Comparative risk assessment is a general framework for evaluating environmental problems that impact human health. Risk assessment does not need to be cumbersome or costly to provide useful insights. Rapid, inexpensive approaches can be considered risk assessment, as long as certain basic concepts are included. There are four generally recognized steps in the human health risk assessment process, as described in the US National Research Council (1983):
Hazard identification is the process of describing the inherent toxicity of a chemical, based on toxicological data from laboratory or epidemiology studies.
Exposure assessment combines data on the distribution and concentrations of pollution in the environment with information on behavior and physiology to estimate the amount, or dose, of a pollutant to which humans are exposed. Exposure is typically estimated by modeling the dispersion of emissions from a polluting source
Dose-response assessment (see Annex) relates the probability of a health effect to the dose of pollutant. It uses statistical or biologically-based models to describe this relationship, using either experimental animal data or epidemiology. Estimated dose-response relationships are readily available for a large number of industrial chemicals and other types of pollutants, and need not be derived separately for each individual country. However, relationships based on site-specific epidemiology data are preferred if available.
Risk characterization, the final step in risk assessment, combines the exposure and dose-response assessments to calculate the health risk estimates, such as the number of people predicted to experience a particular disease, for the population of concern. Risk characterization also describes uncertainties in the calculations, and provides other information to help interpret the results of the analysis.
Comparative Risk Assessment is a simplified, focused methodology for deriving reasonable findings from readily available data. It is used to provide understanding and guidance in the absence of detailed scientific studies and analysis.
Issues in the Use of Risk Assessment
Defining the Scope of the Analysis
An effective risk assessment must have a well-defined scope. The appropriate scope depends on the purpose of the analysis. For example, an evaluation of emissions from a particular industrial facility is likely to concentrate on the health effects on local populations. In contrast, a project to set national environmental priorities may include a broader range of issues, such as the effects of national policies on emissions of greenhouse gases and ozone-depleting substances.
The purpose of most Comparative Risk Assessments is to identify the most important health risks from the point of view of the people affected. Although the options for mitigating risks may be evaluated on a sectoral basis, the initial analysis should consider all types and sources of environmental risk in the ranking.
The analyst must also choose the types of risks and populations to assess. These may include:
type and duration of health endpoint (acute, chronic; cancer, non-cancer; occupational)
special target populations (e.g., children, pregnant women, asthmatics)
ecological effects (populations, unique habitat, biodiversity)
An assessment of a particular industrial project or sector typically begins by describing the source of pollution. From the source of pollution, models of the transport and potential transformation in the environment are used to estimate the concentration of contaminants in air, water or soil. Concentrations in these media are used to estimate human dose, which, combined with dose-response information, predicts occurrence of disease.
For some pollutants, monitored data on concentrations in air or water may be available, obviating the need for fate and transport modeling. In other cases, data on measures of pollutants in the human body (such as blood lead levels) or measures of characteristic clinical responses to exposures (such as elevations in blood enzyme levels) may be available. These may be used as a direct measure of exposure in the dose-response functions, rather than using estimated exposure rates.
Complexity of Analysis
Risk assessment does not necessarily require sophisticated techniques or extensive data collection. Reasonable, practical results can be derived with minimal available information on pollution and the populations exposed to it (see examples below).
For example, in a USAID study in Cairo, an American team worked in Cairo for six weeks with Egyptian counterparts to refine the approach, to identify sources and to collect data. The study relied on existing data without any additional environmental sampling or monitoring.
Such rapid evaluations usually mean greater uncertainty in the results, but are still useful to get a general idea of the magnitude of problems associated with pollution sources and to demonstrate to decision-makers that the problems posed by pollution are real and significant.
Results of rapid assessments are likely to be most valuable when they are used in a relative or comparative, rather than absolute, way. The appropriate complexity of analysis will be influenced by a number of factors. These include the likelihood that additional refinement would resolve the dominant uncertainties in the analysis, budgets, time constraints, availability of data and use of the results.
Quality of Data Required
The quality and quantity of data needed to produce a meaningful analysis will depend on how much uncertainty the analyst is willing to accept.
Ideally one would have high quality local data for all parts of the analysis, including locally-based epidemiology for the dose-response functions. This will rarely, if ever, be the case. However, limited good data can be supplemented by techniques to fill data gaps with reasonable assumptions and extrapolations. For example, data on ambient concentrations of many chemicals are often unavailable, since monitoring is expensive and likely to be directed at only a few constituents. In its place, emissions data can be used in conjunction with environmental modeling systems to estimate concentrations in the environment.
Two such systems developed within the World Bank are the Decision Support System (World Bank, 1996a) and the Industrial Pollution Projection System. The US EPA has also developed and published emissions factors for air pollution sources - these include the AP-42 (US EPA, 1985) for 'criteria' pollutants, and the Toxic Air Pollutant Emission Factors (TAPEF) (US EPA, 1990) In addition, Part III of this Handbook provides ranges for emissions from a number of key industries.
Adjustments for Site-Specific Conditions
Many of the data sources and analytical techniques used in a risk assessment will, by necessity, be transferred from OECD contexts. It may be possible to adjust such data based on a comparison of country-specific conditions to the conditions in countries where the data were derived. For example, epidemiology studies frequently use measures of ambient pollutant concentrations to represent personal exposure. To adjust the results of such studies, the analyst will consider how the relationship between ambient concentrations and personal exposures may differ in the country of interest.
Examples of Use
In Industrialized Countries
Risk assessment has been used during the past decade in a number of OECD countries. In the US, it has been used to set overall environmental priorities for the nation, to guide legislation, and to choose among regulatory approaches. Almost every environmental program within the US Environmental Protection Agency now uses risk assessment to decide regulatory priorities, to perform cost-benefit analysis, or to target enforcement activities. It has been used, for example, to decide which air pollutants to control, which pesticides to allow and which to ban, and to what levels contaminated hazardous waste sites should be cleaned up. In Western Europe, both the European Union and individual countries are now working to adjust risk assessment techniques for application within their contexts.
In Developing Countries/ Transitional Economies
Comparative risk assessment is being used in other international settings as a useful way for regions and countries to allocate limited resources efficiently. For example, the method has been applied on a city-wide basis in Bangkok, Thailand and Cairo, Egypt, where specific recommendations for targeted action (e.g., reducing lead in gasoline, dealing with traffic situations for particulate matter) were identified. The method was also applied in the Silesia region of the Czech Republic and Poland, where it was coupled with an effort to identify realistic cost-effective solutions (US EPA, 1992b, 1994).
Many of the comparative risk assessments performed in developing countries have examined urban areas without significant industrial sources of pollution. These studies have identified a consistent set of priority problems - namely particulate matter air pollution and microbiological disease from water and food contamination. These problems are likely to be of high concern in any rapidly developing area that is both lacking adequate municipal infrastructure and experiencing a rise in industrial activity and traffic volume. Comparative risk assessments performed in such settings may wish to focus resources on examining these problems first, although the specific conditions of each urban area may suggest additional priorities.
Key Issues in Risk Characterization and Priority-Setting
Risk assessment alone cannot establish environmental management priorities. Risk assessment attempts to evaluate environmental problems using objective, scientifically-based measures; risk management considers not only of the magnitude and severity of the health risks posed by pollution, but also the costs and technical feasibility of abatement, the political will and institutional capacity to manage risks. Risk assessment should be viewed only as a first input to the subsequent steps of setting priorities, structuring policies and implementing strategies to address pollution.
The use of risk assessment in cost-benefit analysis and priority-setting has typically meant the use of overall population risk measures (such as the number of cases of disease predicted) as the preferred risk descriptor. However, other measures may be important, such as levels of individual risk, the distribution of risks across the general population and highly exposed sub-populations, identification of special at-risk populations, and consideration of the relative severity of effects characterized. Vital to the interpretation of risk assessments is the identification of major sources of uncertainty. Open, frank description of the uncertainties in the analysis enhances its credibility and provides a context in which the results should be viewed.
The scale and cost of some risk assessments that have been conducted demonstrate that practical application of standard techniques of risk assessment can enhance project design without being overly resource-intensive.
USAID (1993b) presents a typical schedule for conducting an environmental health analysis. The example schedule suggests a project lasting from 4 to 6 months from project planning through the final report. The schedule assumes a full scale analysis of many types of problems; the actual time required may be less for site-specific projects where a narrower set of likely pollution problems can be identified.
The types of consultants needed for a risk assessment will depend on the data available and the problems assessed. If industrial pollution sources are the focus, the project may need environmental scientists/engineers familiar with predicting the fate of emissions in the environment. The exposure assessment, dose-response and risk characterization steps typically require individuals with training in risk assessment, toxicology and/or epidemiology. The task manager may also want specialists familiar with that particular country's governmental and social structure in order to facilitate the collection of diverse data sources.
Possible Sources of data
Environmental Quality Data
The most important sources of environmental quality data are local and regional. When local data are not available, other sources may provide limited information. For example, some international organizations maintain environmental quality data for certain pollutants. These include the UNEP Global Environmental Monitoring Network System. Other organizations may have collected environmental quality data for specific purposes, such as US AID (environmental action plans) or World Bank country reports on environmental management. The World Resources Institute also compiles environmental data on a yearly basis from a variety of sources.
Human Health and Ecological Toxicity Data
International organizations are also good sources of information on hazard evaluations of chemicals, including environmental standards and, for some pollutants, dose-response evaluations. The World Health Organization develops guidelines for acceptable concentrations in environmental media based on protection of human health. Often the background documents supporting these guidelines can provide more information on chemical hazards. The International Agency for Research on Cancer supplies data on the carcinogenic effects of pollutants.
Since risk assessment is widely practiced in the US, the US Environmental Protection Agency is an important source of information on the toxicological information and evaluation methods. The U.S. EPA maintains a centralized, on-line data base containing toxicological information for over 600 chemicals, called the Integrated Risk Information System (IRIS), which can be easily accessed by risk assessment practitioners. Other U.S. EPA documents, such as the scientific documents that support standards for the criteria pollutants (PM10/SO2, lead, ozone, oxides of nitrogen and carbon monoxide), contain substantial reviews and evaluations of the literature on these major air pollutants. A recent World Bank report (Ostro, 1994) summarizes much of this same information, with additional discussion of its applicability to developing countries.
In particular, Ostro (1994) reviews health effects studies commonly used to assess risk from particulate matter and ozone exposure. Performed primarily in North America and Europe, many of these are time-series studies that focus on short-term (e.g., daily) changes in morbidity and mortality in response to short-term changes in pollution concentrations. A peer review of Ostro (1994) pointed out the difficulties of extrapolating these results to developing countries, due to differences in the populations and exposures considered. Furthermore, the peer reviewers expressed concern that the time-series studies capture primarily the acute effects of air pollution on mortality. Short-term fluctuations in mortality due to air pollution episodes may reflect largely the hastening ( by days or weeks) of the deaths of diseased individuals in the population. If so, this component of overall mortality results in fewer life-years lost and may be of less significance to public health than the chronic effects of long-term exposure to air pollution in otherwise healthy individuals.
Two recent cohort studies, Dockery et al. (1993) and Pope et al. (1995), have reported a significant and dramatic association between mortality in the study cohorts and long-term exposure to airborne particulate matter. Because such studies better reflect the morbidity and mortality effects of interest, using the results of chronic effects studies in comparative risk assessment is preferred, when they are available.
Factors for Human Exposure Assessment
Exposure assessment requires the integration of environmental quality data with an estimate of the rate of human contact with contaminated media. This aspect of risk assessment should rely heavily on local data, since it allows an assessment of how particular local conditions and cultural practices affect risk potential. Local data on food consumption patterns, indoor/outdoor activity patterns, types of housing, prevalence of health conditions, etc. can all be important to the assessment process. These data can be obtained from local health department and social service ministries, environmental ministries, NGOs, or from sociological investigations conducted as part of the analysis.
Dose-Response Functions and the Health Impacts of Air Pollution
Few would question that too much air pollution is a bad thing. Not only does air pollution reduce visibility and destroy the aesthetic beauty of our surroundings, it has been generally recognized as a health hazard. The question is not whether air pollution should be controlled but how much should be spent to control air pollution. To answer this question it is necessary to estimate the reduction in health damages that are likely to occur if air pollution is reduced.
Dose-response functions measure the relationship between exposure to pollution and specific health outcomes. By regressing a specific measure of health on a measure of pollution exposure while controlling for other factors, the role of pollution in causing the health effect can be estimated. This estimate can then be used to predict the health improvement corresponding to a decrease in exposure. In short, dose-response functions translate changes in air quality into changes in health.
Both humans and animals have been the subject of studies that examine the effects of air pollution exposure on health. This note will discuss only epidemiological studies, or those based on human populations.
Exposure to Air Pollution
Exposure to air pollution is usually measured in terms of ambient levels of pollutants. Not surprisingly, the pollutants included in the epidemiological literature are limited by the availability of data. Those commonly monitored by environmental authorities can be divided into four categories:
Sulfur oxides, nitrogen oxides, and particulates generated by burning fossil fuels;
Photochemical oxidants (e.g. ozone) created by the interaction of motor vehicle emissions (e.g. hydrocarbons, nitrogen oxides) in the atmosphere;
Other pollutants generated by mobile sources (e.g. carbon monoxide or lead);
Miscellaneous pollutants (e.g. cadmium, lead) generated by localized point sources such as smelters and manufacturing plants.
The Health Outcome
The health outcome is usually very precisely defined. Often, it is expressed as a measure of breathing capacity, such as forced expiratory volume (FEV), forced vital capacity (FVC) or forced expiratory flow (FEF). However, respiratory symptoms (such as cough, phlegm and throat irritation), the incidence of respiratory disease (including bronchitis and pneumonia) and mortality rates are studied as well. Table 1 shows some health effects associated with selected common pollutants.
To date, most studies have examined the effects of acute (short term) exposure to pollution. This should not be interpreted to mean that long term exposure has no effect on health. Long term exposure to low levels of pollution has been shown to affect an individual’s tolerance of short term exposure to high levels of pollutants. Furthermore, questions concerning the relationship between long term exposure and the incidence of cancer and heart disease have been raised. Unfortunately, long-term exposure is often difficult to measure due to the high immigration rates experienced in some urban populations.
A good study will attempt to control for confounding factors, or factors that may contribute to an individual’s likelihood of experiencing the health outcome in question. However, these factors are often not easy to control for and can thus weaken the results of the research.
For instance, although individuals may be affected by a combination of pollutants, the presence of other pollutants may not be incorporated into the study due to the limited availability of pollution data. Other confounding factors include temperature, humidity, physical activity, smoking habits, occupational exposure to pollutants, dietary factors, availability and quality of medical care, and age. The age structure of the population is especially important since children and the elderly are more susceptible to respiratory infection.
Applying the Dose-Response Function
Calculating the total health impact of a proposed pollution control program is relatively easy once the dose-response functions have been estimated. The dose-response equation below, is taken from a 1984 study by Evans et al. which summarizes the results of numerous cross-sectional analyses. The equation relates excess mortality to Total Suspended Particulates (TSP)
Excess Mortality = 0.45 X r TSP X POP
where excess mortality is expressed as an age-adjusted mortality rate per 100,000 persons. Only two other pieces of information are required: the size of the exposed population (POP) and the magnitude of the proposed change in pollution (r TSP) measured in µg/m3.
Ideally, we would like to measure the total life-years saved as a result of an environmental improvement. This can be done only when the dose-response function is estimated separately for different age groups which, unfortunately, seldom occurs.
Recently, dose-response functions estimated for one country have been applied to populations lacking their own epidemiological studies in order to estimate the effects of exposure to air pollution (see Table 2). Although this practice referred to as 'benefits transfer') does provide a rough estimate of the adverse health effects caused by pollution in these previously unstudied countries, it should be applied with caution. Without further testing, there is no reason to believe that the dose-response relationship calculated for one area will be exactly the same for another. Differences in the composition of air pollution, age distribution of the population, access to and quality of medical care, baseline health, education as well as other behavioral and socio-economic variables may cause variations in the response to air pollution exposure.