Scud

This page is a encyclopedia article. The page identifier is Op_en4159
Moderator:Jouni (see all)
Upload data

Environmental health risk analysis: scientific uncertainties in decision-making (SCUD) is a research project in THL in 2005-2010.

Abstract

• Name of the researcher in charge: Jouni T. Tuomisto
• Title of the project: Environmental health risk analysis: scientific uncertainties in decision-making (SCUD)
• Number of person-years: 5 (1.8.2005 – 31.7.2010)
• Site of research: National Public Health Institute, the Centre of Excellence of Environmental Health Risk Analysis, Kuopio, Finland

Objective and brief description:

Risk assessment and risk analysis are becoming more and more important fields when societies try and understand health and environmental effects of various activities. Formal and scientifically based methods are needed, as causal chains of the remaining problems are increasingly complex and span across several scientific disciplines and governmental sectors. For example in the European level, the REACH chemical initiative has changed the focus of chemical risks from single pre-marketing assessments to complex environmental and health assessments of existing substances. Therefore, there is a great need to develop risk analysis methods that provide guidance and scientific understanding to policy-makers. The main challenges lie within the interface of science and policy, and therefore a wide multidisciplinary approach must be used. There are two large research needs: methods for estimating the value of scientific information for decision-making, and methods for explicitly describing the actual decision situations in a way that scientific understanding can be utilised.

The objectives of this research plan are to

1. develop methods for estimating the value of scientific information for decision-making,
2. develop methods for explicitly describing the actual decision situations in a way that scientific understanding can be utilised, and
3. perform comparative risk analyses and risk benefit analyses on topics with high relevance in public health, notably dioxins and fine particles.

The proposed work will utilise several new and advanced methods in the fields of decision analysis, risk analysis and probability theory, such as value of information, Bayesian methods, Monte Carlo, and structured deliberation. It will enhance the scientific understanding and evidence-based practices in the interface of science and public policy. It will have both scientific merit and practical importance, because the risks under scrutiny are selected to be highly relevant for public health and health policy, namely fine particle pollution in air, dioxins in food.

In addition to scientific articles, the documented risk models and structured deliberations will be published as supporting online material to the articles. This will enable and enhance rational scientific and policy discussions on the environmental health risks, and hopefully promote the idea of evidence-based policy-making in this field.

Research plan

Background

Risk assessment and risk analysis are becoming more and more important fields when societies try and understand health and environmental effects of various activities. Formal and scientifically based methods are needed, as causal chains of the remaining problems are increasingly complex and span across several scientific disciplines and governmental sectors. For example in the European level, the REACH chemical initiative has changed the focus of chemical risks from single pre-marketing assessments to complex environmental and health assessments of existing substances. Therefore, there is a great need to develop risk analysis methods that provide guidance and scientific understanding to policy-makers. The main challenges lie within the interface of science and policy, and therefore a wide multidisciplinary approach must be used. There are two large research needs: methods for estimating the value of scientific information for decision-making, and methods for explicitly describing the actual decision situations in a way that scientific understanding can be utilised. The Centre of Excellence of Environmental Health Risk Analysis has done research in both areas.

Cyclic process and value of further information

The concern about a new risk is heavily influenced by the previous knowledge about similar situations and risks. These concerns direct risk assessments to some risks and leave some others with less emphasis. The new information directs pressures for risk management. This affects which risks are studied further. The process is influenced by rational and random factors, as well as by political and scientific ambitions.

Traditionally risk assessment and risk management are kept separately to keep science unaffected by policy. However, the process of risk assessment and risk management is inherently cyclic. The questions whether and what to study have political dimensions, and they should be expressed explicitly to be able to see risks and risk assessments in their context. We therefore propose a risk assessment method that is based on a cyclic approach. A good risk characterisation helps in gaining new perspectives in the policy arena and possibly making new, wiser questions. These can then be answered by subsequent risk assessments. We will use the method of value of information, which estimates the usefulness of particular pieces of further scientific research (see, e.g. Tuomisto et al., 2004d).

Structured deliberation in the science-policy interface

The decision-making process often proceeds faster than model building. The demands for action arise within days after e.g. a discovery of environmental pollution episode. Risk assessment modelling is just unable to cope at that pace. Thus, the use of risk assessments is often limited to standard authority decisions such as pre-market licensing of new chemicals, large construction plans requiring environmental impact assessment, or limit value setting for certain well-defined pollutants. Many other decisions are based on expert opinions collected with ad hoc procedures, or, even worse, on need to show action without knowledge on its usefulness.

Understanding of a particular risk develops simultaneously at various levels and using differently structured methods. At one end there is a "discussion layer": political and public unstructured discussions about risks of a hazard with a wide interest on e.g. economical consequences of available decision options, public health, and social justification equity. At the other end there is a "modelling layer": models dealing with specific questions such as air concentrations of a pollutant from a specific source or risk-benefit analyses of particular actions. There is a need for methods facilitating the flow of information and understanding between the existing layers.

We propose a method of structured deliberation as an intermediate layer in the field of risk assessment. It describes a risk situation in a formal manner by using a diagram method developed by us. It is an enhanced causal diagram that contains items along the causal pathway (or network) from emissions to dispersion to exposure to effects. It is designed to describe also other than causal connections such as values and arguments (Tuomisto, 2004b).

Personal development

I started my research career in the National Public Health Institute, Department of Environmental Health (KTL/YTOS) after graduation from the medical school of the University of Kuopio, 1992. The topic was toxicology of dioxins in rats, especially the effects on feeding regulation, and modulation of effects by two dioxin resistance genes. I also did some work in the field of epidemiology and conducted a large multi-centre dioxin case-control study in the Finnish general population (Tuomisto et al., 2004a;Tuomisto et al., 2004b). However, the thesis dealt with mechanistic toxicology (Tuomisto, 1999).

During that time, I became more interested in applied research, especially environmental health risks. The basis of toxicology and epidemiology proved very useful in the search of a post-doc place, which was found from Harvard University, Boston, USA. Harvard Center for Risk Analysis is within the Harvard School of Public Health, and the scope of the centre was optimal regarding my interests. I joined the group of Dr. John S. Evans, who was studying the health risks of air pollution, notably fine particles (PM). Although the field was new to me, I had a strong support from my home institute (KTL/YTOS) where there are active research groups studying PM epidemiology (Prof Juha Pekkanen), toxicology (Doc Raimo Salonen), and exposure (Prof Matti Jantunen).

In Harvard, I studied risk assessment and decision analysis, and participated a risk assessment study on ozone (Levy et al., 2001). However, the main research area was on the costs of scientific uncertainty in decision-making related to abatement strategies of PM (Carrothers et al., 2004). I also made personal contacts with the most prominent PM epidemiologists and toxicologists in the United States.

During my stay in Boston, we decided to apply for a Centre of Excellence in KTL/YTOS. As we had had established dioxin and PM research for several years, we felt it natural to base the application on those fields. The overarching theme of the Centre became risk analysis. This was a perfect match to my scope of research I was conducting in Harvard at the time. When the application was accepted, I got the leadership of risk analysis research. The main topics were comparative risk assessment of PM and dioxin (Tuomisto et al., 2004c), and coherent utilisation of both epidemiological and toxicological data, among others.

Risk analysis of PM progressed fast, because we got funding from e.g. TEKES (KOPRA project from FINE program). The first results were of direct practical relevance (Tainio et al., 2003;Tainio et al., 2004;Tainio and Tuomisto, 2003). Currently I am supervising four doctoral students on the risks of PM and dioxin (one of the theses is partly dioxin toxicology, one is about the utilisation of risk assessment information in decision-making; the latter is being done over and above a permanent position outside the institute).

I have chaired \'Journal Club\' type regular researcher meetings within the Centre collecting toxicologists, epidemiologists, and exposure assessors together. The aim has been to read new literature, discuss, and develop research ideas. This work has increased collaboration between the existing groups within the house (Hänninen et al., 2004). This has also enabled us to rapidly (three weeks from the idea to manuscript submission) assess comparative risks and make a contribution to a public debate in the journal Science in the field of environmental health. (Tuomisto et al., 2004d).

Five most important publications

Carrothers TJ, Wolff SK, Tuomisto JT, Wilson A, Levy JI, Graham JD et al. Fine particulate air pollution in the US: a preliminary analysis of the value of research. In: European Commission, WHO, European Collaborative Action, editors. Role of human exposure assessment in air quality management. Report on the joint workshop. 2004.

Tainio M, Tuomisto JT, Aarnio P, Hänninen O, Koistinen K, Jantunen M et al. Health effect caused by primary fine particulate matter (PM2.5) emitted from busses in Helsinki Metropolitan Area, Finland. Risk Analysis 2004; in press.

Tuomisto J, Pekkanen J, Kiviranta H, Tukiainen E, Vartiainen T, Viluksela M et al. Dioxin cancer risk - example of hormesis? Non linearity in Biology, Toxicology and Medicine 2004.

Tuomisto JT, Pekkanen J, Kiviranta H, Tukiainen E, Vartiainen T, Tuomisto J. Soft-tissue sarcoma and dioxin: A case-control study. Int J Cancer 2004; 108(6):893-900.

Tuomisto JT, Tuomisto J, Tainio M, Niittynen M, Verkasalo P, Vartiainen T et al. Risk-benefit analysis of eating farmed salmon. Science 2004; 305(5683):476.

Objectives and methods

The objectives are to

1. develop methods for estimating the value of scientific information for decision-making.
2. develop methods for explicitly describing the actual decision situations in a way that scientific understanding can be utilised.
3. perform comparative risk analyses and risk benefit analyses on topics with high relevance in public health, notably dioxins and fine particles

A traditional definition says that risk is a combination of the probability and severity of consequences caused by a hazard. However, in many cases a wider definition for risk is more useful: a particular risk is fully understood only when the population under risk, the causal pathway from emissions to effects and the risk reduction actions are also explicitly defined.

By including actions (i.e. decision options) in the definition of risk, risk assessment becomes more focused on offering guidance to, but also requiring input from, risk management. In this way, many assumptions become more easily justified and explicit. It is also possible to use several useful decision analytical methods such as value of information. The applicability and advantages or disadvantages of the decision-based approach will be explored and utilised in this project.

Structured deliberation is used as a basis for this. It has an advantage of being relatively easily understandable and readable, and it has a basic structure similar to but simpler than mathematical causal models. It is based on influence diagrams (or directed acyclic graphs) that are both intuitive and useful for a basis for e.g. Bayesian modelling. Therefore, translation into a natural spoken language or into a mathematical modelling script is relatively easy, as is required from an interface. Due to this simplicity, it is much faster to build and less data-demanding. It is easy to expand into new areas, as the political discussion proceeds. Because of its structured and formal nature, it requires that many assumptions are made explicit unlike in political rhetoric, and therefore it is easier to identify possibly illogical or conflicting issues. It can be developed further into a form of a mathematical model when needed. Structured deliberation also contains preferences and values, as well as reasoning for e.g. why a particular issue was or was not included in the scope.

Risk comparisons. We will review the methods of comparing risks, which are not directly commensurable. Such methods include monetary valuations, burden of disease concept measured by quality or disability adjusted life years (QALY, DALY, respectively), and expert judgement elicitation. Precautionary principle is a major issue in comparing risks or actions. It can be defined as a systematic assessment and characterisation of uncertainties and their inclusion in decision-making. This is an important theme throughout the project. However, also another definition has been presented: it can be seen as a tendency to make errors towards the safe side. This has similarities with the burden of proof concept, which has strong roots in legal science instead of natural sciences. The implications and differences of the two precautionary principle approaches will be explored. This work has already started within the Centre (Tuomisto, 2004a).

Bayesian methods and expert judgement elicitation have been and will be further used. Bayesian methods (based on concepts of subjective probabilities and updating with new information) are good for dealing with sparse data. In the area of dose-response relations, data on human exposures to hazardous materials is invariably incomplete and often non-existent. Quantitative analysis must therefore have recourse to subjective assessments of experts. Bayesian methods in this field may be roughly characterized as follows. A prior distribution over the parameters of a model is selected, and updated with observations of variables whose values are predicted by the model. If no or few observations are available, expert subjective distributions over observable variables are elicited and used to adapt the prior distribution. The resulting posterior distribution describes the uncertainty after the prior knowledge and the new data.

Existing methodologies will be more widely used across all substance areas in the Centre. This will improve research plans, speed up data analysis, and improve the use and applicability of research data. Examples of such methodologies and new applications are primary component analysis (exposure analysis of dioxins), regression analysis (dose-response analysis in toxicology), Monte Carlo (exposure models of PM and dioxins), data management procedures from epidemiology (all areas).

Variation, uncertainty, and value of information. In the input variables of a risk model, there are always variability (true differences between members of a population) and uncertainty (lack of knowledge about the true value). Although it is well known that these issues must be taken into account, it is usually not easy to do so. In this project we apply and combine several common and new techniques in this area. The first task is to identify the sources of variation and uncertainty throughout the whole assessment. Structured deliberation and iterations are valuable methods in this. The sources are then explicitly separated and quantified and their effects on the risk estimate are fully investigated. Probabilistic modelling (such as Monte Carlo), and Bayesian (including expert judgement) techniques are used.

Value-of-information approach is used to assess which parts of the model and which uncertainties are critical for the outcome. It describes how much the decision is expected to improve, if a particular uncertain variable is known better, or in other words, how much is it worth to pay to reduce a particular uncertainty. Less important parts of the model are kept as simple as possible, while more emphasis is placed on the important parts, and those are developed further. This information is very important, as it can be used to direct decision-making, model development, and further research.

Schedule. We currently have several manuscripts on comparative risk analysis on dioxins and fine particles, on structured deliberation, on value of information of fine particles, and on risks of fine particles due to specific sources. Many of these are expected to be published during the next year, and the active publishing policy will continue in the coming years. By the end of the applied period, I am likely to go and work in a research institute abroad (probably in Europe, if the current large applications prove successful) to further develop the research methods in risk analysis.

Objectives in research training. My two first doctoral students are expected to graduate during year 2006, and the two others during years 2007-2008. The large research community in the Centre produces so much basic knowledge on environmental health that several more doctoral students could start on risk analysis. We are therefore actively applying for new funding from domestic and European sources. There are at least two proposals under preparation in addition to those already submitted and listed in "Other funding". We are also actively maintaining and developing connections to the University of Kuopio in the fields of environment, health, and risk (notably the SYTYKE graduate school).

Ethical questions. No patient or animal studies will be performed. The ethical questions therefore relate to proper data management and protection. The rules and guidelines for data protection and archiving of KTL will be followed. A special emphasis will be put on clear and explicit documentation of the models used. All models, input data, and assumptions will be published together with the results. Only those parts will be refrained from publication that contain non-public material.

Researchers and resources

Current projects

I am leading five (sub)projects, which will be briefly described here. Two of them have separate research funding from TEKES, or the Academy of Finland; the others are directly funded by the Centre of Excellence grants or KTL.

HEAT (Health Effects caused by Urban Air Pollution for the Transport System Plan Scenarios in Helsinki Area) is a project with Helsinki Metropolitan Area Council, Finnish Meteorological Institute, and KTL as partners. It is funded by the Academy of Finland (TervE program). I am responsible for the module that estimates the health effects based on the modelled air pollution exposures. The project funding will end this year; the results will be published next year.

KOPRA (Kokonaismalli pienhiukkasten päästöjen, leviämisen ja riskin arviointiin) is a project estimating the health risks caused by all PM emissions in Finland. Finnish Meteorological Institute, Finnish Environmental Agency, and KTL are partners. I am leading the KTL subproject, which focuses on an exposure assessment tool \'intake fraction\', and on health impact assessment. The project is funded by TEKES (FINE program) and it will end by the end of year 2005. A doctoral student is working full-time on this project under my supervision.

Composite traffic project is closely related to HEAT project. We are exploring the potential for air pollution emission reduction and consequent health benefit by studying novel traffic scenarios in Helsinki Metropolitan Area. This work relates closely to the interface of science and policy. One of the main interests is to explore the role of scientific uncertainties in a decision-making process with multiple decision-makers.

Comparative risks of dioxin have been a main focus of the Centre since its beginning. Although dioxin is an environmental health concern, in Finland the main source is fish, which is otherwise healthy. Thus, the question about competing risks has high public health relevance. One doctoral student is working on this project full-time under my supervision. This project has involved most dioxin researchers from all fields in KTL/YTOS, and a risk-benefit analysis has been published (Tuomisto et al., 2004d). Other aspects in this project have been comparative risks of PM and dioxin on one hand (Tuomisto et al., 2004c), and developmental and cancer risks caused by dioxin on the other hand.

Dioxin and biliverdin accumulation is a project with one doctoral student. The focus is mostly toxicological, studying a unique phenomenon of biliverdin (a heme metabolite) accumulation in the rat liver after a large dioxin dose. This phenomenon was first found during my doctoral thesis work, and it has been subsequently characterised and published (Niittynen et al., 2003). There is a link to dioxin risk assessment because the phenomenon is seen in neither very dioxin-resistant nor dioxin-sensitive rats. It is only seen with rats genetically moderately resistant to dioxin effects. The phenomenon may therefore reveal important mechanisms related to the variation in dioxin risks.

Working environment and collaboration

The Centre of Excellence of Environmental Health Risk Analysis consists of European top researchers in the fields of environmental epidemiology, toxicology, and exposure assessment (about 25 researchers altogether). Close contacts have been a most valuable for developing risk analysis research, as the expertise exceeds the critical mass and enables the usage of up-to-date data and understanding of the underlying causal processes leading to environmental health risks. The institute also contains excellent technical facilities including the basic computer resources, but also advanced programs and researchers and statisticians skilled to use them (programs such as SAS, R, Matlab, WinBUGS, and Analytica are being used on regular basis).

In the Centre of Excellence, I have been actively increasing the expertise of several new methodologies by developing collaboration. Eliciting and combining expert opinions is one such area. These formal interview methods are needed when risk assessment requires data that cannot be obtained by measurements. We have ongoing collaboration in a project "Expert judgement of uncertainty in particulate matter mortality effects" with Dr John S. Evans (Harvard University) and Prof Roger Cooke (Delft University of Technology, the Netherlands). In the project we interviewed the most prominent PM epidemiologists and toxicologists in Europe and in Mexico City. The very first report has been produced (Cooke, 2004), and the main part of the results will be published in the near future. This work is funded by Delft University of Technology, the funding coming originally from the United Nations.

Another area is Bayesian probability methods (using subjective probabilities and probability updating when new data becomes available). We have active collaboration with Prof Elja Arjas (Rolf Nevanlinna Institute, University of Helsinki). This area is under active development in the Centre, and the first results are being completed (Erästö et al., 2004).

The University of Kuopio has a graduate school SYTYKE for environmental health. A large proportion of the students involved come from KTL/YTOS and/or the Centre of Excellence. The senior researchers of the Centre (including myself) are actively teaching courses on risks, toxicology, epidemiology, and exposure assessment. I was recently appointed docent in the Faculty of Pharmacy on toxicology, especially risk assessment.

The Centre has participated in the EU funded network of excellence AIRNET. The network has popularised air pollution knowledge and developed science-policy interface between researchers and policy-makers. In addition to the coordinator (Utrecht University, the Netherlands), KTL has been the only institute participating in all working groups of AIRNET (toxicology, epidemiology, exposure assessment, health impact assessment, science-policy interface). I have participated in the two latter working groups, and I have also been an editor for the science-policy report.

Results

The proposed work will enhance the scientific understanding and evidence-based practices in the interface of science and public policy. It will have both scientific merit and practical importance, because the risks under scrutiny are selected to be highly relevant for public health and health policy. Fine particle pollution is one of the largest remaining environmental health hazards with no easy and clear-cut solutions. The magnitude of dioxin risk at current exposure levels is controversial, but dioxin exposure is related to otherwise healthy fish, and the secondary health effects of dioxin policies may be very large.

All results will be published in high-quality scientific journals. In addition, we will actively publish the input data, assumptions and models in a well-documented and clear way as supporting online material to the articles. This will enable and enhance rational scientific and policy discussions on the environmental health risks, and hopefully promote the idea of evidence-based policy-making in this field.

References

Carrothers,TJ, S K Wolff, J T Tuomisto, A Wilson, J I Levy, J D Graham, J S Evans, 2004, Fine particulate air pollution in the US: a preliminary analysis of the value of research, in European Commission, WHO, and European Collaborative Action (eds), Role of human exposure assessment in air quality management. Report on the joint workshop.

Cooke,RM. Results of expert judgement study on health effects of fine particulate matter for the United Nations Claims Commission. 1-9. 2004. Delft, Delft University of Technology.

Erästö,P, J T Tuomisto, M Tainio. A model for dioxin concentrations in herrings in the Baltic Sea. Society For Risk Analysis Annual Meeting . 2004.

Hänninen,O, J T Tuomisto, T Yli-Tuomi, M Jantunen, 2004, Reduction potential of urban PM2.5 mortality risk using modern ventilation systems in buildings: Submitted.

Levy,JI, T J Carrothers, J T Tuomisto, J K Hammitt, J S Evans, 2001, Assessing the public health benefits of reduced ozone concentrations: Environmental Health Perspectives, v. 109, p. 1215-1226.

Niittynen,M, J T Tuomisto, S Auriola, R Pohjanvirta, P Syrjala, U Simanainen, M Viluksela, J Tuomisto, 2003, 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD)-induced accumulation of biliverdin and hepatic peliosis in rats: Toxicological Sciences, v. 72, p. 1756.

Tainio,M, J T Tuomisto, 2003, Comparing methodologies of six fine particle risk assessments, in J Honkanen and P Koponen (eds), Proceedings. Sixth Finnish Conference of Environmental Sciences: Joensuu, University of Joensuu.

Tainio,M, J T Tuomisto, P Aarnio, O Hänninen, K Koistinen, M Jantunen, J Pekkanen, 2004, Health effect caused by primary fine particulate matter (PM2.5) emitted from busses in Helsinki Metropolitan Area, Finland: Risk Analysis, v. in press.

Tainio,M, J T Tuomisto, P Aarnio, M Jantunen, K Koistinen, O Hänninen, J Pekkanen, 2003, Estimation of the contributions of different sources to average PM2.5 exposure of the adult population of Helsinki, 13th Annual Conference of International Society of Exposure Analysis. Abstract book.: Stresa, Italy, International Society of Exposure Analysis.

Tuomisto,J, 2004a, Is the precautionary principle used to cover up ignorance?: Basic & Clinical Pharmacology & Toxicology, v. 95, p. 49-52.

Tuomisto,J, J Pekkanen, H Kiviranta, E Tukiainen, T Vartiainen, M Viluksela, J Tuomisto, 2004a, Dioxin cancer risk - example of hormesis?: Non linearity in Biology, Toxicology and Medicine.

Tuomisto,JT, 1999, TCDD: a challenge to mechanistic toxicology \[Dissertation\], Kuopio, National Public Health Institute, p. 1-67.

Tuomisto,JT. Pyrkilo diagram method. 2004b. http://www.ktl.fi/risk/

Tuomisto,JT, J Pekkanen, H Kiviranta, E Tukiainen, T Vartiainen, J Tuomisto, 2004b, Soft tissue sarcoma and dioxins - A case control study: International Journal of Cancer, v. 108, p. 893-900.

Tuomisto,JT, M Tainio, J Pekkanen, J Tuomisto. Comparative risk analysis of dioxins and fine particulate matter. Toxicol.Appl.Pharmacol 197\[3\], 169. 2004c.

Tuomisto,JT, J Tuomisto, M Tainio, M Niittynen, P Verkasalo, T Vartiainen, H Kiviranta, J Pekkanen, 2004d, Risk-benefit analysis of eating farmed salmon: Science, v. 305, p. 476.