidential any aspects of the subject they are advising on, if so required by their brief,

including the fact that they are advising.


x

Chairpersons of advisory committees could have further duties, to:

• ensure that every member of the committee is heard and that no view is ignored or overlooked;• refrain from forcing closure where the evidence is compatible with a range of possibilities; and• be responsible for ensuring that the committee acts in accordance with available guidelines.

CONCLUSIONS

The investigations described in this report covered a wide range of situations in which governmentmight use scientific advice on issues involving risks to the public interest. In doing so, it identified acomplex set of interrelated issues that bear on the effectiveness of the scientific advisory system, bothas it exists today and as it might be developed.

The findings of the study include the observations that the existing structures within which scientificadvice is provided to government appear well suited to their function, but that the shortcomings of theadvisory system in the recent past appear to have been related to the way in which the processeshave been conducted. For this reason, a set of universal principles was developed, to guideindividuals and organisations in the course of scientific advisory work and related policy making.

To demonstrate the practicality of these principles, a model process was worked out in greater detail.This is suitable for implementation directly or may be adapted for specific cases.

Several pervasive themes emerged from this work, reflecting issues raised by politicians,stakeholders, civil servants and professional scientists during recent public debates on this subject.These themes address: the broadening of stakeholder inclusion; the proper structuring of policydecisions in the face of risks; and the behavioural code that should guide the actions of individualsengaged in the scientific advisory processes.

The study team is convinced that improvements to the scientific advisory system are achievable andthat these would secure better scientific advice and more soundly-based policies to manage risks tothe public. The outcome of the study is a coherent set of recommendations that give substance to thisconviction.

|O|X|E|R|A| Contents

xi

CONTENTS

INTRODUCTION 1

ANALYSIS OF THE EVIDENCE 5

PRINCIPLES 13

MODEL PROCESS 19

SUPPLEMENTARY NOTES 35

Rights and Duties of Scientific Advisers 35

Legal Issues 37

Decision-taking under Risk and Uncertainty 41

The Selection of Scientific Advisers 44

Scientific Advice in an Emergency 47

LITERATURE REVIEW 49

EXISTING GUIDANCE 59

CASE STUDIES 72

E. coli O157 Food Poisoning in Scotland 73

Margins Around Field Trials of Genetically Modified Crops 79

The Structural Integrity of the Forth Rail Bridge 83

Phthalates in Soft Plastic Toys 86

Total Allowable Catches of Sea Fish 89

The Decision to Construct the Thames Barrier 93

Assessment of Safety Cases for Radioactive Waste Disposal 97

Advisory Committees on Toxic Chemicals 101

APPENDICES 109

Case Study Protocol 109

References 112

Acknowledgements 117

Tender Specification 118

|O|X|E|R|A| Abbreviations

xii

Abbreviations

ACMSF Advisory Committee on Microbiological Safety of FoodsACNFP Advisory Committee on Non-food ProductsACP Advisory Committee on PesticidesACRE Advisory Committee on Releases to the EnvironmentACTS Advisory Committee on Toxic SubstancesBSE bovine spongiform encephalopathyCBI Confederation of British IndustryCEFAS Centre for Environment, Fisheries and Aquaculture ScienceCEP Committee on Expert PanelsCHIP Chemicals (Hazard Information and Packaging for Supply) Regulations 1994CMO Chief Medical OfficerCOSHH Control of Substances Hazardous to Health Regulations 1994COT Committee on Toxicity of Chemicals in Food, Consumer Products, and the

EnvironmentCSA Chief Scientific AdviserCSTA Council of Science and Technology AdvisersDEHP di(2-ethylhexyl) phthalateDETR Department of the Environment, Transport and the RegionsDINP di-isononyl phthalateDTI Department of Trade and IndustryEA Environment AgencyEC European CommissionEU European UnionFAC Food Advisory CommitteeFACA Federal Advisory Committees Act 1972FAI fatal accident inquiryFEPA Food and Environment Protection Act 1985FSA Food Standards AgencyGLC Greater London CouncilGM genetically modifiedHACCP Hazard Analysis and Critical Control PointHMIP Her Majesty’s Inspectorate of PollutionHSC Health and Safety CommissionHSE Health & Safety ExecutiveICES International Council for the Exploration of the SeasILGRA Inter-Departmental Liaison Group on Risk AssessmentLGC Laboratory of the Government ChemistMAFF Ministry of Agriculture, Fisheries and FoodsMBAL minimum biological limitNEPI National Environmental Policy InstituteNGO non-governmental organisationOCT outbreak control teamOECD Organisation for Economic Cooperation and DevelopmentOES occupational exposure standardthe OST Office of Science and TechnologyPDF probability density functionPLA Port of London AuthorityPSD Pesticides Safety DirectorateSCF Scientific Committee on FoodSCHIP Standing Committee on Hazard Information and PackagingSCIMAC Supply Chain Initiative on Modified Agricultural CropsSCTEE Scientific Committee on Toxicity, Ecotoxicity and EnvironmentSSB spawning stock biomassTAC total allowable catchTDI tolerable daily intakeTUC the Trades Union CongressWATCH Working Group for the Assessment of Toxic Chemicals

|O|X|E|R|A| Introduction

1

INTRODUCTION

Terms of reference

1 In late 1998, eight government bodies1

co-sponsored this study, with the objective ofcollecting evidence, based on wide researchand consultation, and identifying good practicein securing and using expert scientific advice.This report is addressed primarily to the eightco-sponsoring government bodies. It may alsobe of interest to other governmentdepartments, executive agencies, and non-departmental public bodies (referred tocollectively in this report as governmentbodies) that use scientific advice in developingpolicy, and to individuals and organisationsthat participate in giving scientific advice or areaffected by related policy decisions.

2 The primary aim of this study is to offerrecommendations for improving the quality ofscientific advice received by government andused in policy development. A consequence ofimproving scientific advice and subsequentpolicy decisions should be increasedconfidence of the public in the scientificadvisory system. The recommendations in thisreport could be applied to external advisersand to internal advisers.

3 This report is the final result of workcarried out by OXERA (Oxford EconomicResearch Associates Ltd) between April 1999and August 2000. The work was coordinatedby a Steering Group, comprisingrepresentatives of the funding departments,and chaired by the Chief Scientist of the HSE.In conducting this study, OXERA has beenassisted by an Advisory Board, appointed toreview the ideas developed during the project,and to advise on the conduct of the study andthe presentation of its results. The AdvisoryBoard members were: Sir Bernard Crossland,Dr George Greener, Dr Jerry Ravetz,Professor Gordon Stewart, Baroness Wilcox ofPlymouth, and Dr Tom Wilkie, and the Boardwas chaired by Dr Dieter Helm, Director ofOXERA.

4 The essential characteristic of thesciences is a rigorous procedure (the ‘scientificmethod’) for testing hypothetical explanationsfor the behaviour of natural or man-madesystems, and predictive models. Scientificmethods are well established, and many

1HSE; the Cabinet Office (Regulatory Impact Unit);DETR; the Department of Health; the OST; the EA;MAFF and the FSA; and the Scottish Executive.

scientific theories are highly successful, andcan therefore inform analysis leading to policydecisions.

5 The area of most interest in this study,however, lies outside the secure region inwhich scientific understanding is complete andundisputed. It is concerned with weakevidence, novel and incompletely known risks,hypotheses, and gaps in data. In this area,there is great scientific uncertainty. In many ofthese situations, scientists will only be able toprovide incomplete information, perhaps astructured judgement or a subjective opinionwithout full supporting evidence.

6 The Interdepartmental Liaison Group onRisk Assessment (ILGRA, 1998a), quotedbelow, recommended that this study beundertaken. The full terms of reference of thestudy are listed at the end of the report.

The time is therefore ripe to reassessthe role of experts in the process ofinforming and adopting decisions witha view to:

• opening up to public scrutiny andpeer review the scientific adviceelicited from experts, and beingclear where scientific judgement oropinion has been applied to convertinformation and expertise intointelligence about risk problems;

• exposing and explaining theassumptions made, and theuncertainties that pervade theassessment of risks and theeffectiveness of possible riskmanagement options;

• adopting appropriate procedures toenable stakeholders and experts tocontribute throughout the process offraming the issue, assessing therisks, identifying risk managementoptions, adopting decisions,implementing the decisions andevaluating the effectiveness of theaction taken;


2

• explaining how expert advice,together with relevant sociological,economic, ethical and politicalconsiderations, contributed to thedecisions made.

To facilitate this work, HSE is currentlysetting up interdepartmental researchto:

• identify and categorise currentpractices within Government foreliciting expert advice; and

• draw up principles of good practicefor the engagement of experts, theelicitation of their advice, and for theincorporation of the advice indecision-making.

7 The Modernising Government WhitePaper (HMSO 1999) calls for the use ofevidence-based policy-making. Other recentwork that is relevant includes the House ofLords report on ‘Science and Society’, theHouse of Commons Science and TechnologySelect Committee investigation of the scientificadvisory system, the work of ILGRA, theguidelines on scientific advice and policy-making produced by OST, and the recent draftcode of conduct for advisory committees.

8 The full tender specification for this studyis reproduced below (page 118).

9 OXERA is solely responsible for thecontent of this report.

Context

10 Government is responsible fordeveloping and implementing policies forcombating natural hazards and regulatingactivities and technologies that present apotential risk to the public or the environment.To do this it must have access to scientificadvice and it must know how to use thatadvice effectively in taking policy decisions.

11 The scientific advisory system has beenscrutinised in recent vigorous public debates,which have mainly focused on public health,the environment and engineering reliabilitystandards.

12 The tenor of some of these recent publicdebates has been adversarial. A number ofexplanations can be proposed for the criticismthat the scientific advisory system andassociated policy decisions have received.These illustrate the scale and nature of theproblem facing policy-makers in this area.

These possible explanations are describedbelow.

13 It could be that there have been a seriesof particularly difficult problems that haveattracted public scrutiny, for which it has beendifficult to formulate policy responses. Thepolicy responses may have been difficult tomake because the risks are not knownprecisely, and the consequences of anincorrect decision are high, either in costs foran industry, or in public health costs.

14 It could be that scientists have not paidsufficient attention to risk and uncertainty inpreparing their advice, or that uncertainty hasnot been effectively communicated to policy-makers, or that policy-makers have ignoreduncertainty in scientific advice. Alternatively, itcould be that policy-makers have rejecteduncertainties as rendering advice useless, andso demanded over-confidence from theiradvisers.

15 It could be that science has been treatedas too difficult for stakeholders to comprehend.They may not have been consulted sufficientlyor some stakeholders may have been givengreater access to the scientific advisorysystems than others.

16 It could be that policy decisions havebeen taken first, and scientific advice thenobtained to justify them. It could also be thatpolicy-makers have been able to manipulatescientific advice by choosing scientists whosupport the government view. They might alsohave been able to restrict the remit of advisersor to capture stakeholders onto committees,and subsume their views into a consensus.

17 These possible explanations representcharacterisations of the origins of publicmistrust, encountered during the study.

Scope

18 This report is intended to apply todecisions that:

• are the responsibility of government, andmust therefore be taken by a Minister, orby a civil servant or appointed body withlegitimate delegated authority to act in aMinister’s name;

• are in part dependent on science, whichis uncertain either because the relevantscientific theory is incomplete orcontroversial, or because the data areinadequate; and


3

• involve potential risks to the public or theenvironment, and therefore call for abalancing of costs, benefits and risks.

19 The report address all stages in theresolution of policy questions in which scienceplays a part:

• identifying problems for which scientificadvice is needed;

• seeking and obtaining scientific advice;and

• building that advice into policy.

20 The UK government has moved towardsa broader definition of scientific advicerecently, illustrated in the May guidelines(Guidelines 2000—Scientific Advice and PolicyMaking (Office of Science and Technology,2000)), which use the definition used by theOffice for National Statistics for itsGovernment R&D. This definition covers thefollowing disciplines: medicine, dentistry,engineering, technology, agriculture, fisheries,forestry, veterinary science; biological,environmental, mathematical and physicalsciences; psychology, geography, economicsand social studies; and humanities. Therecommendations of this study apply to theprocess of seeking and using advice fromexperts in the natural sciences. Theirapplication to other disciplines (for example,the social sciences) has not been tested.

21 This report does not address thegovernment’s strategy for science andtechnology development or funding forresearch, except where it is influenced by theneed to generate advice.

Methodology

22 The first phase of the study wasdedicated to information gathering by meansof a literature search, case studies, andinformal interviews with a wide range ofinterested parties (including NGOs, scientists,policy-makers, and Ministers). The objectivewas to gather diverse examples and viewsprior to the process of synthesis that tookplace in the second phase. This is described inthe ‘Analysis’ section below, and in theevidence sections towards the end of thereport. The study team is grateful to all thoseindividuals, who gave up their time to speak tothe study team (see ‘Acknowledgements’).

23 The evidence gathering influenced thesubsequent work in two ways:

• informally, by exposing the study team toa range of scientific advisory situationsand viewpoints; and

• formally, by providing material fromwhich general principles could beextracted, and which could be used totest emerging recommendations.

24 The ideas in this report were developedsystematically, by extracting observations fromthe evidence, identifying general principles,developing detailed guidance and processesconsistent with the principles, and, finally, bytesting these ideas in progressively widerconsultations.

25 The Steering Group provided valuableguidance as the study progressed, by drawingthe project team’s attention to significantissues that have arisen within government inthe recent past; by facilitating the selection ofcase studies; and by arranging access to theindividuals who were interviewed during theproject.

26 The Advisory Board met three timesduring the project. It commented on theplanned approach to the study at the start ofthe project, and reviewed the evidencegathered during Phase 1. It also commentedon, and discussed in detail, an early draft ofthis report.

27 The recommendations that emergedwere extensively tested by the study team toensure that all were necessary, and that,collectively, they were sufficient to meet thechallenges raised by the evidence collected.Further review ensured that they wereinternally consistent, and that eachrecommendation was addressed to theappropriate participant in the scientificadvisory process, and at the appropriate stagein the process.

Conclusions

28 The conclusions of this study are a set ofrecommendations expressed in the form of:

• the principles, which are fundamentaland comprehensive, and pervade theserecommendations. These principles forma set; any process for obtaining scientificadvice should be consistent with all ofthe principles.

• a model process, which is the mostappropriate way to secure scientificadvice that is fully compatible with theprinciples. However, this does not


4

exclude the use of other processes thatare compatible with the principles.

• supplementary notes, which discuss inmore detail some of the themes thathave emerged during this study, andinclude detailed recommendations forsome aspects of the process.

29 Although no new advisory mechanismshave been proposed (despite attempts toidentify them), the suggestions made in thisreport are radical because they call for

significant changes in the internal operation ofexisting mechanisms. If implemented, theyshould produce advice that is more robustbecause it will embrace uncertainty, willcontain rational justification, and will managebias. This should lead to policy decisions thatare better informed and justifiable. Theserecommendations should help to provide asystem of policy development deserving ofpublic confidence and the support of thescientific community.

|O|X|E|R|A| Analysis

5

ANALYSIS OF THE EVIDENCE30 This study took part in three phases.Evidence was collected in the first phase; inthe second phase it was analysed, andsolutions were proposed and tested; and, inthe final phase, a consultation was held on thefindings, and the emerging recommendationswere refined.

31 The evidence sought included:

• the structure and functioning of thecurrent UK advisory system;

• the structure and functioning of advisorysystems in other countries;

• a sample of the personal experience andviews of advisers, policy-makers,decision-takers, and representatives ofpublic-interest groups and otherstakeholders;

• published reviews;• previous studies;• existing guidance;• the professional practices of civil

servants;• the practical feasibility of different

options; and• cases of the practical application of

scientific advice to policy problems.

32 This evidence was collected in:

• eight case studies;• a literature review; and• more than 65 interviews with policy-

makers, government Ministers, scientificadvisers and others—some relevant toparticular case studies and others ofgeneral relevance.

Observations from the case studies

33 The purpose of the case studies was toprovide evidence against which hypothesesemerging from the rest of the study could betested.

34 The case studies are only concernedwith the way in which scientific advice wassought, provided and used—not with thescientific issues themselves.

35 Recent and ongoing cases that arecurrently subject to review were notconsidered because the participants would nothave been able to discuss the scientificadvisory process in full. Eight cases werestudied in detail, using published reports, andinterviews. Summaries of each case arepresented towards the end of the report andare listed in the table overleaf.

36 The case studies provide examplescovering each of the following attributes ofsituations in which scientific advice is sought:

• uncertainty;• novelty;• criticality;• urgency;• bias;• problem detection;• policy and strategy;• tactical response and regulation;• emergency response; and• information and communication.

The table below illustrates these attributes ofthe eight case studies.


6

Case studies

Unc

erta

inty

Nov

elty

Crit

ical

ity

Urg

ency

Bia

s

Pro

blem

det

ectio

n

Pol

icy

/ str

ateg

y

Tac

tical

/ re

gula

tion

Em

erge

ncy

resp

onse

Com

mun

icat

ion

1. E. coli O157 food poisoning inScotland

x x x

2. Margins around field trials ofgenetically modified crops

x x x x

3. The structural integrity of theForth Rail Bridge

x x x x

4. Phthalates in soft plastic toys x x x x x

5. Total allowable catches of seafish

x x x x x

6. The decision to construct theThames Barrier

x x x

7. Assessment of safety cases forradioactive waste disposal

x x x x

8. Advisory committees on toxicsubstances

x x x

37 Each case study included the process bywhich the policy issue was broken down toyield specific scientific questions; the way inwhich those questions were handled; theselection of advisers; and the way in which thepolicy decision was built up from the scientificanswers, identified uncertainties and otherinputs (see the case study protocol, page109). An attempt was made to probe therobustness of the process by examining whatmight have happened under differentcircumstances.

38 The case studies showed that, evenwhere there was a successful outcome,problems had sometimes emerged because ofthe way in which the scientific advice wassought, used or presented. In each case,aspects of good practice and shortcomingswere apparent.

39 The most striking feature in several ofthe cases was the requirement for theindividual expert or group of experts to take apolicy decision or to propose policy actions,rather than to present scientific facts andscientific opinions. This happened in the

investigation into the E. coli O157 food-poisoning outbreak in Scotland, where theterms of reference included ‘to advise theSecretary of State for Scotland on theimplications for food safety and the generallessons to be learned’ [547–548].

40 The study of the setting of marginsaround GM crop trials illustrated many of thedifficulties. The margins adopted weredetermined by a group from the agricultureindustry [589–596]. That group usedcommercial arguments about crop purity,rather than any wider considerations ofpotential environmental damage.

41 The definition of the margins was widelymisunderstood, with many people apparentlybelieving that the experts had concluded thatthere would be no contamination outside themargin. That was obviously absurd—wind andbees can carry pollen many kilometres—andso the process received little publicendorsement.

42 The margins set by the industry wereimplicit in the considerations by the Advisory


7

Committee on Releases to the Environment(ACRE) of the acceptability of individual trials.ACRE was aware that there was no explicitbalancing of the potential harm that a GMrelease might cause with the potential benefitsthat might result. In the absence of such anexplicit consideration, the members of ACREperceived that they were responsible for therisk assessment and, hence, for making valuejudgements about balancing risks and benefits[587–590, 605].

43 This case study supports the provision ofa model process, the principles and properoperation of which can be checked andaudited. The case study showed severalpitfalls of a weak process—a poor publicperception of the result, and uncertainty in howpolicy should be developed when a review isneeded.

44 In many cases the terms of referencegiven to scientific advisers have demandedthat value judgements be made by thescientific advisers (for example, about thesocial acceptability of risks). It is common inquestions of toxicology to ask for a committeeto deliberate on guideline levels that are safe,in the manner that advisers were asked ‘whatlevel of exposure to phthalates is safe?’ [640,642]. This shows that asking scientific advisers(partly) non-scientific questions is common. Atleast in some cases, this has caused problemsfor advisers, decision-takers and the public:advisers are unsure of their remit; decision-takers are unsure what weight to place onadvice they receive; and the public cannot seethe criteria by which policy options have beenchosen.

45 Another example is the recent Stewartreport on mobile phones and health(Independent Expert Group on Mobile Phones,2000). In response to its terms of reference, itoffered value judgements about theacceptable balance between risk and benefit,and made recommendations for new policiesthat went beyond scientific advice.

46 Across the case studies, the terms ofreference were clearly important both for theadvisers’ understanding of their task, and theusefulness of the advice. When settingfisheries quotas, the advisers were askedsimply to provide figures for the total allowablecatch for the following year. Because of therestricted nature of the advice, policy-makersdid not receive scientific analysis of the trade-off between the consequences for the fishstock and other criteria, such as the variabilityof quotas from year to year. In this case, thescientific question did not reflect the policyconsiderations important to the decision-taker.

47 The policy constraints imposed onadvisers or policy-makers by pre-existingdecisions can limit the terms of reference. Inthe model process presented below [142–213], the terms of reference are iteratedbetween the policy-maker and adviser tosafeguard against misunderstanding and over-restrictive questions [190–196]. It is theprerogative of the policy-maker to define theterms of reference, and to constrain themwithin the bounds of available policy options,with the proviso that the scientific adviser mayask for them to be extended. The terms ofreference could be published so thatstakeholders have an opportunity to commenton whether they are appropriate [167].

48 It is quite common for stakeholderrepresentatives or other non-scientists to beincluded on expert committees to help in theanalysis of these policy questions, as is thecase with the Committee on Toxicity ofChemicals in Food, Consumer Products, andthe Environment (COT) [777–778, 856]. Theysit on these committees as representatives ofthe public interest, or to ensure that adviceproduced by the committee is intelligible to alay person. A minority representation on acommittee does not guarantee that the viewsof the representative will be givenproportionate weight in the advice, and itmeans that the scientific advice will be mixedwith representative views [511]. This studyconcludes that stakeholder interaction is moreeffective if the avenues for interaction with thepolicy-maker are built into the advisoryprocess outside the scientific advisorycommittee itself [151, 157–159].

49 In other situations, committee membersact in a scientific capacity, but have an interestin the issue at hand. For example, themembers of the Pennington Group were not allindependent of the bodies with ongoingresponsibility for managing the outbreak offood poisoning, yet the Group was asked toadvise on the lessons that could be learnt fromthe outbreak [549].

50 The principles and process suggested inthis study do not exclude individuals withinterests related to the case from acting asadvisers. Instead, they apply the commonpractice, followed by many advisory bodies,including COT, that advisers declare theirinterests [782–784]. Since the existing codeson declaration of interests are limited to verybrief rules for public appointments, and do notstate what type of interests should bedeclared, this study presents somesuggestions for the interests that should bedeclared [242].


8

51 Advisers have not always been chosenfor their expertise in the subject; Sir HermannBondi did not have particular expertise in thescientific disciplines bearing on the design of aThames Barrier [706]. There is a role forgeneral scientists to act as advisers whenproblems are unusual, innovative and inter-disciplinary [306–307]. However, in theThames Barrier case, the adviser’s role wasnot purely scientific, but included theevaluation of the costs and benefits of differentoptions. Sir Hermann Bondi was not expert inthese non-scientific issues. While his reportcame to a firm conclusion on a course ofaction that should be taken, it did not providetransparent analysis of the costs and benefitsof all the other possible options.

52 The Thames Barrier issue had a longhistory of debate, during which time somepolicy options had been closed off by otherdecisions that were taken without regard totheir impact on this issue [713]. If a record ofthe early advice on this issue had been mademore widely available, the decision toforeclose options for the Thames Barrier mighthave been more informed. This case supportsthe publication of scientific advice.

53 Any process for obtaining scientificadvice must be able to respond rapidly inemergency situations. In two cases, ongoingfunding was being made available to externalbodies to maintain expertise so that advicecould be provided on demand. These bodieswere the Laboratory of the GovernmentChemist (LGC), in the case of phthalates intoys [645], and the Centre for Environment,Fisheries and Aquaculture Science, in thecase of fisheries quotas [670]. A briefsupplementary note of guidance onemergency advice is included below [331–350].

54 The form of presentation of the advicewas a determinant of its usefulness. Scientistsadvising on fisheries quotas presented theuncertainty in their advice as probabilities andscenarios, which a parliamentary committeefound difficult to comprehend [680]. Theexperts in the nuclear waste disposal casepresented their estimates as a probabilitydistribution function for the parameter theywere estimating [749–751]. The scientificadvice must be intelligible to the policy-makerin order to fulfil its primary purpose insupporting policy decisions [208]. However,the advice may be technical in nature, and itscomplexity may make the details difficult torelate to a lay audience. Thus it is not alwaysappropriate to expect that advisers shouldpresent their advice with a lay audience inmind. Rather, the advice should be

communicated with its primary audience inmind—the policy-maker who commissionedthe advice [208]. Policy-makers could receivetraining, to enable them to analyse scientificinformation and advice; advisers could receivetraining in communication.

55 Finally, it was apparent that differentadvisers faced with the same problem couldcome to different conclusions. This was notedin the phthalates case [649], and in acomparison of the findings of the PenningtonGroup and the Advisory Committee onMicrobiological Safety of Foods (ACMSF)[560]. It shows that scientific advice is not justan objective exercise in gathering evidence,but that advisers will make judgements as partof their advice. In order to make clear to thepolicy-maker where the advice depends on thejudgement of the advisers, the studyrecommends that advisers distinguish thecomponents of their advice that rest onevidence, analysis, judgements and opinion[200].

Lessons from existing guidance

56 In the UK, the Office of Science andTechnology (OST) has published a set ofguidelines for ‘The Use of Scientific Advice inPolicy Making’—the ‘May guidelines’ (OST,1997; the guidelines have recently beenupdates as OST, 2000b). These guidelinescover the selection of advisers, identification ofissues, the publication of advice and theexplanation of decisions. The guidelines arenot prescriptive, and leave considerablediscretion to departments. This studyrecommends much more detailed guidance,and introduces issues not covered in theabove, much shorter, guidelines. Theconclusions of this study are largely consistentwith the May guidelines, except on one veryimportant point. Scientific advisory committeeshave often been asked to take account ofconsiderations wider than the purely scientific.The May guidelines address this byrecommending that non-scientists serve onthose committees. This report, in contrast,recommends that the advisory system shouldbe restructured to reflect distinct contributionsfrom scientific and other advisers. The reportconcludes that quality, transparency andaccountability are otherwise compromised[104–109].

57 The May guidelines are complementedby the Nolan [444–445] and Peach [446]guidelines on standards in public life andappointments to public bodies. There iscontinued debate on the issue ofappointments. The current guidelines permitwide discretion, which lies with civil servants in


9

the preparation of job descriptions andshortlists of candidates, and with Ministers forthe final appointment. The Nolan and Peachguidelines have a focus on executivefunctions, which is not appropriate for scientificadvisers who do not carry any authority to takedecisions.

58 The Nolan and Peach guidelines weredrawn up to make it clear that public servicewould place demands on the conduct ofindividuals of a quasi-contractual status.These concepts apply equally to scientificadvisers. Scientific advisers already havesome duties, arising from legislation, such asthe disclosure of commercially sensitiveinformation, but the principle that publicappointment carries general publicresponsibility has been established by Nolanand Peach. The responsibilities that arecarried across from the Nolan guidelines intothe conclusions of this study on scientificadvice are that advisers should:

• submit to appropriate scrutiny;• be prepared to place the public interest

above any other interest;• declare private interests.

59 The Peach guidelines do not contain adefinition of conflict of interests, but leave thisto individual departments to determine. In thisstudy, a list of potential areas of interest isoffered, but, again, no formula prescribing thelimit for a conflict of interest has beenproduced [242]. The range of interests andpossible advisory scenarios made a definitionimpractical. Instead it is recommended that,where a substantial conflict of interest isapparent, a decision whether or not to appointthat adviser should be documented and shouldinclude reference to the potential conflict ofinterest.

60 In the USA, the operation of the FederalAdvisory Committees Act 1972 (FACA)provides some guidance, especially onaccountability to federal agencies and toCongress. This study considers the locus ofaccountability. In recognition of the role ofMinisters [249–250, 490–495], accountabilityfor policy decisions rests with Ministers.Currently, accountability for the scientificadvisory systems has been placed withdepartmental chief scientists, while the policy-maker is responsible for the quality of thescientific advice in an individual case. In orderto fulfil their Ministerial responsibilitieseffectively, it is recommended that policydecisions be supported by a ‘reasonedopinion’, containing all the evidence andanalysis that led to the decision (andexplaining why other options were rejected).

61 There is a substantial amount ofguidance on the disclosure of information, onrisk communication with the public, and theconduct of civil servants and Ministers [477–496]. It has been used to set this study into thecontext of the existing civil service frameworkand to determine the appropriate delegation ofauthority. The only part that has not been usedgreatly is the communication of risk to thepublic. This is because the study has focusedon the communication of uncertainty betweenthe scientific adviser and policy-maker andupwards to the decision-taker, and not thepolicy-maker’s communication of risks to thepublic, which is covered by an extensive andseparate literature.

Lessons from the literature review

62 The literature provided material fromauthors and institutions with a wide range ofstandpoints; some with a detailedunderstanding of particular disciplines, such asrisk management; others with experience ofthe UK advisory system or the advisorysystems of other countries. The review probedthe relationship between advice (and advisoryprocesses) and the success of policyoutcomes, through papers on the role ofscience, the management of risk, guidelinesfor advisers and their clients, and the socialscience of advisory systems. There is littlepublished work that directly addresses theinterface between science and policy, althoughthere are some papers that compare advisorysystems in different countries, which wereparticularly useful.

63 History shows that the concerns drivingthe present study are not new. In 1961, tenyears after the first US science advisorycommittee had been set up, PresidentEisenhower warned that public policy mightbecome captured by the technological elite—that policy-makers would become reliant onthe advice of experts and that democracywould be eroded [359].

64 Governments have responded to theincreasing need for scientific advice by settingup scientific advisory systems within their civiladministrations, either in central science andtechnology ministries, or as a network acrossdepartments. The scientific advisers assessthe risks arising from new and existingtechnologies, provide the scientific analysis ofpolicy options for government decisions, andsupport the regulation of industry andresearch.

65 The USA is noted for the extent of publicaccess to its advisory system [371–374].Meetings of advisory committees are usually


10

held in public, and both the minutes andpapers are published. Part of the reason forthis approach is the US culture of access toinformation, and part is the federal structure. Ahigh level of consultation is needed to allow allthe US states to participate in the policyprocess, but the US system goes beyond thisin the level of public access it provides. As aresult, there is a powerful audit trail, and clearaccountability to federal agencies [371, 373].

66 It has been suggested that public accesshas constrained the proceedings in a numberof cases in the USA, and there are severalreasons to expect that a high level of publicaccess might have some disadvantages. First,public access to meetings may cause advisersto moderate their deliberations and to tonedown the content of their discussions, perhapsplacing less emphasis on extreme scenarios.Second, it may expose advisers to pressure tochange their advice. Third, it may make itdifficult for advisers to reject views that arecommonly held, or are held by vocal pressuregroups, even if they are not well supported byscientific evidence. The US advisory system iswell trusted by the US public, perhaps partlybecause of its openness, but it has been usinga greater proportion of closed hearingsrecently [372]. In contrast, the UK hasoperated a closed advisory system, but isbecoming more open.

67 Renn (1995) argued that, owing to thedifferent styles of the advisory systems, theUS system needs to work harder to avoidcapture by special interests, whereas theEuropean systems need to engage in morestakeholder consultation.

68 The US government supplements theadvice it receives by a statutory arrangementfor the provision of advice from learnedsocieties. In some countries, expert panels areconvened by learned societies; for example,the Royal Society of Canada has its owncodified procedure for the conduct of expertpanels [467]. Britain has no equivalentarrangement. Learned societies in Britain donot routinely have a formal role in the scientificadvisory process. Although they have anindependent status, learned societies havestrong established interests through theirtraditions, the often close affiliation of theirsenior members with the administration ofgovernment and research funds, and theirselective membership arrangements.

69 Scientific advice may not always be freefrom value judgements, and scientiststhemselves ascribe to political views andmoral values (Weingart, 1999) [384].Furthermore, scientific advisers are sometimes

requested to make a value judgement. Forexample, advisers may be asked to makeassumptions about the tolerability of riskswhen defining regulations. However, thetolerability of risks cannot be determinedobjectively [470]. Thus, the linear model ofproblem definition, advice and decision-making might be developed into a recursivemodel, allowing several rounds of negotiation.In this model, Edwards (1999) notes, the roleof advisers is not merely to advise decision-takers, but to inform, educate and empowerthe public for their own interactions withdecision-takers [386–388].

70 There is an advantage in distinguishingbetween information gathering and makingvalue judgements about welfare distribution.There is a natural break in the policy process,where evidence may be circulated forconsultation before policy advice is prepared.Experts may be needed to gather and analyseinformation, but only a Minister or those actingwith delegated authority can legitimately makevalue judgements.

71 To address these issues, this studysuggests a principle [104–109] that thelegitimate Ministerial function for takingdecisions should be distinguished from thepolicy-maker’s function to formulate policyoptions, and the scientific adviser’s function toprovide scientific advice.

72 Weingart (1999) suggests that policy-makers become ever more reliant on expertscientific advice, and that scientists competestrategically to supply advice [388]. Hesuggests that a solution to this is a moreformalised and prescriptive process ofadvising policy-makers. Scientific advice hasbecome more important in policy as the paceof technological development has increased.The UK government receives a large volumeof scientific advice, and has relied on this inmany high-profile areas. The governmentroutinely seeks advice from many hundreds ofscientific advisory committees and individualadvisers.

73 The advantage of a more formal processis that it is possible to audit and revise itsoperation, and its effectiveness may bemeasured and reported [144, 411–413]. Thepresent advisory processes vary greatly inform, often for good reasons, but there iscurrently no guidance available on commonelements that would constitute best practice[498–500]. It has been apparent fromdiscussions with policy-makers and others thatthe advisory system needs to be flexible inorder to address many different policysituations. A detailed and formal prescriptive


11

process would be cumbersome to operate andmight restrict the policy process unnecessarily[144, 498]. The model process described inthis study is designed to provide a safeguardlevel of good practice under a wide range ofadvisory situations, while retaining sufficientflexibility to be practical and convenient.

74 ILGRA has called for a framework forrisk assessment to be applied consistentlyacross departments [457, 470]. The grouprecommended that the framework shouldsubject scientific advice to peer review andpublic scrutiny, that its assumptions anduncertainties should be explained, and itsprocedures developed to allow stakeholderinput.

75 Public scrutiny of advice and policydecisions is a recurrent theme of all theevidence collected. One of ILGRA’s proposalsis that assumptions and uncertainties shouldbe presented for stakeholder scrutiny. Thisstudy suggests that Ministers publish a‘reasoned opinion’ explaining policy decisionsin detail [0].

76 In seeking to open up the system topublic scrutiny, this study is supporting currentgovernment policy on openness, butchallenging the confusion that arises whenexamining the detail of the Freedom ofInformation Bill in the context of scientificadvice [477–485]. The study concludes thatscientific advice should not be regarded asadvice to Ministers (and hence confidential),and that there should be a presumption ofpublication under the freedom of informationrules.

77 From the sets of guidance and studies ofadvisory systems examined, it was clear thatthere are several models for securing scientificadvice. The common features were as follows:

• all rely heavily on the establishedscientific community;

• they use a range of advisers, from singleexperts to committees;

• balance in the composition of advisorycommittees is important;

• transparency of advice, the explanationof uncertainties, and public access toinformation are important; and

• involving stakeholders in the process isalso important.

78 The differences were in:

• the detailed management of the processby officials or sometimes by a non-

governmental external body, such as alearned society;

• the person to whom the adviser isaccountable, either officials, Ministers, orexternal bodies;

• the publication of evidence, the extent towhich meetings are open or closed, andthe method of communicating the advice;

• the rights of the advisers to have accessto Ministers and independence fromofficials;

• the rights of the advisers to agree theterms of reference;

• the extent to which the procedures of theadvisory system are codified;

• the degree of distinction between policyand scientific advice; although, ingeneral, there is agreement that thedistinction should be made;

• the degree of involvement of Ministers inthe advisory process;

• the models for generating advice,ranging from adversarial to consensual;

• the remuneration of advisers;• the way in which stakeholders and the

public are engaged in the advisoryprocess; and

• views about the value of peer review.

79 The UK system does not offer a cleardistinction between the functions of scientificexperts and stakeholders. In response, thisstudy presents recommendations that providea much sharper distinction between thesefunctions.

80 In recommending a new balance in thegovernment’s approach to securing advice, thevalue of the existing flexible and fairly closedprocess has been compared with the greaterpublic confidence and opportunity for qualitycontrol promised by a more clearly definedand open process.

Lessons from interviews

81 The interviews conducted during thisstudy revealed strong feelings about thescientific advisory process, from advisers,policy-makers, decision-takers and otherobservers. The strongest commentsconcerned the motivation of individuals, andthe incentives or interests that might influencethem. In response, the declaration of interestsand management of bias is an important partof the recommendations in this study [182–189, 242].

82 One of the main concerns of theinterviewees was the reason why particularadvisers are selected, and how government


12

might manipulate the output of the advisoryprocess through the selection of advisers. Aclear distinction of functions in the advisoryprocess has been made in this study. Thelegitimate function of a Minister as thedecision-taker is supported [104–109, 145],and the rights of the adviser and the need for aclear remit for the adviser and the policy-maker are reinforced [240–245, 145]. Thedanger of bias is reduced through stakeholderinput, peer review, and publication.

83 The absence of guidance on theoperation of advisory committees isrecognised (the DTI has recently published aconsultation paper on a code of conduct formembers of advisory committees (OST,2000a)). Several models have beensuggested, such as courts of law andparliamentary select committees. Suggestionsfor guidance to assist committee members andchairpersons are made in this report [239–245], clarifying their rights and duties.

84 Interviewees expressed concern aboutthe capacity of government departments tointerpret scientific advice, and of Ministers tocomprehend it. The recommendations in thisreport support the training of policy-makersand decision-takers in seeking and usingscientific advice (and especially in handlinguncertainty), and offer suggestions on howadvisers can provide their advice in a usefulformat [272–304].

85 Finally, the interviewees emphasised—more than any other topic—the virtue ofopenness and transparency of advice andpolicy decisions [110–122]. While the existingrules for publication were noted, there wasgeneral support for more openness andtransparency. This has been included inrecommendations that stakeholders should beincluded at more stages [see Figure 2, page22]; that aspects of the advisory processshould be open to consultation and published;and that Ministers should provide informationallowing more effective parliamentary scrutinyof decisions [226–227, 0].

Conclusions

86 The evidence showed that, within the UKand further afield, many different advisoryprocesses exist. Some of them are codified

and follow rigid rules, while others arecompletely ad hoc. Nevertheless, the variationdid not disguise several common themes.First, the processes followed similar sequentialsteps; framing the question, seeking evidence,considering the views of interested parties,and reporting conclusions. Second, there wasa common emphasis on the independence ofadvisers, the rights of advisers, and theimportance of communicating uncertainty.

87 There were widely held concerns that, ifthe advisory process is too loosely defined, thefunctioning of the advisory system becomesunreliable, and that some form of codificationis therefore desirable. Even where similarapproaches are taken, the conclusions ofdifferent advisers may be different, and socodification would not prevent variation in theadvice that is received. An explanation of thescientific reasoning, distinguishing betweenfacts and opinions, could therefore be helpfulin resolving disputes about the correctness ofadvice.

88 The evidence also identified manyperceived problems with existing systems,which have been considered in thesuggestions proposed in this report.Notwithstanding these points for which therewas general support, many conflicting opinionsabout the best way for the advisory system tofunction were also apparent. This reflected theabsence of a common agreed basis for theoperation of an advisory system.

89 A strong theme that emerges is that anadvisory system needs some clear basicprinciples according to which it can operate,and against which its operation can be judged.

90 A second clear theme is that somecodification of the advisory process would bepreferable, to provide consistency ofoperation, efficient functioning, robust qualityof outcome, and to permit the application ofthe principles to be audited.

91 This study also recommends an audittrail from the advice through to the decision[120–121, 143, 201]; increased access forstakeholders, in general via the policy-maker[151, 157–159]; and the benefits of balancedcommittees.

|O|X|E|R|A| Principles

13

PRINCIPLES92 These principles express an underlyingphilosophy applicable to all scientific advisorymechanisms, whatever the policy issue maybe. In them is distilled a collection of ideasderived from the evidence-gathering phase ofthis study.

93 The need for a set of principlesapplicable to scientific advice emerged fromthe case studies and initial debates within thestudy team. It became apparent at an earlystage that some issues were being raisedfrequently, where points of principle werebeing advocated.

94 The initial step was to collect a long listof ‘candidate principles’ from all the materialcollated in the first phase of the study, andthen to organise these under thematicheadings. A draft was revised several times inconsultation with the Advisory Board, theSteering Group, and, finally, after presentationat a seminar.

95 Some of the principles may seem self-evident or trivial, but in practice all make animportant contribution to the definition andimplementation of a valid scientific advisoryprocess. Each principle addresses (as far aspossible) a distinct issue. The principles areconsistent and do not overlap.

96 The dominant consideration in settlingthe final form of the principles was the need fora disciplined framework in which theparticipants in the scientific advisoryprocesses can operate. Other schemes wereconsidered, in which there would have beenmore reliance on subjective judgement, butthese were judged to carry greater risks ofabuse and error.

97 The principles guide working practices

and are available as a reference, should theconduct of the scientific advisory process bechallenged.

98 The principles were conceived forapplication to the process of seeking andusing advice from experts in the naturalsciences; their relevance to the socialsciences has not been explored and remainsuntested.

99 The following sub-sections set out theseven principles, which cover, in summary:

• reliance on logic and reasoned argument(rationality);

• distinguishing the functions of the expertscientific adviser, policy-maker, decision-taker, and stakeholder representative(legitimacy);

• ensuring that the process is transparent,auditable, and accessible to allstakeholders (inclusivity andtransparency);

• criteria for selecting advisers andmanaging bias (competence);

• imposing on scientific advisers an over-riding duty to the public interest (over-riding duty);

• acknowledging the limits of scientificknowledge, expressing uncertainty andrecognising this in policy-making(candour);and

• matching the advisory effort to theimportance of the policy issue and thedifficulty of the scientific questions(proportionality).

100 The following sections contain theprinciples, each preceded by the reasoningthat led to it.


14

Rationality

101 Scientists use logic, evidence andreasoned argument in their work, an approachthat is common to science, policy-making,legal decisions, and parliamentary and generalpublic debate.

102 The scientific method has been shown tobe extremely effective at producing theorieswith explanatory and predictive power.Science can make predictions about theconsequences of actions, and changes in thephysical environment through time. It can alsolink causes to effects.

103 Scientific advice can, therefore, informdecision-takers about the potential

consequences of the options open to them.Scientists use a core of firmly establishedinformation. Outside this core, there is a largebody of valuable indicative information that isless firmly established, and, beyond it, on theboundaries of current research, there is verysignificant uncertainty. In these less-secureregions, there is still much that can be said tocharacterise a range of possible outcomes, bystating the risks and uncertainties. It isrecommended that a disciplined approach betaken to the handling of evidence and that thepresentation of this evidence be made bydeductive reasoning from what is known andfrom the partial evidence that is available.

Scientific advice should be founded upon observation and theory, and should describe boththe scientific conclusions and their uncertainty; these should be deduced from the evidenceby reasoned argument.

Legitimacy

104 Scientific advice is one component of apolicy decision. In formulating policy, it shouldbe taken into account, together with economicand social considerations, political constraints,values, and the views of stakeholders.

105 Scientific advisers should be competentto provide advice on particular problems withintheir field of expertise, to analyse theconsequences of a range of policy options,and to assess the probability of variousoutcomes.

106 Scientific advisers are not necessarilyalso qualified or authorised to make policydecisions, or to provide advice that containsjudgements about social values; equally,decision-takers are usually not competent toassess scientific issues.

107 The process of translating a policyquestion into a scientific question, and bringingscientific answers into a policy-making forum,requires particular skills—a combination ofscientific literacy and policy awareness. Thesynthesis and analysis of scientific and other

advice is the professional occupation of policy-makers.

108 Stakeholders, by definition, have aninterest in the outcome of a policy question.One attribute by which the success of a policyoption could be judged is its acceptability tostakeholders. Thus policy-makers may wish toelicit the views of stakeholders during thepolicy-making process. This could includeasking stakeholders to comment on, orchallenge, scientific advice. Stakeholders mayalso possess information that could contributeto the scientific analysis of policy options.However, the contribution that stakeholderscan make to the scientific advisory processshould not include negotiating on behalf oftheir constituencies as a representative.Further, all of these stakeholder functions aredifferent to the function of a scientific adviser.

109 Each function is different, their objectivesare different, and the output from each mustbe treated appropriately in the decision-takingprocess.

The functions of scientific adviser, policy-maker, decision-taker and stakeholderrepresentative should be distinguished.


15

Inclusivity and transparency

110 There are usually many individuals andgroups with a stake in policy decisions,including the public (as the main risk bearersand potential beneficiaries), industry (asproviders of benefits and generators of risks),and pressure groups. These stakeholdersdemand access to the decision-takingprocess, and this includes the scientificadvisory process.

111 Stakeholders have an important part toplay in ensuring that the scientific questionsbeing asked are pertinent to the policy issuesthat are of concern to them. However, activeand direct participation by stakeholders in thegeneration of scientific advice itself could leadto generating scientific advice based on valuejudgements and stakeholder views, rather thanon expert analysis of scientific evidence.

112 As risk issues have become more widelydiscussed, public debate has become moresophisticated and the fact that some situationsoffer no risk-free options has become morewidely recognised. The late announcement ofunwelcome news can undermine publicconfidence, so openness about potential risksis both a beneficial and a practical policy.

113 Care should be taken to avoid a form of‘openness’ that actually obscures keymessages through obscure terminology ordilution in a large volume of detail. Detailshould be available for the expert reviewer, butthe aim of this principle is effectivecommunication with the public. This requirestransparent communication of the key findings,including uncertainty.

114 The convention in the scientificprofession is that new scientific results shouldbe published freely so that they can beexposed to challenge and developed byothers. If this convention were applied toscientific advice, it would impose the normalprofessional discipline upon individualscientists. It might also moderate the way inwhich special interests present their cases,and might stimulate debate and the expressionof alternative viewpoints.

115 In the process of scientific investigationthere may be an initial period when informationis incomplete, there are competing views, andanalysis is inconclusive. The balance ofevidence and argument may shift whilescientific work is in progress. Thus, partlyfinished work may be unbalanced and, hence,misleading. For these reasons, scientificadvice should not be published until it is

complete, at least in draft form—whenbalanced evidence has been collected,opposing views have been addressed, andresidual uncertainties explored.

116 A judgement may need to be made bythe policy-maker as to when scientific advice issufficiently complete that it should bepublished. In addition, some policy problemsmay require advice on more than one scientificquestion. These pieces of advice may not allbe completed at the same time. In somesituations, therefore, a balanced picture mayonly be obtained once all the relevant scientificinputs into the policy-making process havebeen received. The presumption should be forearly publication wherever possible, whether inpart or in whole.

117 Where the developer of a potentiallyhazardous product is obliged under regulatorylegislation to provide information about theproduct and its hazards (for example, in amarketing licence application), the law usuallyaffords some protection of the applicant’scommercial secrets through a limitedconfidentiality restriction. Such restrictions donot extend to essential safety information,such as the precautions to be taken duringtransport and use of the product, but couldcover information such as original laboratoryand trial results on efficacy, because these arecostly to obtain and their publication mightassist the applicant’s competitors. There is apotential conflict between the rights of theapplicant and the rights of the public to accesssuch information. It is therefore recommendedthat the presumption in favour of publicationcontinue to be applied with the absoluteminimum of restriction, and that considerationbe given to creating alternative mechanismsfor protecting commercial interests whileallowing full disclosure of the information onwhich product approval is based.

118 In no case should the existence ofresidual uncertainty be considered reason forwithholding results.

119 However, the process by which thescientific advice is being generated shouldalways be open to scrutiny. This would help toensure freedom from bias and allowstakeholders to contribute to the advisoryprocess.

120 An audit trail of the scientific advisoryprocess could be made available to policy-makers, decision-takers and, ultimately, tostakeholders. This audit trail might cover the


16

generation of the scientific questions; thescientific arguments, calculations andanalyses; alternative scientific opinions anduncertainties; and the scientific appraisal ofpolicy options.

121 Decision-takers may find that it is usefulto explain how scientific advice was taken intoaccount in a policy decision. This wouldconfirm that the scientific advice has beencorrectly understood, and that it provides

assurance to stakeholders that the issues ofconcern to them have been addressed. Itwould provide valuable feedback to thescientific advisers, and, hence, contribute tothe continual improvement of the scientificadvisory system.

122 Despite the above, it is ultimately theresponsibility of government to decide how,when or whether scientific advice togovernment should be made public.

Full information about the process of seeking and using scientific advice should be madepublic, and stakeholders should be encouraged to comment. Scientific advice itself should bepublished promptly in a transparent manner, once it is complete.

Competence

123 Scientific advisers should be selected fortheir competence, which should be appropriateto the scientific question, and should includenot only personal proficiency but alsoawareness of the wider state of knowledge inthe field, and an aptitude for communication.Scientific advisers should show willingness toconsider scientific opinions that are at variancewith their own, and to discuss these in theiradvice. The most distinguished researchscientists may not always make the bestadvisers.

124 Some scientists who have the potentialto contribute useful information to the advisoryprocess, owing to their relevant knowledgeand experience, may be employed byorganisations such as industries or pressuregroups, that have an interest in the policyissues. Such scientists should not necessarilybe excluded from the process, but they shouldadvise as experts rather than negotiators.Their contribution must be managed by thepolicy-makers and balanced by other sourcesof expertise. Their experience, and biases, canbe a useful contribution to the evaluation ofscientific uncertainty.

125 Scientific advisory committees may bebalanced by appointing a group of memberswith different affiliations. This will help todefine the range of scientific opinion on aparticular issue, and may help to prevent

undue bias from the influence of any oneindividual.

126 The Nolan principle that appointeesshould not ‘place themselves under anyfinancial or other obligation to outsideindividuals or organisations that mightinfluence them in the performance of theirpublic duties’ should not be applied rigidly inthe case of individual scientific advisers. Theaim of the Nolan principle should be met inother ways; namely, the declaration ofinterests, the balancing of sources of advice,the over-riding duty to the public interest, andthe explicit communication of uncertainty.

127 The declaration of interests by scientificadvisers is necessary whether they areacademics, employed by industry, oremployed by pressure groups. Possession ofinterests should not be presumed tocompromise an individual’s integrity.

128 In assessing the likely bias of a scientificexpert, policy-makers should consider thewhole of their professional profile(employment, affiliations, consultancy,employment history and experience), and notjust their financial interest in the subject.Scientific preconceptions and adherence toparticular schools of thought may also distortscientific judgement or introduce bias.

Scientific advisers should be selected for their competence. It is not necessary for advisers tobe independent of all interests in the policy question; however, the advisory process shouldensure that interests are declared and any resulting biases are balanced or taken into account.


17

Over-riding duty

129 Scientific advisers may face a conflict ofinterest because they have other occupationsand interests outside their role as advisers.This professional experience of expertscontributes to their ability to provide advice.Experts who are employed in industry oracademia may possess expertise not availablewithin government.

130 Scientific advisers may feel underpressure to bias their advice, for example, tosuit the constituency that nominated them asadvisers (which might or might not be thegovernment), to be consistent with establishedscientific consensus, to lead to a particularpolicy outcome, or to suit the adviser’semployers. Individual advisers should beempowered, and required, to resist suchpressures.

131 The general public is often the bearer ofrisk associated with policy options. The aim ofgovernment in forming policy relating to risks

is the protection of the public or theenvironment.

132 It is difficult to provide a universaldefinition of the public interest. Furthermore,scientific advisers are neither qualified norauthorised to make value judgements, and itcannot be possible to define the public interestwithout making value judgements. Thus,scientific advisers should simply be requirednot to promote any special interest inpreparing their advice. The aims of thisrequirement are to ensure that scientificadvisers recognise that they are forbiddenfrom placing a conscious bias on their advice,and to empower advisers to use thisrequirement to resist any pressure to do so.This requirement addresses only consciousbias, and does not guarantee that advisers willbe free from unconscious bias, or remove frompolicy-makers the duty to balance, orotherwise take into account, all sources of biasin the advice they receive.

The scientific adviser’s over-riding duty is to the public interest. Advisers should not seek topromote any special interests in preparing their advice.


18

Candour

133 There are limits to the extent of scientificknowledge, particularly concerning newtechnologies and the assessment of risks.Scientific advice in some areas is thereforecharacterised by uncertainty and differences ofscientific opinion. For good decision-taking, itis essential that these uncertainties are takeninto account.

134 Scientific advisers could be asked toseek out the sources of uncertainty in theiradvice, and be candid about theseuncertainties, and any differences of opinionamong their peers. They ought to be ready toexplain how the uncertainties arise, whatassumptions have been made, and to providean estimate of the overall uncertainty. Inparticular, they should avoid the temptation toclose out scientific issues prematurely, wherean error could have adverse consequences.

135 The merits of each policy option dependcrucially on the consequences (costs andbenefits) that could flow from it, and on thelikelihood of those consequences. Scientificadvice should therefore address not only the

generally expected (most likely) outcome ofeach policy option, but also the risks ofextreme and, especially, adverse outcomesthat are still within the bounds of possibility.The supplementary note on ‘Decision-takingunder Risk and Uncertainty’ contains a fullerdiscussion of this issue [272–304].

136 Scientific advisers are not, and neithershould they, feel duty-bound to provide asubjective judgement where the evidencecannot resolve a question, nor should they besubject to any pressure to do so.

137 Scientific advisers themselves may notbe the best judges of the uncertainty in theirown advice, although they can make a majorcontribution to this assessment. It is anecessary skill of policy-makers to assessscientific uncertainty by careful drafting ofquestions to scientists.

138 Decision-takers need to accept andunderstand the uncertainty in the advice theyreceive, and to consider it in taking theirdecision.

Scientific advisers and policy-makers should be candid about the limitations of scientificknowledge and should always assess the uncertainty in scientific advice and the risksassociated with each policy option; this information should be taken into account by decision-takers.

Proportionality

139 There is a range of problems for whichscientific advice is necessary. Problems canbe categorised by:

• urgency—a decision may be requiredvery quickly;

• the magnitude of the risks;• the frequency with which problems of the

same type arise;• the public perception of its importance;• uncertainty in the existing science base;• tractability of the problem.

140 Likewise, different ways of generatingscientific advice can be characterised by:

• the time required and cost likely to beincurred;

• the available expertise and capabilities;and

• the degree of certainty achievable.

141 The choice of mechanism for generatingscientific advice is likely to involve a trade-off,for example between the cost of generatingthe advice and the degree of certaintyachievable.

The effort expended on securing scientific advice should be proportionate to the importanceof the policy issue and the difficulty of the scientific investigations required.

|O|X|E|R|A| Model Process

19

MODEL PROCESS

Rationale for a defined process

142 The process described here is a model ofbest practice, designed to be compatible withthe principles described in the precedingchapter. This process is not intended to beseen as mandatory, nor as inhibiting thefreedom of departments to act efficiently. Itneed not be expensive to implement. Theprocess is offered as a model that could beconsidered by departments when deciding howto address an issue on which scientific adviceis needed.

143 A model process brings a number ofadvantages.

• Once a process has been defined it willbe quicker and easier to follow it, whenadvice is required, than to define an adhoc process each time.

• A codified process can be audited. Onceaudited, public confidence in the processcan confer confidence in the advice.

• The definition of the process, and its auditand review, can be published to providepublic accountability.

• A process reduces dependence on thejudgement of individuals. Stakeholdersmay have confidence in a process, ratherthan have to place their trust inindividuals.

• A defined process may help to ensurethat advice is ‘fit for purpose’, since ithelps to avoid ambiguity and uncertaintyabout the functions and contributionsexpected from the participants in theadvisory process.

• If the same process is used acrossgovernment, the outcomes are more likelyto be consistent.

144 The disadvantage of a defined process isthat it may not be sufficiently flexible to coverevery circumstance in which scientific advice issought. The model can be adapted asnecessary, but:

• the adapted process should remainconsistent with the principles;

• the justification for omitting or adaptingparts of the process, and for thealternative adopted, should be recordedso that the operation of the process canstill be audited; and

• even where the pressures of urgency orother constraints, or the close workingrelationship between the scientific

advisers and those whom they advise,preclude the use of a distinct structuredprocess, efforts should be made toconform to the spirit of the model.

Functions

145 Four functions can be identified in thepolicy process: decision-taker, policy-maker,scientific adviser and stakeholderrepresentative. The first three are needed totake policy decisions. Stakeholders have viewsand interests that must be taken into account inreaching a decision. The functions can bedefined as follows:

• Decision-taker: a person with theauthority to take a policy decision. Thismay be a government Minister, or aperson or body with the delegatedauthority to take a decision in the name ofa Minister.

• Policy-maker: a person or organisationcharged with assisting a decision-taker inreaching a decision by providing policyanalysis, generating policy options, or byconducting risk assessment. Policy hasbeen interpreted to include regulation.

• Scientific adviser: a person ororganisation responsible for providingscientific input to policy-making ordecision-making. This includes bothscientists expert in narrow disciplinesrelevant to the problem in question, andbroader-based scientists able to integrateseveral disciplines.

• Stakeholder representative: a person ororganisation representing the interestsand opinions of a group with an interest inthe outcome of a particular policydecision.

146 There is a question over whether it ispossible to separate the functions of policy-maker and scientific adviser. It is argued bysome that the scientific and political elementsof decisions are inextricably linked.

147 If scientific advice and policy-making areallowed to become inextricably mixed together,the resulting input to the decision-takingprocess is muddled and confusing, because it


20

is unclear to what extent the advice depends onthe analysis of qualified experts, and to whatextent it depends on value judgements. This islikely to result in poor decisions, and a lack ofconfidence. Should such advice be published, itwould be liable to be attacked by scientists aspoor science, and also by stakeholders as notrepresenting or taking fair account of theirviews. Such confused advice would also beincompatible with the government’s stated aimof developing evidence-based policy.

148 The model process described here allowsthese functions to be clearly distinguishedwithout compromising the quality of thescientific advice or the decisions that are takenbased on it. It places demands on all theparticipants to recognise and fulfil their roles.

149 In distinguishing these four functions,different people would normally perform eachfunction. However, if there is a good reason forone individual (or group) to enact more thanone function (perhaps because the problem issmall, or specialised, or requires urgent action),that person should be sure which function theyare fulfilling at each point in time. The sameperson may then have to consider the output ofone function as the input to another. In legallanguage, an individual with more than onefunction should ‘direct themselves’ to addresseach function appropriately. Failure to engagein this disciplined process might lead to pooradvice and a poor decision, for the reasonsgiven above. In any case, if the same person(or group) performs more than one function,transparency may be compromised.

150 This overlap of functions may beparticularly challenging if a government-employed scientist with delegated responsibilityfor the decision has to exercise all threefunctions. In practice, a written record of thelogical argument of the case would provide anaudit trail and justification for the decision, ifsubsequently challenged.

Figure 1: Functions andrelationships in a policy decision

decision-taker

policy-maker

scientificadviser

eg, economicadviser

eg, socialadviser

otheradvisers

politicaladviser

stakeholders

challenge to advisers

151 The functions and relationships of theparticipants in the policy process are drawn inFigure 1. The key points are:

• the figure reflects the need for the policy-maker to obtain advice other thanscientific advice, and for that advice to bereflected in the options andrecommendations presented to thedecision-taker;

• stakeholder representatives bring theirconcerns to the policy-maker, not to theadvisers (scientific or otherwise). It is forthe policy-maker to extract from thestakeholder representatives the elementsthat need to be considered by the expertadvisers (including the scientific adviser);

• the scientific adviser has a close workingrelationship with the policy-maker butdoes not interact directly with the otheradvisers, other than through the policy-maker, who may wish to refer issuesbetween advisers or bring them togetherto explore issues which overlap theircompetencies;

• all of the advisers, including the scientificadviser, may be challenged by thedecision-taker. This is essential if thedecision-taker is to be able to takeresponsibility for the decision. Thedecision-taker must be able to questionthe scientific adviser directly, in order tobe satisfied that the scientific advice isrobust (similar lines of communicationmay be necessary for the other expertadvisers); and

• the decision is expressly recognised as apolitical decision, for which political advicemay be needed, and it is the duty of thedecision-taker to reach a balanceddecision in which scientific advice playsits due part, but is not necessarily thedetermining consideration.

Types of advice

152 The use of scientific advice in threesituations was considered in the course ofdefining this process. They are different in kindand in the way that the participants behave.Each of the participants in the advisory processshould maintain a clear understanding of theirresponsibilities in these situations. The threesituations are described below.

153 General advice: the policy-maker asksthe scientific advisers for general advice on thescience of an issue so that this may be takeninto account when defining policy options. Thismay be a preliminary process, to be followed by


21

the preparation of a set of policy options thatmay then be reconsidered by the scientificadviser.

154 Policy advice: the policy options areknown and require a balanced decision to bemade, weighing up the benefits, risks and costsof each option. Scientific advice is a vitalcomponent of the decision but has to be builtinto a legitimate decision-making procedurethat takes account of all other aspects as wellas science.

155 Technical decision: the policy-makermay pose a set of policy options that arecontingent upon scientific inputs, and thedecision-taker may select and endorse theseoptions. Once this has been done, the scientificadvice may legitimately determine the policyoption selected, since a due constitutionalprocess has been adopted. For example,routine decisions on appropriate safetystandards may depend only on technicalanalysis if the political decision about thetolerability of risk has already been made.

156 It is possible that an issue will bediscussed in each of these situations insequence. Initially, general advice is needed toidentify and understand the issue. There maythen be a need to implement policies torespond to it, for which policy advice is needed.Once the issue is well understood, itsmanagement might be purely technical and,with the appropriate policy guidelines in place,might be handed over to technical specialists toadminister. The model process describedbelow is designed to address all three of thesesituations.

Engagement with stakeholders

157 Stakeholders should be entitled tocontribute to most stages of the process.However, they do not have a duty to contributeand cannot therefore be held responsible forany part of the process.

158 Stakeholders have a valuable role inchallenging both the scientific advice and theactions of the policy-maker. Methods to elicitstakeholder contributions have been reviewedrecently within the ESRC Global EnvironmentChange Programme (ESRC, 1999), including:

• focus group: typically 6–10 people,supported by a trained facilitator whoguides the group through a set ofquestions in order to identify the mainconcerns;

citizens’ juries: usually 12–25 people whoare presented with evidence for differentpolicy options. Witnesses may bechallenged and the jury votes to reach amajority decision;

• in-depth groups: longer-term focusgroups;

• consensus conferences: larger citizens’juries;

• stakeholder decision analysis: combiningquantitative multi-criteria techniques andqualitative evaluation; and

• deliberative polling: members of thepublic are polled after receiving a briefing.

159 These techniques appear to beparticularly useful for allowing decision-takersto gauge the views of the general public. Assuch, they are part of the wider decision-making process. These techniques might beuseful in engaging stakeholders with an interestin a policy decision for which scientific advice isbeing sought. This is the subject of ongoingresearch in the social science community andelsewhere, and has not been examined duringthis study.

Nature of the model process

160 This is a benchmark or model process,against which the practicalities of day-to-daypressures need to be set. The process iscompliant with the principles set out earlier (andthe constitutional and other arguments thatunderpin them). There may be other processesthat have equal validity.

161 The process is divided into a set ofdiscrete and bounded steps (see Figure 2overleaf). In practice these may be overlappedor even merged, and may not be clearlydistinguishable. The model is presented as asequence of steps with some explicit feedbackloops. In practice, the stages may undergomany iterations.


22

Figure 2: Model process for securing and using scientific advice

second opinion ifneeded

seek agreement on policy issues

seek agreement on process

Establish the policycontext and options

Detect the issue

Define what the policy-maker needs to know

Choose the scientificadvisory mechanism

Choose the scientificadviser(s)

Agree and confirm thebrief

Prepare the scientificadvice

Communicate theadvice

Prepare advice onpolicy options

Take the decision

publication of draftscientific advice

consultation onactions

Responsibility:

decision-taker

policy-maker

scientific adviser

engagement withstakeholders

emergingissues


23

Detect the issue

162 Policy issues may emerge from manysources, including parliamentary questions,departmental science staff, the media(including academic journals), and the generalpublic. They may also be triggered byaccidents, legal proceedings, patentapplications, or technological innovations.

163 The early detection of a policy issue isdesirable, because it maximises the number ofpolicy options available and the time availableto analyse them. Systematic proceduresshould be employed to maximise the chanceof detecting an issue promptly. This might beachieved by:

• a duty on all standing advisorycommittees to produce regular (perhapsquarterly) briefings on key issues, whichsurvey recent developments within theircompetence. The briefings should beconcise and easily understood by policy-makers;

• a similar duty on all research councils;• a ‘policy-foresight’ exercise, which seeks

to detect issues that might arise beforetheir effects can be felt. This mightinclude systematic modelling ofscenarios and monitoring of scientificdevelopments and their possible effects;or

• an invitation to stakeholders, includingNGOs and other expert interestedparties, to take part in regular reviews ofemerging science and its implications.This might be a role for standingcommittees, provided that their terms ofreference are clearly stated to be toidentify issues, not to offer scientificadvice on policy options.

164 These procedures might be tested byconducting trials.

Recommendation: Procedures should be established and maintained to ensure that issuesdemanding scientific advice are detected as early as possible.

Establish the policy context andpolicy options

165 The policy context of the problemincludes:

• what policy issue has arisen?• what policy options are available and

should be considered? and• who has responsibility for dealing with

the problem (including identifying anddefining the scientific issues, selectingand briefing the adviser, receiving thescientific advice and building it into policyoptions)?

166 Systematic procedures, such as scenariomodelling, might be used to ensure that a widerange of options is explored. Scenariomodelling allows the policy-maker to ask ‘whatif’ questions and is particularly valuable fordetecting extreme possibilities and worstcases. Some scenarios might be contingent ona particular set of scientific circumstances, andthe recognition of the existence of the scenarioallows the correct scientific question to beposed.

167 Stakeholders may make an importantcontribution to this step by ensuring that therange of policy options considered isappropriate.

168 It is legitimate and desirable to frame theissue in terms of a choice of policies, wherethe choice is to be influenced by the scientificadvice. This provides a mechanism forscientific advisers to set their work in contextwithout delegation of the policy decision.Scientific advisers should assist in defining thepolicy options in consultation with policy-makers. This helps to ensure that scientificadvice is relevant to the policy problem byensuring that the right scientific questions areasked.

169 Scientific considerations may help toidentify policy options that were not apparentbefore. It is therefore preferable that policyoptions should not be circumscribed before thescientific advice is available—it is much easierto reject a policy option after the science hasbeen considered than to add one that has not


24

been taken into account earlier. Policy-makersshould be prepared to reconsider the range ofpolicies that are available in the light ofemerging scientific evidence.

170 If some policy options have been ruledout by the decision-taker, then this should bemade clear in the adviser’s terms of reference.

Recommendation: Before seeking scientific advice, policy-makers should identify the policyissues on which the science bears, the policy options that are available, and the scope ofadvice that is needed. Policy options should be kept open as far as possible until scientificadvice has been taken. Policy options should be reviewed in the light of scientific advice toensure that a complete spectrum of options has been considered.

Define what the policy-maker needsto know

171 The policy-maker needs to be competentto specify scientific questions. The questionsshould be refined through discussion betweenscientific advisers and policy-makers, and maybe further refined through challenge bystakeholders. The questions put to scientificadvisers must be restricted to scientific issues,and must admit a purely scientific answer, inorder to maintain the distinction of functions.

172 It may be necessary to pose severalquestions that require a range of scientific

disciplines, and hence a number of scientificadvisers.

173 The policy decision must not bedelegated to scientific advisers by posing thequestion in such a way that the scientificadvice determines the policy option to beadopted. The only exception to this is wherepolicy guidelines establish that a decisiondepends only on scientific advice, and aproper decision-taking process has alreadybeen followed (the technical decision situationdescribed above).

Recommendation: Policy-makers must define the scientific questions that bear on the policydecision. The definition must separate the scientific and policy issues, and pose questionsthat are amenable to purely scientific answers.

Choose the scientific advisorymechanism

174 Four mechanisms for the provision ofscientific advice can be identified.

• Scientific generalist: this is anindividual who can assimilate aninterdisciplinary problem and applysystematic analysis to the interpretationof information. The scientific generalist isusually a highly respected scientist, butnot necessarily an expert in thedisciplines relevant to the issue.

• Single expert: this is an individual whois appointed as an acknowledged, pre-eminent expert in the field.

• Consultancy contract: this may beawarded to a private company, publiclaboratory, or academic research group.It is frequently used where there is muchdetailed work to be done (such as fieldtrials or laboratory research), or whenthe intensity of effort required is beyondthe capacity of the other mechanisms.

• Scientific advisory committee:committees are used to bring together awide range of expertise, usually under anexperienced chairperson.

In each case, some or all of the advisers couldbe government-employed scientists.


25

175 Three other mechanisms wereconsidered.

• Task force: which is often used torespond to emergencies. The authorityfor making policy and taking decisions isdelegated to the task force, whichconsiders scientific matters whenexercising that authority. A task forcecan—and, where possible, should—follow the process described in thisreport. However, because its briefencompasses all of the stages of takinga decision, it cannot be regarded asgenerating scientific advice. Scientificadvice is a subset of the work of a taskforce, and should be obtained by one ofthe mechanisms described above.

• Stakeholder committee orcommission: whose members representthe interests of stakeholders. Thestakeholder committee or commission isnot able to act according to the principlesdescribed in the preceding chapterbecause, by definition, its members areappointed to serve the interests that theyrepresent. The committee exists toinfluence the policy-maker and decision-taker. As such, it can be profoundlyundemocratic, representing only theviews of those who are present and notany underlying public interest. Astakeholder committee may seekscientific advice and use scientificargument, but it does not offer scientificadvice to government—it offers policyadvice. Sometimes, where it has theauthority, it may even be the decision-taker.

• Public inquiry: a scientific advisorycommittee may choose to employ themethods of a public inquiry (openevidence, cross-examination, formalproceedings). A more general publicinquiry usually has a much wider scopethan the preparation of scientific advice,and takes into account non-scientificissues. Like a task force, it may call onscientific advice.

176 Although these mechanisms mayaddress scientific issues and produce advice,they will not produce scientific advice in a way

that is compliant with the principles. They blurthe distinction between the evidence on whichpolicy is based and the policy itself.

177 Policy-makers should make a reasonedchoice of advisory mechanism (that is, theyshould be able to give reasons for their choice,and that choice should be logical, justifiableand traceable). When making the choice, thepolicy-maker should take into account theneed to ensure the appropriate balance ofpure and applied science.

178 The mechanism chosen should beproportionate to the scale, complexity andimportance of the policy issue and to thescientific challenge. When making the choice,policy-makers may wish to characterise theissue according to indicators, such as:

• urgency—how much time is available fortaking advice before a decision isrequired;

• the magnitude of the risks that need tobe managed by a policy decision;

• the frequency with which problems of thesame type arise;

• the public perception of its importance;• the uncertainty in the science relevant to

the problem.

179 The cost of obtaining advice should be aconsideration when choosing the mechanism.When the problem has been characterised, itwill be possible to set a budget for scientificadvice. This should reflect the number ofadvisers that are needed, the experience (andhence authority) that they bring, and theduration of their deliberations. It may need tobe reviewed during the course of theinvestigation.

180 Some of the qualities of each advisorymechanism are summarised in Table 2overleaf and are explored further in thesupplementary notes below.

181 It may also be appropriate to use acombination of mechanisms—for example, ascientific advisory committee may let aconsultancy contract to explore a specificissue in more detail.

Recommendation: Policy-makers should make a reasoned choice of advisory mechanism,taking account of the characteristics of the issue.


26

Table 2: Characteristics of advisory mechanisms

Process Strengths Weaknesses

Scientificgeneralist

Can synthesise many scientific factors

Authoritative

Able to communicate with experts from otherdisciplines

Decisive, swift, inexpensive

Vulnerable to choice of individual

Lack of depth of expertise

Single expert Decisive, swift, inexpensive

Can be radical

Access to top expert

Vulnerable to choice of individual

Can be radical, hence high risk

Quality of advice depends on quality ofexpert

No due process

Credibility

Can be too specialised

Consultancycontract

Capacity—quantity of work, intensity ofeffort, support services

Incentive to do high-quality work and buildreputation

Knowledge of applied science

Can form teams with other consultants

Potentially high direct costs

May not be leading-edge science

May wish to please client

Experts may not cooperate

Profit incentive may cause skimping

Scientificadvisorycommittee

Allows many experts to contribute andachieve a balanced view

Internal debate

Good links to research community

Can carry perceived weight, especially ifstatutory

Can have due process

Not independent from its secretariat

Can be dominated by a strong personality(especially if chairperson)

May lack time and resources, so vulnerableto manipulation by support services

Needs tight brief to keep to task

Incentives for individuals to do high qualitywork are unclear


27

Choose the scientific adviser(s)

182 The selection of scientific advisers isdominated by two drivers:

• the need to obtain a high level ofexpertise; and

• the need to identify and manage biasand conflicts of interest.

183 The basic competencies that aredesirable in scientific advisers include:knowledge of the relevant scientific field,experience, ability to structure problems andanalysis, and time availability. Key individuals,such as committee chairpersons, will alsoneed managerial ability, and skill in thefacilitation of debates and communication.

184 In addition, it is desirable that advisersshould receive training and detailed guidance.The training should enable the adviser tounderstand the place of scientific advice withinthe legitimate decision-taking system, andshould help the adviser to present advice sothat it can be assimilated and understood. Theguidance should include a clear review of theterms of reference and duties of the adviser.

185 The criteria for selection of scientificadvisers depend on the advisory mechanismthat has been chosen. For example, where asingle expert is used, they must be free fromvested interests or bias, and must possess asuitably broad range of knowledge and

expertise. However, the members of acommittee do not need to have all thesequalities individually; collectively, thecommittee should be balanced.

186 It is important to maintain the distinctionbetween the functions of stakeholderrepresentatives and scientific advisersestablished by the legitimacy principle [104–109]. This does not exclude the possibility thata person may act as scientific adviser whilehaving some affiliation to a stakeholder.Indeed, this may be desirable because theycould have relevant knowledge and their viewsmay be used to balance other biases on acommittee.

187 Committees should have a balancedmembership reflecting all relevant schools ofthought, and each individual member must notact except in the role of adviser, and certainlymust not seek to promote any specialinterests.

188 Stakeholders should be invited to makerecommendations for, or comment on, thechoice of advisers. They might also nominateadvisers (but not representatives) if that isneeded to achieve the correct balance.

189 The selection of advisers for each of theadvisory mechanisms is explored in moredetail in the supplementary note, ‘TheSelection of Scientific Advisers’ [305–330].

Recommendation: Advisers should be selected for competence, and bias and conflicts ofinterest should be managed. Advisers affiliated to stakeholders are acceptable, and may bedesirable so that all schools of thought can be included. Advisers should receive trainingand detailed guidance.

Agree and confirm the brief

190 It is essential that the scientific adviserand the policy-maker have the sameunderstanding of the questions that are to beaddressed. This can be tested by the scientificadviser replaying the question in their ownterms. The questions must be stated explicitlyand agreed with the adviser.

191 The scientific adviser should be told thepolicy options and given an explanation of thepolicy background. The scientific adviser maybe asked to evaluate the consequences andlikelihood of all foreseeable outcomes of each

policy option. Even if it is not possible topresent the full range of policy options at apreliminary stage, it should be possible at alater iteration.

192 The distinction between analysingscientific issues and making value judgementsmust be made clear to the scientific adviser.

193 It may not be possible to finalise the briefat the start of the process, for example whengeneral advice is being sought or where datamust be considered before the scientific


28

question can be completely defined. The briefmay have to be developed iteratively.

194 The policy-maker should require thescientific adviser to consider and expressuncertainty. This should include:

• a realistic assessment of the robustnessof all conclusions;

• an candid description of conflictingtheories or evidence; and

• a description of new evidence, which, if itwere to emerge, would cause the adviserto change their advice.

195 If any of the outcomes of a policydecision might have serious consequences,the scientific adviser should be reminded notto exclude unlikely outcomes.

196 When briefing the scientific adviser, thepolicy-maker should make clear anyconstraints on the way in which they shouldwork—for example:

• the procedures to be followed tomaintain an audit trail of the advice;

• how uncertainty and risks should bepresented;

• whether meetings are to be open orclosed;

• when draft reports are to be prepared;and

• whether any draft scientific advice shouldbe published, or if there is to be anycommunication with the press or generalpublic while the work is in progress.

Recommendation: The brief for the adviser must state the scientific questions and should listthe policy options that are available. The terms of reference should make clear the scope ofthe advice that is sought, including the need to present uncertainty and the consequences ofpossible outcomes.

Prepare the scientific advice

197 Preparing scientific advice is not thesame as conducting scientific research. Thelatter is an iterative process of generatinghypotheses and testing them againstevidence. It depends on the intellectualcuriosity and creativity of the researchers,moderated by publication and peer review.

198 The preparation of scientific adviceinvolves the synthesis of known science withestimates of what is unknown or uncertain,directed towards the resolution of a policyissue. It depends on the ability of the advisersto appreciate the policy needs, and to reflectcandidly the full scope of scientific views, andnot just their own scientific position. Theadvisers may recommend, or even conduct,further research in the course of preparing theadvice, but the two activities should be clearlydistinguished.

199 Scientific advisers should maintain thehighest professional standards of scientificpractice when preparing their advice, but theyalso need to be aware that the preparation ofscientific advice is different from the processby which science itself is pursued. Theyshould, for example, follow good practice inassessing evidence. They should giveappropriate weight to partial (either incomplete

or biased) evidence, and be aware of thedangers of errors of omission, bias, andpreconception. Scientific advisers shouldcomment directly to the policy-maker,separately from their advice, on any scientificaspects of an issue that lie outside the termsof reference. This could be part of the iterationof the terms of reference.

200 At all times the scientific adviser shouldremain aware of the basis of the advice. Foursuch bases can be recognised, all of which arewithin the scope of scientific advice but whichmerit different levels of certainty.

• Observation: empirical evidence that isunambiguous and uncontentious,although it may still be open to differentinterpretations.

• Formal analysis: which should lead to aconsistent result, regardless of whoconducts the formal analysis.

• Reasoned judgement: the outcome of adisciplined approach to a problem,whereby deductions are made byextrapolation or extension from formalanalysis.


29

• Opinion: simply an assertion (ie, abelief), the value of which dependsentirely on the integrity, competence andcredibility of the party expressing theopinion.

201 An audit trail should be maintained,which makes clear how each element of theadvice was derived and on which of the fourbases it depends.

202 Scientific advisers should have asystematic procedure for identifying andhandling uncertainty. They should consider alltypes of uncertainty (inadequate data,incomplete theory, conflicting theory,limitations of theory) and should characterisetheir results accordingly. Sensitivity analysisshould be used to test how the advice mightchange if new evidence emerges, or if a theorythat has been used is incorrect. This shouldresult in a statement of what new evidence, if itwere to emerge, would require the advice tobe changed. Advisers should assess theconsequences of uncertainty for each possiblescenario, or each policy option.

203 Where more than one hypothesis iscredible, the scientific adviser must not decidewhich is the more probable and offer that asthe sole advice. The alternative hypothesesmust be presented, together with the evidenceand analysis, and the weight to be attached toeach. In addition, where there are gaps inscientific evidence, it should be indicated hownew evidence might support or contradict thehypotheses.

204 Dissenting advisers should beencouraged to present their viewsconstructively. Committee chairpersons shouldensure that those views are considered,accurately reported and given appropriate

weight. Minority reports should be accepted ifthe dissenting views cannot beaccommodated. It is not the place of scientificadvisers to achieve consensus where differingviews exist.

205 Where scientific advisers receiveadministrative support from the governmentdepartment that appointed them, they shouldnot be reluctant to call on assistance from thesecretariat.

206 Scientific advisers should ensure thatthey contribute the scientific elements of riskanalysis (such as quantification of thehazards). Scientific advisers should not applythe precautionary principle or any hiddensafety margin. They may wish to draw theattention of the policy-maker to precautionaryoptions, but the application of any valuejudgement by the scientific adviser, such as tothe degree of precaution that is appropriate,usurps the role of the decision-taker.

207 Even if the policy-maker indicates thatthe advice will not be published, the scientificadviser should:

• prepare the advice to a standard thatwould be acceptable for publication;

• work constructively to consider thecomments of stakeholders and otherswhom the policy-maker uses tochallenge and review the advice; and

• present their advice in a report or otherofficial communication to the policy-maker, and not in informal briefings;however, it may be necessary anddesirable for the adviser to explain theirreasoning in person to the policy-makeror decision-taker.

Recommendation: The scientific adviser should follow the good scientific practice of testinghypotheses, and should apply this to the full range of possible outcomes, explaining thebasis of advice offered. The adviser should maintain an audit trail of advice and shouldpresent advice, of publishable quality, undistorted by value judgements.


30

Communicate the advice

208 The primary audience is the policy-maker, and subsequently the decision-taker. Itis the responsibility of the scientific adviser toensure the clarity and comprehensibility of theadvice so that it may be fully understood bythe recipient of the advice. The scientificadviser must be prepared to explain the advicein person to the decision-taker.

209 This duty may be particularly demandingif the advice involves a large amount data ordifficult scientific principles. There may bemerit in non-scientists or non-specialistscientists working with the scientific specialiststo test the accessibility of the advice and tohelp with its presentation.

210 It is important that the scientific advice isreviewed, to challenge any assumptions,identify omissions or errors, and to engagewith the wider community. The policy-makermight publish a draft of the advice in sufficienttime to allow it to be modified before beingfinalised. If the model process described herehas been followed, that advice should not fallwithin the definition of ‘advice to Ministers’,and therefore there is a presumption ofpublication. For important issues, a pressannouncement that the draft is available mightbe made, including highlights of the advice.

211 There is a difference between publishinga considered draft and exposing all of the workin progress to public scrutiny. The latter maybe acceptable, but there are many cases

where public concerns would be unnecessarilyfuelled by early hypotheses that are laterrejected. Too much public exposure at anearly stage may cause scientific advisers to beexcessively cautious, and thus compromisethe quality of the advice.

212 Where scientific advisers are called on topresent and explain their advice to the generalpublic, they must restrict their commentsstrictly to science and the scientific advice.They should not ‘second-guess’ the decision-taker, nor should they seek to close off policyoptions. However, if the scientific advisercommunicates the advice properly to thepolicy-maker, the policy-maker should beperfectly able to communicate it subsequentlyto the decision-taker and thence to the public.

213 However, scientific advisers should notmoderate their advice in anticipation of publicreaction. Uncertainties and conflicting opinionsshould be identified explicitly and not hiddenunder false certainty.

214 All participants in the advisory system,particularly advisory committee chairpersonsand policy-makers, should be trained intechniques for posing questions and operatingthe various advisory mechanisms. The CivilService College should be encouraged to takea lead in developing training material andensuring that its courses for senior civilservants and Ministers include the handling ofscientific advice—and especially uncertainty.

Recommendation: The scientific adviser should present the advice in full, in such a way thatit will be fully understood by the policy-maker and decision-taker. If asked to present andexplain the scientific advice to the general public, the scientific adviser should restrict theircomments to science and the scientific advice.

Prepare advice on policy options

215 The policy-maker is responsible formanaging the process by which scientificadvice informs policy. Many other sources ofinformation and advice may have to be takeninto account, and the policy-maker is requiredto synthesise all of these into a set of policyoptions and supporting analysis to bepresented to the decision-taker. In particular,the policy-maker must expressly consider thevalues and demands of the public and thepolitical objectives of the government, as well

as expert advice from scientists and otherspecialists.

216 This process of synthesis is a key skill ofpolicy-makers and it would be inappropriate inthis report to produce general guidance onpolicy-making. Scientific advice is not differentin principle from any other input to the task ofpolicy-making. However, it may be that manyof the problems that have arisen with policiesfor which science is important have resultedfrom the attribution of undue weight to the


31

scientific advice, relative to the other inputs topolicy-making.

217 Many of the policies for which science isimportant involve risk, where policy-makersare responsible for risk assessment, buildingon the risk analysis provided by the advisers.One of the causes of risk is uncertainty, andthe level and cause of uncertainty should bemade clear.

218 The policy-maker must be able to makeuse of advice that carries explicit scientificuncertainty. In addition, there may be severalpossible outcomes of a given policy. It istherefore necessary to include in the scope ofthe synthesis process the identification andcharacterisation, for each policy option, of therange of possible outcomes. This will requirethe policy-maker to consider the other policyinputs as well as scientific advice.

219 Policy-makers are sensitive to thedanger that the way in which they conductsynthesis and present the results can have aprofound effect on the range of options that isoffered and the choice that the decision-takerwill make. They can help to guard against biasby consulting stakeholders and presenting tothem the range of options that are beingconsidered, the advice that has been receivedand its significance. This should help to

assess the likely outcomes of the options andensure that the full range of options has beenincluded.

220 If consultation or advice from otheradvisers identifies options that had not beenconsidered by the scientific adviser, it may benecessary to put them to the scientific adviserand seek further advice. Policy-makers ordecision-takers may also call for a more formal‘second opinion’.

221 When presenting policy options to thedecision-taker, the policy-maker should makeclear the implications and robustness of thescientific analysis. This might be achieved bycharacterising the policy options according to:

• expected and possible worst-caseoutcome;

• the degree of reversibility of the option;• sustainability—whether the option can be

sustained in the long term;• any precautionary arguments that may

be relevant;• the reliability of the key assumptions and

evidence; and• the source of the advice, and the weight

that should be attributed to that source.

Recommendation: The policy-maker is responsible for synthesising the advice fromscientific advisers, other advisers and more general inputs, such as the values and demandsof the public and the political objectives of the government. The synthesis should result in aset of options for the decision-taker, together with an assessment of the possibleconsequences of option, and their likelihood.

Take the decision

222 The decision-taker has a duty toconsider the scientific basis and all otheraspects of the policy options presented by thepolicy-maker. The decision-taker should notallow (or encourage) the scientific adviser toprescribe the range of policy optionsconsidered.

223 It is suggested that the decision-takershould:

• always consider the consequences of a‘do-nothing’, or ‘business-as-usual’ policyoption;

• consider all the policy options in terms oftheir costs and benefits, and risks andlikelihood of success;

• avoid the use of absolute statements (orquestions), such as ‘safe’, when it ismore appropriate to balance costs, risksand benefits;

• consider the reversibility of the options,and the evidence that might require adecision to be changed; and

• check for the consequences of false-positive and false-negative errors.

224 The decision-taker must understand theadvice and its uncertainties. The decision-taker must be satisfied that the scientificadviser understands the issues and that thepolicy-maker has correctly interpreted thescientific advice. If any doubts arise, thescientific adviser may be required to give a


32

briefing in person so that the decision-takercan gauge the credibility of the adviser and ofthe advice.

225 The decision-taker must at all timesensure that the decision is justifiable. There islikely to be strong public feeling that thereshould be no secrecy surrounding scientificadvice relating to risks to which the public maybe exposed. The decision-taker may wish tomake public the reasoning that led to thedecision.

226 It has been suggested by many peopleconsulted in the course of this study that therewould be great merit in encouraging decision-takers to make public the reasons for thedecision. The decision-taker has no legal dutyto give reasons for the decision, exceptthrough their accountability to Parliament, andmay, if reasons are given, be vulnerable tolegal challenge if those reasons areconsidered irrational. However, theadvantages of giving reasons are:

• scientific advisers, and their contribution,can be acknowledged when a decision isannounced;

• the way that scientific advice relates topolicy can be revealed, placing thedecision-taker in a stronger position todefend the decision;

• the decision is more likely to beacceptable to the public, especially whenit is contentious;

• the uncertainties can be made explicit—this helps to protect the decision-taker if

the policy has to be changedsubsequently, perhaps because newevidence is presented; and

• in the event that the decision is foundwith hindsight to be sub-optimal, it ispolitically much easier to reverse thedecision if the reasoning and advice onwhich it was based were known andhave since changed.

227 Presented below is a suggestion whichmight provide a helpful structure for the fullexplanation of a decision:

• the decision was about X;• this is important because of Y;• the options were a, b, c …;• advice was received from p, q, r …; they

were chosen because …;• their advice was …, and hence the

consequences and risks of each optionwere…;

• in reaching the decision, account wastaken of … (eg, international treatyobligations, long-term policies,implications for other areas of policy);

• the criteria adopted when consideringthe advice were …;

• where there was conflicting advice, pand q were accepted (and r rejected)because...; and

• hence the decision is … because …(summary of earlier points, values thatwere applied and weighting that wasgiven to conflicting arguments).

Recommendation: The decision-taker must take the decision, not allow it to be taken by thescientific advisers. The costs, risks and benefits of each option should be expresslyconsidered. The decision-taker must be satisfied that the scientific advice is sound, mustunderstand it, and must be able to explain, and defend, a reasonable decision. The decision-taker should consider stating advice and explaining the reasoning that led to the decision.

Audit and maintain the process

228 The decisions that this process isintended to inform are taken by, or in the nameof, Ministers. Therefore Ministers must beaccountable to Parliament for the decisionsand the scientific advisory system thatsupported them.

229 Three activities could contribute to themaintenance of the scientific advisory process:

• auditing the operation of the scientificadvisory system to ensure that it iscompliant with agreed guidelines;

• training those who provide, and thosewho use, scientific advice to ensure thatthey understand and fulfil their duties;and

• reviewing, adapting and maintaining theadvisory system to ensure that it remainsfit for purpose.


33

230 There are four checks on the quality ofscientific advice: quality assurance, peerreview, audit, and parliamentary scrutiny.

Quality assurance

231 During the preparation of advice by thescientific advisers, the policy-maker couldreview progress, to ensure that advisers arekeeping to the timetable and to their remit. Thepolicy-maker could also check more generallythat the scientific advisers are following goodpractice, consistent with the recommendationsin this study. This is a quality-assuranceexercise for the policy-maker. It assures thepolicy-maker that the final advice will meettheir quality requirements, and that theprocess by which it has been delivered isrobust.

Peer review

232 In many cases, the policy-maker maywish to consult to check that the advice theyhave received from the advisers is complete,and to provide a formal opportunity for thescientific community to comment. One way ofdoing this is to circulate the advice to otherexperts for peer review. The selection ofexperts might be drawn from the same list ofexperts from which the scientific advisers weredrawn. Peer review is used by most editors ofprofessional journals, especially researchjournals, to certify the quality of the scientificpapers. It is not a completely reliable method.Peer review cannot prevent falsification ofevidence, errors in analysing unpublisheddata, or poor experimental technique. Wherepeer review fails, according to the opinionsexpressed to this study, it is commonlybecause the reviewer has not understood thepaper, or the author of the paper has beenmisleading in their claims. There arenumerous examples where peer-reviewedpublished research was later discredited.

233 In considering the value of peer reviewfor checking the quality of scientific advice, thedifference between scientific research andscientific advice is important. Scientific adviceanalyses policy options, and does not alwaysinvolve original research, and may includeexpert judgements and opinions. It is moreapplied, and may be more subjective thanscientific research submitted for publication. Itmay also be more forthright in its considerationof uncertainty and less stylised in its use oflanguage. As a result, scientists acting as peerreviewers may not be effective in their reviewunless directed to particular aspects of theadvice. For example, they might be asked toexamine the adviser’s survey of evidence andcomment on whether any evidence has been

omitted. They might also be asked to considerthe set of hypotheses considered by theadviser and comment on whether it iscomplete. The reviewer should be given theterms of reference of the scientific adviser sothat they can place their comments in context.

234 In interpreting the reviewer’s report, thepolicy-maker should bear in mind that thepurpose of the peer review is to widen therange of possible advice to be considered,rather than constrain it. This appliesparticularly where an adviser has adoptedmore unconventional thinking, which is out ofline with the scientific consensus. There havebeen many cases where new scientificthinking has emerged slowly, partly because itis incompatible with the previous scientificconsensus. The policy-maker should thereforereflect on the fact that the peer-review processhas a tendency to reinforce the currentscientific consensus.

235 Peer review is only a weak assurance ofthe quality of advice. This is because, althoughthe reviewers may be accomplished, the timedevoted to peer review is short, and peerreview in scientific journals does not have areliable record of screening out poor-qualitywork. Furthermore, peer review is appropriatefor science research but less useful forscientific advice. Peer review of advice doesnot appear to be common practice in the UK.

Audit

236 To ensure that the scientific advisorysystem, including the communication ofadvice, remains fit for purpose, it isrecommended that departmental chiefscientists should arrange for it to beperiodically and independently audited andreviewed. The results of these audits andreviews, and any changes to the system,should be reported to the department’smanagement board and to the OST.

Parliamentary scrutiny

237 Ministers may be called to account fortheir decisions before Parliament. Parliamentoften exercises this right, but interviewees inthis study commented that decisions are notalways explained well. Parliamentary scrutinydoes not provide a routine challenge todecisions because it is usually only usedwhere a decision is disputed or an adverseevent triggers the review of previousdecisions. Ministers are under an obligation toexplain to Parliament how they reachdecisions based on the information presentedto them, but in practice, where a statement ismade at all, it is to show that there was


34

evidence to support the chosen policy. This isusually done by explaining the links betweenthat policy and some evidence collected in thepolicy-making and advisory process. It doesnot always include: discussion of alternativepolicy options that were rejected; the level ofuncertainty in the advice; or the balancing ofinterests and values which support theselection of the chosen option rather than anyof the rejected options. There is strong support

for the production of a ‘reasoned opinion’ thatdoes all of these things. It could become themain output of the scientific advisory andpolicy process, [226–227]. The reasonedopinion need not be a chronological account ofthe scientific advice and policy process, butwould, instead, explain the policy options andhow they related to the evidence available tothe Minister.

Recommendation: Government should establish and operate a system for auditing andmaintaining the scientific advisory system, and for training those who work within it.

|O|X|E|R|A| Supplementary Notes

35

SUPPLEMENTARY NOTES238 The following sections containsupplementary notes that discuss in moredetail some of the themes that have emergedduring this study, and include detailedrecommendations for some aspects of themodel process.

Rights and Duties of ScientificAdvisers239 The rights and duties proposed in thissection could be incorporated into the terms ofreference of advisers, or into a code ofconduct.

Rights

240 Scientific advisers should have a right to:

• be given terms of reference which set aquestion that is amenable to a scientificanswer and that does not demandbroader policy questions to beaddressed by the adviser;

• be heard by other members of theadvisory team or committee;

• submit a minority or dissenting report if,as a member of a committee, they areunable to reach a common position withthe other members of the committee;

• be allocated sufficient resources (bothtime to prepare their advice andadequate support from a secretariat) toensure that they can answer thequestion satisfactorily and fulfil theirduties. This does not allow advisers tocall for indefinite time or other resources,but does allow them to decline to offeradvice when, in their opinion, insufficientresources have been made available toallow adequate advice to be preparedand delivered; and

• be allowed to record a personalstatement which sets in context theirdeclarations of interest and theirunderstanding and acceptance of theirduties.

Duties

241 Scientific advisers should a duty to:

• act with integrity;

• put the interests of the public above anyother, including personal interests, theinterests of their employer, or theinterests of the person whom they areadvising;

• disclose, to the person proposing toappoint them, any interest which mightbe seen to have the potential to distorttheir judgement [242];

• disclose, to the person proposing toappoint them, any prior scientific positionwhich could be seen as potentiallycausing the adviser to prejudge theissue;

• conduct their advisory work in theexpectation that the findings might beplaced in the public domain, and in sucha way that they would be understood bypolicy-makers and the advisers’ peers;

• present and explain the advice to policy-makers and decision-takers if asked;

• identify any sources of uncertainty in theadvice being given and assess theirsignificance;

• state what new evidence, if it were to befound, might require the advice to bechanged;

• avoid any methodology which introducesbias to the scientific advice. Forexample, a clear statement ofuncertainties is more transparent thanappeal to the precautionary principle,which should only be applied by policy-makers during policy synthesis;

• act in an efficient manner, within thetimescale agreed with the policy-maker;

• keep confidential any aspects of theissue, if so required by their brief,including the fact that they are advising,although the principles demand of thepolicy-maker the presumption that theadvisory process will be conductedopenly;

• recognise that the questions for whichscientific advice is sought are different inkind from the type of questions that theyface as practising scientists. Scientificadvice may have to address questions


36

which are not amenable to certainanswer;

• use good scientific techniques ofobservation and logical deduction, butnot exclude inconclusive evidence orunproven hypotheses, which might berelevant; and

• express their advice in line with theterms of reference.

242 The interests to be disclosed by advisersmight include:

• membership of professional bodies andinstitutions;

• membership of political or charitableorganisations;

• directorships;• material shareholdings;• their employers and those of their close

relatives;• previous employers;• sources of funding for research activities;• advisory and other appointments;• journal editorships; and• a full list of publications.

243 The adviser should then be askedwhether they expect that any of these interestsmight be compromised or enhanced by thefindings of their advice. They should also beasked whether any of their interests wouldinfluence the way in which they produce theiradvice and the judgements that they may haveto make about the scientific issues.

244 The policy-maker should then make ajudgement as to whether the advice might be,or might be seen to be, compromised by theadviser’s interests. In making this judgement,the policy-maker might consider the interestsdeclared in the light of any statement that theadviser has made setting their interests incontext of their experience.

Additional duties of chairpersons

245 Chairpersons of advisory committeeshave further duties. They should:

• ensure that every member of thecommittee is heard and that no view isignored or overlooked, using, ifappropriate, a structured process whichensures that all views are captured andexplored;

• ensure that advisers consider not onlythe most likely outcome but also lesslikely possible outcomes, if these mighthave severe consequences;

• refrain from forcing closure, preferring tokeep open all possible hypotheses toensure that any diversity of opinionamong the members of the committee isaccurately reflected in the advice; and

• be responsible for ensuring that thecommittee acts in accordance withavailable guidelines.


37

Legal Issues

Constitutional background

246 The recommendations in this report areintended to apply to decisions that are takenby government. The constitutional basis ofgovernment must therefore form their startingpoint and they must be consistent with theconstitution. In particular, the central principleof these recommendations, which is thedistinction between the functions of decision-taker, policy-maker and scientific adviser,derives directly from constitutional principles.

247 Although there is no single documentthat contains the British constitution, statutes,court precedents, conventions and other textsset out rules which govern the behaviour of UKgovernments. It is inevitable that these leaveareas that are unclear. This is both an asset,because it allows the constitution to respondand evolve, and a threat, because it leavesroom for abuse. Accordingly, it is mostimportant that all those engaged ingovernment respect and consider theconstitutional basis of their actions carefully,since there are no rigid and prescriptive rulesto guide them.

248 The first principle is the supremacy ofParliament. In practice, this may becircumscribed by practical realities (forexample, an Act which rejected EU legislationwould be legal in the UK but might have suchgrave economic consequences thatParliament would not pass it). However, allother forms of authority derive from, and aresubject to, the Acts of Parliament. Thisincludes actions by, or in the name of, theCrown, where royal prerogative has beenpermitted by Parliament to remain.

249 Ministers act in the name of the Crownbut, given the supremacy of Parliament, areconstrained to act within the law. This requiresthem to take decisions which are within theirlawful powers. They are also required to actrationally. A further requirement is that theymust not fetter their powers by taking adecision which prejudges a subsequentdecision or pre-empts legitimate options. It isnot legitimate to take a decision using onepower that pre-empts a decision that is yet tobe taken under another power (William Cory vCity of London Corporation, 1951). SimilarlyMinisters may not delegate their decisionswithout legitimate authority.

250 Ministers are accountable to Parliamentfor their actions. Ministerial practice has been

codified as the Ministerial Code, the relevantparts of which are quoted below.

251 Ministers are served by civil servants.The concept of the professional service wasfirst developed in 1854 in the Northcote andTrevelyan Report, and put into practice byGladstone in 1870. Its principles were restated(and its text reproduced) in the Fulton Reportof 1968 (Fulton, 1968). Civil servants aredefined (Tomlin, 1931) as:

servants of the Crown, other thanholders of political or judicial office,who are employed in a civil capacityand whose remuneration is paid whollyand directly out of moneys voted byParliament

252 Ministers may legitimately delegateadministrative decisions to civil servants(Carltona Ltd v Commissioner of Works,1943). Constitutional law recognises that it isnot feasible for the Minister to take everydecision personally and that civil servants maytake decisions in the Minister’s name. Whenthey do so, they are constrained by law as ifthe decision had been taken by the Minister,and are subject to the same constitutionalconstraints.

253 The distinction of functions emergesnaturally from these principles. The practicalimplications of this constitutional argument areset out below in terms of the duties imposedon decision-takers and policy-makers.

Duties of decision-takers

254 The decision-taker (whether a Minister ora legitimately delegated civil servant) has aduty to take decisions that satisfy the law. Thisis not as self-evident or trivial as it might seem.With regard to the substance of the decision,the decision-taker is only answerable toParliament. However, the process by whichthe decision is taken is subject to challenge inthe courts, by way of judicial review. Therecommendations in this report have beendesigned to define a process that, if followedrigorously, would be robust against challenge.

255 The circumstances under which thedecision-taking process can be challenged byseeking judicial review were summarised byLord Diplock in 1985 (Council of Civil ServiceUnions v Minister for the Civil Service, 1985):


38

Judicial review has I think developed toa state today when … one canconveniently classify under threeheads the grounds on whichadministrative action is subject tocontrol by judicial review. The firstground I would call ‘illegality’, thesecond ‘irrationality’ and the third‘procedural impropriety’. …

By illegality as a ground for judicialreview I mean that the decision takermust understand correctly the law thatregulates his decision taking powerand must give effect to it. …

By irrationality I mean what can now besuccinctly referred to as ‘Wednesburyunreasonableness’. It applies to adecision which is so outrageous that nosensible person who had applied hismind to the question to be decidedcould have arrived at it. …

I have described the third head as‘procedural impropriety’ rather thanfailure to observe the rules of naturaljustice or failure to act with proceduralfairness towards the person who will beaffected by the decision. This isbecause susceptibility to judicial reviewunder this head covers also failure byan administrative tribunal to observeprocedural rules that are expressly laiddown in the legislative instruments bywhich its jurisdiction is conferred, evenwhere such a failure does not involveany denial of natural justice.

256 For a decision to be legal, amongst otherrequirements:

• a decision-taker must take into accountall relevant considerations and disregardall irrelevant considerations. Thisrequires the decision-taker to seekappropriate scientific advice and toconsider it when reaching the decision;

• a decision-taker must not fetter theirdiscretion, for example by makingcontractual commitments whichpredetermine subsequent decisions; and

• a decision-taker to whom the exercise ofa discretion has been entrusted must notdelegate the exercise of that discretion toanother unless clearly authorised to doso. This delegation may be neitherexpress nor implied, nor may it be toanother part of government (Lavender vMinister of Housing and LocalGovernment, 1970), but it may be to a

civil servant (Carltona v Commissionersof Works, 1943).

257 All of these can be illustrated withexamples of illegal behaviour in seekingscientific advice. If the decision-taker hasdemanded of scientific advisers a dichotomouschoice (for example, by asking them aquestion such as ‘is it safe?’) where thescience is uncertain, and therefore hasignored the uncertainty in the answer, they willhave failed to take account of a relevantconsideration. By making a pre-emptivedecision before obtaining considered scientificadvice, the decision-taker may have effectivelyfettered their discretion to take an informeddecision in the future. By asking (or allowing)the scientific advisers to apply theprecautionary principle, the decision-taker hasdelegated the decision. All of these givegrounds for challenge.

258 The need to avoid the charge ofirrationality is often interpreted as requiring adecision-taker to take a decision that lieswithin the spectrum of possible reasonabledecisions. Provided that it lies within thatspectrum, it cannot be challenged on theground of irrationality. This ground, whichoverlaps with the first ground of illegality, isoften cited where the decision-taker hasintroduced irrelevant considerations (such aswider political or moral objectives, as in thedecision to grant aid to the Pergau damproject).

259 Procedural impropriety is particularlyconcerned with decisions that are unfair, forexample because an affected party was notgiven an opportunity to set out a case, orwhere there is ‘a real danger of bias’ (R vGough, 1993). It also applies where adecision-taker has failed to follow a statutoryprocedure, for example by not consultingappropriately. It is easy to imagine how, forexample, the manipulation of the membershipor terms of reference of an advisory committeeto achieve a desired result could amount toprocedural impropriety.

260 Ministers are also subject to theMinisterial Code:

Ministers have a duty to give fairconsideration and due weight toinformed and impartial advice from civilservants, as well as to otherconsiderations and advice, in reachingpolicy decisions; a duty to uphold thepolitical impartiality of the Civil Service,and not to ask civil servants to act inany way which would conflict with the


39

Civil Service Code; a duty to ensurethat influence over appointments is notabused for partisan purposes; and aduty to observe the obligations of agood employer with regard to termsand conditions of those who servethem.

261 The first part of this gives effect to therequirements of a legal decision, which isimmune from challenge by way of judicialreview.

Duties of policy-makers

262 Policy-makers are usually civil servantsor, if not, act subject to the Civil Service Code(Cabinet Office, 1999a) (which, for example,applies to special advisers). The code states:

Civil servants should serve theirAdministration in accordance with theprinciples set out in this Code andrecognising:

• the accountability of civil servants tothe Minister or, as the case may be,to the Assembly Secretaries and theNational Assembly as a body or tothe office holder in charge of theirdepartment;

• the duty of all public officers todischarge public functionsreasonably and according to thelaw;

• the duty to comply with the law,including international law and treatyobligations, and to uphold theadministration of justice; and

• ethical standards governingparticular professions.

Civil servants should conductthemselves with integrity, impartialityand honesty. They should give honestand impartial advice to the Minister or,as the case may be, to the AssemblySecretaries and the National Assemblyas a body or to the office holder incharge of their department, without fearor favour, and make all informationrelevant to a decision available tothem. They should not deceive orknowingly mislead Ministers,Parliament, the National Assembly orthe public.

263 Where civil servants are legitimatelydelegated to take a decision in the name of theMinister, that decision is immune to legalchallenge other than by way of judicial review,

and the heads listed above apply to theprocess.

Implications for scientific advisers

264 A list of rights and duties of scientificadvisers is set out in the preceding section ofthis report [239–245]. The two key duties arebased on the legal considerations above.

265 There are no established codes ofconduct or rules of procedure for expertadvisers to government. However, there arerules for expert witnesses in courtproceedings, which could provide a guide.These were explained by Lord Justice Ottan(The Ikarian Reefer, 1993):

Expert evidence presented to the courtshould be, and should be seen to be,the independent product of the expertuninfluenced by the exigencies of thelitigation.

An expert witness should provideindependent assistance to the court byway of objective unbiased opinion inrelation to matters within his expertise.An expert witness in the High Courtshould never assume the role of anadvocate.

An expert witness should state thefacts or assumptions upon which hisopinion is based. He should not omit toconsider material facts, which coulddetract from his concluded opinion.

An expert witness should make it clearwhen a particular question or issuefalls outside his expertise.

266 The core principle is codified by thecourts in the Civil Procedure Rule 35.3:

(1) It is the duty of an expert to help thecourt on the matters within hisexpertise.

(2) This duty overrides any obligationto the person from whom he hasreceived instructions or by whom he ispaid.

267 Expert witnesses are granted immunityfrom any civil action arising from anything thatthey say in court or write in their preparationsfor court. This is intended to give them thefreedom to act fearlessly and honestly, withoutthe fear of actions for, for example,negligence, if they feel obliged to make astatement which is not in their client’s interest.


40

268 By analogy with the rights and duties ofexpert witnesses, it is recommended that:

• scientific advisers should have an over-riding duty to the public interest, whichtakes precedence over their duty to theirclient (who has sought their advice) andto any employer;

• scientific advisers to government shouldbe protected as far as possible from anycivil action arising from their work asadvisers brought by a third party, or bythe government department whichappointed them.

269 During the research leading to thisreport, some scientific advisers stated thatthey had felt constrained by the risk of legalaction in the event that their advice wassubsequently found to be wrong (through nofault of the adviser), or were to be misused ormisinterpreted. An incomplete solution to thisproblem would be for the government toindemnify advisers and underwrite legal costs.However, this would not protect advisers’reputations or personal costs when facing aprosecution. Even a small likelihood ofincurring these large personal costs may detersome experts from advising.

270 An indemnity might not protect anadviser who wished to retreat (resile) fromprevious advice without being accused ofnegligence. This might discourage advisersfrom changing their advice when new andbetter information became available.

271 It could at least be made clear toadvisers what their position is with respect toimmunity. The Biocides ConsultativeCommittee model might be adopted (BiocidesConsultative Committee Information Pack):

The government has indicated that anindividual committee member who hasacted honestly and in good faith will nothave to meet out of his or her personalresources any personal civil liabilitywhich is incurred in the execution oftheir committee function, save wherethe member has acted recklessly.


41

Decision-taking under Risk andUncertainty272 This supplementary note presents adiscussion of a number of related conceptsand issues: policy options; scenarios,consequences and outcomes; uncertainty;risk; and the optimisation of decision-makingunder uncertainty. The concepts presentedhere originate from risk-management practiceand are particularly relevant to the integrationof scientific advice, with its uncertainties, intothe decision-taking process.

273 All policy decisions involve considerationof multiple factors, such as price effects,environmental harm, ill-health orunemployment. Professional policy-makers inthe civil service are expert in handling thesemulti-faceted problems and in presentingoptions to decision-takers. Ministersthemselves are used to considering thebalance that often has to be struck betweenvarious interests and concerns. If all theconsequences of a decision were perfectlypredictable, the decision would strike abalance between the interests of the affectedstakeholders, and there would be nouncertainty—even though there might bewinners and losers.

274 However, some of these policy-makingproblems involve an additional factor—risk.This term is used to refer to situations where,for example, the consequences of a policydecision are not certain. It expresses both thelikelihood of a harm or loss, and the severity ofsuch consequences. This supplementary noteaddresses the extensions to existing policy-making approaches that are necessary for theproper handling of risk.

275 Where there is risk, there is a need toconsider the alternative scenarios that mightflow from each policy option. Each scenario isa possible future—some being more likelythan others—and a likelihood can beexpressed for each scenario. In addition, eachoption has costs and benefits for stakeholders.

276 Awareness of a risk allows a risk-management response: for example,monitoring of events as the impact of thechosen policy starts to be felt, and correctiveaction if an unwanted scenario appears to beunfolding.

277 Uncertainty will also affect many of thenon-scientific inputs to the policy decision.However, the scientific component may be asignificant source of uncertainty in problems

involving novel technologies, health andenvironmental risks.

278 Therefore, policy-makers need topossess, or acquire, expertise in assessing theuncertainty in each piece of scientific advicethey use, and in assessing the correspondingrisks to stakeholder interests.

279 Decision-takers are reputed to bereluctant to confront uncertainty, and unwillingto explain decisions that admit any risk.However, the case studies have shown thatuncertainty is an irreducible feature of manypolicy problems. Moreover, the decision-takerswho participated in this study asserted thatthey wished to be advised of any scientificuncertainties, and that they were willing to takethese into account in their decisions.

Policy options

280 Policy options should span the range offreedom within which the decision-taker couldreasonably exercise discretion, otherwise thedecision could be open to challenge.

281 Because of scientific and otheruncertainties, the ultimate consequences ofchoosing one policy option today will dependon which scenario actually unfoldssubsequently. In general, each policy optionwill have several possible scenarios, and themerits of the options will depend on the spreadand valuation of all possible future scenarios.

282 Where major risks are present, it isespecially important to consider a wide rangeof options. This is because the multiplicity ofscenarios stemming from each option makes itmore difficult to determine, early in theassessment, whether one particular optionshould be eliminated from the frame.Therefore, options should be preserved andexamined well into the assessment, and notdiscarded until the risks have beenestablished.

283 Any list of policy options should alwaysbe kept open to additional entries and shouldnot circumscribe the scientific advice.

284 Uncertainty may affect scientific advicein several ways, including:

• experimental observations, subject toinstrument error;


42

• scientific models that work well underlaboratory conditions, but do not explainnatural phenomena, or requireextrapolation beyond the region in whichthey have been tested;

• competing theories that yield differentpredictions;

• ignorance—of an important factor; and• unproven assumptions in the analysis.

285 Although science is founded on arigorous discipline of experiment andreasoning, scientists can only offer anapproximate model of reality, whose predictiveability depends on the proximity between thepolicy problem under consideration and theapplications in which the theories have beentested.

286 Uncertainties also arise fromrandomness or chaos in the real world.

287 Where a risk arises from scientificuncertainty, it may be reduced by pilot studies,monitoring programmes, or other furtherresearch.

Option evaluation: scenarios,consequences and outcomes

288 A proper evaluation of the merits of asingle policy option requires consideration ofthe events that may occur subsequently, orphenomena that may come to light. A singlescenario is defined as a sequence of suchevents. The events comprising each scenariomay have probabilities attached to them, andthe combination of these will be the probabilityof the entire scenario.

289 Each scenario will therefore bedescribed by at least a probability and aconsequence.

290 The challenge in addressing risk-bearingscenarios in this way is that the amount ofinformation to be considered by the decision-taker is large.

291 To limit the volume of informationgenerated, the dominant characteristics ofeach of the options should be identified andthe appraisal effort concentrated on thesefeatures. For example, the expected (ie, mostlikely) outcome should be included, as shouldany outcome with severe consequences, sothat the precautionary principle can beproperly applied by the decision-taker. In somecircumstances, it may only be possible toidentify the dominant characteristics after fairlydetailed investigation.

292 The brief to the scientific advisers shouldbe expressed in terms such as the following.

• Consider the following three (forexample) policy options.

• Identify and evaluate the consequencesthat might flow from each option.

• Consider the robustness of the scientifictheories and data (includingassumptions, external events, naturalvariations) on which the models chosenare based. Identify what variations ofview exist among scientists expert in thefield, and what the implications would befor the assessment if any of thesevariations turned out to be well founded.

• Against each combination, evaluate theconsequences and likelihood of itsoccurrence, given that the policy optionin question had been selected.

293 This form of brief embodies an importantchange of perspective: the scientific adviser iscalled upon not only to advise on how theybelieve the world to be, but also on the state ofscientific knowledge.

294 The policy-maker would have to extract,from all the advice tendered, the principaldifferences between the options. Policy-makers should receive formal training inanalysing uncertainty and risk.

295 It is inadvisable to reduce theconsequences into single indices of merit foreach option. Indexing will average thevariation in outcomes and thus conceal lowrisks of severe consequences. For example,an average expressing the expected value oflosses associated with the option does notconvey any information about the maximumpossible loss.

296 It is therefore not appropriate to usesolely the customary measure of risk (theprobability of occurrence of an adverse eventmultiplied by its cost) in the context of policydecisions.

297 The characteristics of options that relateto risk and uncertainty must be described. Forexample, suppose that a particular technologyproposal carried some uncertainty concerninghealth risks. There might be three scenariosflowing from a decision to allow the proposal toproceed:

• no health effects will occur;• health effects will occur, but will be

detected by early monitoring and theprogramme is abandoned; or


43

• health effects will not be detected untilthey are widespread and irreversible.

If there is no great confidence that the thirdscenario can be eliminated then the policy-maker could advise that health effects ofepidemic proportion are a worst-caseconsequence of any decision to proceed.Otherwise, the worst case could be the secondscenario.

298 Decision-takers have a duty to considerthe worst-case scenario.

299 If it is necessary for this complexity to bereduced then the policy-maker should judgewhen the probability of a scenario is so low inrelation to the consequences that the scenarioneed not be considered. This judgementshould be consistent across all policy areas,and is tantamount to a risk-acceptabilitycriterion.

Taking decisions

300 Once the options have been assessed,the conclusions will be presented to thedecision-taker.

301 Certain decision-taking criteria may beapplied thereafter:

• optimisation—maximising the overallexpected benefits;

• sustainability—keeping future optionsopen, avoiding irreversible actions;

• equity—fair treatment of all stakeholders;• the precautionary principle—ie, where

there is uncertainty and there are majorpotential harms, decision-takers shouldact with caution.

302 These recommendations have beendesigned to ensure that decisions are rationaland justifiable. Good decisions are those that:

• are well balanced between risks andrewards;

• are equitable;• are sustainable over time;• keep future options open;• carry little or no chance of catastrophic

outcomes.

303 It is important that the precautionaryprinciple is only applied once—at the end ofthe entire process. If precaution is applied atintermediate stages in a chain of reasoningthen the final conclusion may be far morecautious than the decision-taker intended.

304 The EC has recently issued aCommunication on the precautionary principle(Commission of the European Union, 2000),and the practice recommended here is in linewith this.


44

The Selection of Scientific Advisers305 The criteria for the selection of scientificadvisers depends on the advisory mechanismchosen, because each mechanism usesdifferent methods for securing adequatemanagement of bias and uncertainty. Theseare discussed separately under the sub-headings below.

Scientific generalist

306 This class of scientific adviser is typicallyused when the issue calls for a synthesis ofvarious scientific aspects that interact strongly.The latter type of expert should possess:

• ability to comprehend the scientificknowledge in the relevant fields;

• knowledge of the spread of competentopinion in the fields, and willingness toaddress the associated uncertainty;

• intellectual rigour and analytical skills—particularly in the synthesis ofinterdisciplinary problems;

• personal credibility related toprofessional standing, open-mindednessand freedom from attachment toparticular schools of thought; and

• communication skills.

307 The scientific generalist is a class ofscientific adviser that has been little used inrelation to scientific advice for policy on riskissues. Their value in this context arises fromthe fact that most issues of risk to public orenvironmental health are inter-disciplinaryproblems, and call for a degree ofunderstanding not only of scientificmechanisms but also of more general matters.An important factor is the evaluation of theconsequences of policy options, where theidentification of potential risks calls for athought process more common in thegeneralist than in the specialist.

Single expert

308 This class of scientific adviser is typicallyused when there is a need for expertunderstanding of a specialist subject, or wheresome element of a wider policy issue isreducible to a highly specific scientificquestion.

309 The particular qualities sought in such anexpert are:

• leading expertise in the requiredspecialism;

• familiarity with the state of scientificknowledge in the field, and with therange of opinion held by competentexperts;

• personal credibility related toprofessional standing, open-mindednessand freedom from attachment toparticular schools of thought; and

• willingness and ability to address theuncertainty.

310 The selection of the single expert mustbe closely allied to the definition of thescientific question, as the expert should not beplaced under any temptation to stray beyondtheir field of competence, to apply valuejudgements, or to advise on policy aspects.

311 It is the duty of the policy-maker toprovide the necessary scrutiny and challengeto single experts. They must be able torecognise when a single expert has not fullyjustified their opinion, or has wrongly assessedthe uncertainty. If necessary, a secondspecialist expert may need to be selected.

Consultancy

312 In addressing consultancies, the authorsof this report declare a conflict of interest,being themselves consultants. The followingparagraphs are informed by their experienceof acting as consultants to a wide range ofclients under a variety of tendering andcontractual frameworks, and from observationof their clients.

313 In this context, ‘consultancy’ may be anyorganisation, whether academic, commercialor publicly owned, that sells the time of itsscientific staff. Consultancies includeuniversities, research institutes, commercialconsultancies, and government researchlaboratories. All of these types of organisationare similar, in that they depend on contracts tomaintain their staffing levels, and this givesthem a vested interest which must be takeninto account in their selection and in evaluatingtheir advice.

314 It makes little difference whether theorganisation is in the public or private sector.However, some organisations may havecharters that specifically enshrine professionalprinciples, such as publication rights orrequirements of non-interference of clients inscientific judgements. These policies shouldbe examined before any consultancy isengaged to provide scientific advice togovernment.


45

315 Consultancies are suitable where thereis a scientific problem that calls for a largeamount of research or investigation, orparticular facilities, beyond the resources ofunpaid advisers or government scientists.Therefore the capacity to deliver theseservices is an important criterion for selection.

316 A second, equally important, criterion isquality. There is tension between cost andquality in this area. There are consultancieswhose commercial strategy is to win contractson price, but which only provide a basic levelof service. Often, their clients want this level ofservice. Even a top-class team can producepoor work if they are under-funded. Neithercase would meet the needs of government inprocuring scientific advice on a matter of highpublic interest.

317 Therefore, the selection of a consultancyto provide scientific advice is closely allied tothe question of funding. There are twoapproaches that can work well. The first is bypre-qualification of a shortlist of candidateorganisations on capacity and quality ofpersonnel but not on price, followed by acompetitive tender process. The second is forthe client government body to declare itsbudget for the work and seek the besttechnical proposal.

318 The latter option is especially appropriatewhen the scope and depth of the work is opento widely varying interpretations, and bidderswould be guessing what level of effort isproportionate to the policy issue at stake. Italso has the considerable advantage thatbidders are not tempted to pare down theirprice to levels that would make it difficult forthem to deliver a satisfactory piece of work.They are able to offer their best proposal forthe fixed price, and hence this practiceencourages selection of the most competentteam. The expenditure by the client, moreover,will reflect their judgement of proportionality tothe importance of the problem.

319 In evaluating offers, the client bodyshould satisfy itself that the consultant will actin accordance with available guidelines forproviding scientific advice to government.

Advisory committee

320 This section covers standing committeesand ad hoc committees formed for a specificpolicy-making purpose.

321 There are two types of committee thatcurrently handle scientific advice in the UK.First, the scientific advisory committee, whichis comprised of experts (who may have

stakeholder connections, usually in balance).Second, a policy advisory (or even decision-taking) committee, which is a pure stakeholdercommittee, and act as a negotiating forum.

322 The policy advisory committee shouldnot be considered a source of scientific advice,but it often handles such advice and frequentlyhas a policy-making or even a decision-takingfunction. It should not offer scientific advice.Instead, it should obtain such advice fromanother source, such as a scientific sub-committee—the Advisory Committee on ToxicSubstances (ACTS) and its technical sub-committee, the Working Group for theAssessment of Toxic Chemicals (WATCH), isan example. The following paragraphs discussthe selection of members of scientific advisorycommittees.

323 The exclusion of applicants withaffiliations or interests has sometimes servedwell when the issue has a high profile and isso contentious that an extremely high degreeof independence is required. However, itcarries the disadvantage of severely restrictingthe field of candidates, and possibly excludingthose whose views would aid the evaluation ofthe science and its uncertainty. Often thedeclared interest that would cause a candidateto be rejected is trivial, and represents only asmall proportion of the individual’s overall workand sources of income. Accepted candidatesmay have other, more subtle, biases.Increasingly, practitioners in this area arecoming to realise that there is no such thing ascomplete independence.

324 This leads to the concept of the balancedcommittee in which individuals withstakeholder affiliations are not excluded, butthe overall balance of competing influences iscarefully maintained. This format has the twinadvantages of allowing the full diversity ofscientific opinion to be expressed, which isgenerally welcomed by stakeholders (includingindustry and NGOs), and assisting theevaluation of uncertainty and the possibleadverse consequences of policy decisionssubject to risk.

325 The members of scientific committees,whether they have stakeholder affiliations orotherwise, would still owe their primary duty tothe broad public interest, and the chairpersonis responsible for ensuring that they conductthemselves accordingly. As individuals, theywould be allowed to express their sincerelyheld views and their personal biases, but notto act consciously as negotiators on behalf oftheir stakeholder interests.


46

326 In the selection of individuals for suchpositions, it should be borne in mind that manyexperts are engaged by stakeholders on thebasis of their reputation for probity andindependence, or because the stakeholderwants an internal ‘devil’s advocate’. There istherefore a marked difference between, forexample, a lifetime’s employment by acompany and being hired by that company asa consultant. The former could build in deepand unconscious prejudices, while the lattermight generate conflicts of interests. The realnature of stakeholder affiliations is therefore asubtle matter, and the selection processshould not only list all affiliations, but alsoexamine the professional relationship betweenthe expert and any affiliated organisation.

327 It is particularly important that thechairperson of a scientific advisory committeeshould be free from conflicts of interest.

Government scientists

328 There are many circumstances in whichthe best scientific adviser under any of thesefour mechanisms may be one employed withinthe civil service. One common situation iswhere a government body has to deal with alarger number of small cases every year, andhas developed specific expertise for thatpurpose that cannot be obtained elsewhere.

329 Another example is where there is noother academic or commercial motive forcreating a centre of expertise. Equally, thegovernment may wish to create a centre ofexpertise that is free from commercial orstakeholder influence.

330 It may, however, be necessary toconsider whether government itself (or adepartment of government) has a stakeholderinterest in an issue through havingresponsibility for policy on the issue.Government scientists should be consideredas candidate advisers on a par with externaladvisers, and subject to the same rights andduties (recognising that they are also subjectto the Civil Service Code). It may beundesirable for the Chairperson of a scientificadvisory committee to be a civil servant fromthe department responsible for policy in thearea on which advice is being sought.


47

Scientific Advice in an Emergency331 Disasters are unpredictable in theirtiming, and action must often be taken veryquickly if loss of life, impairment of health ordamage to the environment is to be minimised.It is important that the actions taken areeffective and that they do not add to thedanger in unexpected ways. For thesereasons, reliable scientific advice may beneeded.

332 The problem of providing appropriatescientific advice in an emergency differs fromthe standard case for several reasons:

• advice will be needed extremely rapidly;• the greatest uncertainties may be due to

lack of information from the field, ratherthan a gap in fundamental scientificknowledge; and

• the emergency-response measures mustbe practical and capable of rapiddeployment.

333 This supplementary note explores thedifferences between emergency response andmore general scientific advisory problems. Itsrecommendations are based on the evidencefrom the E. coli case study; on a short reviewof the UK response to the airborne radioactivecloud from the Chernobyl reactor disaster; andon good practice in emergency planning inother spheres.

Preparedness

334 Successful emergency responsedepends on prior preparation, characterisationof potential threats, definition of functions andresponsibilities, and identification of theresources that might have to be mobilised. Allgovernment bodies should have emergency-response plans in place, which should bebased on a risk assessment of the specifichazards within their domain. A good exampleis the Scottish Office (now Executive)Department of Health’s plan for food-borneoutbreaks of disease, which was activated inthe E. coli case.

335 However, such risk assessments willonly provide a broad description of theforeseeable cases. Many emergencies arenovel, or at least present uniquecharacteristics, so there may be a need foradditional scientific advice during anemergency.

336 As far as possible, scientific resourcesthat might be needed should be identified in

advance and, if necessary, generic tools andmethods developed. An example of the latteris the service in long-range airborne pollutanttrajectory modelling that is now available fromthe Meteorological Office, but which was not inexistence at the time of the Chernobyldisaster.

337 There may also be a need foridentification of emergency-response functionsfor departmental scientific staff, and on-callaccess to advisory committee members (suchas applies with COT, for example) or forexternal contractors.

Detection and issue identification338 The identification of an emergency is notusually difficult. However, in unprecedentedsituations—well exemplified by the Chernobylcase—there may be a lack of appreciation ofthe scale of the event. Another problem maybe the identification of the particularmechanisms by which the event might causeharm.

339 It is therefore good practice forgovernment bodies to consult among a varietyof scientific advisers at an early stage, in orderto ensure that the full implications of emergingevents are quickly identified and thatinformation-gathering efforts are correctlytargeted. Usually, these advisers wouldalready be known to government bodiesthrough their routine advisory work.

340 Some emergency situations present aproblem of diagnosis. For example, in the E.coli outbreak in central Scotland, there was aneed to identify the source of thecontamination (which was achieved quitequickly) and to map out the distribution routes(which was a much slower process).

341 The main technical problems will belikely to take the form of quantitativeapplication of predictive models based on well-understood principles.

Organisation

342 Disasters often involve more than onegovernment department, so there is a need forcoordination across departments and provisionof consistent scientific advice to the relevantMinisters and to the public.

343 The UK response to the Chernobylincident called for a combination of


48

meteorological, environmental, agricultural andradiological protection expertise that wasprovided by a single integrated team servingall departments. This team dealt with allscientific matters, and its existence ensuredthat a consistent and comprehensive technicalbrief was provided. The balancing of interestsbetween departments was done in a differentforum; namely the Civil Contingencies Unit,which is hosted and staffed by the HomeOffice and operates like a stakeholder body.These arrangements provide a good model foremergency situations, fully consistent with theprinciples of these recommendations.

344 Since the required actions will be tacticalsolutions that cannot be delayed for research,effective emergency response calls for anexecutive team of people with command andoperational experience, and with personalcontacts in relevant sectors. Improvisation andinformal communication is often the mosteffective way of achieving quick results. Thismeans that parties directly involved in thematter may be best placed to act as advisers.

345 This executive team may be separatefrom the providers of scientific advice,because there is no need for the scientificadvisers to be permanently engaged in theemergency response itself. However, insmaller cases, the two functions could becombined.

346 The contribution of interested partiesneeds careful management and supervision,however, because these parties may beconcerned about the legal liability of theirorganisation or staff for the consequences ofthe event. This may inhibit timely provision ofinformation, and affect the completeness oraccuracy of that information.

Information to the media and public

347 Provision of the necessary resources forbriefing of the media and communicating withthe public should be an integral part of theemergency-response plan.

348 Those at risk have a right to informationabout an incident. Communication is also avaluable instrument for reducing theconsequences of an incident. Also, thepressure of public and media enquiries maybecome a major difficulty for the emergency-response team and for their scientific advisers.In the Chernobyl case, for example, the inter-departmental scientific team set up a team ofsix staff to deal with telephone enquiries thatwere beyond the capability of the departmentalpublic relations offices. The scientific teamprepared standard responses to the mostcommon questions, which were issued todepartments, while the more difficult questionsfrom individual members of the public werehandled directly by the scientific team. Therewas strong media interest, and the scientificteam referred press enquiries to thedepartmental press offices for handling bypublic relations professionals.

Post-event review

349 It must be borne in mind that theadequacy of the emergency response willusually be revised as part of the subsequentinvestigation of any serious incident. Thescientific advice provided will be part of this,and a record must therefore be kept of allcommunications and advice, including theinformation on which the advice was basedand the reasoning.

350 Any difficulties encountered should berecorded at the time, with a view to futureimprovement of the emergency-responsesystem.

|O|X|E|R|A| Literature Review

49

LITERATURE REVIEW351 This chapter describes the ideas andevidence gathered by a review of publishedliterature, and is divided into two sections:

• theory and practice;• concepts and ideas.

352 The review seeks to describe what isknown about the relationship between howexpert advice is generated and incorporatedinto policy, and the quality, or perceivedquality, of the decisions made. This particularrelationship has not previously beenaddressed directly in the literature. However,there is literature on historical or internationalcomparisons of scientific advisory structures,on guidelines developed on the basis of casestudies, and on social theoretical work on therelationships between science, policy-makers,and the public. This material has beenreviewed with the aim of distilling currentthinking on the link between process andpolicy outcome. In addition, ideas that mightbe useful for analysing the various processesfor obtaining expert scientific advice, andexisting or proposed guidelines, have beenextracted from the available literature.

353 Relevant literature has been identified bysearching bibliographic databases, bybrowsing journals with relevant coverage, andfrom suggestions made by researchers activein this field.

354 Existing literature that is relevant to thiswork falls into a number of categories,including:

• reflections on the role of science andscientists within society;

• analyses of the assessment, perception,communication, and management ofrisk;

• comparisons between the scientificadvisory structures in different countries;

• guidelines and recommendations foradvisers or receivers of advice; and

• social scientific analyses of the theory ofadvisory systems, and of the principlesaccording to which they operate.

355 There is no significant body of literaturedirectly addressing the link between theprocess of advising policy-makers and thesuccess of the policy outcome. This reviewattempts to address that link by drawingtogether work from the fields described above.

356 Existing guidelines on scientific adviceand policy-making are reviewed in thefollowing chapter.

Theory and practice

357 The role of science and the expertscientific adviser in public life has beenevolving continually. However, one aspect ofthe relationship between science and societyhas changed significantly in the twentiethcentury. A strong science base becameessential for military success, and moregenerally, for strong economic growth. Forexample, after the Manhattan Project, and thedevelopment of the nuclear weaponsprogramme, the relationship betweenscientists and policy-makers became moreexplicit in the USA. A Science AdvisoryCommittee was established in 1951, initiallywithin the Office of Defence Mobilisation, andsoon reported directly to the President as thePresident’s Science Advisory Committee(Weingart, 1999). In contrast to earlier times,when individual scientists or benefactorsfunded research, science was now funded withpublic money.

358 Formal systems for providing scientificadvice were established in many countries,and scientists were perceived as independentfrom the political process (Johnston, 1993).

359 The newly raised profile of science gavesome cause for concern. PresidentEisenhower warned in 1961 that public policymight be captured by the scientific-technological elite (Weingart, 1999). Therewere two concerns: first, that scientists wouldbe able to exploit their public profile to obtainincreased funding for research; and, second,that policy-makers would become reliant onthe advice of individuals, and thataccountability and democracy would beeroded.

360 The relationship between science andpolicy-making has two components. ‘Policy forscience’ refers to government funding forresearch, and the contribution of scientific andtechnical progress to economic growth.‘Science in policy’ refers to the use of scienceby policy-makers in developing policygenerally (see, for example, Bondi, 1992). Thefocus of this study is ‘science in policy’.However, there is some overlap whenresearch is commissioned directly by policy-makers in order to address a policy question,and when individual advisers carry outgovernment-sponsored research. Although


50

this division has been emphasised by somecommentators, it is almost always blurred inpractice.

361 Since the first formal scientific advice topolicy-makers in the 1950s, a number ofchanges have taken place (Johnston, 1993).

362 Science itself has become the object ofacademic study. Discourses on the nature ofscientific knowledge, the process by which it isacquired, and on the human nature ofscientists themselves, have lessened theperceived objectivity and authority of expertscientific advisers. Scientists are no longerheld in awe.

363 Various controversies and disasters(especially surrounding civilian and militarynuclear programmes) have highlighteduncertainty and conflicting opinions withinscience, and the subject of scientificcompetence is now debated. Scientificarguments have served on occasion tosupport both sides of an argument, leading topolarisation rather than to a consensus(Nelkin, 1975). Scientists are no longerregarded as infallible.

364 At the same time, science andtechnology has become a much moreimportant fundamental requirement foreconomic growth.

365 The frameworks within which scienceand policy interact have evolved diversely indifferent countries. There is considerablevariation in the relative importance ofindividual advisers, who advise Ministers orPrime Ministers directly; expert bodies (suchas learned societies), who provide advice fromoutside government; and scientifically qualifiedofficials, who provide advice from withingovernment. The representative features ofthese frameworks are briefly described in thefollowing section.

Structures

366 The structures of the advisory systems indifferent countries have been reviewed (see,for example, Stein and Renn, 1998; Johnston,1993; and Skoie, 1993). Broadly speaking, theorganisation of the provision of expertscientific advice has either been centralised inone ministry for science and technology, ordecentralised across all departments requiringadvice and having the funding to commissionresearch. A new structure is emerging in anumber of countries, combining features ofboth centralised and cross-departmentalorganisation. In the UK and Australia, forexample, departments are responsible for

scientific issues affecting their work, but thereis also central coordination on cross-departmental issues and a central mechanismfor scientific advice at Cabinet level (Johnston,1993; the OST, 1998a).

367 The underlying function of the structuresis broadly similar in each country. The 1963report of the OECD’s Committee on SciencePolicy (Skoie, 1993) recommended thatcountries should:

• formulate a national policy for science;• ensure coordination of scientific

activities; and• integrate science with general policy.

368 The diversity of structures in placeshows either that no recognised optimumarrangement has yet been found, or that thearrangements for the provision of advicenecessarily depend on the local style of policy-making.

369 Renn (1995) has described four culturalstyles in the use of expertise in policy-making.These styles can be summarised as follows.

• Adversarial—the process is open topublic scrutiny; the policy selection mustbe justified scientifically; there areprecise rules of procedure; and evidenceis required. The emphasis is on evidenceand knowledge, not judgement. Dueprocess is used to resolve conflictingpositions. The process is contingent onclaims of objectivity of method.

• Fiduciary—there is public input, but nopublic control; there are few rules ofprocedure; and the system depends ontrust. The emphasis is on backgroundknowledge, in-house expertise, andbureaucratic efficiency. The processuses personal contacts and networks.

• Consensual—the process is open to‘members only’; negotiations are closed;the procedure is flexible; and internalconsensus is the goal. The emphasis ison scientific reputation, expertjudgement, and personal status.

• Corporatist—the process is open tointerest groups and experts; there is highpublic visibility, but little public control;procedural rules are strict; and the aim isto sustain the trust of the decision-making body. The emphasis is onjudgement and political prudence,


51

impartiality of experts, and negotiationonly within limits set by scientific experts.

370 Renn suggests that the advisory systemin the USA is broadly adversarial, in Japan it isconsensual, in northern Europe it iscorporatist, and in southern Europe fiduciary.However, calls for increased openness andpublic participation in Europe and Japan, andfor increased focus on communities in theUSA and elsewhere, are leading to a moremediatory style. This may involve morestakeholder consultation than is currentlycommon in Europe or Japan, while seeking toavoid the capture by special interests thatoften occurs in the USA.

371 The USA is unusual in that its expertadvisory committees, reporting to federalagencies, are covered by blanket rules ofprocedure, and there is congressionaloversight of their work. Scientific advisorycommittees advising policy-makers in the USAare subject to the FACA. This requires thatcommittees have a charter specifying missionand objectives, are certified as balanced bythe federal agency to which they report, andpublish minutes of all meetings. Committeemeetings are usually (but not always) open tothe public. In addition, all committees arerequired to provide Congress with an annualreport on their work. The FACA was recentlyextended to cover (in modified form) the workof the National Academy of Sciences.

372 It seems to be generally felt that theFACA has been successful in promoting publictrust in the advisory process without undulyrestricting the availability of good scientificadvice (Stein and Renn, 1998). Although theprinciples behind the FACA are broadlyaccepted, some of the practical details, suchas the increasing prevalence of closedmeetings, are not.

373 Although the FACA provides foropenness in the US system, this has not, inpractice, led to greater public participation inthe process. Stein and Renn (1998) suggestthat participation in the advisory process is lowbecause the public have sufficient trust in theoperation of the FACA. Since the workings ofadvisory committees are placed on record,there is a powerful audit trail that can be usedto review decisions (Shapiro, 1990).

374 The development of regulations, distinctfrom policy-making, is not covered by theFACA. Public consultation on new regulationsis compulsory (Stein and Renn, 1998).

375 The National Academy of Sciences, theNational Academy of Engineering, and the

Institute of Medicine are obliged by statute toprovide advice to Congress. Over 200 reportsare issued each year, for which the academiesare paid at cost.

376 The scientific advisory system in the UKis described in a memorandum from the OSTto the House of Commons Select Committeeon Science and Technology (the OST, 1998a).Departments typically obtain expert scientificadvice through advisory committees, inaddition to their in-house scientific resources.Also, the OST and the government’s CSAhave a centralised responsibility for cross-departmental issues, and for fosteringcooperation between departments. Theworkings of the system are guided by theprinciples set out below [441–443]. The UKgovernment has no statutory arrangement forthe provision of expert scientific advice fromindependent learned societies, unlike the USgovernment [375].

The uses of scientific advice

377 The needs that policy-makers have forscientific advice are well recognised. Smithand Halliwell (1999) identify:

• monitoring, measurement andassessment for regulatory purposes (forexample, health and safety regulations,fishing quotas);

• supporting negotiation of internationalstandards and trade agreements; and

• providing advice to Ministers on complextechnological issues, such as nuclearpower, BSE or xenotransplantation.

378 Beckler (1992) identifies a number ofreasons why governments require externalscientific advice, including the following:

• a policy problem raises questionsexceeding the capacity of in-houseadvisers;

• a problem is of such importance that anindependent assessment is required;

• advice is needed for the longer term,such as advice on global climate change(in such cases, expert advice may berequired because the policy timescale islonger than the political cycle, orbecause projections far into the futurerequire the very best experts in order toreduce the uncertainty); or

• a problem is of such an esoteric naturethat only specialised academics cancontribute.


52

379 Boehmer-Christiansen (1995) givesfurther reasons for policy-makers to takeexpert advice—it provides:

• a source of authority and legitimacy forofficials;

• a mechanism for legitimising, delaying oravoiding action;

• an opportunity to label unpopular policiesas unavoidable;

• a cover-up for policy change—sincescience underlying advice may evolve,dependent policy U-turns may beexplained; and

• a mechanism for centralising decision-making—since advice may be tooexpensive for local or regional policy-makers.

380 These points are illustrated by Boehmer-Christiansen with reference to European policyon transboundary air pollution. The nature ofthe scientific advisory system in general withinthe EU is discussed by Stein and Renn (1998).

Theory

381 Two models that have been proposed todescribe the relationship between scientistsand decision-takers are the decisionist modeland the technocratic model (Weingart 1999and references therein).

382 The decisionist model considers therelationship as follows: a supposedly factual,objective domain of science is separated fromthe value-laden domain of politics. Thus,scientists pass objectively determined facts tothe politician, who uses them to developpolicy. Policy is outside the domain of thescientist because it depends on a set of valuesor social norms that cannot be determinedscientifically. In the decisionist model,information flows only from the scientists to thepoliticians, and there is no loss of democraticcontrol, since the advice of the scientists isobjective and factual (legitimised by the statusof the scientists within their peer group), andthe decision-takers are elected politicians.However, this model ignores the fact that mostadvice is sought on problems that are not wellunderstood and draw the scientists away fromthe certainty of objective facts. In these cases,expert advice must contain subjectivejudgement and opinion.

383 The technocratic model recognises the‘scientification’ of politics. The objectivity ofscience guides the politician in making policydecisions. Scientific rationalism replaces thesubjectivity of politics. However, thetechnocratic model fails to take account of

uncertainty, ignorance and conflict betweendifferent scientific disciplines or schools ofthought. The technocratic model alsoundermines representative democracy. Ifdecisions taken by politicians are to a largeextent dictated by scientific analysis of thevarious policy options, democratic control canno longer be exercised.

384 According to Weingart (1999), both thesemodels fail, either as a description of reality, oras a guide to improving current practice,because the linear sequence of politicalproblem, scientific advice, and politicaldecision, as described in the models, is not agood description of the real world. Scientiststhemselves may be required to define oridentify the problems given to them to solve,and scientific advice may not always be free ofvalues (the degree to which scientificknowledge itself is value-laden may bedebated, but scientists certainly ascribe topolitical views and hold moral values). As anexample, safety regulation may requireexperts to define the scope of regulation, andthen to assess risks in a way that must includeassumptions about the tolerability of risks. Thetolerability of risks cannot be determinedobjectively [459–461].

385 These linear models of problemdefinition, advice, and decision-taking can bedeveloped into a more realistic ‘recursivemodel’. This allows several rounds ofnegotiation between experts and decision-takers as the problem is defined and refined,and as possible solutions are tested forpolitical and technical fitness (Weingart, 1999;Jasanoff, 1990). This description of theprocess avoids the problems of the decisionistand the technocratic models. However,Weingart identifies a second problem: that ofever-increasing demand for scientificexpertise, which is discussed below [388].

386 This more realistic recursive model isalso discussed by Edwards (1999), whoaddresses the role of the public in thedecision-taking process. The decisionist andtechnocratic models referred to above restrictthe role of the public to the election ofrepresentative politicians. Experts interactdirectly with decision-takers, but not the public,and decision-takers interact with the public atelection time. Edwards, referring to the ideasof Habermas, describes the function of the‘public sphere’. The negotiation betweenexperts and decision-takers may take placeagainst the background of public agendas, andmedia discussions of the problems beingaddressed. Alternatively, this function for thepublic sphere can be extended further: thepublic sphere acts as an intermediary between


53

the experts and the decision-takers. Thissituation arises if experts and decision-takersappeal directly to the public, to further strategicaims. Decision-takers wish to further politicalagendas and respond to popular concerns;experts may have their own strategicagendas—attracting funds for research, andpromoting themselves as advisers. Taken toan extreme, the public sphere is the arena inwhich experts and decision-takers interact witheach other, in full view of the public. Thefunction of experts is not merely to advisedecision-takers, but to inform, educate andempower the public for their own interactionswith decision-takers.

387 Edwards (1999) also suggests that thisdescription of the function of the public spheremay explain the apparent paradox (Weingart,1999) that policy-makers’ demand for expertadvice remains high, despite the public loss offaith in that advice.

388 Weingart (1999) suggests that policy-makers become ever more reliant on expertscientific advice (scientification of politics), andthat scientists compete strategically to supplyadvice (politicisation of science). This isanother example of the mixing of ‘policy forscience’ and ‘science in policy’. If both thedemand for, and the supply of, scientificadvice are subject to inflationary pressures,the casualty may be quality. Weingartsuggests as a solution formalising the processof advising policy-makers. On the demandside, this could take the form of commonprocurement arrangements across differentbranches of government (or, internationally,across different governments). Alternatively,on the supply side, this could involve themonopolisation (and quality control) of thesupply of advice, by some kind of NGO.

389 Several authors have recognised that, oncontentious issues, scientific arguments andexperts may support both sides of a policydispute. This position is most likely when thepolicy debate is conducted in an adversarialmanner, which tends to polarise the debate,and gives equal weight to the view of themajority and minorities within the debate. vanEeten (1999) describes such cases as a‘dialogue of the deaf’, because the two sidestalk past each other. Resolution throughargument is impossible because the two sidesare arguing from different premises, and thepremises themselves are based on values.

390 It is instructive to consider theweaknesses in stylised resolutions to thesepolicy problems (after van Eeten, 1999).

• Choose the better argument: thisapproach can only work if the two sidesof the debate are arguing from the same(or compatible) premises. If they are not,it is impossible to choose one of thearguments without making a judgementabout the validity of the premises. This islikely to be a value judgement, not ascientific judgement.

• Get the big picture: it is plausible that,in such disputes, the two arguments mayboth be partly right—if the two sidescould step back a little, they would seethe bigger picture, which includes bothpositions. This synthetic approachassumes that the two sides are differentpieces of the same puzzle.Unfortunately, there is no guarantee thatthe two sides are not different pieces ofdifferent puzzles (Roe, 1994).

• Let politics decide: while politicianshave a mandate from the public to takedecisions, science may have a valuablecontribution to make to the search for anoptimum policy outcome. Thus, allowingpolitics to decide without recourse toscientific advice is a failure. This failuremirrors the failure of the technocraticmodel described above; just as it isunacceptable to treat political choices asamenable to scientific decision-making,so it is unacceptable to treat conflictsbetween different sources of scientificadvice as subject to uninformed politicalchoice.

• Participatory processes anddiscourse: there is much support forincreased stakeholder participation incontroversial policy decisions. Althoughthis may clarify opposing positions, thereis no guarantee that any resolution of thedisagreement will result from opposingsides understanding each other’spositions more clearly.

• Finally, it is suggested that, whilescientific advice cannot necessarilyprovide answers in these difficultsituations, it may be used to reassessassumptions, test arguments, and re-frame the policy question.

Concepts and ideas

391 From the literature discussed above, anumber of concepts and ideas that are central


54

to the successful provision of expert scientificadvice to government can be identified. Theseconcepts and ideas are drawn out and brieflydiscussed further in the following sections.

Openness

392 Smith and Halliwell (1999) identify adrive for increased openness as one commonfeature of the way in which many advisorysystems throughout the world are changing.The drive for openness in the provision ofscientific advice is part of wider calls for moreopenness in government generally.

393 Openness has two components: allowingthe public access to information used bypolicy-makers in arriving at their decisions;and allowing the public greater opportunity tocontribute their views as inputs to the advisorysystem.

394 A debate is under way in the USA over arequirement for data produced by federallyfunded research programmes to be published.This requirement applies to all research that isfederally funded, including research atuniversities or other independent institutions.Hahn (1999) argues that this requirement isimportant because it allows checking of theanalysis on which new regulations are based.If data are not published, it is not possible forthe analysis of the data to be properly audited,and inefficient and burdensome regulationsbased on flawed analysis may be enacted.

395 Openness in risk assessments may helpto prevent barriers to trade, as it would then beharder to raise such barriers on spurioushealth and safety grounds (ILGRA, 1996).However, as may be seen elsewhere in thisstudy, a risk assessment may depend onfactors other than technical assessment ofprobabilities and consequences of hazards.Thus, openness per se may not preventdispute.

396 On the other hand, publication of ascientific risk assessment might precipitatecalls for regulation before other factors (suchas the tolerability of the risk) can be taken intoaccount (ILGRA, 1998a, b). In this situation, itis important that experts and decision-takersplan how they will communicate theirassessment of risks.

397 The Freedom of Information Bill (Houseof Commons, 1999a) is likely to restrict anymoves to publish advice given to Ministers.Clause 34 of the draft bill, presented toParliament on November 18th 1999, statesthat information will not be disclosed if itwould, or would be likely to, ‘prejudice the

maintenance of the convention of collectiveresponsibility of ministers of the crown’, or‘inhibit the free and frank provision of advice orexchange of views’. This is thought likely torule out the publication of policy advice fromofficials. However, background papers andscientific analysis produced for governmentare likely to be in the public domain.

Transparency

398 An advisory system can be open withoutbeing transparent. The system is open ifrelevant documents are published (forexample, minutes of committee meetings, andreports received by Ministers). However, thesystem is only transparent if it is a specificgoal of the system to ensure that the public isable to understand the workings of the system.This is only likely to be achieved ifcommunication with the public is part of theadvisory system. Transparency is unlikely tobe achieved if all of the relevant documentsare simply placed on a web site.

399 Advice may not be transparent to thepolicy-makers to whom it is addressed, letalone to the general public. Transparency ofadvice could also involve distinguishing thecomponents of the advice which depend onwidely accepted fact, judgement, or opinion,and describing as fully as possible theassumptions or analytical methods on whichany conclusions rest (National EnvironmentalPolicy Institute, NEPI, 1998). The identificationand explanation of uncertainties are also arequirement for advice to be transparent. Itmay also be good practice to identify wherenew evidence might alter the conclusions ofadvisers, and where monitoring should beundertaken to ensure that the assumptions onwhich advice rests remain valid (Council ofScience and Technology Advisers, CSTA,1999).

Independence

400 The value of expert scientific advice maydepend to an extent on the identity of theadviser as well as the content of the advice.However, independence is neither easy todefine nor to achieve. Advisers may beindependent from particular special interests,stakeholders, industries, political parties, oreven scientific schools of thought. Absoluteindependence, however, may be neitherachievable nor desirable. The selection ofadvisers is more commonly dealt with in termsof conflict of interest and balance, discussedbelow.


55

Balance

401 Various guidelines call for advice to bebalanced (the OST, 1997), or for committeesthemselves to be balanced. Although the term‘balance’ is not always explicitly defined, it isoften taken to mean that experts should bechosen to represent a range of views, or evena range of constituencies. In some cases, thisrange may be an attempt to cover the differentscientific viewpoints that might be held by thewider scientific community on a particularpoint. However, in other cases, it may be anattempt to cover the range of views of special-interest groups.

402 Thus, one balanced committee mighthave both toxicologists and epidemiologists asmembers; another might be balanced byhaving a government ecologist as well as anecologist funded by an environmental pressuregroup.

403 The question of proper balance istherefore linked to the representation ofstakeholders on expert advisory bodies, aswell as to the question of whether stakeholderinput into policy-making should be separatefrom the scientific input to policy-making.

404 The US experience of the requirement ofbalance under the FACA suggests that, inorder for a requirement of balance to besensibly interpreted, it must be tightly defined.A number of lawsuits have resulted fromarguments over whether the requirement inspecific cases could be satisfied without theinclusion of special-interest representatives(Shapiro, 1990).

405 The process of selecting balanced expertpanels under the Committee on Expert Panels(CEP) in Canada might lead to the selection ofa group of experts who admit to a range ofviews on controversial issues which touch onthe proposed work of the expert panel.

Remuneration

406 The question of whether advisers shouldbe paid for their advice is not discussed asoften as other issues surrounding the provisionof expert scientific advice. It is commoninternationally for advisers either to be unpaidor to receive token sums, although expensesare met and the government typically suppliesadministrative support. It appears that advisersare content with the reward of professionalkudos, influence, and the sense of fulfilling acivic duty (Stein and Renn, 1998), although itmay be that not all potential advisers canafford to advise for free. Smith (1997)recommends that experts be paid at

commercial rates, once allowance has beenmade for contingent intangible benefits.

Liability

407 The question of liability of advisers forthe advice that they give has only recentlybecome an issue. The Committee onStandards in Public Life, under thechairmanship of Lord Neill, commissioned astudy of personal liability in public-serviceorganisations. The conclusions of this reviewwere that the extent of liability is not alwaysdefined, and that the situation requiredclarification (Hambley, 1998). This issue mayfeature in the outcome of the ongoing Phillipsinquiry into BSE.

Legitimacy

408 Two possible failings of the scientificadvisory system are described above [381–384]—policy-makers may be tempted to usethe apparent rationality of experts to advise onmatters beyond science, or they may falselyassume that advice from experts is free fromall value judgement.

409 These dangers may perhaps be avoidedif the advisory process is opened to publicscrutiny, and if the public are given moreopportunity to provide input into the policy-making process.

410 The nature of consensus is discussedbelow [423–424]. It can be argued (in asomewhat circular fashion) that legitimacy canbe achieved by reaching consensus among asufficiently wide group. It may be difficult todecide when the usual processes ofrepresentative democracy are sufficient toempower decision-takers legitimately tobalance the views of experts and others, and,when necessary, to use more direct forms ofconsultation.

Process

411 Smith and Halliwell (1999) argue thatguidelines of process provide a counter to thelack of public confidence in the credibility ofscientific advice; an assurance to decision-takers that the advisory system is robust andwill be able to withstand future legal challenge;an effective means to ensure that theprecautionary principle is incorporated intoadvice; and quality control over the advisoryprocess, where implementation of guidelinescan be followed. Being able to demonstratethat an acceptable process has been followed


56

could provide a defence against judicial reviewor liability claims.

412 Many of the sources discussed aboveassert the importance of procedure in seekingscientific advice and incorporating it intopolicy. It may be intuitively obvious to somethat an auditable procedure will increase theacceptability of the policy outcome. However,others may not be so easily convinced,particularly if the ‘straightjacket’ of proceduremay, in particular cases, compromise theability of experts to advise in the way thatseems to them most effective. Onecommentator, Perri 6 (1999), described theargument between those advocating theeffectiveness of proceduralism and theirdoubters as follows: ‘Since each side couldoffer only anecdote dignified by the title ofcase study, it is hard to assess the evidencefor proceduralism.’

413 Shrader-Frechette (1990) provides anargument in support of proceduralism.Discussing risk-assessment methodologies,Shrader-Frechette argues that democratic,ethical and political factors need to be takeninto account. Although the technicalcomponent of a risk assessment could bejudged against an appeal to ‘absolute’standards of scientific reasoning, the othercomponents cannot be so judged.Philosophers hold that moral and ethicaldecisions cannot be judged against an appealto explicit rules, but only by comparison withother decisions, generally held to have beencompatible with accepted moral and ethicalvalues. Thus, rules of procedure would allowexamples of advice giving to be judged againstthe rules, or, equivalently, against a body of‘case law’ of previous examples of expertadvice, as in the legal system.

Terms of reference

414 Several of the guidelines reviewed aboverecommend that advisory bodies should beable to negotiate the terms of reference underwhich they work, and the wording of thequestions that are put to them.

415 If a group of experts is to be asked toanswer a very narrow technical question, it isunlikely that the question itself can be framedwithout input from the experts. On the otherhand, if the question is framed more broadly,experts may need to negotiate with policy-makers in order to narrow the scope of thequestion such that it can be answeredscientifically.

Uncertainty

416 Uncertainty arises from non-predictabilityinherent in the real world, and from gaps inscientific knowledge (or disagreementbetween experts). It is increasingly recognisedthat science, particularly at the frontiers ofknowledge, is uncertain. This type of uncertainscience may be the science that is of mostrelevance to policy-makers. Consensus maymask uncertainty (Beckler, 1992, and see423). However, a best estimate, together withan indication of the range of possibilities, canbe presented to policy-makers. On the otherhand, communication of uncertainty to policy-makers may lead to the rejection of scientificadvice, or to advisers being pressured to maskthe extent of uncertainty (NEPI, 1998).

417 Irrespective of the degree of uncertaintyestimated at the time advice was given,science does not stand still, and even scientificideas with broad consensus support (almosthaving the status of facts) may be shownsubsequently to be incomplete or even wrong.This has led some to suggest that policieswhich are heavily dependent on scientificadvice should be reviewed within a periodspecified at the time of originalimplementation, in order to ensure thatchanges in scientific understanding can betaken into account (CSTA, 1999). Forexample, the Delaney Clause famously forbidsthe US Food and Drug Administration toapprove any substance as an additive if itinduces cancer in laboratory animals. This wasbased on the consensus at the time that thereare relatively few carcinogens, all of which aredangerous at any dose. It is now thought thatthere are very many carcinogens, some ofwhich present negligible hazards if exposure isvery low. The result is that the US Food andDrug Administration has, on occasion, simplyignored the Delaney Clause (NEPI, 1998).

Scientific judgement

418 The distinction between scientificevidence, analysis, judgement and opinion isconsidered vital by the authors of several ofthe sets of guidelines described above (forexample, the OST, 1997; Smith, 1997; Smithand Halliwell, 1999). Hahn (1999)distinguishes between data and analysisprimarily on the grounds that analysis is proneto errors and must therefore be checked. Inmoving from scientific evidence, throughformal analysis and reasoned judgement toopinion, uncertainty increases. However, atthe same time as uncertainty increases, theextent to which non-scientific valuejudgements become incorporated may alsoincrease.


57

419 Bacon (1998) recognises this inidentifying scientific judgement as becoming ‘asubstitute or competitor for policy judgement’.Transparency in the advisory process, whichshould expose the relative contributions ofevidence, analysis, judgement and opinion,may resolve this problem.

Formal methods

420 Experts on scientific advisorycommittees may be asked to resolveuncertainty, draw together results fromdifferent studies, or reconcile differences ofjudgement or opinion. A number of formalmethods for achieving this have beendeveloped, ranging from the Delphi techniqueto Bayesian statistics.

421 For example, Cooke (1991) describes anumber of formal quantitative methods fordealing with uncertainty in expert advice.These methods focus on risk assessment.They are highly technical and thereforeinaccessible to most policy-makers. As such,they may be relevant to the work of particularscientific advisory committees or groups ofexperts, but not to expert advice in general.

422 Bayesian analysis allows, for example,an existing estimate of a parameter to beupdated by new experimental results. It alsoallows a weight to be assigned to the newresults, which could be based on the degree ofrelevance that the particular experimentalsystem has to the parameter of interest, or thedegree of confidence that the expertconducting the analysis places on the newexperimental results. The use of Bayesianmethods has been illustrated by Lilford andBraunholtz (1996), using as an example thecontroversy over the health risks associatedwith some contraceptive pills.

Consensus

423 Policy-makers often ask groups ofexperts to reach consensus in answeringquestions put to them. For example, of theseveral hundred reports generated each yearby the National Academies of Sciences in theUSA, only half a dozen or so are notconsensus reports. It may be easier to reachconsensus when answers are expressed as arange. The wider the range, the greater is thelikelihood of reaching consensus.

424 Where experts are asked to assesspolicy options, consensus may be harder toreach. Alternatively, experts may disagreeabout the application of particular methods ortheories in analysing scientific evidence. TheRadioactive Waste Management Advisory

Committee (1999) gives as a definition ofconsensus ‘the achievement of a sufficientconcurrence of view at various stages tolegitimise a decision to proceed with aparticular course of action’. It alsorecommends that the process of achievingconsensus must be open and transparent, andthat, in the particular case of the disposal ofradioactive waste, the consensus must beachieved across a wider group of people thanjust expert scientists.

Stakeholders

425 The need to consider ethical, moral orother value judgements along with expertscientific judgement has been discussedabove in the context of risk assessment, andalso features in several of the sets ofguidelines covered in this review. The RoyalCommission on Environmental Pollution(1998) discusses this question in its 21streport on setting environmental standards, andrecommends that the public should beinvolved in the formulation of policies, ratherthan simply being presented with a choice ofpredetermined options.

426 Although there is broad agreement in theliterature that the views of stakeholders needto be taken into account in the advisoryprocess because, at least in some cases, theexpert advice cannot be value-free, there is avariety of views on how this could beachieved. Some consider that stakeholdersshould be represented on expert scientificcommittees (Consumers’ Association, 1999),perhaps as non-voting members (Shapiro,1990).

Communication

427 Communication is often neglected aspart of the risk-assessment and managementprocess. ILGRA (1998b) recommended thatmore effort be put into appropriatecommunication of risks and risk-managementstrategies. Lack of communication during andafter risk assessments carried out by expertcommittees may have contributed to the lackof public trust in the workings of suchcommittees.

428 However, debate continues over whetherexpert scientific risk assessments can beimproved by better communication of theresults of the risk assessment, or whetherinvolving the public in the process of riskassessment itself is also necessary if thequality of the assessment is to be improved(see, for example, Stirling, 1998). Riskcommunication may sometimes appear to bean attempt to educate the public away from


58

beliefs about risk and control of risksconsidered ‘irrational’ by experts. Although theviews of the general public on risk may beunscientific, describing the general public asirrational may appear to an extent tautological(Shrader-Frechette, 1990). The reasons fordivergence between expert and lay views of agiven risk are usually clear—and far fromirrational—but may have nothing to do withscience.

Trust

429 It is widely perceived that a lack of publictrust in government in general, and ingovernment scientists in particular, is asignificant hindrance to implementing effectivepublic policy (for example, Worcester, 1999).In the particular field of managingtechnological risks, it is often accepted that thepublic perception of the risk depends, amongother things, on the degree of trust in whichthe risk assessors and managers are held(Slovic, 1993). As discussed above, increasedopenness, public accountability, andadherence to rules of procedure havefrequently been called for in order to increasepublic confidence and trust in the advisoryprocess. However, calls for more opennessseem to be made more on the basis of appealto self-evident truths of human nature than onthe evidence of actual experience.

430 Slovic (1993) compares the perception ofrisk from nuclear power in the USA and inFrance. While, in both countries, the risk isperceived to be high, the French publicexpresses a high level of trust in theauthorities charged with managing the risk, butaccept that they themselves have little controlover the risk. On the other hand, the US publichave very little trust in the nuclear industry orthe regulatory authorities, yet feel that they(the public) are able to intervene to a certainextent in the risk-management process. Slovicnotes that the anti-nuclear movement inFrance was slow to reach the politicalmainstream, and that the French governmentdid not permit public intervention in theprocess of formulating or implementing policy.It may be that both the closed French systemand the more open US system flow from thestyle of decision-making that operates in eachcountry, but there is no support in thiscomparison for the thesis that increasedopenness builds trust.

Quality

431 Support for the peer-review process inassuring quality is not universal. Somesuggest that peer review simply ensures thatonly results in accord with the prevailing

consensus within a peer group are accepted(NEPI, 1998). Hahn (1999) articulates seriousconcerns about the peer-review process, andrefers to studies where already published (andpeer-reviewed) results were subjected to re-analysis which discovered errors, and wherespoof results containing mistakes were sent toreviewers who did not find the mistakes. Hahnargues for publication of data to allow a muchwider review of analysis and results dependenton the data.

432 As well as arguing for publication of thescientific data on which new regulations arebased, Hahn (1999) also calls for all analysisthat results in economically significantregulations to be repeated by an independentagency before the regulation can be enacted.He asserts that ‘replication is a key to ensuringthe quality of the results.’ However, this doespresuppose that the data itself is to be trusted,and he thereby makes a distinction betweendata and analysis, to a certain extent mirroringthe distinction between fact, judgement, andopinion. These distinctions are not universallyrecognised, however [418–419].

Peer groups

433 Scientists tend to specialise, and fields ofstudy to fragment. The peer group for aparticular scientific problem may numberaround a dozen individuals (NEPI, 1998).

434 Scientific advice produced by individualswithin a peer group is likely to conform to theconsensus within the peer group. However,advice on the same problem from differentpeer groups may conflict.

Policy for science or science in policy

435 The importance of the distinctionbetween policy for science (ie, the allocation ofpublic money for research), and science inpolicy has been made many times since theHaldane Report of 1918 (the OST, 1998a;Wilkie, 1991). The OST, in its memorandum tothe House of Commons Select Committee onScience and Technology, states that thisprinciple of separation has been reinforced bythe recent changes in the structure of thepublic-sector research establishments, and theincreased emphasis on a ‘customer–contractor’ relationship.

436 Nevertheless, considerable overlapremains between the administration of policyfor science and science in policy. This isillustrated, for example, in the role of the CSAand the Council for Science and Technology(the OST, 1998a).

|O|X|E|R|A| Existing Guidance

59

EXISTING GUIDANCE437 This chapter presents a summary ofexisting guidance on scientific advice and ongeneral policy-making. Governmentdepartments have been active in reviewingthis area over the last few years, particularly inrelation to the Freedom of Information Bill,which the Home Office has led; reviews ofpolicy-making, led by the Cabinet Office;guidelines for scientific advice, led by OST;and risk assessment, led by the HSE.Departments have prepared guidance for civilservants and Ministers, and, to a lesser extentfor advisers, as well as commentaries on theshortcomings of certain policy issues, theirpotential remedies, and lessons to be learnt toimprove the linked scientific advisory andpolicy processes.

438 This review begins with the currentguidance on scientific advice in the UK, USA,Canada and New Zealand. It then draws outthemes from a wider body of guidance onpolicy-making, risk communication, andscientific advice, reflecting the principles andprocess section of the report.

439 There are several sets of publishedguidelines that are a codification of currentpractice (or current best practice), or havebeen drawn up from case studies or inresponse to consultation on the generalchanges in the standards expected of publicbodies and those in public life.

May guidelines

440 In 1997, the OST published a note by SirRobert May, the UK CSA, entitled ‘The Use ofScientific Advice in Policy Making’ (OST1997). The note contains guidelinesaddressed to departments and agencies ofgovernment that have responsibility for theirdetailed implementation. The MinisterialScience Group, which includes members fromall government departments with an interest inscience, has as one of its key priorities theimplementation of the guidelines. The Grouphas asked departments to report annually tothe CSA on their implementation of theguidelines. In addition, the CSA anddepartmental chief scientists meet regularly toreview emerging issues (the OST, 1998a).

441 The guidelines are presented underthree key principles, and are summarisedbelow.

• Identifying issues—departments shoulduse a wide variety of sources to aid earlyidentification of issues, including external

research and reports by special-interestgroups. Capacity to recognise andrespond to unforeseen issues that ariseis also required, and exchange ofinformation on cross-departmental orinternational issues will be handled bythe OST.

• Building science into policy—the bestavailable scientific advice should besought. This should include independentadvice (for example, from ‘eminentindividuals, learned societies, advisorycommittees, experts outside the UK, andconsultants’). Experts from a range ofdisciplines, not necessarily scientific,should be asked to review evidence froma range of viewpoints. Data should bemade available to the general researchcommunity as soon as possible.Scientific advisers should help to frameand assess policy options. Fundingpriorities should be kept under review inlight of the need for advice, and abalanced view should be reached in atransparent way that is consistent acrossdifferent policy areas.

• There should be a presumptiontowards openness—scientific evidenceand analysis underlying policy decisionsshould be published. Scientists shouldbe encouraged to publish associatedresearch work. Uncertainties and policyoptions should be presented to the publicin a manner that is consistent with thescientific advice. Departments shouldconsider giving scientists a leading rolein presenting their advice, with Ministersor officials describing how policy wasderived from the advice. In addition,early communication with interest groupsand other governments may beappropriate.

442 The guidelines were drawn up followingwide consultation with governmentdepartments (seen as part of the process ofopening up Whitehall that has taken place inrecent years). Progress in implementing theguidelines has been reported on twice (OST,1998b), and the CSA announced a review inDecember 1999 to determine whether anychanges or additions might usefully be madeto the key principles. This review involvedconsultation both within government andexternally. The revised guidelines have beenpublished as ‘Guidelines 2000—ScientificAdvice and Policy-making’ (OST 2000b). The


60

guidelines have attracted praise from overseasobservers (Smith and Halliwell, 1999; Steinand Renn, 1998; CSTA, 1999). It appears thatthe UK is unusual in having published suchguidelines, although they are, for the mostpart, aspirational rather than prescriptive, andconsiderable discretion in interpretation is leftto departments. Furthermore, it is not clearhow implementation of the guidelines can bemeasured, although departments havereported specific case-study examples in theirannual implementation reports.

443 The ‘May guidelines’ are currently underreview. At the time of writing, consultation onan updated version of the guidelines isongoing.

Nolan Principles and Peach Guidelines

444 The Committee on Standards in PublicLife (the Nolan Committee) was set up in 1994in response to general public concern. Inparticular, there was concern over allegationsof payments to members of Parliament fortabling of questions in Parliament; that ex-Ministers had been employed by firms whichthey had regulated when in office; and overfraud or mis-spending in quangos (quasi-autonomous non-governmental organisations).The remit also included all holders of publicoffice, non-departmental public bodies, andboth civil servants and advisers to Ministers(Committee on Standards in Public Life, 1995).

445 The Nolan Committee devised ‘SevenPrinciples of Public Life’:

• selflessness—decisions should be takensolely in the public interest;

• integrity—individuals should not beunder any obligation to outsiders thatmight confer influence;

• objectivity—appointments and contractsshould be given on merit;

• accountability—holders of public officeshould be accountable for their decisionsand must submit to appropriate scrutiny;

• openness—decisions should be takenopenly, and access to information shouldonly be restricted ‘when the wider publicinterest clearly demands’;

• honesty—private interests should bedeclared, and conflicts of interest shouldbe resolved to protect the public interest;and

• leadership—the principles should besupported by leadership and example.

446 Following the recommendations of theNolan Committee, Sir Leonard Peach was

appointed Commissioner for PublicAppointments, and guidelines were issued todepartments on how appointments to publicbodies should be made (Office of theCommissioner for Public Appointments, 1998).Appointments to bodies, such as the AdvisoryCommittee on Novel Foods and Processes,are now made in line with the ‘Nolan andPeach’ guidelines (Rooker, 1999).

447 Departments are required to draw up jobdescriptions for each post, and candidates areshortlisted according to their merit against thejob description. Equal-opportunity principlesare applied. The guidelines require that theappointment process be scrutinised by anindependent assessor, who will monitor theapplication of the guidelines.

448 The guidelines also address conflicts ofinterest. Such conflicts should be declared anddocumented, although the definition of aconflict of interest is left to individualdepartments, and it is acknowledged thatsome appointments will be more sensitive thanothers in this regard.

449 The selection of individuals fromshortlists prepared by departments remainsthe responsibility of Ministers. In recentpractice, appointees to advisory committeeshave been ruled out on grounds of beingindustry employees, although not if they areacademics partly funded by industry. There iscontinued debate on this issue. Some arguethat this interpretation of appropriate standardsof independence unnecessarily rules outexperts who have a contribution to make to theadvisory process, and others argue that theinterpretation does not go far enough indistancing advisory committees from theinfluence of industry (House of Commons,1999a).

USA

450 Stein and Renn (1998) argue thatscientific advice to government in the USA ismore open than in any other country becauseof the operation of the FACA [371–374].

451 Beckler (1992) has written about the useof scientific expertise from outsidegovernment, based on the experience ofsupporting the provision of scientific advice atthe highest level within the US government, atthe OECD, the National Academy of Sciences,the Carnegie Commission on ScienceTechnology, and government. The followingrecommendations are described.

• Decision-takers should recognise whenadvisers should be called in from outside


61

government, and should haveprocedures in place for their selection.

• Uncertainty and differences of view arecommon in science, and the decision-taker needs to ensure that differentviews are taken into account, both indecision-making and in deciding thecomposition of advisory groups.

• Recognising that, to a greater or lesserextent, values are inevitably embeddedin expert advice, advisory panelsaddressing policy issues should becarefully balanced with respect to field ofexpertise, sector, and political inclination.

• Advisory bodies should have a well-defined remit, but should be free toreformulate questions or initiate study ofemerging problems, with the consent ofthe policy-maker.

• Both advisers and policy-makers shouldbe alert to the possibility of conflictinginterests between an adviser’s privateand public work.

• Policy-makers need to be supported bypolicy analysis as well as expertscientific advice.

• The advisory process should be openand transparent, as far as is feasible.

• Decision-takers should provide feedbackto advisory committees.

452 Beckler asserts that capacity for policyanalysis is ‘inversely proportional to the levelof decision making’, and that scientific adviceand support for detailed policy analysis islacking at the highest level of policydevelopment, particularly on cross-departmental issues. It may be that the recenttrend in the UK for the appointment ofspecialist advisers to Ministers (for example,economic advisers) is driven by this lack ofestablished advisory capacity at the higherlevels of government.

Canadian guidelines

453 The Canadian CSTA producedguidelines in 1999 for the use of scientificadvice in a report entitled ‘Science Advice forGovernment Effectiveness’. This report drawsheavily on the work of May (the OST, 1997),Smith (Smith, 1997; Smith and Halliwell,1999), and Beckler (1992). The guidelinesinclude the following recommendations.

• Departments should maximise the use ofexpert advice to identify and address‘horizontal’ issues across departments.

• ‘Traditional knowledge’ of local peoplesshould be given consideration, anddecision-takers should balance the many

views they receive (this point mayillustrate the extent to which guidelinesmust recognise the political landscape).

• External advice should be sought whenproblems exceed the in-house expertiseof departments, when there is a range ofscientific opinion, or when there is apossibility of controversial policyoutcomes that could benefit fromindependent advice to strengthen publicconfidence.

• Decision-takers should be open to bothsolicited and unsolicited advice.

• All advisory processes should be subjectto due diligence, including rigorousinternal and external peer review.

• Scientific advice must be supported bypolicy analysis; scientists should be ableto analyse the consequences of theiradvice; and there should be a strongcoupling between advisers anddepartmental policy analysis functions.

• Advisers and decision-takers mustdistinguish between scientific fact andjudgement or opinion.

• Decision-takers should report back toadvisers on how decisions were made,and involve advisers in the formulation ofpolicy, in order to maintain the integrity ofadvice throughout the decision-makingprocess.

454 In addition, the CSTA report proposesdetailed guidelines on openness and review.

• Decision-takers should explain theirdecisions, and how advice was used inreaching the decision.

• Decisions could be explained at publicmeetings; scientific advisers could beasked to explain their advice; andofficials could explain how the advicewas secured, and how policies wereframed in the light of the advice.

• In controversial cases, decision-takersshould balance the need for timelydecisions with the need for effectiveconsultation.

• Departments should institutionalise areview process, so that once decisionsare made, a follow-up exercise provideswritten responses to recommendationsgenerated by advisers.

• Policy decisions should be reviewed todetermine whether advances in scientificknowledge change the basis of scienceadvice used in arriving at the decision. Atthe time decisions are taken, the latestintended date for a review should beannounced.


62

455 Finally, the CSTA report recommendsthat the implementation of the guidelines beaudited. This could be achieved by periodicreview by an external body, or by the creationof a permanent ‘oversight’ function.

Risk perception, assessment, managementand communication

456 It is widely recognised that risk is acomponent of many high-profile, contentiouspolicy decisions where scientific advice isrequired. As such, risk issues are important tothe work of all major government departmentsin the UK. ILGRA has responsibility forkeeping under review developments in thefields of risk perception, assessment,management and communication, andidentifying common approaches that can bedeveloped by all departments.

457 ILGRA (1996) identified a need for theframeworks for risk assessment in differentdepartments to be logically consistent witheach other. Otherwise, for example, differentdepartments may currently use differentdegrees of caution in risk assessment.

458 ILGRA (1998b) has also identified theimportance of increasing public awareness ofrisk. The Internet and the general increase incommunication worldwide have raised andbroadened awareness of major accidents andthe sensitivity of the public to risk issues, inthat greater reporting means that issues attractmore attention more quickly. Thus,governments or regulators may be called uponto act to control exposure to risks more often,earlier, and in a more controversialatmosphere.

459 Furthermore, there is growingacceptance among regulators that publicperception of risk depends not only on theexpected physical harm arising from hazards,but also on wider societal concerns attachedto exposure. This adds a complication to riskassessment. A group of expert scientists mayarrive at a consensus assessment of aparticular risk which depends on their ownvalue judgements about the wider socialissues attached to the risk exposure. Thesevalue judgements may very likely differbetween experts and the general public (aswell as being different for different sections ofthe public).

460 Further concerns about the transparencyof the risk-assessment process, particularlywhen it involves the use of experts on advisorycommittees, were identified by the ILGRA sub-group on the setting of safety standards. Thesub-group recommended that the terms of

reference for expert advisory committees beclarified, to affirm that it was the function of thecommittee to advise, rather than to setstandards. Decisions would then be informedby the expert advice as well as by other inputs,such as societal concerns.

461 Finally, ILGRA (1998a) makes thefollowing recommendations on the function ofexperts, in order to counter the problemsdescribed above:

• expert scientific advice should be opento public scrutiny and peer review, and itshould be made clear where judgementor opinion is used;

• assumptions underlying advice, as wellas uncertainties, should be exposed andexplained;

• procedures should be adopted to allowstakeholder as well as expert input; and

• all considerations underlying decisions,including expert scientific advice, shouldbe fully explained.

New Zealand expert panels

462 The New Zealand Ministry of Research,Science and Technology commissioned areview of ‘Expert Panels for Provision ofScientific and Technological Advice forDevelopment of Public Policy’ (Smith, 1997).This followed the establishment by the NewZealand government of expert panels on BSEand the Tussock Moth. The review, carried outby Dr William Smith of the University ofAuckland, recommends best-practiceguidelines for the design and operation ofpanels of experts reporting to governmentdepartments.

463 The recommendations were preparedafter interviews with members of the expertpanels, officials, and Ministers. In addition,some interviews were conducted in the UK.The expert panels described by Smith arepanels set up by a department of government.Expert panels operated by learned societiesare discussed below [466–467].

464 The review concluded that forindependent scientific advisory panels toprovide useful and timely advice, it isnecessary that:

• quality-control mechanisms should bebuilt into the scientific advisory panelprocess, both in the selection of panelmembers and in its provision ofinformation and advice;

• the composition of panels should reflectthe nature of the issue to be addressed


63

and the breadth of judgement required.Panels should be carefully balanced interms of disciplinary expertise,institutional allegiance and stakeholderinterests;

• panels should be established with clearauthority and accountability, and shouldbe insulated from any attempt tomanipulate or dictate working methodsor the content of advice.

465 In support of these imperatives, a seriesof more detailed recommendations weremade, the more innovative of which aresummarised below.

• The terms of reference of expert panelsshould be a matter for debate andapproval of the panel, in consultationwith Ministers.

• Panels should have access to anyanalysis of their advice provided byofficials to Ministers.

• All independent scientific advisory panelsshould report directly to Ministers.

• Panels should have the right to excludeofficials from their deliberations.

• Panels should explicitly debate thefunction of officials in the drafting ofadvice to Ministers.

• The responsibilities accompanyingmembership of a panel should be clearlycommunicated to members onappointment.

• Panels should have a right to directaccess to Ministers when required.

• Specific guidelines should be drafted forthe conduct of the chairpersons of suchpanels.

Independent expert panels

466 Expert panels may be convened outsidegovernment ministries and departments. It hastraditionally been a function of learnedsocieties to advise governments in this way.For example, the Royal Society of London hasan explicit objective to ‘promote independent,authoritative advice, notably to UKgovernment, on science and engineering-related matters, and to inform public debate’(Collins, 1998). Most learned societies provideadvice to their national governments with orwithout a request for advice having beenreceived, although arrangements for fundingdiffer.

467 The Royal Society of Canada does notreceive government funding. When reports arecommissioned and funded by governments or

other organisations, expert panels are rununder the auspices of this Society. TheSociety is unusual in that it has a strictlycodified procedure for the organisation andconduct of such panels. The followingprocedures and guidelines are summarisedfrom its ‘Expert Panels: Manual of ProceduralGuidelines’ (The Royal Society of Canada,1998).

• The sponsoring organisation cannotdictate who does or does not sit on thepanel.

• The scope of the task is agreed afternegotiation between the CEP and thesponsor.

• In selecting members of the panel,composition and balance are optimised,and conflicts of interest are identified.The CEP aims from the outset to select apanel which will have the skillsnecessary to carry out the task assignedto it, and the balance of views required toensure that the final report of the panel iswidely accepted.

• In addressing balance, the CEP may askprospective panel members to list theirviews on issues connected with the taskof the panel. Balance may be achievedby having representatives of differentviews on the panel, or by appointingindividuals without strong views oncontroversial questions.

• Issues of communication anddissemination of the report areconsidered explicitly by the CEP at anearly stage in the process.

Europe

468 The work of the EC is supported byhundreds of scientific advisory committees.There are no universal guidelines covering thework of these committees at present, althoughthe European Parliament has considered thisissue (Stein and Renn, 1998).

Themes

469 The UK departmental literature began toemerge in the early 1990s, from roots 20 yearsearlier in Lord Rothschild’s 1971 report, ‘TheOrganisation and Management of GovernmentR&D’. Sir John Fairclough, CSA, recognisedthat ‘Science and technology is increasinglypervasive. There are very few policy areasupon which it does not impact.’ (Fairclough,1990). There was a strong recognition thatscientific advice would play a central role inmany areas of policy-making. There is no


64

reason to doubt that, at that point in time, hewas referring to natural and physical sciences,and technology. This is the definition used inthis study. The purpose of his advice to thePrime Minister of the day was ‘to ensure thatscience and technology advice is fullyintegrated into the wider consideration ofpolicy by Departments’. This review ofsubsequent reports and guidance shows howthis objective has been translated intoguidance over the last decade.

470 ILGRA, in its discussion of riskassessment and management, identifies areaswhere reform could be considered, andtouches on the themes reviewed in thissupplementary note.

Traditionally Departments andAgencies have operated under theassumption that, with the assistance ofexperts as necessary, they woulddefine the problem, assess the risks,identify risk management options, andadopt decisions. Typically ... thedecision adopted was justified on thebasis of reliance on the bestindependent scientific advice.

Such an approach is becomingincreasingly untenable for manyreasons. First, as already mentioned,Departments and experts may notframe the problem in the same way asthe stakeholders. Second, there is atendency for experts throughout thedecision-making process to substitutetheir own value judgements for those ofthe stakeholders. Third, stakeholdersmay feel disenfranchised if they havelittle or no opportunity to express theirvalue judgements. And finally,assurances offered on the basis ofobjective science often implode intouncertainty because of unreliable orincomplete data, modellinguncertainties, debatable underpinningassumptions, or conflicts of scientificjudgement in interpreting data (ILGRA,1998a).

471 The general principles espoused by theCabinet Office’s Better Regulation Unit havebeen taken up by ILGRA in its guidance togovernment departments on risk assessmentand risk management, ‘namely transparency,accountability, targeting of action, consistencyand proportionality’ (ILGRA, 1998a). Theseprinciples form the foundation of the currentcivil service guidance on the handling of risks.

Legitimacy

472 This principle emerges as a constanttheme in the history of guidance. Sir JohnFairclough recalled the ‘remarkably sound’principles set out in Lord Rothschild’s report,which ‘continue to underlie presentarrangements, although his recommendationswere not formally accepted’ (Fairclough,1990). Lord Rothschild recommended, aboveall, ‘clear identification and rigorous separationof the roles of customers for scientific adviceand the contractor who provided it, whether in-house or outside the department’. He alsothought that ‘advice should be given by orstrongly informed by leading people fromoutside the civil service.’ Sir John Faircloughconcluded, ‘The key remains a rigorousapproach to the customer/contractor principlecoupled with strong external input.’

473 This theme continues to receive supportin more recent work, for example ILGRA’sguide on risk communication, which warnsthat:

The risk of inappropriate transfer ofresponsibility for decision making toexperts needs to be guarded againstcarefully (ILGRA, 1998b).

474 ILGRA explained in more detail thereason for its conclusion:

The role of scientific advisers is wellestablished in the regulatory process,largely through their position onGovernment advisory committees.However, there is evidence of a lack oftransparency in how experts reachdecisions about the level of risk posedby particular hazards. The sub-groupjudged that thought could usefully begiven to clarifying the terms ofreference for such groups so as tomake clear that their role is to assistdecision makers and not to setstandards as such. This would help toensure and make apparent that otheruseful inputs, such as costs andbenefits, are also taken into account inthe formulation of policy. This wasparticularly important. Once an expertgroup’s judgement on the balance ofthe scientific arguments is published,pressure may be created forGovernment to regulate, even if it isunjustified in cost–benefit terms.

475 It does not appear that this advice wasfollowed in the Stewart report on mobilephones (Independent Expert Group on MobilePhones, 2000).


65

476 The Select Committee on Science andTechnology took up the theme byrecommending that ‘Ministers should obtainadvice on these other, non-scientific, issuesbut should not seek such advice from thescientific advisory system.’ The governmentdid not address this recommendation in itsresponse to the Committee’s report (Scienceand Technology Committee, 1999).

Disclosure of information

477 The disclosure of information is subjectto a large volume of guidance, not all of whichis consistent. In support of the disclosure ofinformation, several powerful statements havebeen made.

478 First, in laying down recommendationsfor risk communication, ILGRA noted:

in certain fields, the provision ofindependent advice via expertcommittees is widely distrusted as partof the system. This may be becausesome of them have a history of takingtheir decisions behind closed doors.(ILGRA, 1998b)

479 Second, the government has supporteda presumption of publication, which issummarised in the May guidelines;‘departments should publish the scientificadvice and all relevant papers’ (OST, 2000b).The Freedom of Information Bill requiresauthorities to adopt a scheme for publication,stipulating the information they will placeproactively in the public domain, and agreeingthe scheme in advance with the InformationCommissioner. However, the Bill includesimportant exemptions to the presumption ofpublication or access.

Information held by a governmentdepartment is exempt information if itrelates to—

(a) the formulation or development ofgovernment policy....

(b) would, or would be likely to inhibit—

(i) the free and frank provision ofadvice, or

(ii) the free and frank exchangeof views for the purposes ofdeliberation

(House of Commons, 1999a)

480 At present, access to Governmentinformation is described in the ‘Code of

Practice on Access to GovernmentInformation’, and in the related ‘Guidance onImplementation’.

The code commits departments to giveinformation on the factual andanalytical background to new policies,but there is an important distinctionbetween the process by which adecision or policy has been reached(which remains confidential) andexplanation of the basis of the decisiononce reached (which should be as fulland open as possible) (Cabinet Office,1997, Home Office, 1998).

481 The Code also expresses an aim to‘improve policy-making and the democraticprocess by extending access to the facts andanalyses which provide the basis forconsideration of the proposed policy’, and hasa specific requirement to ‘publish facts and theanalysis of facts which the Governmentconsiders relevant and important in framingmajor policy proposals and decisions’.

482 This distinction may be difficult to apply,but does address ILGRA’s concern, outlinedabove, provided that a clear distinction ismade between external advice, and internalpolicy-making. Protection for the latter isexplained as follows:

The justification for confidentiality ofinternal opinion, advice, recom-mendation and deliberation is the needto ensure that matters can bediscussed candidly and frankly withingovernment, and a full record keptwithout taking account of the possibilityof publication within any period of timeduring which the material might remainsensitive . . .

Exposure of differences betweenMinisters, between Ministers and theircivil servants or between civil servants,could prejudice working relationshipsand effective discussion of policy ... Itis not the intention, however, towithhold this class of information onlywhere internal differences anddisagreements would be revealed . . . Itis important that reasonableexpectations of confidentiality arepreserved.

483 It is not clear where the line betweenpublication and confidentiality will be drawn,and there can only be consistency with theMay guidelines, above, if there is a clearseparation between advice that is purelyscientific, and policy advice. Indeed, thespecial status of scientific advice is


66

recognised, but not to the extent of apresumption of publication.

There is less need for confidentiality inrespect of advice from expert advisorycommittees, ... where the availability ofthe assessment will enhance publicdebate and understanding ... or madeavailable for peer group review.

484 In fact, the position of scientific advisorycommittees is vague:

Advisory committees themselves arenot usually within the jurisdiction of theOmbudsman, .... the Ombudsman maybe asked to investigate complaintsconcerning refusal by departments todisclose expert advice on which theyhave relied.

485 The guidelines published to date havenot given detailed guidance on the publicationof information. There is an exception in theadvice on the publication of information withinthe biotechnology regulatory framework,although its interpretation may be quiteflexible.

subject to confidentiality andcommercial sensitivity, publishagendas in advance of meetings andminutes thereafter together with suchexplanations as will enable the readerto understand the issues, argumentand basis of decisions taken . . . whereinformation is confidential orcommercially sensitive, write it in sucha way as to make as much informationpublic as possible without infringinglegitimate concerns (Cabinet Office,1999f).

486 This might be an indication of howscientific advisory committees in general willbe treated in the future.

Over-riding duty

487 The principle of a duty to the publicinterest is applied within the draft Freedom ofInformation Bill to the disclosure ofcommercially confidential information, andmight therefore be applied to scientific advice.The bill states that, with respect to publicationof advice, where there is an exemption forcommercially confidential information, theexemption from disclosure for commercial andconfidential information may be set asidewhere there is a clear and outweighing publicinterest (HMG, 1997). There is clearly a finedividing line between the operation of thisclause, and the main exemption in the bill.

The Government therefore considersthat where information is supplied to apublic authority in circumstances wherea duty of confidence arises, it shouldnot be required to disclose theinformation if to do so would constitutea breach of confidence actionable[under common law] by the supplier ofthe information.

488 Furthermore, the Public Disclosure Act1998 also gives some protection to advisersthrough the ‘whistle-blowing’ conditions. It alsoapplies to the public sector, supplementing theCivil Service Code. Under the Civil ServiceCode:

it is preferable to raise the matterinternally [with the appeals officer ofthe department] if appropriate andpractical, or with Civil ServiceCommissioners ... Otherwise, thedisclosure qualifies for protection underthe Act, if in the reasonable belief ofthe worker making it, it tends to showthat one or more of the following hasoccurred, is occurring, or is likely tooccur:

• the endangering of an individual’shealth and safety

• damage to the environment

• deliberate concealing of informationtending to show any of the above.

(Cabinet Office, 1999a).

489 Whether the term ‘individual’ also appliesto unnamed members of the public in ageneral policy situation is not clear. However,similar protection could be offered explicitly toexternal advisers.

Accountability

490 Ministers are solely accountable toParliament for the decisions they take:

The constitutional position of civilservants in relation to Ministers is suchthat:

Ministers alone are accountable forthe information given to Parliament;

civil servants have no final authorityto decide what information shall bemade available to Parliament(Cabinet Office, 2000a).

491 The guidance on accountability ofMinisters to Parliament is well established. It


67

complements the guidance on disclosure ofinformation, but contains a crucial exemption:while Ministers and their officials are generallyrequired to provide full information, thereasons for confidentiality include:

internal discussion and advice:projections and assumptions relating tointernal policy analysis; analysis ofalternative policy options andinformation relating to rejected policyoptions (Home Office, 1998).

492 This could be interpreted to mean thatonly justification for chosen policies is requiredto be published, and not their alternatives(subject to the public-interest provisionsabove). This is an extremely important clause.It is clear that this is the intendedinterpretation, for:

The aims of the Code are:

to improve policy-making and thedemocratic process by extendingaccess to the facts and analyseswhich provide the basis for theconsideration of proposed policy;

to protect the interests of individualsand companies by ensuring thatreasons are given for administrativedecisions, except where there isstatutory authority or establishedconvention to the contrary (HomeOffice, 1998).

493 It can be argued that there is anestablished convention not to give justificationfor the selection of one policy option aboveanother.

494 Otherwise, the rules for Ministerialaccountability mirror the other rules on publicdisclosure. It is noted that they include explicitreference to the public interest. The guidanceis contained in the Resolution on MinisterialAccountability (adopted by the House of Lordson March 20th 1997).

That, in the opinion of this House, thefollowing principles should govern theconduct of Ministers of the Crown inrelation to Parliament:

(1) Ministers have a duty toParliament to account, and be heldto account, for the policies,decisions, and actions of theirDepartments and Next StepsAgencies;

(2) Ministers should give accurateand truthful information toParliament...

(3) Ministers should be as open aspossible with Parliament, refusing toprovide information only whendisclosure would not be in the publicinterest (Cabinet Office, 2000a).

495 It is also relevant to this report that therules on disclosure, particularly in relation toscientific advice, are more flexible than inother areas. For example, the Cabinet Office’sguidance to civil servants explains whyMinisters may not normally have access to thepapers of previous Administrations.

If Ministers believe that there issufficient evidence of an unsatisfactorystate of affairs, it is open to them, afterconsultation with Ministers of theprevious Administration, to appoint asuitable person to inquire into theevents concerned, giving themnecessary access to papers. The BSEInquiry is a recent, large-scale exampleof such an inquiry.

Thus scientific advice should be madeavailable to succeedingAdministrations, and to facilitate this—to make it possible without recall tospecial measures—scientific advicecould not be given to Ministers, but topolicy-makers instead, so that it doesnot attract privilege (Cabinet Office,2000a).

496 The importance of the accountability ofthe process was implied in the conclusions ofSir John Fairclough. He recommended thatdepartments should ‘check periodically thatthe structures and mechanisms in place arematched to need’, and that such checks needto be made ‘against objective criteria’. No clearcriteria exist in current guidance. Herecommended that ‘every four yearsthereafter, Permanent Secretaries should usethese criteria to review the arrangements inplace for obtaining science and technologyadvice within their Departments’ (Fairclough,1990).

Uncertainty

497 The expression of uncertainty in policyadvice and decision-taking is only discussed ingeneral terms in guidance. The focus of theguidance is on public perception andcommunication with the public, rather thaninternally within government. This is illustratedby the principles of risk communicationpublished by ILGRA (1998b), ‘listen to


68

stakeholders, tailor messages, and managethe process.’ The same report concludes that‘Across Government, scientific input to theregulatory process is invariably strong.’

Process

498 There have been several attempts todescribe the policy process as a process

model. Most recently, the Cabinet Officereview of professional policy-making foundthat a policy process model was not receivedwarmly by experienced policy-makers becausethey felt that it did not accurately reflect therealities of policy-making (Cabinet Office,1999e). The report presented a model of thepolicy process in context, which is reproducedbelow.

Figure 3: The policy process in context

What evidence isneeded and/oravailable to test the‘real -world’problem?

What are the desired policy outcomes?

Which are the most effective outputs forachieving these outcomes?

Who are the key stakeholders and howshould they be involved?

What are the needsand views of thosethe policy seeks toinfluence / affect?

What evidenceis available,relevant anduseful?

What have theexperiences of othercountries been?

What are the risksto the policy andhow can they bemanaged?

What is the impact of possiblesolutions on equalopportunities, business, women,environment, etc?

How can different solutions be tested?

How does the problem / policy fit withgovernment manifesto / priorities?

What policy conflict / priorities need to beresolved?

Is a cross-cuttingapproachneeded?

How canevidence bestbe presented?

How and when shouldkey political

Are Ministerssigned up?

What isthestrategyforpresentingpolicy?

Who needs to be told what, when and how?

How can stakeholders be kept committed and involved?

What are the quick wins?

Who else within government needs to beinvolved and how?

What is the impact of devolution?

What is the role of the EU?

How should front-line staff be involved?

What sort of cross-cutting intervention isrequired (if any)?

What is the impacton other existingand developingpolicies?

What are the costs /benefits of differentoptions?

What evaluation systems andperformance targets are needed?

What are thealternatives tolegislation andregulation?

What training and support forfront-line staff is needed?

What ISchanges areneeded?

What needs tohappen to ensurepolicy becomesself-sustaining?

How/when shouldpolicy effectivenessand contribution tocorporate objectivesbe reviewed?

WIDER PUBLICCONTEXT

POLITICALCONTEXT

ORGANISA-TIONALCONTEXT

Understanding the problem

Developingsolutions

Putting solutions intoeffect

Testingsuccess andmaking itstick

POLICYPROCESS

representativesbe involved?

Source: Cabinet Office, 1999e.


69

499 Earlier, in 1990, Sir John Fairclough hadconcluded that ‘it is no longer [appropriate] toattempt to define a prescriptive model in theway that the Rothschild report did’ because ofthe diversity of the advice required. He wenton to note that a flexible process model isappropriate, but should remain within the firmcontrol of the chief scientist. This leadsnaturally to ILGRA’s conclusion that:

Departments therefore need to developways of explaining the scientific andpolicy background to decisions to thepublic, and to establish how best toincorporate both expert judgementsand society’s preferences into thedecision-making process (ILGRA,1998b).

500 The order in which this occurs may becrucial. As an example of the confusion thatcan arise if the process is not clear, considerthat the guidance in Sir John Fairclough’smemorandum is ‘to review and contribute tothe presentation of Departmental policieshaving science and technology content, bothnationally and internationally’. This could beinterpreted to mean the review of policies thathave already been set out. The decision andpolicy should flow from the advice, rather thanin the other direction, although ex-postevaluation of policy effectiveness is a valuablefunction.

Framing the question and terms ofreference

501 The definition of scientific questionsrelating to policy issues has not received widetreatment in existing guidance. The Treasury,in its guidance on policy appraisal (HMTreasury, 1997), notes that the first stage is toset clear policy objectives with departmentalMinisters. ILGRA identifies the link betweenthe framing of the question, the publicperception of risk, and stakeholder concerns:‘the regulator must find out how much the risksmatter, to whom and why, and agree anagenda with them’.

502 The Cabinet Office guidance (1999e)gives policy-makers detailed notes on theissues that they should consider, but does notoffer similar guidance for scientific advisers.This guidance, issued to policy-makers by theCabinet Office, is summarised below, andrecommends that policy-makers refer to:

• a statement of intended outcomes at anearly stage;

• contingency and scenario planning;• evidence of taking into account

government’s long-term strategy;

• the work of international agencies, andpolicy in other countries;

• the origin of the issue;• why previous policy options failed;• the use of pilots (trials);• new research that could be

commissioned;• feedback from implementers of current

policy;• cross-cutting objectives and inter-

departmental issues;• a programme of ongoing review;• mechanisms to remove, reverse or

modify policy.

503 ILGRA has provided a checklist for theconsideration of policy and risk (ILGRA,1998a). It is designed to help identify the risksto which a policy response may be needed,and to assess potential public reaction tothese risks. ILGRA notes that ‘risks aregenerally more worrying ... if perceived to bepoorly understood by science, ... and subjectto contradictory statements from responsiblesources.’ It is surprising that there is not moreguidance on preparing terms of reference forscientific advisers.

504 The importance of setting the correctterms of reference is demonstrated by theadvisory failures that can occur otherwise. Forexample, the government’s response to theScience and Technology Committee discussesGM crops (House of Commons Science andTechnology Committee, 1999). The initialscientific research did not take into accountthe concerns of a stakeholder, English Nature,so another research programme was laterundertaken to address this. The report impliesthat the committee did not identify legitimateconcerns and did not take into account theviews of stakeholders.

Sources of advice and scientific data

505 The sources of scientific advice weregiven a taxonomy by Sir John Fairclough, whocited ‘Home grown Chief Scientists, full orpart-time CSAs drawn from industry oracademe, independent advisory committees,consultants, scientists and engineers fullyintegrated into Departments at the workinglevel’. This study did not find otherclassifications of sources of advice thatexpanded upon this list. There is sparseguidance for policy-makers on the selection ofadvisers.

506 There is little discussion of sources ofdata in the guidance literature. The Scienceand Technology Committee suggested thatACRE and Advisory Committee on Non-food


70

Products should not use anything other thanan applicant’s own data in their assessments,and supported evidence to the committee fromNovartis that ‘regulation by review ofcompanies’ risk assessments works well inpharmaceutical and pesticide regulation.’ Thegovernment broadly endorsed this view in itsresponse (House of Commons Science andTechnology Committee, 1999). However, thisapproach does not acknowledge the value ofevidence from other sources, the importanceof a balance of access to the advisoryprocess, and the ability of companies to editevidence they present. Neither does itrecognise the much more tightly defined rulesand greater degree of rigour in analysisrequired for pharmaceuticals, and the clearerprecedents explaining how the burden of proofoperates in pharmaceuticals testing.

507 It has also been suggested thatcommittee work and research should becoordinated with other similar workinternationally.

Selection of advisers and conflicts ofinterest

508 The Science and Technology Committeerecommended that no interests had a right tobe represented on scientific committees, andthe government endorsed this view (House ofCommons Science and TechnologyCommittee, 1999). They both rejected thesuggestion that scientists’ integrity isautomatically compromised by associationwith industry. If advisers with associationswere to be excluded, the government woulddeprive itself of access to many of the bestadvisers (House of Commons Science andTechnology Committee, 1999). The committeeasked for clear guidelines on the disclosure ofinterests, including annual disclosure,transparent procedures for review, and clearcriteria for decisions on whether interests arematerial. The government responded thatthorough safeguards are in place to ensurethat interests are declared, but did not answerthe other points. No guidance on assessingthe materiality of interests was found.

509 There is some guidance in the Code ofPractice for Public Appointments (Office of theCommissioner for Public Appointments,undated). This contains few details, and manyof the recommendations within it are difficult tointerpret. They are summarised below:

• the ultimate responsibility forappointments rests with Ministers;

• no appointment shall take place withoutfirst being scrutinised by a panel whichmust include an independent assessor;

• Board members must be committed tothe principles and values of publicservice;

• the principles of open Government mustbe applied to the appointments process;and

• the appointments procedures need to besubject to the principle of proportionality.

510 The Code has several shortcomings. Itdoes not explain what independent means, orwhether, if the rest of the assessors are notindependent, the independent assessor hasthe right of veto. There is no guidance onwhether shortlists for appointments should besubject to public consultation before theappointments are made. The principles andvalues of public service are not articulated orreferenced and principles of open Governmentare not elaborated.

511 The Science and Technology Committeerecommended that scientific advisorycommittees should draw one-fifth of theirmembership from lay experts with abackground in ‘other, not necessarily scientific,disciplines, to ensure that the evidence issubjected to a sufficiently questioning reviewfrom a wide ranging set of viewpoints’ (Houseof Commons Science and TechnologyCommittee, 1999). The government felt, inresponse to this recommendation, that arequirement to have lay members was notappropriate, but that it should be at theMinister’s discretion.

512 Sir John Fairclough (1990) laid anemphasis on external advice that is notapparent in more recent treatments of thesubject. His criteria for selecting advisersincluded:

the Departments should haveindependent advice of the highestcalibre in order to complement internaladvice, to act as a check and abalance to internal advice, and tointroduce a wider perspective . . .Where the Chief Scientist is drawnfrom the career civil service, this isparticularly important.

Rights of advisers

513 The Civil Service Code containsprotection for civil servants, as listed below.This might be considered for application toscientific advisers. The civil servant has a rightto appeal:


71

where ... he or she is being required toact in a way which:

• is illegal, improper, or unethical;

• is in breach of constitutionalconvention or a professional code[which would need clarification inthe case of scientific advisers];

• may involve possible mal-administration;

• is otherwise inconsistent with thisCode. (Cabinet Office, 1999a)

Function of the secretariat

514 The secretariat is bound by the CivilService Code. It has a duty to the government(Administration). Its duty to the advisorycommittee, if any, is not clear. The code isover-ridden by ‘existing statutory or commonlaw obligations to keep confidential, or todisclose, certain information’. According to theCode, ‘Civil servants should not withoutauthority disclose official information whichhas been communicated in confidence withinthe Administration, or received in confidencefrom others.’ There is an important question asto whether they can pass on this information toadvisers. Secretariats may also need guidanceon the implementation of the Freedom ofInformation Bill with respect to scientific advice(Cabinet Office, 1990).

515 The Science and Technology Committeerecommended that the secretariat should be ofsufficient size to ensure the efficient working ofthe committee. In its reply, the governmentsaid that the secretariat’s principal functionwas to ensure that decisions on whether tolicense releases are taken based on ACRE’sadvice (House of Commons Science andTechnology Committee, 1999). This may be asubtle difference from a principal function insupporting the committee itself.

516 The government has stated thatsecretariats should have sufficient scientificcompetence to understand the issues they arehandling (House of Commons Science andTechnology Committee, 1999). In this case,training should be available. As Sir JohnFairclough (1990) noted:

There will be an increasing need notjust for top level science andtechnology advice but for respectedand competent scientific and technicaladvice at all levels within Government.And this will need to be fully integrated

into the machinery that informs widerpolicy making and implementation.

Consultation and involvement ofstakeholders

517 The Cabinet Office (2000b) has providedguidance on how to conduct consultations ofdifferent sorts. It recommends consultationfrom the start of the planning process for anew policy or service. With respect to thebiotechnology regulatory framework, thegovernment has asked the strategic advisorybodies to ‘undertake consultations/issueconsultation documents about specific issuesand publish details, ensuring that respondentsknow that their views have been listened toand acted upon if appropriate, or if not, whynot’. The Cabinet Office notes that ‘It isdesirable to keep as full an account aspossible of responses, formal and informal, toconsultations; both to ensure that everyone’sview is fairly considered, but also to helpaddress any allegation of privileged access.’

518 It is not apparent that this has been thecommon practice in the past. In laying downrecommendations for risk communication,ILGRA (ILGRA, 1998b) noted ‘it is theexception rather than the rule thatcommunication is treated as an integral part ofrisk management policy.’

519 The consultation process is not justabout framing the question, it is part of a two-way dialogue. The Home Office (1998) hasrecommended that ‘Decisions ... should bepublished promptly ..., with a summary ofviews expressed ... and clear reasons forrejecting options that were not adopted.’

Conclusions

520 The review of guidance has revealedimportant areas where no guidance isavailable, particularly to scientific advisers,where guidance is ambiguous or contradictory,and instances where it is not followed, as wellas cases of good practice and clear guidance.There is therefore a strong case for the studyto examine the guidance and recommendimprovements in both the guidance itself andits implementation.

|O|X|E|R|A| Case Studies

72

CASE STUDIES

Introduction to Case Studies

521 The following sections of this reportdescribe case studies investigated asexamples of the working of the scientificadvisory process.

522 Throughout the case studies, the focusof attention was on the process of securingexpert scientific advice and on the use of theadvice in decision-making, but not on whetherthe advice given by experts or the decisionstaken were correct.

523 Within each case study, the origin of thepolicy question, the definition of the scientificquestion, the selection and briefing ofadvisers, the process of generating scientificadvice, and the use of that advice in policy-making were all considered.

524 The case-study histories are based ondocumentary records, published bygovernment or public inquiries, or published

privately, and on a series of interviews withkey individuals involved in the cases. The casestudies are only concerned with the way inwhich the scientific advice was sought,provided, and used—not with the scientificissues themselves. The case studiesnecessarily contain a mixture of factual recordand the personal perspectives of thoseinvolved. These are recorded withoutcomment or attribution.

525 Some of the case studies examineissues that are ongoing, and on which newscientific advice is being sought, and newpolicy being developed. The focus of this studyhas been on the advisory processes, ratherthan the advice itself, so the case studiesconcentrate on the historical elements of theseissues, rather than the more immediate andcontentious aspects.

|O|X|E|R|A| Case Study 1

73

E. coli O157 Food Poisoning in Scotland

Origins of the Issue

526 Many strains of the bacterium E. coli canbe present in the human gastro-intestinal tractwithout causing any illness. Certain strains,however, produce toxins that are extremelyharmful, especially to young children and theelderly. One such strain is E. coli O157:H7,which was the agent responsible for anoutbreak of food poisoning in central Scotlandin late November 1996 that resulted in severalhundred infections and 18 deaths.

527 This particular outbreak was traced to abutcher’s shop, where poor hygiene and poormaintenance of separation between cookedand raw meat are thought to have led to theincident. Furthermore, unknown toenvironmental health officials, a substantialwholesale business was in operation at thesame premises, and thus the potential scale ofthe outbreak was not immediately apparent.

528 E. coli O157 was first associated with asignificant outbreak of infection in the USA in1982, which was eventually traced tocontaminated hamburger meat. There was amajor incident in Washington State in 1993, inwhich 700 were infected and 4 died, while, inthe UK, isolated cases were also observed asearly as 1982. By 1995 it had been noticedthat there was an unusually high incidence inScotland. An outbreak in West Lothian in 1994affected 100 people, and one child died; theoutbreak was traced to contaminated milk.

529 E. coli O157 is found in the gut of cattle,sheep and pigs, in which animals it does notappear to be pathogenic. It is believed that themain reservoir is cattle, and that approximately15% of these animals may harbour theorganism.

530 The source of human infection is almostexclusively contaminated cattle faeces. Thereare also instances of human-to-humantransmission, due to handling soiled clothing,for example. A common route is consumptionof meat that has become contaminated withfaeces during the slaughtering process, andhas not been properly cooked. However, directcontact with farm animals may also sometimesbe a route of infection.

531 At the time of the central Scotlandoutbreak, all of the above was known topublic-health experts in Scotland, although fewgeneral practitioners or environmental healthofficers would have had professionalexperience of E. coli O157 food poisoning.

The symptoms of infection by this organismare common to a number of other diseasestates, and confirmation of diagnosis is bymicrobiological testing.

532 This case study examines the roleplayed by expert scientists in the discovery ofthe outbreak and the response to it—both incontainment and in subsequent developmentof policy to prevent its repetition.

Framing of the Question

533 One of the special features of this caseis that the scientific advisory process wastriggered by the occurrence of anemergency—an outbreak of serious foodpoisoning. Thus, one of the key problems wasdetection. A subsequent problem was thecorrect identification of the cause and sourceof the outbreak, and the final problem was todevelop policy for the prevention of futureepisodes.

534 Although each of these issues wasexamined in the course of this case study, itemerged that the first two problems weresolved by procedures that already existed,without additional special recourse to expertscientific advice. These problems are onlycommented on briefly here. The third problem,however, did prompt expert scientific advice tobe sought, and is developed in much moredetail.

535 The Health Boards and the laboratorystaff and general practitioners had policies andprocedures in place for detecting outbreaks offood poisoning.

536 The requirement for fundamentalscientific advice was minimal because thepathogenic organism had been identified atthe detection stage, and its characteristics andtypical modes of transmission were wellknown. However, experts from severaldisciplines were engaged to assist indetermining the source of the outbreak so thatit could be controlled.

537 The requirement for policy advice on thefuture safety regime was a problem that couldbe addressed more formally, but still with adegree of urgency. It was an issue that hadalready been considered by standing advisorycommittees [543, 552], but this outbreakstimulated new advice.


74

538 The questions posed by the ScottishOffice, as will be seen below in the terms ofreference of the Pennington Group [547–548],sought recommendations for action, and werethus not entirely limited to scientific issues.

539 Although there were many calls for apublic inquiry, the accident was formallyinvestigated through the mechanism of a fatalaccident inquiry (FAI). Owing to the legalframework of such inquiries (the FatalAccidents and Sudden Deaths Inquiry(Scotland) Act 1976), matters relating to the255 non-fatal infections were not considered,and the Sheriff did not have authority to makerecommendations. In addition, theconcurrency of the Pennington Group’s workwith the criminal proceedings against thebutcher at the centre of the case and hiscompany may have restricted the freedom ofthe Group to comment on those aspects of thecase that were sub judice.

540 It appears that, despite the conflictingpressures of this complex arrangement ofinquiries, all the relevant facts did emerge.

Selection and Briefing of theAdvisers

541 The detection of the outbreak may beviewed in two parts: first the detection ofpractices that were hazardous, before theevent; and second, the detection of the eventitself.

542 Local authority environmental healthofficers were responsible for carrying out aregime of inspection and monitoring ofbutchers’ shops. Although the effectiveness ofthis regime was a subject of the PenningtonGroup’s inquiries [548], the detection ofhazardous practices itself did not requireexpert scientific advice.

543 An emergency-response plan wasalready in place that provided for the formationof an outbreak control team (OCT),responsible for responding to any suddenepidemic identified in the human population.This plan had been set out by the ScottishOffice Department of Health Advisory Groupon Infection in 1996 (Scottish Office, 1996).

544 This document set out the statutoryrequirements for Health Boards and localauthorities to cooperate over outbreak control,and defined their shared and individualresponsibilities. The Health Board informedthe Environmental Services Department ofNorth Lanarkshire Council about the

occurrence on the first day of the emergencyand an OCT was formed the following day.

545 The OCT for this incident comprised thefollowing.

• A Consultant in Public Health Medicine(Communicable Diseases/EnvironmentalHealth) from Lanarkshire Health Board(Chairman).

• The head of Protective Services, NorthLanarkshire Council.

• Representatives of:

− the Lanarkshire Health Board;− local hospitals and National Health

Service Trusts;− the Scottish Centre for Infection

and Environmental Health;− the Wishaw Health Centre; and− the Scottish Office (as observers,

from November 28th).

546 It was a tactical team of people withexecutive powers and good local knowledge.The team was effectively chosen in advance ofthe outbreak, following the procedures of theemergency-response plan. The team wastherefore not necessarily independent of allthe organisations responsible for theprevention of food poisoning.

547 In a separate development, thePennington Group was appointed by theScottish Office, only six days after thediscovery of the outbreak, with the followingterms of reference:

to examine the circumstances whichled to the outbreak in the central belt ofScotland and to advise the Secretary ofState for Scotland on the implicationsfor food safety and the general lessonsto be learned.

548 The Pennington Report adds, in relationto the terms of reference:

We were asked to examine the presentknowledge of E. coli taking intoaccount scientific research in this area,and the adequacy of presentarrangements for, and guidance on,handling food poisoning outbreaks.The Secretary of State asked us to lethim have any priority recommendationswe wished to make by the end of 1996.

549 All the members of the Group wereappointed by the Scottish Office. TheChairman, Professor Hugh Pennington, was,and remains, a prominent academic and a


75

leading expert on the organism involved in theoutbreak. He did not play any part in theselection of the other members, who were:

• Dr John Cowden, ConsultantEpidemiologist, Scottish Centre forInfection and Environmental Health;

• Dr Helen Zealley, Director of PublicHealth, Lothian Health Board, and Chair,Scottish Group of Directors of PublicHealth;

• Mr John Summers, Director of Technicaland Leisure Services, Moray Council,and Immediate Past President of theRoyal Environmental Health Institute ofScotland;

• Mrs Ann Foster, Director, ScottishConsumer Council;

• Dr Rosalind Skinner, Head of TheScottish Office Department of Health,Public Health Policy Medical Division;

• Mr Stephen Rooke, Chief Food andDairy Officer, The Scottish OfficeAgriculture, Environment and FisheriesDepartment.

550 The establishment of the Group as aseparate entity from the existing UK-widebody—the ACMSF—may have reflected theneed for a focus of expertise in Scotlandwhere E. coli O157 was unusually common. Inscientific terms, there was no obstaclebecause the required expertise was readilyavailable in Scotland.

551 There was one consumer representative,and all of the other members were eitherpublic servants or academics. There was noveterinary expert. Overall, therefore, the groupwas not constituted to be well balanced,although its members had first-handknowledge of the issues.

552 The composition of this Group wasdifferent in character from that of the standingACMSF, which had a broad mix ofmembership from local government, clinicalmedicine, medical science, the food industry,public health and consumer-interestbackgrounds. Members of the ACMSF were,at the time, predominantly, but not exclusively,expert scientists. Central-governmentrepresentation was relegated to a second tierof ‘assessors’ who attended meetings.However, the ACMSF working group onverocytotoxin-producing E. coli (of which themost important is O157:H7) had eightmembers broadly reflecting the composition ofthe parent committee, but with a greateremphasis on scientific expertise. There was aconsumer-interest representative on thisworking group.

Preparation of the Advice

553 The Pennington Group took both writtenand oral evidence from all of the interestedparties, visited the butcher’s premises andinterviewed members of the OCT. Itinterviewed senior officials from the variousregulatory departments and agencies involvedin both prevention and emergency response. Italso studied the histories of previousoutbreaks.

554 It considered papers submitted by itsown members and, more generally, it reliedupon their background in the field. There wasa feeling within the Group that there was anopportunity to make a change—in particular, toreverse the deregulatory tendency that hadcharacterised the control regime in the yearsbefore the outbreak. There was also a senseof mission—to improve food safety. The Groupsat from early December 1996 and producedits final report on April 8th 1997.

555 An interim report was issued on January15th 1997, addressing research into theincidence of E. coli O157 in animals,surveillance of pathogens in food, enforcementmeasures and outbreak management. Onenforcement, the Group recommended:

• legislation on selective licensing of foodpremises;

• physical separation, within foodpremises, of raw and cooked meat;

• measures to promote awareness of thehealth risks;

• acceleration of the introduction of theHazard Analysis and Critical ControlPoint (HACCP) system—the emergingglobal methodology for assessing andcontrolling food-related risks.

556 It was felt that the OCT had acted swiftlyand with vigour, and was ultimately successfulin containing the outbreak. The PenningtonReport stated that:

the guidelines, and the arrangementsput in place locally for outbreakmanagement and control, apparentlyworked reasonably well in practice—albeit that the nature and scale of theoutbreak presented a stern test andchallenges to the system.

557 Nevertheless, the OCT at the timeidentified gaps in scientific and technicalunderstanding that hindered its work.

558 The final report of the Pennington Groupproduced further recommendations, including:


76

• increased awareness of E. coli O157among farm workers, and precautions atfarms to avoid transmission of infectionto farm workers;

• improvements in cleanliness of animalspresented for slaughter, and theadoption of HACCP and other measuresat abattoirs to minimise the risk ofcontamination of carcasses;

• pending implementation of HACCP,arrangements for selective licensing ofpremises not already covered by theMeat Products (Hygiene) Regulations1994;

• licensing to ensure: staff training, record-keeping to facilitate product recall, andphysical separation of raw and cookedmeat, and of equipment and staffinvolved in their preparation and sale;

• various measures relating to policy andresources for enforcement, research,surveillance and outbreak control.

559 The report stated that, while the Grouphad studied the central Scotland and otheroutbreaks exhaustively, it had also taken intoaccount more general concerns about theincidence of food poisoning, and the widersocio-economic impacts of the preventativemeasures it was proposing.

560 The Pennington Group recommended, inits interim report, that the government requestthe ACMSF to review the Pennington Group’sguidance on cross-contamination. This wasdone, and the report by the Working Group setup by ACMSF for this purpose was appendedto the Pennington Group report. In summary, itconcurred with the idea of expediting theimplementation of HACCP, but it said that aprescriptive interim regime would send thewrong messages to duty-holders and wouldlull them into the false sense of security thatcode compliance offers. It did not accept thatspatial separation of raw and cooked meatswas necessary in all food businesses. Thistechnical point reduced to a dispute as towhether separation in time (subject to suitabledisinfection procedures) could be equallyeffective as separation in space. The ACMSFWorking Group felt that the achievement offood safety had more to do with the level ofawareness of personnel involved in theindustry than with adherence to rules. It feltthat hazard analysis would identify the keyprocesses for which separation wasnecessary, and would therefore offer a greaterassurance than any prescriptive scheme.

561 The Pennington Group, in its final report,commented on this as follows:

We welcome the broad level ofagreement that clearly exists betweenour Group and the ACMSF and thesupportive nature of the WorkingGroup’s report. We have consideredvery carefully the points that have beenmade about the issue of separation.We cannot share the ACMSF’sconfidence in existing food safetylegislation, however rigorouslyenforced, as an adequate means ofprotection for the public against E. coliO157. For that reason, and for thereasons set out earlier in this Chapter,we therefore remain persuaded of theneed for the physical separationmeasures we have specified and of theneed for licensing to bridge the gapuntil HACCP is universallyimplemented.

Use of the Advice in the PolicyDecision

562 The interim report of the PenningtonGroup was received on January 15th 1997and a statement was made in Parliament bythe Secretary of State for Scotland the sameday. The interim recommendations wereaccepted in full. Directions were given to therelevant Scottish Office departments, HealthBoards and local authorities to review theirprocedures and to respond in time for theirviews to be taken into account during theremainder of the Group’s investigations.

563 When the final report was published onApril 8th 1997, a further statement was madeby the Secretary of State for Scotland, inwhich it was again declared that all of therecommendations were accepted. On theissue of licensing, the statement confirmedthat immediate action had been taken toensure that the existing powers were fullyutilised and that regulatory attention wasfocused on those premises presenting thegreatest risk. A commitment to decide on thefunding of local authorities’ increasedenforcement activities was promised as soonas possible after the general election on May17th 1997 (by then imminent).

564 The statement went on to confirm thegovernment’s intention to implement,throughout the UK, selective licensingarrangements for certain premises not alreadycovered by the Meat Products (Hygiene)Regulations 1994. In effect, this appeared tomean that the residual area of differencebetween the Pennington Group and theACMSF had been decided in favour of theformer.


77

565 The licensing measures have not yetbeen implemented, although they are due tocome into force in October 2000.

Observations

566 Two members of the Group had directand ongoing involvement in the control of theoutbreak—their interest in that matter couldhave caused a conflict, but it does not seem tohave done so. Another potential difficulty lay inthe Chairman’s involvement in research in thefield, which could have been perceived as aconflict of interest affecting therecommendations of the Group that pertainedto research.

567 There was considerable press interest inthe progress of the outbreak, and the work ofboth the OCT and the Pennington Group.

568 The FAI report favoured a smaller teamfor the OCT, and suggested that, in view of thelimited personnel resources available to somelocal authorities, it might be better to have acentral team to cover the whole of Scotland.Such a team would be better qualified, wouldhave authority to cross local-authority borders,and powers to close down businessoperations, etc. Such a group could also beindependent of the local bodies (although stilldependent upon them for information).

569 Much has been made of the differencesof professional opinion between thePennington Group and the ACMSF, andbetween the Scottish authorities and those inEngland and Wales. In this study, a highdegree of agreement between these partieswas observed, although there was a differenceover the use of prescriptive standards as ashort-term solution.

570 The Pennington Group stronglyendorsed the HACCP approach as the long-term solution, thus aligning policy in Scotlandwith that in much of the rest of the world. Thiswas also the solution favoured by ACMSF andMAFF. The proposal to introduce licensing ofcertain premises based on certain prescriptivestandards was seen as ‘a short-term measurewhile HACCP systems are being introduced’(Pennington Group report, Chapter 4).

571 Traditionally, safety regulation in manysectors has been prescriptive, setting outminimum standards of behaviour that can bereadily verified by inspection. More recently,‘goal-setting’ approaches have come intofavour, in which the emphasis is on overallperformance, and the methods are left moreopen. In the latter regime, as applied to food

safety, responsibility is firmly pinned on theparticipants in the supply chain, and the focusof regulation by government is on the auditingof management systems, with reduced levelsof policing performance in premises. TheHACCP system is a management tool thatduty-holders can use to discharge theirobligations under such a regime, and theinspectors can audit it.

572 This case was characterised byambiguity about the function of key advisorybodies, regulation and stakeholder interests.For example, both the ACMSF and thePennington Group comprised a mix of expertsand stakeholders. The two groups came todifferent conclusions on aspects of theregulatory mechanism, possibly because ofthe different mixture of stakeholderrepresentatives on the two committees.

573 The Pennington Group came to its taskwith many years of experience and well-formed views. This was seen as a virtue in asituation where urgent action was required,although it inevitably brings a risk of bias, or offailure to assimilate evidence objectively.

574 The Scottish Office appointed the Group,and the procedure does not appear to havefollowed any particular guidelines, such asthose of Nolan.

Robustness of the Process

575 The public interest was representedindirectly through the very few consumerrepresentatives on the various committees.The FAI sat in public, but its powers weresignificantly restricted, especially in the area ofrecommendations. Thus, public participationwas very limited, and, if the outcome had beenless clear-cut in assigning blame for theoutbreak, the public may have been less readyto accept subsequent policy changes.Therefore, this process may not suffice forsome other outbreak where the blame for theincident could not be assigned.

576 The recommendation by the PenningtonGroup to pursue the licensing regime has notyet been implemented (although it is expectedthat it will be implemented in Scotland inOctober 2000). There was insufficient generalsupport for licensing, especially as atemporary measure that did not fit well with theagreed long-term solution.

577 The Group recommended a specificcourse of action, instead of limiting itself toscientific issues. In part, this was because theGroup had never seen itself as purely an


78

expert body, more as an ad hoc working party,with an objective to recommend new policy.This ambiguity risked the following possibleoutcomes:

• the Group is actually an expert scientificbody, but nevertheless makes policyrecommendations dependent onexpertise it does not possess;

• the Group is actually a stakeholdergroup, and gives advice that appears tobe scientific, but is in fact partially drivenby policy aims. This distorts itsinterpretation of the science, and risksthe development of policy that conflictswith scientific consensus and may beineffective.


79

Margins Around Field Trials of Genetically Modified Crops


578 A great deal of controversy surroundsthe developing, testing, licensing, marketing,and growing of GM food plants, and the use ofGM crops in food products. The policyquestions facing decision-takers are complex.Scientific advice, along with many otherconsiderations, has informed the developmentof policy in this area.

579 This case study on margins around fieldtrials of GM crops has been carried out inorder to illustrate the process by whichgovernment has taken scientific advice toanswer a particular, well-defined question onthe width of margins around fields where GMcrops are being grown on a trial basis. Thisparticular question has been chosen as anexample of how government has takenscientific advice and used it to inform policy-making; it is not intended to cover all theissues raised by government’s handling of thedebate on GM food. The wider debate on GMfood is beyond the scope and remit of thisproject, which examines particularly theprocess by which government obtains expertscientific advice, and incorporates it into policydecisions.

580 Margins, also called separationdistances, define a minimum gap between GMcrops and other crops. They are designed tomaintain a low incidence of cross-fertilisationof non-GM crops outside the delimited trialareas, and to provide a physical separationfrom natural habitats. Volunteers (plants thatgrow from seeds that drift across fieldboundaries or are spilled, or plants that regrowfrom previous years’ use of the land), are alsocontrolled by management of the site bothduring and after the trial.

581 The licensing of GM crops, for trials orcommercial production, is carried out inEngland by the DETR and MAFF acting jointly.Their concern is that the GM crop shouldcause no harm to the environment. Consentsfor experimental trials may include conditionsthat specify the width of the margins, and mayalso prescribe what may, or may not, beplanted in neighbouring fields. Consents forcommercial production are only awarded tothe owner of the seed on an EU-wide basis bymember states acting jointly, and are ageneral permit to allow the seed to be sold tofarmers. However, consents do not govern thebehaviour of the consent-holder’s customers(farmers).

582 A critical issue for the agriculturalindustry is the purity of seed varieties. MAFF isconcerned that there should be no commercialimpact on other farms from trials orcommercial production of GM crops. This isquite separate from any consideration of thehealth or environmental implications of trials ofGM crops, which are governed by thelicensing process. MAFF, however, has noauthority to impose conditions on growers.

583 These functions of the DETR and MAFFare linked under legislation and by aMemorandum of Understanding, and in thatboth aims—protecting the environment fromvolunteer plants, and protecting neighbouringcrops from contamination—depend (partly) onthe width of margins around the field.


584 There are two separate questions.

• What margin is necessary to protect theenvironment during a trial?

• What margin is necessary to protect thecommercial interests of neighbouringfarms during production?

585 The first is posed by DETR/MAFF toACRE; the second has been addressed byMAFF.

586 DETR/MAFF pose the first question toACRE explicitly as part of the process oflicensing a trial. The applicant proposesmargins, and ACRE may either accept orincrease them, or recommend that theapplication be rejected [594, 602].

587 DETR/MAFF have to ensure safety forhuman health and the environment. It invitesACRE to apply the precautionary principle, butexpressly states that the committee is notmandated to weigh up the benefits. TheEuropean and UK legislation requires thedecision on permitting a release to be takensolely on the basis of safety of human healthand the environment. Any potential benefitscannot be taken into account. This isanalogous to the Medicines Control Agency,which does not consider efficacy. There is thusno attempt at risk–benefit analysis. UK policywas explained in March 2000 by Mr MichaelMeacher, Environment Minister, who said ‘Iwant to make it perfectly clear that there will


80

be no commercial growing of GM crops in thiscountry until we are satisfied that there will beno unacceptable effects on the environment.We can only find this out by testing under farmconditions’.2

588 However, ACRE’s perception is that itought to consider benefits. Concern has beenexpressed that ACRE is widely expected todecide whether the benefits of a proposed GMproduct outweigh the risks, which is outside itscompetence.

589 MAFF became aware of industrialconcerns about cross-contamination, andconsulted the industry in 1997, in order toanswer the second question given above. Thetrade association, the Supply Chain Initiativeon Modified Agricultural Crops (SCIMAC),offered to investigate the issue, and preparedguidelines [596]. MAFF allowed the industry toframe the question of separation distances foritself, arguing that this would be more effectivethan waiting to create new regulations.

590 ACRE is aware of the second questionon the commercial purity of seed, in that itassumes that applicants have prepared theirapplications in accordance with the SCIMACguidelines. If different separation distances arerequired to protect the environment, ACREcan specify these.


591 The DETR, MAFF and the devolvedadministrations appoint experts to sit onACRE. A number of new members wereappointed to ACRE in 1999 following theexpiry of the previous Committee’s term ofoffice. New ACRE members were appointedby strict adherence to Nolan principles. Thevacancies were advertised widely, and anindependent sift panel was appointed to judgethe applications against pre-agreed andpublished criteria, and to draw up a long list ofpotential candidates, solely on that basis.Interests (such as employment by the agri-biotech industry or pressure groups) were notconsidered. Ministers then selected expertsfrom this list. The press reported a ‘purge’ ofinterests—the selection of only experts withoutany industrial affiliations from the long list.

592 ACRE members are paid the standarddaily allowance (around £150 in 1999) for

2 http://www.cabinet-office.gov.uk/gm-info/1999/gmcrops.htm

attendance at meetings, and they receivetravel expenses. In practice, they do muchunpaid work in preparation for meetings.

593 MAFF used its own staff, in the ChiefScientist’s Unit, to comment on the draft Codeof Practice on crop management, prepared bySCIMAC. MAFF also commissioned the JohnInnes Centre to research the specific issue ofthe relationship between GM crops andorganic farming. MAFF also took advice fromthe DETR.


594 A critique of each application is preparedby DETR staff for ACRE. This is a scientifictask, in which the staff pre-filter theapplications before ACRE’s assessment, toreject those which are bound to fail for reasonsof incompleteness, or other errors in preparingthe application. ACRE can recommend thatMinisters accept or reject an application, and itcan recommend amendments.

595 A crucial scientific judgement is whetherthe GM crop has a general survival advantagethat would lead to amplification of geneticstock in subsequent generations. Such risksare always considered as a specificrequirement of the legislation that has been inforce since 1993. To date, it has always beenconcluded that the genetic modification onlyprovides an advantage in the context of amanaged environment (eg, herbicide toleranceis specific to a single herbicide, and otherherbicides or natural hazards affect GM andnon-GM crops equally). However, there may,in the future, be applications for crops modifiedto be resistant to fungus or invertebrate pests,which could contribute to the risk ofamplification.

596 SCIMAC prepared a draft Code ofPractice covering many aspects of theintroduction of GM crops. It identified andmandated good practice in crop rotation andseparation to avoid contamination (a non-pejorative term, meaning the unintendedpresence of a foreign plant or part of a plant).The industry has a long-established set ofseparation distances to allow different varietiesof the same species to be sold as certifiedseed. The aim is typically 98% purity—that is,up to 2% of certified pure seed may be otherthan the intended variety. Similar argumentshave been used for organic crops, where theseparation distance between certified organiccrops and non-organic crops is such thataround 98% of the harvested crop will satisfythe organic growing rules.


81

597 The SCIMAC draft Code of Practicerecommended that the same separationdistances should be used between GM andnon-GM (including organic) crops as is alreadyused between certified seed varieties. This isbased on the assumptions that:

• the same level of purity is commerciallyacceptable;

• GM crops have no survival advantagethat would lead to amplification ofcontamination in successive generations.

598 MAFF staff reviewed this and found thatthe assumptions were acceptable. MAFFasked SCIMAC to include in the final versionof the Code of Practice an explicit statement ofthe separation distance required, rather thanjust stating that it was the same as thatrequired for certified seed.

599 The SCIMAC Code of Practice requiresdifferent separation distances for differentseeds, and for different levels of certification ofthe same type of seed. The required distancesrange from 6 to 600 metres. As the industryassociation, SCIMAC is able to force farmersto follow its guidelines by withdrawing accessto GM crop varieties following infringements,through its control of the supply of seed.

600 A study undertaken by the John InnesCentre, commissioned by the Soil Association,concluded that a separation margin of sixmiles would be needed. This impractical figurewas the result of setting effectively a ‘zero-risk’standard. This may be contrasted with theorganic agriculture industry’s position withregard to cross-contamination of organic bynon-organic crops, which is far less extreme.The zero-risk target, if accepted as necessary,would be a major challenge to the GM industrybecause it is impossible to eliminate allcontamination—a single bee may fly manymiles and deposit one grain of pollen from aGM crop onto an organic crop. The study bythe John Innes Centre recognised this:

No system for the field production ofseed can guarantee absolute geneticpurity of seed samples. Very rarelylong distance pollination or seedtransfer is possible, so any criteria fororganic crop production will need torecognise this. There has always beenthe possibility of hybridisation and seedmixing between organic crops and non-organic crops. Organic farmingsystems acknowledge the possibility ofspray or fertiliser drift from non-organicfarming systems, and procedures areestablished to minimise this.

601 The study also drew attention to thepractice of sourcing organic seed from othercountries where the production criteria may bedifferent. The overall conclusion is that someform of separation distance, together withother measures such as cleaning ofimplements, will have to be adopted by theorganic agriculture industry.


602 ACRE’s recommendation has alwaysbeen accepted by the DETR and MAFF.

603 If ACRE recommends that a particulartrial be permitted, it is on the basis that nosignificant harm to the environment will resultfrom the trial.

604 MAFF’s policy decision was to adopt theSCIMAC Code of Practice. When it waspresented to the public, a widespreadmisunderstanding developed that theseparation distances would ensure that nocontamination would occur. The John InnesCentre study was then seen to underminegovernment policy because it stated expresslythat there would be contamination.

Observations

DETR

605 The function of ACRE is crucial, althoughit is not clear that ACRE perceives its ownfunction in the same way as the DETR. Theabsence of any formal risk–benefitassessment, together with Ministerial wishesnot to allow any GM product to be grownunless it is safe, indicate that consent is onlygiven if risks are negligible. This is also arequirement of the legislation. It could beargued that, if there are risks, benefits shouldbe assessed, and only if the benefits outweighthe risks should consent be granted.

606 Expert panels sometimes have lay orrepresentative members; ACRE used to haveone member from an environmental andpublic-interest group. The new ACREmembers have no industry links, and wereselected from a list of experts (leaving no roomfor a non-specialist). The Committee has norepresentatives or lay members.

607 The organic agriculture industry has notaccepted the principle of an acceptable levelof contamination, so their concerns have notbeen answered by the use of margins.


82

MAFF

608 MAFF has scientific expertise within itsChief Scientist’s Group, which was able tocomment on the draft SCIMAC Code ofPractice. It was also able to commission theJohn Innes Centre to carry out an independentscientific analysis.

609 The agriculture industry recognises theneed for sensible regulation to maintain purityof crops. It therefore funded andcommissioned its own scientific advice,leading to a policy for government to endorse.Self-regulation is arguably quicker andcheaper than statutory control, and, in thiscase, has introduced commercial penalties fornon-compliance that would be difficult toimplement within the legal system. Theindustry welcomed Ministerial support andcooperated with MAFF to ensure that thedecision was one that the Minister couldendorse.

610 The problems resulting from the JohnInnes Centre report showed that a more openexplanation of the risks, although difficult topresent, would have pre-empted thesubsequent embarrassment. The policy hadbeen portrayed (and possibly presented) assetting separation distances which ensured

that there would be no contamination, whereasthe policy had always been that there wouldbe an acceptably low level of contamination.


611 The findings of this process (ie, consentscontaining separation distances) were, in theevent, challenged by pressure groupsopposed to genetic modification in principle,whose position was that the acceptable levelof cross-contamination should be zero. Thathas left the issue unresolved because thewider issues surrounding the acceptability ofGM food have not been addressed.

612 Surprisingly, the reliance on the SCIMACguidelines has not attracted much challenge,but it remains a weak point in the processbecause of the lack of independence ofSCIMAC in the matter.

613 The two government departmentsinvolved necessarily had different approachesto setting separation distances, which weredifficult to explain to a wider audience.


83

The Structural Integrity of the Forth Rail Bridge


614 The Forth Bridge was built over 100years ago. Its maintenance has been part offolklore—as soon as the painters finish, theystart again at the other end. In the 1990s therewas public unease about the state of thebridge. The maintenance regime had changed,the bridge appeared to be rusty, and materialhad fallen from it.

615 The issue was of great politicalsensitivity because of the recent privatisationof the rail network, and because this Scottishlandmark was under the stewardship of acompany with headquarters in England. Theroutine assessment and approval of thestructural integrity of a bridge became thesubject of hostile local press attention.

616 Stephen Norris, then Minister withresponsibility for transport and roads, took upthe issue, in response to questions raised inParliament. He requested advice from HMRailway Inspectorate, within the HSE.


617 The Minister informally posed thequestion of the safety of the bridge to theDirector General of HSE. The latter, with HMRailway Inspectorate, identified the essentialissues and framed two questions.

• Is the bridge safe to carry the weight ofthe trains that use it?

• Is the future of the bridge secure?

618 The first question required a conditionsurvey (to establish the current state of thestructure), and a structural analysis (toestablish whether the existing structure metcurrent design codes). The second required ananalysis of the maintenance regimeestablished and operated by Railtrack.

619 HSE conducted an initial inquiry using in-house expertise to establish whether therewas a substantial case to consider. Thisconvinced Railtrack (which is responsible formaintaining a safe bridge) that it should takeaction. This is standard practice whereactivities are regulated by a safety-caseregime. The process is one of constructivechallenge and dialogue in which either theduty-holder satisfies the regulator that themeasures in place are suitable and sufficient,or the regulator acts to secure remedial action,either by persuasion or compulsion.


620 HSE recognised that it was Railtrack’sduty to provide a safe bridge and, therefore,that Railtrack should take the lead in checkingand demonstrating that the bridge was indeedsafe.

621 HSE and Railtrack agreed that:

• they would address the first questionjointly and set up a joint steering group tooversee the project;

• Railtrack would appoint and pay aconsultant (Pell Frischmann) to carry outthe condition survey and structuralanalysis;

• Railtrack would remain responsible fordecisions based on the consultant’sresults;

• HSE would place its own experts (fromthe Health and Safety Laboratory andelsewhere) on the technical sub-committee of the steering group.

622 The consultant appointed was a frequentcontractor to Railtrack. The reputation of theconsulting company, and the professionalstanding of the individual engineers, were notin doubt. It could be argued that it would havebeen difficult to find a consultant that had thenecessary experience and was alsocompletely independent of Railtrack.


623 It was quickly apparent to the jointsteering group that a full structural assessmentwas not necessary to answer the questionsconcerning the safety of the bridge. Theexperts from all three parties (Railtrack, HSE,and the consultant) jointly agreedapproximations and assumptions that werereasonable and necessary to answer the twoquestions without complex modelling.Conventional engineering practice is to makeconservative approximations, although thereappears to have been no formal attempt toestimate the reliability of those approximationsin this case.

624 It was concluded that the answer to thefirst question was that the bridge was safe tocarry the weight of the trains that use it, but


84

that the answer to the second question wasthat the maintenance regime was notadequate to ensure its future. Consensus onthe definition of a new regime was reached,although there were some initial differences oftechnical outlook and methodology betweenRailtrack and HSE. Again, such differencesare part of the challenge and dialogue thatcharacterises the relationship betweenregulator and duty-holder in a safety-caseregime.


625 HSE issued a formal ImprovementNotice to enforce the immediate actions of therevised regime. Railtrack subsequently revisedits procedures for approximately 30 majorassets.

626 HSE prepared a short report thatsummarised the work carried out and theconclusions reached. It stated in absoluteterms that ‘the Bridge is safe, in its currentcondition, to carry Railtrack’s present loading.’The report included an engineering annex thatsummarised the consultant’s report and statedthe approximations that had been made.

627 The Minister accepted these withoutqualification, and HSE held a press briefingand published the report. The report appearsto have been accepted by the local public andno further action was required.

Observations

628 This case is instructive for severalreasons.

• HSE acted both as the link between thepolicy-maker and the expert scientificadviser (in this case, the duty-holderunder a safety-case regime) and as theregulator in producing evidence thatrequired the duty-holder to take action.

• Railtrack had responsibility formaintaining the bridge in a safe stateand this duty extended to the provision ofsuch information as the regulator (HSE)needed. The potential conflict of interest,arising from the duty-holder providingadvice on whether its duties were beingadequately discharged, was resolved bythe personal standing of the engineers(primarily those of HSE, but also those ofRailtrack and its consultant), and thestatutory authority of HSE.

• The methodology and the final reportwere based on the engineeringjudgement of all parties.

• HSE’s technical expertise was crucial, inauditing the work carried out by Railtrackand its consultant, and in endorsing theconclusions.

• HSE performed twin functions ofconstructive cooperation with the duty-holder (and its consultant) and as formalenforcement authority. Both parties wereaware that HSE held the power of veto.

• The outcome of the process wassuccessful.

629 The advisory process was not a simpleone-to-one relationship of adviser anddecision-taker; a network of experts wasinvolved. HSE helped the Minister to frame thequestion, audited work of the experts, andframed the conclusions for the Minister, actingas a policy-maker. The scientific informationwas generated by Railtrack and its consultant,in a process that was closely audited by HSE,represented by its own technical experts onthe steering group that managed the process,and the sub-committee that dealt withtechnical issues.

630 This interaction between HSE andRailtrack was necessary because railwaysoperate a safety-case regime, where the duty-holder must decide what it is going to do andsubmit its case for approval by the regulator.

631 The process was largely open, in that thepublished report to the Minister explained themethodology, assumptions andapproximations, and named all of the parties.It was not open, in the sense that there wereno stakeholders present when the work wasbeing conducted. HSE, as policy-maker andregulator, had to judge whether theapproximations and assumptions made in theanalysis were acceptable, weighing the riskthat the assumptions were invalid against thecost of a more exhaustive inquiry.

632 The Minister reached the conclusion thatthe report would be acceptable to the public.Subsequent public reaction bore this out,which suggests that the model of a trustedregulator, backed up by statutory regulatorypowers, was able to produce an outcomeacceptable to the general public and the pressin this case.


85


633 If the outcome of the approximatestructural analysis had been less clear-cut orindicated that the safety margin was smaller,then there would have been a need for muchmore detailed surveys and stress calculations.It is not obvious that the same team wouldhave been able to do this work, or how thefinancial resources would have been found.The extra costs could have been considerable,and this might have put pressure on the teamto avoid this course of action. The studieswould have taken longer, and the question ofwhether the bridge should be closed duringthe investigation would have to have beenaddressed. This would have been a decisionto balance a short-term safety risk againstconsiderations of commercial and publicdisruption.

634 HSE engineers and the Railtrack/PellFrischmann engineers could have disagreedabout methodology or results. It is not clearwhether the process actually employedprovided for this eventuality. If discussions onmethodology or results were based on theprofessional standing of the individualengineers involved, it is not clear that theengineers acting on behalf of the regulatorwould have prevailed in any disagreement, orhow the process could ensure that this wouldhave happened.

635 The Minister or the general public mightnot have accepted an outcome based, evenpartly, on the work of a firm of engineers,however reputable, which worked frequentlyfor Railtrack. Given the type of work involved,it is not clear that a frequent contractor toRailtrack would necessarily have been moresuitable (on grounds of competence or cost,for example) than any of a number of civil orstructural engineers with experience of bridgedesign and structural analysis, outside thepool frequently employed by Railtrack. Suchexperts could have been engaged on an‘arm’s-length’ basis, perhaps through HSE.

636 A key factor in this case was theconfidence of policy-makers in the ability ofHSE to engage in technical dialogue with theduty-holder. This was based on HSE’spossession of sufficient in-house expertise toensure that technical questions are properlyaddressed by duty-holders and any externalconsultants engaged by HSE itself. HSE hadaccess to sufficient expertise, in this case, toaudit the work carried out by Railtrack and PellFrischmann.


86

Phthalates in Soft Plastic Toys


637 The safety of phthalates used asplasticisers for PVC has been kept underreview by the DTI and the LGC for manyyears. Historically, two phthalates have beenused in children’s toys: di(2-ethylhexyl)phthalate (DEHP) and di-isononyl phthalate(DINP). DEHP is no longer used in toys orchildcare articles which are intended to bemouthed by children.

638 The present concerns arose in April1997 when the Danish EnvironmentalProtection Agency announced the results oftests that showed high levels of migration ofphthalates from children’s teethers. The Danesnotified the EC, and the machinery of ECregulation began to react. This case studylooks at how the UK government dealt with,and sought advice on the issue, in light of newinformation from the EC.

639 Responsibility for this issue rests with theConsumer Safety Unit of the DTI. Action onthis issue at an EC level is ongoing.


640 Two questions were posed by the DTI toestablish whether action was required.

• What level of exposure to phthalates issafe?

• What quantities of phthalates migratefrom children’s toys or other items thatare intended to be, or might reasonablybe expected to be, mouthed?


641 The selection of advisers was automatic.The Department of Health EnvironmentalChemicals Unit, in the Environmental andHealth Branch, advises the DTI’s ConsumerSafety Unit on the health effects of chemicalsin consumer products. The Unit has access toscientific advice on issues relating to toxicity ofchemicals from COT. In addition, expertadvice on the specialist areas ofcarcinogenicity and mutagenicity can beprovided by the sister committees coveringthese areas. These committees areestablished by the Chief Medical Officer(CMO) for England and provide advice to theDepartment of Health and to other government

departments, on request, relating to thetoxicity, carcinogenicity and mutagenicity ofchemicals. Formal accountability lies with theCMO and advice is passed through him to thedepartments. In practice this is done by thesecretariat, which is provided by theDepartment of Health.

642 COT was not asked to advise specificallyon the question of phthalates in consumerproducts. However, it had already (in the late1980s) advised on a range of phthalates in thecontext of their use as plasticisers in food-contact materials, and, more recently (in1996), in the context of the presence ofphthalates as contaminants in infant formulae.

643 The working of COT is examined in moredetail in case study eight [769–860].

644 The second question was addressed tothe LGC. Formerly part of the DTI, the LGCwas privatised in 1996 and provides research,analytical and advisory services to both thepublic and private sectors. It has a contractwith the DTI for services, including providing itwith advice on a wide range of issues, and hadbeen tracking the problem of phthalates inPVC for many years. It was contracted by theDTI to devise a method for the measurementof migration rates and to conduct migrationtests. In parallel, the LGC has, with DTIapproval and funding, participated indiscussions and trials with other laboratories inEurope, looking at various test methods. TheLGC has made presentations to the EC andothers on the test methods it has developed.

645 The LGC–DTI brief falls within the formalcontract for the LGC to support the DTIconsumer-research programme on chemicalmatters. The relationship between the two hasbeen in place for more than ten years, and thiscontinuity has allowed the LGC to maintainexpertise and awareness of the issue. TheLGC also sometimes attends EU meetings,usually as advisers to the UK delegation.

646 The LGC has also had contact with thephthalates industry. One of its key staffchaired a panel of experts (the WeinbergCommittee), convened on behalf of, andfunded by, US companies engaged in makingor using phthalates. Although this attractedsome hostile comment, the LGC and the DTIwere satisfied that it created no conflict ofinterest.


87


Toxicology

647 COT had previously advised onphthalates in the context of phthalates in foodcontact materials and as contaminants ininfant formulae. In this case, its approach hadbeen to support the use of tolerable dailyintake (TDI) values that had beenrecommended by the EC’s expert advisorycommittee, the Scientific Committee on Food(SCF)—although, in the case of two phthalates(butylbenzyl phthalate and dibutyl phthalate),further studies were recommended. The TDI isderived from doses that produce no effect inanimal studies, divided by a 100-folduncertainty (or safety) factor.

648 In the case of phthalates in soft toys, theDepartment of Health’s advice to the DTI’sConsumer Safety Unit was based on thepreviously expressed COT view that the TDIsrecommended by the SCF were acceptable.

Migration

649 While it is difficult to reproduce themouthing action on toys, the LGC hasdeveloped laboratory-based procedures thatproduce release rates that closely match thoseobserved in human studies by the Dutchcontract research organisation, TNO. Thesecorrespond to significantly lower levels ofmigration than those reported using a differenttechnique in Denmark, which initiated thecurrent concerns. The Danish results have notproved repeatable elsewhere.


650 The Department of Health’s advice to theDTI (based on the opinions of the COT andSCF relating to phthalates in food) was thatthere were no health concerns if exposureswere below the TDI. The LGC’s advice onmigration suggested that exposures wereindeed below the TDI. The DTI concluded thatphthalates as currently used in children’s toysin the UK pose no risk to human health.However, this conflicted with an ‘opinion’ of theEC expert advisory committee, the ScientificCommittee on Toxicity, Ecotoxicity and theEnvironment (SCTEE), which found that therewas concern with DINP because the margin ofsafety was only 75. A subsequent opinion onSeptember 28th 1999 found that the exposuremeasurements were insufficiently reproducibleto form the basis of regulation.

651 The DTI concluded that no ban wasnecessary, but that migration limits should beintroduced. Other European countries do notagree, and several have banned phthalates inchildren’s toys. The EC is divided on the issue.The Directorate General for Consumer Safetyand Public Health supports a ban onphthalates. The Directorate General forIndustry does not agree that a ban is justifiedby the evidence.

652 It appears that the DTI has followed UKscientific advice on migration and acted inaccordance with a reasonable view of theEuropean advice on toxicology. The DTI hasnot been subject to the same pressure on thistopic from activists, such as Greenpeace orFriends of the Earth, or from phthalatesmanufacturers, as its counterparts in othercountries.

Observations

653 Several issues emerge from this casestudy:

• the status of a private contractor asscientific adviser;

• situations in which a private contractorrepresents a government department ininternational discussions andnegotiations;

• the relationship between an adviser anda regulated industry;

• the composition and function of COT;• the relatively small opposing voice in the

UK.

The function of the LGC

654 The LGC continued to act as the DTI’sadviser of choice after it was privatised. It hasa constitution that prohibits capture bycommercial interests. It does enter intocontracts with industry, and, for example,provided the chairman of the WeinbergCommittee. The LGC appears to have beensuccessful in avoiding conflicts of interest.

655 The DTI chooses to continue to fundresearch and monitoring (referred to as‘intelligence-gathering’) by the LGC whichmaintains the Laboratory’s expertise andcapability. If this research were to bedispersed across many contractors, it might bethat no single laboratory would maintain the


88

critical mass needed to provide adequateadvisory services. The corollary is thatdirecting funding to only one contractor mayrestrict the choice of adviser in the future.

The function of COT

656 COT comprises a group of experts,mainly drawn from academia, which respondsto questions raised by the Department ofHealth or other government bodies. Thequestions are framed by the secretariat. Themembers receive only nominal payment. A listof candidate committee members is drawn upby the secretariat after open advertisementand selection by an independent sift panel thatmakes recommendations to the CMO.Constitutionally, the Committee serves theCMO, who appoints its members from thecandidate list and receives (and is responsiblefor) the advice. The working of COT isconsidered in more detail in case study eight[769–860].


657 A striking feature of the phthalates casein the UK is that the government has not comeunder public pressure on the issue, in contrastto the situation elsewhere in Europe. Only veryrecently has the issue come to the attention ofpressure groups, and the scientific advisorysystem in the UK has operated largelyunchallenged either by pressure groups or bythe phthalates industry.

658 The UK policy remains that phthalates(other than DEHP) in toys are safe, providedthat TDIs are not exceeded. It has beendifficult to obtain reproducible results fromexperiments designed to produce an estimateof exposure through mouthing toys.

659 It might be expected that any processleading to a ban would also have consideredthe alternatives to phthalates, and, hence, thewider consequences of a ban. The plasticsindustry has pointed out that some alternativeplasticisers can make plastic toys easier totear or bite through, with a consequentlygreater risk of small pieces breaking off andcausing choking.

660 There could have been criticism that theadvisory system of the DTI and theDepartment of Health had not compared therisks from phthalates with risks fromalternatives. This would require the synthesisof a wider range of expertise than has beenused to date. The question to put to expertscientific advisers would have to change from‘are phthalates safe?’ to ‘what are the possibleways of making chewable toys that areacceptably safe?’. Although the UK hasconsistently called for any proposal for a banto be accompanied by a consideration of theconsequences of alternatives, this questionhas not been addressed by the advisorysystem.


89

Total Allowable Catches of Sea Fish


661 For at least 100 years it has beenrecognised that it is possible to destroy fishstocks by over-fishing, and that it is thereforeimportant to collect scientific data on fishstocks and catch sizes. Most parties involvedin the fishing industry now accept thatregulation to prevent over-fishing is essential.The Common Fisheries Policy is one of thecore policies of the EU.

662 MAFF has lead responsibility forrepresenting UK fishing interests in the EU.The Scottish Executive Rural AffairsDepartment also plays an important role.

663 There have been attempts to control thesize of the catches using many differentrestrictions, including reducing net sizes,increasing mesh sizes, decommissioningboats, limiting engine size, limiting the dayswhen boats may be at sea, restricting areasthat are fished, and limiting annual landings ofeach species. The aim of regulation has beento maximise short-term production, as well asensuring the long-term future of the fishery.More recently, advice has been applied interms of a more precautionary approach.


664 The relationship between fisheriespolicy-makers and their scientific advisers hasevolved over time. Historically, politicianssimply sought the advice of eminent scientiststo tell them what the policy should be. Theynow seek advice on the amount of fish thatcan safely be killed each year. This leads tothe question: what is the safe total allowablecatch (TAC) from each fishery for eachcommercial species?

665 ICES is asked to define reference orthreshold levels for the number of fish that canbe killed, and the minimum size of the stock,such that there is a high probability that thepopulation will not crash.

666 Within Europe, this question is put to theInternational Council for the Exploration of theSeas (ICES). Fisheries scientists fromindividual countries meet under the ICESumbrella to consider the data each hasavailable from national monitoringprogrammes. ICES reports form the basis fornegotiation of TACs by EC Ministers.

667 A recent report of the House ofCommons Select Committee on Agriculturestated:

We believe that it is important that theadvice from ICES is demonstrablyscientific and that it is not the body bestplaced to offer objective economicanalysis (eighth report, paragraph 20).

The paragraph continues:

However, since it is accepted thateconomic factors should be taken intoaccount in making decisions onallocations, it is clearly preferable thatadvice on this aspect is publishedbefore the recommendations are put tothe Council of Ministers. The EC, incollaboration with member states,should continue to developmethodologies for the inclusion ofeconomic criteria for evaluatingbiological recommendations for fishstocks.

ICES does not (currently) considereconomic criteria in recommending aTAC.

668 Although in the past, ICES hasconsidered whether socio-economic issuescould be incorporated into its advice on TACs,at present there is no formal mechanism forthis to happen within ICES.


669 The scientific adviser is ICES, which ismade up of scientists from each of its 19member countries. The scientists work atICES through a structure of committees togenerate a single consensus view. Thegovernments of each of the member countriestherefore have the same advice on which tobase their negotiations.

670 The UK contributes to ICES via theFisheries Research Service in Scotland, theCentre for Environment, Fisheries andAquaculture Science (CEFAS)—a ‘next steps’agency of MAFF—and the Fisheries ResearchLaboratory in Northern Ireland. Scientists fromthese agencies conduct research for theirgovernment departments, and othercustomers, including private industry. MAFFfunds are administered by its Chief Scientist;


90

there is separate funding for research and forthe monitoring work and preparation of advice,although, in practice, the two overlap. There isa strong customer–contractor relationship inforce, and it is recognised that research isessential, both to improve the science(especially modelling), and because it isbelieved that it is not possible to retain goodscientists without providing an opportunity toundertake basic scientific research.


671 The UK government scientists are activemembers of ICES, which generates an annualreport on the state of each commercial speciesin each fishing area. The input data is sparse:there are a small number of research ships,but most data comes from declared landings,which can be inaccurate and incomplete, andmay be biased.

672 The key concept is the safe minimumbiological limit (MBAL). A simple interpretationis that a stock is secure if it is above theMBAL. ICES, in its latest annual report,adopted the precautionary principle for the firsttime.

673 In order to retain harvesting within safebiological limits, there must be a highprobability that:

• the spawning stock biomass (SSB) isabove the threshold where recruitment isimpaired (recruitment takes into accountall the factors that affect the rate ofresupply of the stock, such as breedingrates and age distribution, and conditionof the stock);

• the fishing mortality (which includes allfish deaths caused by fishing, not justlanded catches) is below that which willdrive the spawning stock under the safebiological limit.

674 Two thresholds are defined to representthese two constraints: Blim (the SSB belowwhich recruitment of new fish to the populationis insufficient to maintain the population), andFlim (the mortality above which the SSB will fallbelow Blim). Two further thresholds aredefined, Bpa and Fpa, which represent the safethresholds after applying the precautionaryapproach. The difference between the limitingand precautionary thresholds reflects theuncertainty in the data and population models,and varies between species and fisheries.

675 ICES does not set TACs. It estimates theprobability (which is low at Fpa and high at Flim)

that the stock will collapse at various mortalityrates. It also does not attempt to estimate thevalue of F that would give the maximumsustainable yield (Fmsy, which is adopted bythe United Nations Food and AgricultureOrganisation). It makes no commercial,political or economic statements in its advice.

676 The precautionary estimates, Fpa andBpa, were first introduced in 1998.

677 An example is the advice given for cod inthree sub-areas (North Sea, Eastern EnglishChannel, and Skagerrak) for 1999 (ICES,1999). The overall advice on management isas follows:

ICES recommends that fishingmortality in 1999 should be reduced toF=0.60 (below the proposed Fpa),corresponding to expected landings of147,000 t in 1999 in order to bringSpawning Stock Biomass above theproposed Bpa in the short term.

678 The table then presents a range ofvalues of F from 0.4 to 0.8, with theirimplications for landings and the likely SSBthat will result in the year 2000. The medium-term effects are expressed as a probabilitythat SSB will be greater or less than Bpa. Flim is0.86.

679 This analysis includes a set ofmanagement options and estimates of theimplications of selecting different levels ofTAC.

680 This presentation has been criticised bythe House of Commons Select Committee,which recommended that ICES be encouragedto ‘present its advice on TACs in a moregenerally comprehensible form’ (House ofCommons, 1999b). It is hard to see how itcould do so and still quantify the extent ofuncertainty in the underlying science. Byexplicitly quoting the uncertainties and thedifference between scientific best opinion anda precautionary position, it gives the policy-maker information on which to make a rationaltrade-off between commercial interests,conservation interests, and the future of thefishery.

681 Scientific advice is also taken whenconsidering technical measures to be adoptedin order to put TACs into effect. These include,for example, restricting access to spawninggrounds or setting minimum mesh sizes fornets. Scientific advice is sought on how theserestrictions might affect recruitment andmortality of the stock.


91


682 ICES estimates the likely consequencesfor the fish stock of different levels of mortality.The members of its committees, as scientistsemployed in member states, may also adviseon the likely effect of proposed deviations fromthe TACs recommended by ICES. The policy-making process (at an EU level) translates thisadvice into controls on fishing.

683 Concern about this process has beenexpressed by the scientists involved and bythe House of Commons Select Committee onAgriculture. Both have called for a more openand direct process. There is concern that thepolicy-making process in total does not makegood use of data that include explicit estimatesof uncertainty, and that it distances thescience from the decision-making process.Some of the political compromises that havebeen adopted may conflict with the logicunderpinning the scientific advice. Forexample, the eventual decision on TACs hassometimes reflected a political desire tosmooth out large changes in TACs from yearto year. However, given a highly non-linearsystem, such as a fish population, suchsmoothing may have profound consequences.

684 Fishermen, too, are concerned that therehave been sudden and unexpected changes inpolicy, and would like to see scientificevidence at an (even) earlier stage of theannual policy cycle. There is also concern inthe fishing industry that, because the policy isset annually, it lacks continuity.

Observations

685 At one level the setting of TACs could bea model of how to secure and make use ofscientific advice. The government employsprofessional advisers, which it keeps informedby funding research. The advisers work incollaboration with many other scientists togenerate a consensus scientific view. Theinternational negotiations can then proceed ona common, agreed scientific basis. Thescientific advisers expressly state theuncertainty, leaving to the politician thequestion as to how much risk should beaccepted. The scientific advisers studiouslyavoid recommending a policy or taking intoaccount any non-scientific issues (such as theeconomic consequences of the advice).

686 The level of detail in the current scientificadvice is not sufficient to indicate theconsequences of a wide range of policy

options, yet even the level of detail that isprovided is thought by some politicians to behard to comprehend.

687 The introduction of express levels ofuncertainty, and of the impact of applying theprecautionary approach, provides new andmore powerful data on which to base policyoptions. However, scientists believe that theconsideration of other factors (economic,social and political) at a later stage in the (non-scientific) advisory process changes thescientific question that would have beenasked.

688 Alternative models are used in otherfisheries (for example, in some single-speciesfisheries in South Africa). Three policy aimscan be defined: to conserve stocks, tomaximise catches, and to stabilise TACs. Therelative weight assigned to each objective is apolitical decision. Fisheries scientists are thenable to model the effect of different levels ofTAC, to score the outcomes against the threeobjectives, and, hence, to optimise the TACagainst the stated political goals. On the basisof the outcome of this modelling process,politicians may reconsider the relative weightsattached to each objective, and requirescientists to recalculate options on the newbasis. This model has not yet been used in themajor European fisheries. It has the potentialadvantage that it allows the best use of all ofthe scientific resources. All interested partiescan contribute to an explicit debate to draw upthe weights assigned to the three policyobjectives.

689 A number of stocks in the North Sea areat extremely low levels.

Robustness

690 The advisory process is robust, in thesense that it achieves scientific consensus onthe probability distribution of outcomes, but itis vulnerable to challenge on other grounds.The science delivers a probability distributionof outcomes, but the policy-makers complainthat they find these difficult to understand, andthe policy-making process cannot currentlymake good use of them. There is therefore aneed to develop ways of expressinguncertainty that decision-takers can acceptand use with confidence.

691 The advice provided is for one year only,even though the underlying scientific modelsrun over several years into the future; and thepolicy might be more effective if it were set forseveral years at a time.


92

692 There is a danger that an eventualdecision might not reflect the scientific advicebecause of a false interpretation of theinformation about uncertainty. If theuncertainty is misunderstood as a ‘safetymargin’ then politicians may wish to use part ofthat margin in pursuit of another policyobjective.

693 A more complex process could beimagined, which would be more robust. Itwould require the policy-maker to agree withthe decision-taker in advance a measure ofthe effectiveness of a policy, which would takeinto account not only the yield of the currentyear, but also future yields, the risk ofpopulation collapse and the commercial needfor stability in TAC from year to year. Once thismeasure had been agreed, the scientistswould have a more detailed brief againstwhich to prepare their advice, and the advicewould be more useful to policy-makers.


93

The Decision to Construct the Thames Barrier


694 Control of the flooding of the RiverThames began in Roman times, but the eventthat triggered the decision to construct theThames Barrier was the 1953 disaster, inwhich 300 people were drowned in easternEngland, and about 3,000 in the Netherlands.

695 Central London was not affected by the1953 flood, but sea defences were overtoppedalong much of the eastern coast. In response,these defences were raised, and when, in1965, a tide of similar height recurred, thewaves lapped the tops of the parapets incentral London, but there was no majoroverflow.

696 The historical record of extreme tidesreveals a general upward trend in record tideheights. The underlying cause of this is nowknown to be the progressive lowering of theland in south-east England relative to sealevel, and the raising of sea level caused bythe melting of polar ice.

697 The policy set in the late nineteenthcentury had been to make the flood-defencelevel one foot above the highest recorded tide.

698 The capital expenditures for a barrierand associated bank raising were estimated in1965 to be of the order of hundreds of millionsof pounds at today’s prices, and the costs of aserious flood, such as had occurred inHamburg in 1962, were in the order of tens ofbillions.


699 There were two distinct questions.

• What is the appropriate defence level touse as the design basis for the seadefences of the Thames Estuary and theeastern coast?

• Is it right to construct a barrier across theRiver Thames, rather than the alternativeof raising its banks?

700 There was a trade-off to be madebetween the incremental cost of raisingexisting defences and the one-off cost ofconstructing a barrier. If the barrier were to bebuilt, the sea defences downstream of thebarrier would have to be raised not only tomatch it, but also to deal with the additionalsurge levels that would be caused by the

artificial damming of the river against the risingtide.

701 The design height of the defences wascrucial because, if the required increase wassmall, bank raising would be preferred; alarger increase would favour the barrier (byobviating the need for unsightly andimpractical bank raising in central London).


702 Within three months of the 1953 flooddisaster, the government set up aDepartmental Committee of the Ministry ofHousing and Local Government, chaired byLord Waverley, who had been HomeSecretary during the Second World War, andsubsequently Chairman of the Port of LondonAuthority (PLA). The Committee was given afree hand by the government to develop apolicy that would solve the problem, and itenjoyed wide discretion to advise as it thoughtfit.

703 It reported after a year that furtherresearch was needed. In 1954, the ThamesTechnical Panel was established, comprisingthe following:

• the Chief Engineers of:

− the Ministry of Housing and localgovernment

− the Ministry of Works− MAFF− London City Council− the City of London− three London Boroughs (West

Ham, Woolwich and Barnes)− four river authorities (PLA, Lee,

Essex and Kent);

• an Admiralty representative;• a Trinity House representative;• a Ministry of Transport representative;• the Director of the Hydraulics Research

Station.

704 This Panel was both a stakeholder bodyand an expert body. Unlike the WaverleyCommittee, its terms of reference were clearlyspecified:


94

• to make a survey of the existing flooddefences;

• to estimate the cost of raising them;• to advise whether, having regard to

navigational and other requirements, thequestion of a structure across the rivermerited further consideration;

• to consider the appointment of aconsulting engineer and the terms ofreference.

705 The Thames Technical Panel and thetwo firms of consulting engineers appointed onits recommendation3 (Rendel Palmer & Tritton,and Sir Bruce White, Wolfe Barry & Partners)spent the next seven years developingfeasibility studies of the various options. By1965, several schemes had been worked outin detail, and it was estimated that the costswould be between about £250m and £420m intoday’s prices. However, 12 years after thelast flood, the project had lost momentum andnew impetus was needed.

706 The government sought advice from thethen CSA, Sir Solly Zuckerman. Herecommended that a report be commissionedfrom a senior independent scientist, and, indue course, Professor Hermann Bondi (laterSir Hermann Bondi) was appointed to thistask. Professor Bondi is an astronomer who atthe time held the Chair of Mathematics atKing’s College, London. He is not an expert onflood defences, but he was considered expertin the comprehension of complex technicalissues and in the application of scientificmethods to such problems.

707 He carried out his investigation alone. Itappears likely that he was appointed on thebasis of the personal recommendation of SirSolly Zuckerman. Professor Bondi has beeninfluential in providing scientific advice to theUK government, and has written widely aboutthe scientific advisory process [360].

708 By the time of Professor Bondi’sappointment, the Panel and the consultingengineers had carried out a great deal of work,which had generated information about theflooding risk and the costs of flood-defenceschemes. Professor Bondi’s task was todetermine whether a clear recommendation foraction could be made on the basis of thisinformation.

3 The Thames Technical Panel recommended thatthree firms be appointed, but the governmentreduced this to two.


The Waverley Committee

709 Although the Waverley Committee wasconcerned mainly with policy and finance, itmade two major decisions that defined thefunctional requirements of the flood-defenceproject.

• Any barrier across the river would haveto be movable so that the interests of thePLA and ship operators would bepreserved, and to avoid the risk ofdisturbing the natural flushing action ofthe tides and the stability of bottom silts.

• The Committee pronounced on thestandard of protection that was required.The standard was to withstand a flood ofthe level of that in 1953 where areas orproperty of high value would be at risk,and, in lower-value areas, a standardthat would reasonably have beenconsidered adequate before 1953.

710 Both of these recommendations werenon-scientific; they set out the aims but not themeans, and they struck a balance between theconflicting interests of river users and coastal-land occupiers, but without carrying out aformal cost–benefit analysis.

The Thames Technical Panel

711 The 1954 Thames Technical Panel hada purely scientific and engineering remit.Acting on behalf of central and localgovernment, the Panel, and its successorsteering committees, sponsored studies of:

• underlying meteorological and oceano-graphic causes of surge tides;

• surge tide statistics and trends;• siltation and estuarine hydraulics;• existing defences;• cost estimates for various elevations of

defences;• negative effects of raising river banks in

central London (amenity and cost);• engineering feasibility and costs of

alternative barrier designs.

712 No single study made a holisticassessment; as gaps in knowledge becameapparent, studies were undertaken to fill them.

713 The Panel favoured a barrier located inLong Reach (ten miles downstream of theeventual site in Woolwich Reach) until 1961,when the PLA declared that, consequent upon


95

its own decision to authorise new jetties to bebuilt near to the site, Long Reach would haveto remain unobstructed in order to allowvessels using the jetties to turn around. ThePLA proposed an alternative location 2 milesupstream at Crayfordness. This radical andunexpected change in the PLA’s demands notonly rendered several years of workredundant, but also presented the engineerswith unprecedented technical challenges,owing to the size of the movable barrier thatwould be required at this location.

The Bondi Report

714 Professor Bondi had a large quantity ofinformation available from the precedingscientific and engineering studies, and advicefrom the people directly involved. Heconsidered the calculations that had beenmade of the risk of flooding.

715 The actuarial calculations of the timeindicated a low probability of flooding.However, because the cost of a major flood incentral London had been estimated at £20billion, the annual probability of flooding wouldneed to be extremely low to make theinvestment of between £250m and £420muneconomic (figures in 1999 prices).

716 There was also a commonly held viewthat the consequence of flooding was so largethat the risk was unacceptable.

717 Professor Bondi, recognising themagnitude of the consequences of a flood,and the uncertainties extant in the statistics,decided upon a precautionary approach. Hisreport concluded:

The extremely severe effects on the lifeof the country of a tidal surgeappreciably higher than that of 1953 (or1965) makes it appropriate to takepreventative measures, even thoughthe probability of a high surge is notgreat.

718 The Bondi report went on to say that thisprecautionary standpoint might not apply inother cases where the available data weremore robust and there was less uncertainty.

719 The report also observed that cost–benefit considerations had not been properlyapplied earlier in the project, especially inrelation to the demands of the PLA regardingnavigational requirements, which had beenpresented, and received, as non-negotiable.Ideally, these demands should have beensubjected to a test of cost-effectiveness; abetter overall decision might have been

reached, in which the combined interests ofthe various stakeholders would have beenmaximised.


720 The Bondi report was accepted in full bythe government and the necessaryexpenditure was authorised.

721 The newly formed Greater LondonCouncil (GLC) took over responsibility for theproject and the Thames Barrier came intoservice in 1982. Its final cost was £530m in1982 prices (£1.2 billion today). In the periodup to 1998, it had been closed against surgetides on 33 occasions, which is in line withexpectations. As envisaged in the earliestdays of the project, further provision will haveto be made to ensure adequate protection tocentral London in the year 2100 and beyond,and a working group commenced initialdeliberations in 1999.

Observations

722 In the discussion of the first three papersduring the 1977 Institution of Civil Engineersconference on Thames Barrier Design (variousauthors, 1978), Dr D. E. Wright of Sir WilliamHalcrow & Partners questioned whetherProfessor Bondi’s view that the consequencesof a flood dictated the decision was a sufficientanswer because it begged the question as towhat standard of protection should beprovided. Dr Wright said:

Despite Sir Hermann’s remark, thestandard of protection that is to beprovided by the barrier is still finite.

He posed the issue of the degree ofprotection as a problem of costs versusbenefits.

723 In their reply, the authors (Horner,Beckett and Draper, 1978) provided details ofthe cost–benefit calculations of 1970 (threeyears after the Bondi report). They stated thatthe annualised cost of flood damage wasestimated at about £5m, which correspondedto a return period of about 400 years on a £2billion loss. £5m per year discounted at 10%yields a net present value of about £55m,which was about the same as thecontemporary estimate of the cost of theBarrier and associated bank raising (£61m).This is a crude method of discounting.


96

724 They went on to say:

[discounting] tends to favour low costprotection against short return-periodevents. A high benefit–cost ratiomeans more frequent flooding. This ofcourse is not what the public want. Asa result of the flooding of the tributariesof the Thames in the summer of 1977,the cry has been heard again andagain that action must be taken toprevent a recurrence of this kind ofevent. The additional cost on the wholeproject in the GLC area of less than10% to raise the standard of protectionfrom the 100 year return period to the1000 year return period is thereforereadily accepted.

725 Professor Bondi was not required toundertake any original scientific work, butmade a decision based on the evidencepresented to him.

726 The reliance on one individual to createan overarching assessment was evidence of alack of direction among the rest of theparticipants, and a lack of a clear vision of thedecision-making principles that the advisoryprocess was supporting.

727 There was no ambiguity about functionsof the Waverley Committee (policy) and thetechnical panel (science), but the caseillustrates the need for a process of synthesisas an interface between the scientific andengineering investigation and the policy-making functions.

728 The case also illustrates the need tomanage stakeholder interests openly and inbalance. Although the PLA defended itsstance vigorously at the 1977 Institution ofCivil Engineers conference, other partiesdescribe the PLA as unyielding and dogmatic.There was no direct representation of thepublic interest.

729 Uncertainty played a considerable role inthis case—especially uncertainty in thefrequency of extreme tides—and encouraged

the precautionary approach of ProfessorBondi. This sufficed for the main decision onwhether or not to build the barrier, but it wouldnot have sufficed for the other key decisionthat was needed—the setting of the new flood-defence level. The latter could only have beenresolved by balancing cost and risk reduction.


730 The trust that was accorded to ProfessorBondi in 1965 might not be so readily given toa single scientist today, no matter howeminent or competent.

731 The advantage of using a single expertwas that the advice could be very clear andpersuasive. However, reliance on a singleexpert to recommend a solution ran the risk ofreaching an idiosyncratic or impracticalconclusion.

732 A modern cost–benefit analysis wouldhave justified the Barrier, on the basis of thefigures available at the time. However, if thebalance between costs and benefits hadproved marginal, the consequence-basedargument that was used in 1965 might be anacceptable deciding factor in a modernassessment.

733 The advisory process made very slowprogress, during which time flooding couldhave occurred. A simple calculation of extremetide probability could have been used to set anoptimum time horizon for decision-making andconstruction, but this was not done. It ispossible that the slowness, and the recourseto a single expert for advice, may have beendue to institutional factors: although manyinterests were involved in the decision, nosingle organisation had clear overallresponsibility for defining and solving theproblem.

734 The Waverley Committee set thestandard of protection before full costings hadbeen made, and pre-empted full cost–benefitanalyses of all available solutions.


97

Assessment of Safety Cases for Radioactive Waste Disposal


735 When nuclear power was firstintroduced, it was recognised thatintermediate-level waste would requiredisposal. Two options for disposal existed atthat time: disposal at sea (ie, ‘dilute anddisperse’), and deep geological disposal (ie,‘concentrate and contain’—which would evolveover geological time to ‘dilute and disperse’,but with much of the radioactivity havingdecayed). Although the risks associated withsea dispersal had been assessed assatisfactory on environmental and publicsafety grounds, this disposal option was soonfound to be politically unacceptable, along withall forms of waste disposal at sea. Geologicaldisposal became the preferred option, with theassumption that it posed no serious technicaldifficulties. By the late 1970s, however, it wasapparent that deep disposal did presenttechnical challenges. Two broad challengesarose.

736 First, over time, chemical action bygroundwater leads to the deterioration of thecontainment system, ultimately allowingleaching of radioactive material into thesurrounding rock. The key requirement wastherefore to design a containment systemsufficiently robust to ensure that the wastewould be contained for longer than the periodrequired for most of the radioactivity to decaynaturally.

737 The second challenge was to preventaccidental or deliberate human access to thewaste.

738 The containment period required forsubstantial decay of the radioactivity is greaterthan 100,000 years.

739 UK Nirex Limited was charged withdevising a disposal solution and was requiredto prepare and submit a disposal safety casefor approval by regulators. The regulatorswere:

• the Field Operations Directorate of HSE,responsible for the safety of the workersand others affected by the works duringconstruction and pre-closure;

• the Nuclear Installations Inspectorate,part of HSE, responsible for theoperation of the site after constructionand pre-closure;

• Her Majesty’s Inspectorate of Pollution(HMIP), from April 1996 part of the EA,responsible for authorisation of disposalof radioactive waste, includingassessment of the post-closure safetycase for the disposal facility.

740 This case study concerns specifically theactions of HMIP in preparing itself to review apost-closure safety case, up until the formationof the EA in April 1996. Subsequently, in 1997,the EA published guidance on theauthorisation of radioactive waste disposal(Radioactive Substances Act 1993, Guidanceon Requirements for Authorisation.)


741 HMIP intended to assess any safetycase by means of a detailed scientific andtechnical examination of the case submitted byNirex. Nirex was responsible for preparing thecase and for devising the methodology bywhich it would be prepared, since no completemethodology then existed in the UK. HMIPhad undertaken and published a number ofpartial hypothetical assessments, and therehad been significant research in Canada andSweden. However, there was no establishedapproach to assessing the safety of a nuclearwaste disposal facility, or producing a safetycase. Furthermore, there was no experience ofconstructing such a facility in the UK and littleexperience in other countries. The regulatorsfor underground disposal of nuclear waste hadno experience of reviewing safety cases.

742 Thus, HMIP had to preparemethodologies for assessing the safety case.The question put to expert advisers was,therefore, framed in a general way—toconduct research into waste disposal safety, inorder to provide the tools for assessing a casewhen it was eventually submitted.


743 HMIP sought to develop a team ofscientific experts who were independent of thenuclear industry. HMIP had no laboratory orsource of expertise of its own, and thereforerelied on contractors (consulting companies,academic departments and laboratories, whichwere either publicly owned or had been


98

privatised). HMIP recognised that it would notbe possible to prevent individual staff movingbetween employers, but it sought to usecontractors that had worked exclusively for theUK regulators or had experience fromoverseas disposal programmes.

744 HMIP developed its own expertise, andthat of its contractors, largely by setting inplace research programmes that addressedthe uncertainties in the science andtechnology underpinning safe disposal. Theresearch programmes were funded by HMIPand managed by its staff. The work inpreparation for the assessment of Nirex’ssafety case was managed by HMIP staff, butwas funded by Nirex under a voluntary, cost-recovery agreement with HMIP. Part of thefunding provided by Nirex was used by HMIPto develop site-specific models forindependent quantitative assessment ofNirex’s safety case.

745 In practice, it was difficult to maintain thestrict separation between contractors workingfor the industry and those that worked for theregulator within a finite pool of experts. Most ofthe available experts had worked for both theregulator and the industry. Rejection of theplanning application for the rockcharacterisation facility in March 1997 resultedin reduced funding and shrinking capabilities inthe UK in the field of deep geological disposal.


746 The advisers undertook research todefine the parameters that would have to betaken into account in a safety case, thescenarios under which the parameters mightvary, and hence to assess the consequencesthat might result. Once models had been builtincorporating these features, it would then bepossible to apply those models to proposedfacilities.

747 Although the primary responsibility forbuilding these models lay with Nirex, HMIPcommissioned its researchers to carry out aparallel exercise in preparation for evaluatingNirex’s eventual submission.

748 Much of the model building involvedestablishing the likely values (and associateduncertainties) of several key parameters (forexample, the likely flow rate and chemicalcomposition of groundwater). The researchersengaged in a systematic knowledge-elicitationexercise to determine likely values for theseparameters.

749 This was designed to make the best useof the available expertise to estimate aprobability density function (PDF) for theparameter. Uncertainty is made explicit by thisprocess since the PDF reflects both theintrinsic variability between, for example,different rock types and formations, and theuncertainty in the state of scientificunderstanding (or disagreements betweenexperts).

750 The process has been extensivelydocumented, and derives from work originallyconducted at the Stanford Research Institutein the 1950s and taken up in the UK, inparticular by the London School of Economics.Key elements of the process are:

• a group of experts is convened, typicallyfor one day, to address a specificproblem;

• their aim is to produce a consensusestimate of the PDF of each variableunder consideration (this means thatuncertainty is explicit and expressedquantitatively);

• each expert makes an independentestimate without consulting the others;

• these estimates are then pooled anddebated—outliers are not smoothed outby averaging, but retained and debated.

751 A trained facilitator manages theprocess, and some of the experts taking partmay have expertise that is only peripheral tothe problem. The overall aim of this process isto avoid capture by one powerful personality orby force of argument. By making each expertcommit to a view, it requires a consciousdecision by that expert to change the view inthe light of pressure from the others. Thisensures that ‘heretical’ views are not sweptaway and only disappear as a result of areasoned argument. If no reasoned argumentcan be found, the non-conforming viewbroadens the PDF that results.

752 This process shares many propertieswith the Delphi approach to sampling expertopinion, but differs in that there is only oneformal round and the process is much moreintensive. It also draws on the approach ofsystematic innovation, in that it attempts tocapture unconventional ideas and debatethem, not reject them out of hand.

753 The scientific groups engaged in thisprocess tested their draft methodology in anassessment of a hypothetical facility and site(Harwell, in an exercise called Dry Run 3).


99

754 While HMIP’s research programme wasunderway, Nirex decided to postpone forseveral years any application to develop awaste disposal facility, until results from theplanned Rock Characterisation Facility locatedat the preferred site could be evaluated.Funding for HMIP’s research programme waswithdrawn, and funding resources were thenconcentrated on data collection and on thedevelopment of Nirex’s own assessmentmethodology, rather than on the developmentof parallel methodologies. The EA argues thatit is the function of the regulator to beindependent, and this may include repeatingpart of the work supporting a safety case.HMIP did not take some of Nirex’s work intoaccount because it was in draft form, andHMIP believed that its function was to respondto the Nirex safety case when finallysubmitted, not to become involved in itspreparation. HMIP wished to work in parallelwith Nirex in order to arrive at a position whereit could assess a safety case independently ofthe process by which the case had beengenerated.


755 No policy decision on construction of adeep geological repository has yet beenrequired because the planning application forthe rock characterisation facility submitted byNirex to the planning inquiry was rejected. Thisbrought Nirex’s preparations for a deep-disposal facility at Sellafield to an end. Nirex iscurrently engaged in researching those issuesof concern in a generic safety assessment (ie,non-site-specific), and in the provision ofadvice to the nuclear industry on wastepackaging and transport.

756 The problem of the management ofintermediate-level waste was debated recentlyby a House of Lords Committee. TheCommittee recommended the setting up of a‘nuclear waste management commission’ anda ‘radioactive waste disposal company’. It isunlikely that any decision on setting up suchorganisations will be made before a new policyon radioactive waste management has beenannounced. The process for defining the policywill be initiated in late 2000 when thegovernment plans to issue a Green Paper forwide consultation with stakeholders. In theshorter term, however, BNFL, operator of thelow-level waste facility at Drigg, is currentlypreparing a post-closure safety case for thesite for submission to the EA in late 2002. TheEA plans to review the authorisation for Driggand, as part of this process, will evaluateBNFL’s post-closure safety case and decide

whether the case meets the regulatoryrequirements, as set out in the publishedguidance.

Observations

757 This case raises several importantissues:

• the concept of parallel and independentteams;

• methods for finding consensus amongexperts, including the use of PDF as anexpert output;

• the need for a scientific capability withinthe regulatory authority;

• the funding of regulatory research by theregulated industry;

• the fundamental principle of evaluatingthe safety case for each site in isolation,without considering the widerimplications of not approving anydisposal route.

Parallel independent teams

758 The selection and maintenance of anexpert team to evaluate the safety case that isindependent of the team preparing that safetycase is attractive, but proved difficult inpractice because there was only a small poolof experts from which both teams had to berecruited. Further, there appears to have beeninsufficient funding to support a second expertteam. Several individuals moved betweenorganisations working on opposite sides of theregulatory divide.

759 Nirex felt that the regulators providedlittle feedback on the developing safety case.The process was informal, although well-documented, and there was no fixed reviewperiod. The approach taken contrasts stronglywith the approach taken by the regulators inSweden, who worked with the industry duringthe preparation of a disposal case. The USAadopts a similar (if somewhat more strict)confrontational approach to the UK, but it canbe argued that this is more appropriatebecause there is sufficient funding availablefrom industry in the USA to support severalexpert teams.

Expert consensus

760 The consensus-building process wasformal and appears to have been effective.One test was conducted where two separategroups of experts addressed the same issueand reached similar conclusions. The processappears to have much to commend it when


100

compared to the working of a conventionaladvisory committee, where much of the controllies with the chairperson and secretariat, in theabsence of guidance as to committeeprocedure in the committee’s terms ofreference.

Scientific capability of the regulator

761 The regulator relied upon contractors toprovide the broad range of scientific expertiserequired to assess a safety case for disposal.To reduce the project-management load on itsstaff, HMIP selected two lead contractors tomanage the review work and the independentquantitative assessment studies. HMIP staffhad developed the required experience andexpertise in directing assessment studiesthrough the Dry Run exercises over a period ofabout 12 years. Some argue that the regulatorrelied upon contractors to provide too much ofthe scientific expertise, to the extent that it didnot have sufficient expertise of its own tomanage or direct its contractors effectively.

Funding of regulatory research

762 Part of the research conducted toprepare the evaluation team was funded byNirex, under a voluntary levy. Nirex had nocontrol over the conduct of the research, but itcould determine the level of funding. Supportwas withdrawn before HMIP had been ablefully to assess and endorse its consultants’reports, so these reports have never receivedofficial regulatory status or endorsement.

763 This funding arrangement contrasts withother regulated industries, where the applicantpays the cost of assessing the safety case, butthe assessment procedure is already clearlydefined (eg, the safety of medicines under theMedicines Control Agency). Each applicantpays the marginal cost of an assessment, notthe cost of developing the assessmentmethodology in the first place. It has beenargued that site-specific methodologies arerequired for the assessment of nuclear wastedisposal.

Case-by-case evaluation

764 The scientific advisers were asked togenerate a methodology to assess a safetycase for disposal at a specific site. They werenot asked to assess the best overall solutionfor waste disposal and were not asked tocompare the safety of alternative sites.

765 It is necessary to break down a complexissue into separate manageable parts.However, in this case, it can be argued thatthe main choices were closed down too earlyin the process.


766 The consensus-building techniquesstand out as exceptionally well considered, butthere was a shortage of expertise within HMIP.The process proved unsustainable in practicebecause it was too long drawn-out and therewas not enough continuity of work to sustaincritical mass in the team.

767 The process illustrates the difficulties ofadopting a competitive approach as against acooperative approach. Each method hasadvantages and disadvantages. Thecompetitive approach used here avoids therisk of regulatory capture, but it needs financialcommitment and, in this case, was alwaysdependent on the goodwill of the industry. Thedisadvantage of such a lack of true financialindependence is not so much that the financemay be withdrawn—that at least would be anopen fact—but that the threat of withdrawalmay subtly influence the research findings.

768 The main weakness in the HMIP positionwas that it had too few expert staff in-house toenable it to carry out the overall synthesis ofthe research programme that was a centralrequirement. This left HMIP dependent uponexternal contractors, with the inevitableproblems of loss of continuity and knowledge-retention.


101

Advisory Committees on Toxic Chemicals769 This case study examines onemechanism by which scientific advice isprovided to government, rather than a specificissue. The advisory committee system is suchan important mechanism for obtaining expertscientific advice from outside government thatit merits individual treatment.

770 The scope of this case study is restrictedto a particular group of committees that adviseon matters relating to toxic chemicals. Thecommittees provide advice on similar subjects,but have different advisory functions andreport to different government departments.The committees studied are:

• COT, which reports to the Department ofHealth and the FSA;

• ACTS, which reports to the HSC; and• the Advisory Committee on Pesticides

(ACP), which reports to the PesticidesSafety Directorate (PSD), an agency ofMAFF.

771 While dealing with related scientificissues, these committees have distinctfunctions and, therefore, differentcompositions and methods of working. Thiscase study explores these differences andhow they match the requirement for advice,and identifies common factors among all thecommittees.

Committee on Toxicity of Chemicalsin Food, Consumer Products andthe Environment

Terms of reference and referral of issues tothe Committee

772 Prior to April 2000, COT was aDepartment of Health committee, reporting tothe CMO. Following the establishment of theFSA, the COT secretariat became a jointactivity of the Department of Health and theFSA. Other departments (MAFF, HSE and theDETR) continue to attend as ‘assessors’—theymake sure that broader implications are takeninto account and that powers are not beingexceeded.

773 COT’s main function is to provide experttoxicological assessment of specificchemicals, which is done principally byevaluating reports in the scientific literature. Italso declares allowable daily intake levels.COT does not consider itself to be a policy-

maker and restricts itself to providing scientificadvice.

774 COT normally examines issues referredto it by departments (the Department of Healthand the DETR), or other interested bodies(HSE and the Food Advisory Committee,FAC). The FAC is primarily a stakeholderrepresentative committee.

775 However, COT’s terms of referencepermit it to initiate its own investigations, whichit has done on several occasions (for example,on peanut allergies). COT also has a duty toprovide advice in an emergency. It is engagedfollowing an initial approach to thechairperson, who mobilises a small workingparty of the members. There are mechanismsfor postal exchanges of data and opinions sothat meetings are bypassed for the sake ofurgency. These tasks may be concluded in aslittle as three days.

Committee membership and selectionprocedures

776 The chairperson and members areappointed by the CMO following a selectionprocess conducted by the Department ofHealth. There is a shortage of applicants—potential recruits have to be sought out—butthe pool from which they are drawn is smallbecause of the expert character of theCommittee. Recruits mostly come fromacademia, and exceptionally from industry;however, in the latter case, they are appointedfor their expertise rather than their stakeholderperspective.

777 There is one non-expert (lay) member.When this innovation was first proposed, therewas a widespread view among the membersof the committee that COT was such a highlytechnical committee that a lay member wouldbe unable to contribute and would feeldisadvantaged. However, in practice, thefunction of the lay member has emerged ashighly valued because the lay member is ableto challenge the arguments presented by theexpert members of the Committee. Thepresence of the lay member causes thescientific members to express themselvesmore clearly, and this is a valuable disciplinethat assists the general functioning of thegroup. The consensus now is that the adventof the lay member has been entirely beneficial.

778 In one of the COT working groups therewere two lay members, and this is considered


102

to have worked even better because of thesupport they are able to provide to each other.

779 Members are appointed for fixed terms,usually three years, and are eligible forreappointment. Thus, on average one-third ofthe membership comes up for renewal eachyear.

780 Members are paid a nominal honorariumfor each meeting attended, plusreimbursement of expenses.

781 The secretariat is run by the FSA.

782 The COT procedural rules requiremembers to declare their relevant interests,and these are then published in the COTAnnual Report. When necessary, thechairperson will exclude a member fromparticipating in the debate on a specific issue,on the grounds of conflict of interest.

783 Members’ interests are classified aseither ‘personal’ or ‘non-personal’, where thelatter implies an interest only through themember’s employer. Conflicts are classified as‘specific’ or ‘non-specific’, according towhether the individual has an interest in thespecific chemical that is the subject of aparticular review.

784 The publication of members’ interestshas resulted in public accusations of biaswithin the part of the Committee, based on thefact that some 70% of members have at sometime or other undertaken work on behalf ofindustry—either in previous employment or asconsultants. The members strongly deny thatthese connections influence their professionaljudgement.

Method of working and form of scientificadvice

785 The COT secretariat includes a scientificteam of about four or five toxicologists, whocarry out the major tasks of retrieving scientificliterature, assessing its significance anddrafting an overall assessment. TheCommittee itself meets about six or seventimes a year to consider these issues anddetermine its programme.

786 There is a great deal of uncertainty inmuch of what COT has to do, which arisesfrom the nature of the scientific data on whichthe Committee relies. COT expresses its viewon the uncertainty in the science non-numerically; for example, in its report on CSsprays, it drew attention to the fact that ‘theavailable data did not, in general, raiseconcerns regarding the health effects of CS

spray itself.’ COT also recommended follow-upstudies.

787 COT has published a policy on opennesson its web site. The policy has the followingfeatures:

• it refers to the ‘common law ofconfidentiality’ in justifying its policy ofmaintaining confidential any informationprovided to it that is of a confidentialnature; in practice, this mostly coversresults of toxicological trials submitted byindustry;

• it commits to publishing agendas,finalised minutes, agreed conclusionsand statements;

• it commits to the appointment of a laymember;

• it commits to the holding of some openmeetings on specific topics;

• it stresses the need for all documentsreleased to be finalised and agreed bythe Committee.

788 Within the policy on openness there is aprocedure for handling confidentialinformation. Companies submitting data mustidentify information that they consider to beconfidential, and state their reasons. Thereremains a problem that the results oftoxicological trials are inherently commerciallysensitive because they may lead to regulatoryintervention. There will be inevitable pressurefrom companies to keep such resultsconfidential until some time of their ownchoosing.

789 Companies are requested to release fullcopies of their submitted reports for deposit atthe British Library, but there is no obligation onthem to do so.

790 COT publishes, with its sistercommittees on carcinogenicity andmutagenicity, an Annual Report. The maincontent of this report is a set of summaries ofits scientific findings and reasoning.

Use of the scientific advice in the policydecision

791 COT’s advice is not the only scientificinput to policy-makers. The other majorscientific source is the lobbying by industry.COT values its independent status because itfeels this gives its views more weight withpolicy-makers, in comparison with industrylobbying.

792 COT sometimes provides advice to otheradvisory committees with a more sectoral


103

mandate, for example the Veterinary ProductsCommittee, the ACP, and the Committee onSafety of Medicines. These committees havemainly scientific membership, but are moreconcerned with application of products thanwith fundamental science.

793 In the main, COT does not receive muchfeedback on the way in which its advice hasbeen used. The messages are indirect andoften considerably delayed.

794 One particular case (vitamin B6) is ofgreat relevance to the present study becauseof the issues it raises concerning the interplaybetween scientific assessment andconsideration of wider political issues.

795 COT treated vitamin B6 like a foodadditive, and recommended restrictions. Itrecommended that the allowable daily intakeof vitamin B6 from food supplements shouldbe 10mg, reaching this conclusion on thebasis of:

• quantitative results of one study, which itsaid had ‘some methodologicaldeficiencies’, but which it felt it would beunwise to ignore in the light of othersupporting human and animal data;

• qualitative support from studies at muchhigh doses and quantitative scaling (by asafety factor of 300) from trials in dogs athigh doses.

796 COT’s advice was accepted by the FAC,and the government proposed to follow thepolicy advice of FAC to treat all foodsupplements with a daily dose in excess of10mg of vitamin B6 as medicines. Aconsultation document to this effect stimulateda strong response because it would have hadserious effects on the dietary supplementindustry and, it was argued by some, on thefreedom of the individual.

797 The House of Commons AgricultureCommittee, in its fifth report (House ofCommons, 1998), also took issue with therestrictions, on the basis that they were notscientifically justified. It criticised in detail themethodological deficiencies of the study thatCOT had used, and cited alternative contraryevidence—which was in turn rejected by COT.

798 Finally, the decision was made to takeno action to change the status of vitamin B6supplements.

Advisory Committee on ToxicSubstances

799 ACTS advises the HSC, which is astatutory stakeholder body with representationfrom the Confederation of British Industry(CBI), the Trades Union Congress (TUC), localauthorities and other interested parties—theso-called ‘tripartite’ constitution. HSE is theexecutive arm of HSC and provides policyadvice to HSC; in its own right, it promotescompliance with health and safety legislation.


800 The terms of reference of ACTS are toadvise HSC on:

(a) matters relating to the prevention,control and management of hazardsand risks to health and safety ofpersons arising from the supply or useof toxic substances at work, with dueregard to any related risks toconsumers, the public and theenvironment;

(b) other associated matters referred toit by the Commission/Executive.

801 The general practice in HSC advisorycommittees is that their subject areas arealigned either with industry sectors (industryadvisory committees), or with generic topicalareas (subject advisory committees). ACTS isone of the latter. The scope of HSCcommittees also aligns broadly with segmentsof legislation for which HSC and HSE areresponsible. In the case of ACTS, the majorstatutes are:

• the Control of Substances Hazardous toHealth Regulations, 1994 (COSHH);

• the Chemicals (Hazard Information andPackaging for Supply) Regulations, 1994(CHIP).

802 Other topics are dealt with on an ad hocbasis (examples are asbestos and lead, whichhave their own statutory provisions). TheCommittee also assists in developing the UKresponse to EU proposals on occupationalexposure limits and other matters relating tochemicals. In general, HSC does not makespecific referrals to ACTS—the Committeesimply sets its own programme within its termsof reference, taking into account advice fromHSE about legislative and hazard priorities.ACTS submits its work programme to HSC forapproval.


104


803 The constitution of ACTS wasdetermined by HSC and is an extension of thetripartite model. Its membership is as follows:

• a chairperson who is a senior HSEofficial;

• four nominees of the CBI;• four nominees of the TUC;• two local government representatives

(nominated by the Local GovernmentAssociation (LGA) and the Council ofScottish Local Authorities (COSLA));

• one environmental interestsrepresentative;

• one consumer interests representative;• five independent experts.

804 HSE provides a secretariat, and much ofthe detailed work of the Committee is carriedout by related units within HSE.

805 The nominees of the CBI, the TUC, theLGA and COSLA are proposed by thoseorganisations and approved by HSC. Theother members are more difficult to recruitand, in practice, the HSE secretariat searchesfor them using informal channels ofcommunication, but with the overall intentionof complying with Nolan principles. Thesecretariat seeks to balance the skills andexperience of the independent members inparticular.

806 Although many of its members areexperts, ACTS remains primarily a stakeholderbody. It is necessary for the stakeholdernegotiations to be largely resolved at theACTS (and other comparable HSC advisorycommittees) stage, otherwise HSC would beoverwhelmed with work. Nevertheless, manyof the stakeholder nominees are in factexperts as well—they are often professionaloccupational hygienists or health and safetyspecialists. Thus, the major stakeholderdivision within the Committee (between theCBI and TUC) is to some extent bridged bythis shared background.

807 WATCH provides scientific advice, butACTS nevertheless has five independentexpert scientists as members. In practice,these independent expert scientists facilitatethe debate between the stakeholders, help inresolving differences, and maintain a constantchallenge to the process. They also provide abalance to the CBI representatives, who havemore powerful technical support than the otherstakeholders.

808 Process challenge is a valued function ofthe environmental and consumerrepresentatives, and, in the longer term, theirpresence is expected to result in a strongeremphasis on the broader context of the workof ACTS, and improvements in itspresentation.

809 Regarding declaration of interests, somemembers evidently represent the interests oftheir nominating constituencies. For theindependent members, declarations of specificinterest relating to a substance under thecommittee’s scrutiny are made as appropriate(for example, a member whose researchinstitute is undertaking a programme of workon that substance).

810 ACTS is reconstituted every three years.This process includes consideration by HSC ofthe need for the Committee, and of itsmembership, terms of reference and workprogramme.

811 The detailed work, of which there is aheavy load related to the setting ofoccupational exposure limits and labellingrequirements related to COSHH and CHIP, isdelegated to sub-committees, of which the twomost important are:

• WATCH—the main scientific/technicalsub-committee;

• and the Standing Committee on HazardInformation and Packaging (SCHIP)—atripartite consultation forum.

812 This case study only examines WATCH.It is a technical expert committee thatconsiders highly detailed risk assessments ofindividual chemicals carried out by cross-HSEteams drawn from a variety of disciplines. Italso proposes to ACTS the level at whichoccupational exposure standards should beset.

813 WATCH meets three times per year andits functions are defined by ACTS. The termsof reference are:

To consider in relation to chemicalsand ill-health the technical aspects ofhazard identification and character-isation, occupational exposureassessment, risk characterisation andthe uncertainties involved in theseissues; also to consider the technicalaspects of associated occupational riskmanagement measures; all of this inaccordance with a programme of work


105

agreed by ACTS; and to makerecommendations to ACTS, HSE and,where appropriate, to SCHIP, based onsuch technical considerations. WATCHwould also, where appropriate,consider broader risks to the publicfrom occupational use of chemicals.

814 The reference to uncertainties is a recentamendment originated by ACTS.

815 The membership of WATCH iscomposed by discipline as follows:

• toxicology 7 members• occupational hygiene 4 members• occupational medicine 2 members• epidemiology 1 member• chemical pathology 1 member• psychology 1 member

816 Members do have stakeholderaffiliations, but the way these operate is theconverse of the situation in ACTS itself. OnWATCH, members contribute as experts first,and representatives second. The members aremostly individuals whose usual employment iswithin their scientific discipline, rather than insenior management. Personal or corporateinterest in any item on the agenda is routinelydeclared.

817 In recruiting members to WATCH, HSEseeks a high level of scientific expertise andan overall balance of skills in the group.


818 ACTS meets three times a year. It doesnot make any scientific judgements—these aredelegated to WATCH. The aim, as in the HSC,is to reach a consensus.

819 ACTS considers a wide range of issuesrelating to the control of chemicals in theworkplace. These include generic healtheffects, scientific methodology, classificationand labelling, risk assessment andcommunication, as well as recommendinglevels for the setting of occupational exposurelimits under the COSHH regulations.

820 In respect of occupational exposurestandards (OESs), consideration ofappropriate limits is undertaken againstpublished ‘indicative criteria’. These criteriarequire that scientific evidence is available toidentify an exposure level that would not beharmful on the basis of continuous workplaceexposure, with reasonable certainty; thatshort-term exposures above that level would

not be likely to cause harm; and thatcompliance with the OES would be reasonablypracticable. WATCH makes arecommendation as to whether an OES canbe set for a particular chemical under theseguidelines.

821 Chemicals are referred to WATCH byACTS through an annual programming cycle.Newly introduced substances are identified toHSE by the suppliers, while the chemicals thatare already in supply are selected by aprocess of prioritisation, as there is a largebacklog. The priorities are proposed by theHSE officials on the basis of scale of use,initial judgements on the hazard, publicconcern (as expressed in enquiries fromParliament, NGOs, the media or privateindividuals), and EU or other legislativeobligations.

822 The Notification of New SubstancesRegulations require suppliers of any chemicalnot already on the notified list to make anotification to HSE of its proposal tomanufacture or import the substance, andvarious other information concerning itshazards, precautions, packaging and labelling.

823 HSE specialists present a detailedtechnical assessment, which includes a criticalreview of the literature, evaluation ofimplications of animal testing for humantoxicity, uncertainty, methods of exposuremeasurement, and an assessment of the riskof using the substance in the workplace.WATCH reviews this work and determineswhat action to recommend to ACTS.

824 There is a need for judgement indeciding which papers from the scientificliterature should be used, and how theextrapolations from animals to humans shouldbe made. This is probably the main areawhere uncertainty enters the process.

825 The committees apply a safety factorwhen they recommend an exposure limit. Thisfactor, which is judged for each case, reflectsthe nature of the extrapolation from animals tohumans, and the quality of the availableevidence. The way in which the factor is builtup from the uncertainties in the assessment isexplained in a published summary of criteria(EH64).

826 There are groups doing comparablework in Germany, the Netherlands,Luxembourg and Scandinavia, and theseshare data and the results of theirassessments. Further relevant work is done inthe USA and elsewhere.


106


827 Summaries of risk assessments arepublished in draft form for a three-monthperiod of public consultation, after which theyare revised to a final form and added to thepublished compendium EH40, which supportsthe COSHH regime. At this stage they areapplied to practice and precautions inworkplaces, and the inspection activities ofHSE field operations inspectors. A summary ofthe scientific and technical justification for thelimits is published in both the consultationdocument and in EH64.

828 Where WATCH finds that the indicativecriteria for an OES are met, it recommends thelevel of an OES to ACTS. An analysis ofcomments received through the publicconsultation exercise is also included in thematerial put to HSC. Where the criteria are notmet, the chemical must be considered againby ACTS. This is because the regulation ofthese more difficult cases involves striking abalance between risk and cost, which is seenas a matter for resolution by stakeholdersrather than technical experts. These casesmay be regulated by setting a ‘maximumexposure limit’.

829 HSC usually accepts therecommendations of ACTS. Occasionally,owing to lobbying at the HSC level, there maybe discussions about the more serious cases(the timing of introduction of the benzenemaximum exposure limit was an example).

830 The advice of ACTS is presented to HSCalongside the advice of HSE officials, whichmay be different. Members of HSC may, inaddition, be briefed by their own advisers.

831 A regulatory impact assessment isrequired for those chemicals that require amaximum exposure limit. The experience withthese assessments, which are similar to cost–benefit assessments, is that compliance costsare easy to estimate, but the risk-reductionbenefits are much more difficult.

Advisory Committee on Pesticides

832 ACP was originally set up in the early1950s and was put on a statutory basisthrough Section 16(7) of the Food andEnvironment Protection Act 1985 (FEPA).

833 ACP advises on the use of pesticidesand reports to Ministers (the Secretaries ofState for Health and Environment, togetherwith the Minister for Agriculture, Fisheries and

Food, and the Ministers of the ScottishExecutive and the Welsh Assembly). It dealswith two main groups of products: pesticidesfor agricultural applications, and a variedgroup of products for non-agriculturalpurposes, such as timber preservatives andanti-fouling agents. It is supported by the PSDand HSE.


834 Section 16(7) of FEPA requires ACP togive advice to Ministers, either whenrequested or otherwise, on any mattersrelating to the control of pests in furthering thegeneral purposes of Part III of the Act. Thesegeneral purposes include protection of humanhealth, animal and plant health; safeguardingof the environment; the securing of safe,efficient and humane methods of controllingpests; and making information aboutpesticides available to the public.

835 Under Section 16(9) of the Act, Ministersare required to consult the ACP regarding newor amended:

• regulations;• approval, revocation or suspension of

pesticide licences;• conditions to licences.

836 ACP usually considers pesticides thatare referred by other bodies, but it is also ableto initiate its own investigations. These usuallyexamine generic issues.


837 The membership of the ACP, as at 1999,comprised an independent chairperson plus13 independent members (all senior scientists)and two lay members. There are sixdepartmental ‘assessors’, representing theDepartment of Health, PSD, the WelshAssembly, the Scottish Executive, HSE andthe DETR, whose main function is to ensurethat their departments reach a decisionwhether to accept or reject the advice fromACP. There are also 18 ‘advisers’ whosefunction is technical, and a support staff of fourscientific and four policy staff.

838 Members are required to declare theirinterests in any company or organisation withan interest in the subject area of theCommittee, or such as would be likely to


107

prejudice their performance of their functionson ACP. Declarable interests includeshareholding, current consultancy contracts,employment, and non-executive directorships.

839 The register of interests is published inthe Annual Report. It is the recruitment policythat the overwhelming majority of members,including the chairperson and the deputychairperson should have no such interests todeclare. In 1997, only three members hadsuch interests. The Committee has beenlargely free from criticism of bias.

840 Appointments are formally made byMAFF Ministers, with agreement from theother departments. In the past, recruits had tobe actively sought out, and this was done byACP and PSD, and the interesteddepartments. From 1999, the Committee hasused the full Nolan procedures (advertisement,applications, interviews, etc) for selection ofnew members. It has been found thatadvertisements generate plenty ofapplications.

841 Members are appointed for three-yearterms with a maximum of two terms. They arepaid an attendance fee of about £140 (in1999) per meeting plus expenses, and this isdeemed to cover any background readingtime. The Committee meets seven or eighttimes per year.


842 The main work of the ACP is to considerapplications for approval of new pesticidesunder the Control of Pesticides Regulations1986 (COPR), and retrospective reviews ofapproved pesticides. All approvals must bepassed by ACP; however, in practice, thereare many small technical re-applications thatare processed by PSD or HSE following theframework and precedents established byACP.

843 When offering a new pesticide forapproval, or when responding to a requestfrom PSD in relation to an approved pesticide,manufacturers must submit a detailed dossierfor examination initially by the PSD/HSEscientific staff and then by ACP. For a newactive substance (as opposed to areformulation of an existing active substance),the evaluation might take 12 months. Thereare about 12–15 of these cases each year,and about 75% are eventually approved atsome level.

844 Manufacturers have a legal obligation toreveal their health risk and efficacy data. The

typical course of events is that a new pesticidefirst obtains a provisional approval, subject toa requirement to obtain more evidence;approval at first application is a relatively rareevent—there was only one in 1997.

845 The selection of pesticides forretrospective review is made by PSD, whichhas a prioritisation scheme for this purpose.Also, the EU has a pesticides assessmentprogramme in which PSD participates. In thefirst round, there were 90 cases, of which theUK handled 11.

846 The scientific staff (from PSD or HSE—as appropriate to the envisaged use of thepesticide) evaluate the dossier provided by themanufacturer and present a paper to ACP.The Committee plays a quality-assurancefunction with respect to the PSD scientificwork. In addition, the ACP chairperson is amember of the PSD Ownership Board.

847 There may be a risk of bias in theoriginal data submitted by manufacturers. Fornew chemicals, this is not considered to be amajor problem because the demands of theevaluation procedure are so great that much ofthis work has to be contracted out toindependent laboratories. Also, there issufficient contact with other national regulatorsthat data and findings can be checked—particularly for pesticides that have alreadybeen approved.

848 In seeking a balance of scientific dataand interpretation, the principal counterweightto the industry view is provided by thePSD/HSE scientists.

849 ACP is a purely scientific committee, witha low public profile and a well-defined function.This format was deliberately chosen inpreference to a stakeholder body. The laymembers were appointed as a result of abroad initiative for all advisory committees inthe Department of Health and MAFF. At theend of 1999, they had only recently started tomake an impact on the scientific process itself.The content and style of the discussions hadnot changed much since they joined theCommittee, and there was some disjunctionbetween the lay contributions and those of thescientists.


850 Formally, regulatory decisions underFEPA and COPR are made by Ministers. Inpractice, the great majority of these decisionsare made within PSD and HSE, on behalf ofMinisters, following the advice of ACP.


108

851 However, PSD says that its Ministerswish to be advised of uncertainties. Ministerswant to be able to challenge the advice, andone of the best ways of doing this is to pressfor openness about uncertainties.

852 ACP does consider a miscellany ofspecial or generic issues, in addition to itsmain work on pesticide approvals. In 1997,there were 19 of these, including such topicsas: poisoning incidents; practices in grainstores; feasibility of epidemiological researchamong pesticide users; use of buffer zones forrisk management; and pesticides useminimisation. Some of these issues touch onpolicy matters. However, from the contents ofthe Annual Report it appears that theCommittee usually restricts its advice toscientific and technical aspects. Itsrecommendations for action, however,occasionally go beyond purely scientific adviceto include policy judgements.

Observations

853 The composition of COT and ACPmembership is similar, and the main differenceis the approach to members’ interests. Itappears that ACP’s policy of excludingapplicants with interests in the pesticideindustry has served it well, although theproblems that COT has had in this area maybe undeserved.

854 ACTS, unlike COT and ACP, is openly astakeholder body, reflecting the composition ofits parent, the HSC. This Committee deals withthe problem of identifying the scientificconsensus by delegating its scientific functionentirely to a technical sub-committee. Eventhis sub-committee has strong stakeholderaffiliations, but its modus operandi is scientificdebate.

855 ACTS and its sub-committee, WATCH,are an example of structural separationbetween the scientific contribution and thenegotiation between stakeholders. This formatallows the scientific constraints to be definedfirst, the stakeholders then challenge thescience, and subsequently makerecommendations.

856 The contribution of the lay members onCOT has been very valuable, while on ACPthey have been slow to find a role. It is not

obvious why this is, because the twocommittees have very similar functions andconstitutions. There are various public-interestmembers on ACTS, but it is debatable whetherthese can be considered lay members.Certainly, the arrangements on ACTS forensuring a balance of stakeholderrepresentation are quite elaborate andcarefully engineered.

857 The terms in which all three committeesexpress their sense of uncertainty appearopaque. There are several possible reasonsfor this:

• a desire not to confuse the decision-taker;

• a desire not to allow latitude for potentialcritics to exploit;

• a basic lack of expertise in handlingprobabilities;

• a desire not to appear vague orindecisive.

858 Some of the committees, ACTS andWATCH being an example, do consideruncertainty in a systematic way, and theirmethodology is published. However, theconversion of uncertainty into safety factorsobscures the uncertainty itself, in the finaldeterminations. This may have led toinconsistencies in regulation of different risks,to over-cautious positions in some cases, andto insufficiently secure positions in others.

859 All three committees depend heavily ontheir scientific and policy support units. Theseunits actually wield significant power becauseof their agenda-setting function and theircontrol over the detailed work of the scientificevaluations of chemicals.

860 Finally, some or all of the committeesappear to be dependent on data supplied bythe industries whose products they areregulating. The data supplied is often treatedas confidential, having been so identified bythe supplier. This may expose the process tobias if there is no effective check on thedependability of the data supplied.

|O|X|E|R|A| Case Study Protocol

109

Case Study Protocol

Case-study interview procedure

861 This note guided the study teammembers in the conduct of interviews for casestudies.

862 It contains procedures and checklists.The former were intended to enforce a degreeof consistency between the various casestudies and to ensure that they could beassessed systematically during the secondphase of the project. The latter were intendedto be more flexible; they were an aid to theinterview process itself and helped to ensurethat all the key issues were raised during theinterviews.

Definition of the cases

863 The scope of each case was defined bythe decision-making issue that gave rise to it.This issue usually involved scientific and non-scientific elements. The study focused on thescientific elements, but with due regard to thewider context of the case.

864 The subject matter of each caseincluded:

• the framing of the decision-making issue;• the framing of the scientific questions on

which advice was sought;• the advice itself—its basis and the form

in which it was expressed; and• the utilisation of the advice.

Sources of data for the case studies

865 Interviewers prepared for interviews byreading suitable background material so thatthey had reasonable command of the subjectmatter.

Interviews

866 At least two interviewees were seen foreach case. Typically, at least one was aprovider of scientific advice and another was auser of that advice.

The interview procedure

867 The material exchanged during theinterviews was, by default, not confidential andmay appear in this report. However, if aninterviewee wished to provide some

information under a condition of confidentiality,that was accepted and the condition has beenrespected.

868 Both interviewers took notes. In general,it was not intended that interviewees wouldreview the notes, because this might haveintroduced bias. None of the observationswere attributed to individual interviewees.

869 The above ground rules were stated tothe interviewee at the outset of the meeting.

870 It was not intended to discuss thescience itself, only the process. However,where interviewees wished to illustrate theirresponses with scientific examples, thesewere recorded for illustration purposes.

871 Interviewees were in general seenindividually, with the particular aim of capturingthe different perspectives of providers andusers of scientific advice. However, this wasnot an absolute rule.

872 At the outset, the background and statusof the interviewee was ascertained andrecorded, covering the following:

• position;• role in the advisory or decision-making

processes;• organisation (government, advisory

committee, government researcher,academia, consultant, etc);

• independence;• applicability of a professional ethical

code;• remuneration in relation to the advice;• stakeholder interest;• scientific school of thinking—previous

declared positions on the subject.

873 The interview then examined the issueslisted in the checklists below, as appropriate tothe type of interviewee. The checklists wereused as an aide-mémoire, not as a rigidquestionnaire. However, the meeting notesidentified the topic under discussion, by itsheading in the checklist.

874 An opportunity was provided for a freecontribution from the interviewee, on the widersubject of the science advisory process,beyond the specific scope of the case study.


110

Checklists

The problem identification process

• How did the problem come to yourattention?

• In hindsight, were there any priorwarning signs or other indicators thatcould have picked up the problemearlier?

The selection of experts (questions mainlyfor decision-takers)

• How was the adviser selected?• What were the criteria for selection?• Was a search made for alternative

sources of advice?• How did the selection process handle

bias and conflict of interest?• Did the providing organisation or person

change during the course of the advisoryprocess? If so, why?

Other channels of advice (questions mainlyfor decision-takers)

• Was any advice provided on anunsolicited basis?

• Was any advice offered but declined orignored?

• Was there active public or media interestat the time of the advice or the decision?

• Were there precedents from previouscases?

• What was the role of stakeholders,lobbies, etc?

Briefing of experts

• What was the brief given to the expert?• How was this brief developed?• What background information was given

to indicate the context in which theadvice was to be provided?

• Did the requirement change during thecourse of the work?

• Did the brief request scientificinformation, or recommendations?

• Did the brief address uncertainty?• What resources were provided to the

adviser?• What timescale or deadlines were set?

The scientific analysis and preparation ofadvice (questions mainly for providers ofadvice)

• Problem definition—was the brief clearand complete, and did it providesufficient context?

• In hindsight, was the brief properlyinterpreted?

• Was the problem affected by scientificuncertainty:− was it a novel issue?− was the basic science (theory)

missing?− was data to populate the theory in

short supply?

• Was the advice required urgently? Whatwere the deadlines/time available forresponse?

• Was there any bias in the brief?• Did the way the brief was presented

influence the answer?• Was enough information available and

was any withheld?• Did it unduly circumscribe the scope?• Were sufficient resources used/provided

for?• What process was employed to generate

the advice (eg, literature search,computer simulation, experiment, peerreview, etc)?

• Who participated in the advisory process(eg, stakeholders, academics,consultants, committees, etc)?

• Would you characterise the process asany of the following: inclusive,consensual, partisan, or dominated byone individual?

• In the analysis, was the process risk-based, and, if so, did it addressalternative outcomes or alternativetheories/schools of thought?

• How would you characterise your advice,in terms of the following or anycombination?− formal analysis− structured judgement− opinion

• In the advisory response, wasuncertainty expressed, and, if so, how?

• Did the advice constitute arecommended decision, or confine itselfto scientific matters?

• Were you satisfied with the way youradvice was used?

• In hindsight, should anything have beendone differently? (Another question is:Was the answer right? Although this is


111

not for reporting, it may illuminate thisaspect.)

The decision-making process

• What was the decision required?• What type of decision was required?

− policy/strategy− tactical management/regulation− emergency response− public information

• What was at stake?• Which of the following decision-making

paradigms were invoked in what wasactually decided?− precautionary principle− optimisation (eg, cost–benefit

analysis, etc)− sustainability− risk management (probability,

consequences, etc)− political judgement by elected

representatives

• What other information and externalfactors, besides the scientific advice,were used in the decision?

• How were risk and uncertainty weighed?• Was there any conscious aversion to

large adverse consequences?• What was the justification given for the

decision?• What feedback was given to the

scientific advisers?

Evaluation of the outcome—was it a goodprocess?

• Was the scientific advice suitable andhelpful to the decision-making process?

• Which particular features worked well?• What could have been improved?• Overall, were you satisfied with the

process?

|O|X|E|R|A| References

112

References

ACMSF (1995), 'Report on Verocytotoxin-Producing Esherichia coli', London: HMSO, ISBN 0 11321909 1.

Bacon, J., in HSE (1998), 'Memorandum to the Science and Technology Committee-The ScientificAdvisory System'.

Bate, R. (1999), 'What Risk? Science, Politics & Public Health', Butterworth-Heinemann.

Beck, U. (1992), 'Risk Society: Towards a New Modernity', Sage Publications.

Beckler, D.Z. (1992), 'A Decision-maker's Guide to Science Advising', Technology in Society, 14, 15-28.

Boehmer-Christiansen, S. (1995), 'Reflections on Scientific Advice and EC Transboundary PollutionPolicy', Science and Public Policy, 22:3, 195-203, June.

Bondi, H. (1967), 'London Flood Barrier', a report prepared for the Ministry of Housing and LocalGovernment (unpublished).

Bondi, H. (1992), 'Bridging the Gulf', Technology in Society, 14, 9-14.

Cabinet Office (1997), 'Open Government: Code of Practice on Access to Government Information:Guidance on Interpretation', second edition.

Cabinet Office (1998), 'Quangos: Opening the Doors'.

Cabinet Office (1999a), 'The Civil Service Code'.

Cabinet Office (1999b), 'Overview of Regulatory and Advisory System'.

Cabinet Office (1999c), 'Policy-maker's Checklist: Using Impact Assessment and Appraisal'.

Cabinet Office (1999d), 'The Advisory and Regulatory Framework for Biotechnology: Report from theGovernment's Review'.

Cabinet Office (1999e), 'Professional Policy Making for the Twenty First Century', Strategic PolicyMaking Team.

Cabinet Office (2000a), 'Directory of Civil Service Guidance. Volume 2: Collected Guidance'.

Cabinet Office (2000b), 'Code of Practice on Written Consultation: Draft for Comment by 7th July2000'.

Cabinet Office (1999f), 'The Advisory and Regulatory Framework for Biotechnology: Report from theGovernment's Review.'

Cabinet Office (undated), 'The Better Regulation Guide and Regulatory Impact Assessment'.

Carltona Ltd v Commissioner of Works (1943), 2 All ER 560.

Collins, P. (1998), 'The Role of Academies in Advising National Governments', Interacademy Panelon International Issues, September.

Commission of the European Union (2000): 'Communication from the Commission on thePrecautionary Principle', Brussels, February 2nd, COM(2000) 1.

Committee on Standards in Public Life (the Nolan Committee) (1995), 'Standards in Public Life', May.


113

Consumers' Association (1999), 'Confronting Risk-A New Approach to Food Safety'.

Cooke, R. (1991), 'Experts in Uncertainty: Opinion and Subjective Probability in Science', OxfordUniversity Press.

Committee on Toxicity of Chemicals in Food, Consumer Products, and the Environment (1999),'Organophosphates', November 26th.

Council of Science and Technology Advisers (1999), 'Science Advice for Government Effectiveness(SAGE)', May 5th.

Council of Civil Service Unions v Minister for the Civil Service (1985), AC 374, 432-433.

Cox, G. L. (1998), 'Determination into the E. coli O157 Fatal Accident Inquiry', Sheriff Principal of theSheriffdom of South Strathclyde, Dumfries and Galloway, Airdrie, August.

Department of Health (1999), '1997 Annual Report of the Committees on Toxicity, Mutagenicity andCarcinogenicity of Chemicals in Food, Consumer Products and the Environment', September.

Department of Health, Committees on Toxicity, Mutagenicity and Carcinogenicity of Chemicals inFood, Consumer Products and the Environment (1999), 'Statement on 2-chlorobenzylidenemalononitrile (CS) and CS Spray', September.

Edwards, A. (1999), 'Scientific Expertise and Policy-making: The Intermediary Role of the PublicSphere', Science and Public Policy, 26:3, June.

van Eeten, M.J.G. (1999), '"Dialogues of the Deaf" on Science in Policy Controversies', Science andPublic Policy, 26:3, June.

ESRC Global Environment Change Programme (1999), 'The Politics of Food', Special Briefing No 5,www.gecko.ac.uk

Fairclough, J. (1990), 'Scientific Advice in Departments', internal memo to the Prime Minister

Fulton (1968), 'The Civil Service', Report of the Committee 1966-68, HMSO.

Gilbert, S.K., and Horner, R.W. (1984), 'The Thames Barrier', Thomas Telford Ltd, London,

Hahn, R.W. (1999), 'A Proposed Solution to Concerns over Public Access to Scientific Data',Testimony before the Subcommittee on Government Management, Information and TechnologyCommittee on Government Reform, US House of Representatives, AEI-Brookings Joint Center forRegulatory Studies, July.

Hambley, E., in Committee on Standards in Public Life (the Neill Committee) (1998), 'PersonalLiability in Public Service Organisations: A Legal Research Study', June 12th.

Health and Safety Executive (1998), 'Risk Assessment and Risk Management: Improving Policy andPractice within Government Departments', Second report prepared by the Interdepartmental LiaisonGroup on Risk Assessment.

HM Treasury (1996), 'The Setting of Safety Standards', The Public Inquiry Unit, HM Treasury,London.

HM Treasury (1997), 'Appraisal and Evaluation in Central Government', London, The StationeryOffice.

HMSO (1999), ‘Modernising Government’, Cm 4310, March.

Home Office (1998), 'Code of Practice on Access to Government Information'.


114

Home Office (1999), 'Freedom of Information: Consultation on Draft Legislation'.

Horner, R.W. (1978), 'Historical Background to the Thames Barrier', Proceedings of Conference onThames Barrier Design, October 5th 1977, Institution of Civil Engineers, London.

Horner, R.W., Beckett, A.H., and Draper, R.C. (1978), 'Sites and Schemes Considered', Proceedingsof Conference on Thames Barrier Design, October 5th 1977, Institution of Civil Engineers, London.

House of Commons (1998), House of Commons Select Committee on Agriculture, Fifth Report,www.publications.parliament.uk/pa/ cm199798/cmselect/cmagric/753/75302.htm

House of Commons (1999a), 'Freedom of Information Bill', November 18th.

House of Commons (1999b), House of Commons Select Committee on Agriculture, Eighth Report,www.publications.parliament.uk/pa/cm199899/cmselect/cmagric/141/14102.htm

House of Commons Science and Technology Committee (1999), 'Scientific Advisory System:Genetically Modified Foods-Government Response'.

HSC (1998), 'Raison d'être for Advisory Committee on Toxic Substances', Paper HSC/98/146,September 8th.

HSE (1996), 'An Assessment by HSE of the Structural Integrity of the Forth Rail Bridge'.

HSE (1999), 'EH40/99 Occupational Exposure Limits 1999', ISBN 0 7176 1660 6 HSE Books.

The Ikarian Reefer (1993), 2Ll L Rep 68.

Independent Expert Group on Mobile Phones (2000), 'Mobile Phones and Health', NationalRadiological Protection Board.

Interdepartmental Liaison Group on Risk Assessment (1996), 'Use of Risk Assessment withinGovernment Departments', HSE.

Interdepartmental Liaison Group on Risk Assessment (1998a), 'Risk Assessment and RiskManagement: Improving Policy and Practice within Government Departments', HSE, December.

Interdepartmental Liaison Group on Risk Assessment (1998b), 'Risk Communication: A Guide toRegulatory Practice', HSE.

International Council for the Exploration of the Seas (1999), 'Report of the Working Group on theAssessment of Demersal Stocks in the North Sea and Skagerrak', October (ICES CM 1999/ACFM:8).

Jasanoff, S. (1990), 'The Fifth Branch: Science Advisers as Policymakers', Harvard University Press.

John Innes Centre (1999), 'Organic Farming and Gene Transfer from Genetically Modified Crops',May.

Johnston, R. (1993), 'Designing an Effective Science and Technology Policy Advisory System for theAustralian Government', Science and Public Policy, 20:2, 87-96, April.

Lavender v Minister of Housing and Local Government (1970), 1 WLR 1231.

Lilford, R.J. and Braunholtz, D (1996), 'The Statistical Basis of Public Policy: A Paradigm Shift isOverdue', The British Medical Journal, 313, 603-7, September 7th.

Lord Rothschild (1971), 'The Organisation and Management of Government R&D'

Lord Tomlin (1931), 'Report of the Royal Commission on the Civil Service', Cmd 3909, 1931.


115

MAFF and HSE (1997), 'Advisory Committee on Pesticides Annual Report 1997'.

May, R. (1999), 'Genetically Modified Foods: Facts, Worries, Policies and Public Confidence,' DTI.

National Environmental Policy Institute (1998), 'Enhancing Science in the Regulatory Process(ESRP): A Guide for Congress, Federal Agencies, Policy-makers and Others Interested in Improvingthe Use of Scientific Knowledge in Public Policy Decisions', Autumn.

Nelkin, D. (1975), 'The Political Impact of Technical Expertise', Social Studies of Science, 5, 935-954.

Neustadt, R.E. and Fineberg, H.V. (1978), 'The Swine Flu Affair: Decision-Making on a SlipperyDisease', US Department of Health, Education and Welfare.

Office of Science and Technology (1997), 'The Use of Scientific Advice in Policy Making: A Note bythe Chief Scientific Adviser'.

Office of Science and Technology (1998a), in House of Commons Select Committee on Science andTechnology, 'Select Committee on Science and Technology Minutes of Evidence, MemorandumSubmitted by the Office of Science and Technology', July 31st.

Office of Science and Technology (1998b), 'The Use of Scientific Advice in Policy Making:Implementation of the Guidelines', July.

Office of Science and Technology (2000a), ‘A Consultation Document on a Code of Practice forScientific Advisory Committees’, July.

Office of Science and Technology (2000b), ‘Guidelines 2000—Scientific Advice and Policy Making ',July.

Office of the Commissioner for Public Appointments (1998), 'The Commissioner for PublicAppointments' Guidance on Appointments to Public Bodies', July.

Office of the Commissioner for Public Appointments, 'The Code of Practice for Public Appointments'.

'Opinions' of the SCTEE, April 24th 1998, November 26th-27th 1998, and September 28th 1999.

Pennington, T.H. (1999), 'Recent Experiences in Food Poisoning: Science and Policy, Science andthe Media'.

Pennington, T.H. (1999), 'VTEC: Lessons learned from British Outbreaks' manuscript submitted to theJournal of Applied Microbiology.

The Pennington Group (1997), 'Report on the Circumstances leading to the 1996 Outbreak ofInfection with E. coli O157 Infection in Central Scotland: The Implications for Food Safety and theLessons to be Learned', The Scottish Office, ISBN 0 11 495851 3.

Peters, B.G. and Barker, A. (1993), 'Advising West European Governments: Inquires, Expertise andPublic Policy', Edinburgh University Press.

R v Gough (1993), 2 All ER 724.

Radioactive Waste Management Advisory Committee (1999), 'The Radioactive Waste ManagementAdvisory Committee's Advice to Ministers on the Establishment of Scientific Consensus on theInterpretation and Significance of the Results of Science Programmes into Radioactive WasteDisposal', April.

Renn, O. (1995), 'Style of Using Scientific Expertise: A Comparative Framework', Science and PublicPolicy, 22:3, 147-56, June.


116

Roe, E. (1994), 'Narrative Policy Analysis: Theory and Practice', Duke University Press, Durham.

Rooker, J., in House of Commons Select Committee on Science and Technology (1999), 'SelectCommittee on Science and Technology Minutes of Evidence', April 26th.

Rothschild, (1971), 'The Organisation and Management of Government R&D'.

Royal Commission on Environmental Pollution (1998), '21st Report: Setting EnvironmentalStandards', October.

The Royal Society of Canada (1998), 'Expert Panels: Manual of Procedural Guidelines', version 1.1,December.

Scottish Office (1996), ‘The Investigation and Control of Outbreaks of Food-borne Disease inScotland’.

Shapiro, S.A (1990), 'Public Accountability of Advisory Committees', Risk: Health, Safety &Environment, Summer.

Shrader-Frechette, K. (1990), 'Scientific Method, Anti-Foundationalism and Public Decision Making',Risk: Health, Safety & Environment, Winter.

Skoie, H. (1993), 'Science and Technology Advice to Governments-A Norwegian Perspective',Science and Public Policy, 20:2, 79-86, April.

Slovic, P. (1993), 'Perceived Risk, Trust, and Democracy', Risk Analysis, 13:6, 675-82.

Smith, W. (1997), 'Review of Expert Panels for Provision of Scientific and Technological Advice forDevelopment of Public Policy', Ministry of Research, Science and Technology, Report no. 61, May.

Smith, W. and Halliwell, J. (1999), 'Principles and Practices for Using Scientific Advice in GovernmentDecision Making: International Best Practices', Report to the S&T Strategy Directorate, IndustryCanada, January.

Stein, J.A. and Renn O. (1998), 'Transparency and Openness in Scientific Advisory Committees: theAmerican Experience', working document for the STOA Panel, November.

Stirling, A. (1998), 'Risk at a Turning Point?', Journal of Risk Research, 1:2, 97-109.

Supply Chain Initiative on Modified Agricultural Crops (1999), 'SCIMAC Code of Practice on theIntroduction of Genetically Modified Crops', May.

Weingart, P. (1999), 'Scientific Expertise and Political Accountability: Paradoxes of Science inPolitics', Science and Public Policy, 26:3, June.

Wilkie, T. (1991), 'The Unfulfilled Dream', British Science and Politics Since 1945, BlackwellPublishers.

William Cory v City of London Corporation (1951), 2KB 476.

Worcester, R.M. (1999), 'Seeking Consensus on Contentious Scientific Issues: Science andDemocracy', Foundation for Science and Technology, July 12th.

Various authors (1978), 'Discussion of Papers 1-3', Proceedings of Conference on Thames BarrierDesign, October 5th 1977, Institution of Civil Engineers, London.

6, P. (1999), The Ditchley Foundation, 'Understanding, Managing and Presenting Risk in PublicPolicy', Ditchley Conference Report No. D99/06, May 7th-9th.

|O|X|E|R|A| Acknowledgements

117

Acknowledgements

OXERA would like to thank the individualslisted below, who kindly agreed to discuss thisproject, and to contribute their ideas andopinions.

• Ron Amann, Centre for Managementand Policy Studies, Cabinet Office

• Elizabeth Atherton, UK Nirex plc• Philip Balls, Fisheries Research

Requirements, Scottish Executive• Declan Barraclough, EA• David Bench, MAFF• Elaine Brown, Scottish Executive• Joe Brown, Science and Technology

Unit, Scottish Executive• Paul Burrows, DETR• Julian Braybrook, LGC• Michael Clarke MP, House of Commons

Science and Technology Committee• Alan Cooksey, HSE• Peter Collins, Royal Society• Mark Courtney, Cabinet Office• Karl Cunion, DETR (Strategy and

Corporate Services Group)• Keith Davies, Royal Academy of

Engineering• Paul Davies, HSE• Sue Davies, Consumers’ Association• Brian Doble, Royal Academy of

Engineering• Robin Fielder, COT secretariat,

Department of Health• Chris Fisher, Joint Food Standards and

Safety Group, MAFF• David Fisk, DETR• Lord Flowers• Robin Foster, HSE• Jack Gow, Council of Scientific and

Technical Institutions• Alan Gray, Institute of Terrestrial

Ecology, Wareham• Peter Graham, Strategic and Analytical

Support Directorate, HSE• John Grieg, Department of Health• John Gummer MP• Christine Hennessy, Consumer Safety

Unit, DTI• Julie Hill, Green Alliance• Tom Horlick-Jones, Centre for

Environmental Strategy, University ofSurrey

• Lord Hunt, Department of Health• Neil Jackson, Scottish Executive

• Geoff Kirkwood, Imperial College ofScience, Technology and Medicine

• Peter Lachmann, Academy of MedicalSciences

• William Leiss, Royal Society of Canada• Peter Levein, Scottish Executive• Ian Lomas, DTI• Stuart MacDiarmid, New Zealand

Ministry of Agriculture and Forestry• David Mallon, Agricultural and Biological

Research Group, Scottish Executive• Sir Robert May, the OST• Michael Meacher, DETR• Jim McQuaid, HSE• Ian Meadowcroft, EA• Ann Millar, Scottish Executive• Nigel Ogilvie, Railtrack• Tony O’Hagan, Sheffield University• Patrick O’Sullivan, UK Nirex Plc• Ron Oxborough, Imperial College of

Science, Technology and Medicine• Douglas Parr, Greenpeace• Angela Patel, Department of Health• John Perry, Consumer Safety Unit, DTI• Hugh Pennington, Department of

Medical Microbiology, University ofAberdeen

• Edmund Quilty, the OST• Naomi Rees, Joint Food Standards and

Safety Group, MAFF• Peter Robinson, Quantisci• Lord Sainsbury, DTI• Rosalind Skinner, Public Health Policy

Unit, Scottish Executive• Kevin Stokes, Fisheries Systems

Modelling Group, CEFAS, Lowestoft• Brian Thompson• Roger Vine, EA• Jim Wildgoose, Rural Affairs

Department, Scottish Executive• David Williams, Pesticides Safety

Directorate, York• Peter Winterbottom, Fisheries III, MAFF• Simon Wisbey, UK Nirex plc• Frank Woods, Professor of Medicine,

University of Sheffield• Roger Yearsley, EA

|O|X|E|R|A| Tender Specification

118

Tender Specification

Background

In assessing risks and adopting appropriaterisk-management strategies, governmentregulators need, among other things, sound,authoritative expert scientific advice to informdecision-making.

While expert scientific input is essential,experience (perhaps most recently exemplifiedby food and medicinal safety concerns)indicates that there is often a lack oftransparency as to whether the role of thescientific expert is to advise about facts (ie,what is known or not known about an issue, orto express a value judgement about the actionthat should be taken by government).

In addition, health or safety risks (particularlyemergent risks) are often associated withsignificant uncertainty arising from, forexample, unreliable or incomplete data,modelling uncertainties, debatableunderpinning assumptions, or conflicts ofexpert judgement or opinion in interpreting theresults.

The pervading uncertainty and the potentialconfusion surrounding the role of scientificexperts may be compounded by the publicperception that, to varying degrees:

• experts engaged to provide advice arenot independent;

• the process by which expert advice iselicited is not transparent;

• the expert view of risk is out of tune withthat of the public;

• resort to expert judgement or opiniondoes not provide reassurance, but

• leads to the spectacle of mediaconfrontation between experts with verydifferent views.

These considerations point to the need forgovernment departments to be able todemonstrate that they follow good practicewhen assembling expert scientific advice thatinforms decisions.

Aims

To identify good practice for securing andusing expert scientific advice.

To provide a sound framework, based on thedominant assumptions of openness andtransparency, on which the UK government

can assemble authoritative expert scientificadvice that is robust when exposed to publicscrutiny.

To enhance trust and confidence in theprocesses of risk assessment, managementand communication by:

• opening up to public scrutiny the expertadvice elicited;

• exposing and explaining theassumptions made, and the uncertaintiesthat pervade both the assessment ofrisks and the effectiveness of possiblerisk-management options;

• making clear where expert judgementhas been applied to convert informationand expertise into intelligence about riskproblems, or where the uncertainties areso large that the expert advice isessentially a matter of opinion;

• adopting suitable procedures to engageas appropriate both stakeholders andexperts; and

• explaining how expert scientific advice,together with the relevant sociological,economic and political considerations,contributed to the final decision made.

Methodology and outputs

In Phase 1, to undertake a mapping study to:

• identify and categorise current practiceswithin government for securing expertscientific advice, and incorporating it intopolicy; and,

• review what is known generally about therelationship between how expertscientific advice is incorporated intopolicy and the quality, or perceivedquality, of the decisions made.

In Phase 2, to probe in greater depth and todraw out principles of good practice (includinghow to avoid pitfalls) for:

• the engagement of scientific experts (ie,selection, remit, independence, etc);

• the elicitation of their advice (ie, framingof issues, support provided, avoidance ofbias, characterisation and reporting of

|O|X|E|R|A| Tender Specification

119

uncertainty, resolution of conflict,presentation of advice, etc); and,

• the incorporation of expert scientificadvice in the wider decision-makingprocess.

Prospective contractors are invited to devisethe approach and methodology they considerappropriate. The sponsors will arrange

opportunities for the successful contractor tointerview selected:

• departmental scientists;• external scientific experts who advise

government, including chairpersons andmembers of advisory committees;

• departmental policy customers.

Printed and published by the Health and Safety ExecutiveC30 1/98

Printed and published by the Health and Safety ExecutiveC3 10/00

CRR 295

£20.00 9 780717 618521

I S B N0-7176-1852-8

CONTRACT RESEARCH REPORT 295/2000

Documents

Transcript of CONTRACT RESEARCH REPORT 295/2000