Click for next page ( 208


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 207
7 Current and Anticipalred Applications INTRODUCTION The previous chapters in this report dealt with the basic principles and methodological elements of exposure assessment. To illustrate the state of the science and its application to the mitigation of deleterious effects on health or nuisance effects, this chapter analyzes some current and emerging problems of exposure to environmental contaminants in the form of case studies: vola- tile organic compounds, environmental tobacco smoke, polycyclic aromatic hydrocarbons, lead, acidic particulate matter, substances in buildings that cause occupancy complaints (sick-building syndrome), chemicals released from manufacturing facilities, and radon. These do not represent all the important issues but illustrate the state of the science in particular areas, such as biologi- cal markers, multiroute exposure, and personal monitoring. Each section addresses the completeness and results of the approaches in question, the sophistication of the methods used, the requirement for improvement or redirection, the misapplication (if any) of results, and the use of scientific results in making regulatory decisions. Discussions of several of the case studies in the context of exposure through environmental media other than air, such as water, food, or soil, relate to the general framework for exposure assessment discussed in Chapter 1. Accordingly, approaches to assess exposure through inhalation should be considered within the framework of total exposure, which accounts for all exposures a person has to a specific compound regardless of environmental medium. Therefore, strategies to reduce air exposures to a given contaminant should consider exposures due to other media. If other media are found to contribute significantly to the total exposure even after air exposures are re- duced, agencies responsible for or groups experienced with the other medium should be apprised of the issue and play an active role in the development of integrated exposure reduction strategies. 207

OCR for page 207
208 ASSESSING HUMAN EXPOSURE Unless the hazard of a contaminant is unique or the source of the contami- nant exposure is well characterized, it is difficult to conduct an assessment on one contaminant out of a group present in specific microenvironments. When the contaminant does not have a unique health effect, it is necessary to identi- fy those situations where populations have important exposures. Once the exposure is assessed, that information should be used to perform studies to establish the magnitude of the health outcome from exposure in those situa- tions. These case studies focus on the development of a new paradigm for e~o- sure assessment in risk assessment, risk management, epidemiology, and the application of clinical intervention. The conclusions focus on broad implica- tions for the discipline of exposure assessment, notable advances, and remain- ing needs. The committee hopes that these case studies will stimulate con- tinued or accelerated development of basic principles of exposure assessment and suggest ways to improve the investigations required for specific air con- taminants and general problems. 1 VOLATILE ORGANIC COMPOUNDS Introduction Some volatile organic compounds (VOCs) such as benzene, formaldehyde, and vinyl chloride-are classified as hazardous because of their role in human carc~nogenicity. This discussion deals mainly with VOC exposure of the U.S. population in general; occupational exposure is not specifically considered. The discussion examines EPA's current approach to assessing exposure as part of regulatory investigations of selected VOCs as air contaminants and the advances made by EPA's Total Exposure Assessment Methodology (TEAM) study in evaluating human exposure to VOCs. Benzene is used to examine an eyposure-assessment dichotomy found between the TEAM study and EPA's regulatory investigations. Current Approaches to Exposure Assessment Under the Clean Air Act EPA is required, under Section 112 of the Clean Air Act, to establish National Emission Standards for Hazardous Air Pollutants (NESHAP) that provide an ample margin of safety to protect the public from harmful expo- sure to VOC contaminants. NESHAPs are set by considering major source

OCR for page 207
CURRENT AND ANTICIPATED APPLICATIONS 209 categories of emissions, determining exposures, calculating health risks associ- ated with each contaminant, and focusing regulation on categories with the greatest risk potential. EPA's selection of source categories is often based on the assumption that sources emitting the greatest amounts cause the greatest exposures. Outdoor stationary sources (e.g., chemical plants and petroleum refineries) are usually identified as the greatest contributors to exposure. The EPA approach to exposure assessment relies heavily on modeling and uses little, if any, actual monitoring data. The human-exposure model combines source emission rates with atmospheric-dispersion equations to predict concentrations of VOC contaminants at various receptor sites In the general population and test the effectiveness of various emission-control strategies. Modeling extremely long-term exposures, as is required for a NESHAP risk assessment for exposure to carcinogens, presents several major difficulties. The typical practice is to measure or model the concentration of a contami- nant at one time and determine lifetime exposure by multiplying that concen- tration by a fixed number of years, e.g., the average human lifetime. Model input data are source locations and estimated emission characteristics, popula- tion census data, and meteorological data. It is assumed that population density remains unchanged for 70 years and that ambient concentrations are constant for 24 hours/day throughout the assumed lifetime. , ~ However, the nature of sources of exposure can change substantially over a lifetime. Large facilities commonly have a design life of 30 years, so consid- erable change can be anticipated in the sources over the 70-year human life- t~me. In addition, individual time-activity patterns can vary substantially over very long periods. In the United States, people change their place of resi- dence often, and few live in the same place over a lifetime. Recent studies of exposures to some VOCs cast considerable doubt on the NESHAP modeling approach and showed clearly that most people's exposures depend far more on their activities than on whether they live near an industri- al source of benzene emissions. The TEAM study has shown that in many circumstances focusing on industrial sources is ineffective in determining human exposure to select VOCs (Wallace, 1987~. Total Exposure-Assessment Methodology Study Overview An assessment of human exposure to airborne VOCs has been carried out through the TEAM study. The program originally intended to develop tech

OCR for page 207
210 ASSESSII`JG HUMAN EXPOSURE niques to measure total human exposure to a broad range of toxic chemicals, including selected volatile and semivolatile organic compounds and metals, but analysis of those chemicals in air, water, and food presented serious method- ological problems except for a group of VOCs (Wallace, 1987~. An implicit hypothesis was that the observed personal exposures to selected VOCs could be related to point sources (e.g. from industry) and that the farther one moved from these sources the smaller the observed exposures would be. Stated another way, this implicit hypothesis was that there is no difference between VOC exposure estimates made from stationary monitoring networks and from direct personal-exposure measurements as made in the TEAM program. For the small group of VOCs measured, the hypothesis has been rejected. The TEAM study measured exposure to selected VOCs directly with per- sonal monitors that were worn by subjects. The monitors were designed to be small, and to permit unobtrusive but accurate and precise sampling. Monitoring of VOCs is complex, because VOCs are typically found at trace levels. Contamination and artifact problems can affect the reliability of the data, and the applied analytical methods generally require laboratory-based instruments (Moschandreas and Gordon, in press). An extensive quality- control and quality-assurance program was carried out to ensure the proper interpretation of data. Sufficient sample size and probability sampling were used to support inferences regarding the target population and to permit the extrapolation of results to the general population. (Probability sampling is an experimental design that provides unbiased estimates of statistics, including precision, by weighting probability of selection, stratification, and clustering.) The TEAM study measured 2=hour personal exposures to 20-35 target VOCs in air and drinking water, including halogenated alkalies, alkenes, and aromatic compounds. Subjects were monitored in urban (heavy and light industry) and rural environments. In addition to personal samples, concurrent outdoor samples were collected from the backyards of a subset of the subjects. A comparison of matched indoor and outdoor samples showed that the con- centrations of most of the chemicals were higher indoors. That conclusion has been confirmed by other studies that analyzed for a comparable set of VOCs (Molhave and Molter, 1979; Jarke et al., 1981; Seifert and Abraham, 1982; De Bortoli et al., 1984; Gammage et al., 1984; Lebret et al., 1984; Monteith et al., 1984~. In particular it should be noted that Molhave and Moller (1979) found higher concentrations of benzene indoors than outdoors.

OCR for page 207
CURRENT AND ANTICIPATED APPLICATIONS 211 Measurement Methods The sampling system used a single-tube containing Tenax sorbent through which a known volume of air was drawn with a personal sampling pump. The adsorbent and pump were combined in a vest that was worn by the test sub- ject. Two consecutive 12-hour samples were collected (6 a.m. to 6 p.m. and 6 p.m. to 6 a.m.~. While the subject slept and bathed, the vest was placed carefully in a convenient location. Because most subjects remained at home overnight, the overnight samples were considered indoor samples. Outdoor samples were taken simultaneously near the house. The indoor-outdoor relationships were then established. The disadvantages of Tenax are that it will not retain very volatile compounds (vinyl chloride and methylene chloride) well and it cannot be used to trap reactive compounds (such as formalde- hyde). Samples were thermally desorbed from the Tenax onto a gas chro- matograph, where the analyses were separated, and then detected using mass spectrometry, which is highly specific and sensitive. Recently, the TEAM study employed canisters for the indoor measurements. Biological Markers At the outset of the TEAM study, blood samples were taken at the end of the sapling period and analyzed for the selected VOCs. However, the inva- sive nature of the sampling and poor detection limits associated with the analysis of blood led to the discontinuation of the technique. Fortunately, breath samples were also taken at the end of the sampling period. Breath sampling involved the use of a special spirometer in which the person exhaled approximately 20 L of air into a Tedlar bag, the contents of which were passed through the same type of Tenax traps as used in the air sampling. The same analytical techniques were used for the breath samples as for the air samples. The breath studies showed significant correlations with the personal- monitoring analyses for all 11 prevalent chemicals and showed no correlation with outdoor-air analyses (Wallace, 1987~. To understand the relation of the breath analyses to the air measurements, it is necessary to know the rates of absorption, distribution, metabolism, and elimination of the analyses in the body (physical pharmacokinetics). In funda- mental studies, subjects remained in an exposure chamber for a specified period breathing selected VOCs at specified concentrations. The subjects then left the chamber and their respired breath was analyzed repeatedly after specific periods to establish the half-life of the VOCs in the blood. Half-lives ranging from a few hours (benzene) to 21 hours (tetrachloroethylene) were

OCR for page 207
212 ASSESSING lIUMAN EXPOSURE observed (Gordon et al., 1985~. Similar results have been seen by Jo et al. (in pressb) for chloroform. The half-lives can be used to determine the most appropriate sampling time for the use of breath measurement as an indicator of exposure. Questionnaires Two questionnaires were used. The first was a household questionnaire, which included age, sex, occupation, household characteristics and activity characteristics of the participant and other members of the household. The "formation was used to obtain a probability sample of subjects and to ensure the inclusion of highly exposed subjects in the studies. The second question- naire involved a 2lhour recall and was administered immediately after the end of the 2=hour monitoring period. The participants were asked whether they had been exposed to potential sources of target chemicals. Monitoring data were then compared with data from the second questionnaire. Variables related to smoking, occupation, home characteristics, personal activities, and automobile travel were found to be the most important determinants of expo- sure. Benzene concentrations were 30-50% higher in homes of smokers than in homes of nonsmokers. Subjects were heavily exposed to benzene (over 1 mg/m3) when filling automobile gas tanks; benzene exposure could often be related to automobile use, which also includes time spent inside of an automo- bile compartment (Wallace, 1989~. Models No models were used specifically to assess exposure in the TEAM study. However, the use of pharmacokinetic models was considered essential for the proper use and interpretation of breath measurements as indicators of eypo- sure. A simple two-compartment model accounted for the effect of the initial breath concentration and the residence time of VOC measured in the TEAM study within the body (Wallace et al., 1983~. The model successfully predicted the time needed for clearance of tetrachloroethylene from the body when compared with the chamber studies mentioned earlier (Gordon, 1985~. Benzene Results of the TEAM study indicate that personal benzene-exposure con

OCR for page 207
213 o I: ,, o o - C ~ ~ Y o . E ~ ~ _ 3 o O i/- s~ 'A ~ . ~- I: C _ o E _c :, c ,, 3~ C) C) o .. ~ ~ 0 C)In as o o- X CryC.) ~ to ~ C o C , ~ ~ _ ~- I,, CL ~C).>

OCR for page 207
214 ASSESSING HUMAN EXPOSURE centrations exceed ambient outdoor concentrations (Wallace, 1989~. Figure 7.1 shows industrial sources represent about 14% of total emissions of ben- zene, but their contribution to exposure is relatively small~nly about 3% of the total. Thus programs and regulations to reduce emissions from major stationary point sources could affect, at most, 3% of total exposure nation- wide. Nevertheless, a recent rule-making has established national emission standards for benzene from industrial source categories: maleic anhydride plants, ethy~benzene-styrene plants, benzene storage, equipment leaks, and coke by-product recovery plants. Other larger indoor and personal sources of exposure are not covered by this rule-making (EPA, 1988d). Exposures from active smoking, involuntary smoking, products in the home, and personal activities such as driving or painting have been estimated to account for more than 80% of nationwide exposure to benzene. The sources of exposure la- beled Motor vehicles (outdoor air)" do not include personal use, such as driving or riding in an automobile; such uses are included in Motor vehicles (travel).~ (Note that the TEAM subjects were drawn from areas with little use of wood stoves or kerosene heaters, which are potentially important sources of expo- sure to benzene (Wallace, 1989~. These important sources of exposure must be re-evaluated and considered for regulation and education. In addition, similar types of integrated analyses are necessary for other VOC contami- nants, which may have both indoor and outdoor sources. Recommendations To incorporate all significant exposure findings into future rule-makings for other hazardous VOCs, exposure analysts and risk managers need to Interact. Regulatory investigations should not be limited to some readily identifiable and measurable point sources that might have insignificant impacts on expo- sure. The findings of TEAM are at odds with conventional approaches used to control VOC exposure. Therefore a major rethinking of the approaches used to identify public health risk is warranted. Exposure analysts must con- tinue to refine techniques that can identify important sources of contaminant exposures, whether those sources are indoors or outdoors. The VOCs examined in the TEAM study were almost exclusively in a single exposure medium (air), were chemically stable, and had a volatility that permitted their effective collection and concentration with the sorbent Tenax New analytical techniques should be developed to broaden the range of ana- lytes that can be collected and measured, so that the "T" in TEAM will actually stand for "Total," and not for "Targeted compounds," as is now the case. In particular, greater attention should be given to analyzing for highly reactive

OCR for page 207
CURRENT AND ANTICIPATED ~4PPLIC:ATIONS 215 compounds. Passive dosimeters (Lewis et al., 1985) that match the time resolution of active monitors should continue to be developed, because they are less expensive and usually more convenient to wear. Better microenvironment monitoring data and time-activ~ty data, including quality assurance aDd quality control, are needed to improve the modeling of VOC exposures. ENVIRONMENTAL TOBACCO SMOKE Introduction The health herds associated with smoking have received extensive study and are well letdown. Thus, it is not surprising that there is now a growing concern that exposure to environmental tobacco smoke (ETS) might affect the health and comfort of nonsmokers. The health and nuisance effects of so- called involuntary smoking have been extensively reviewed in a National Re- search Council report (NRC, 1986) and in a report of the Office of Smoking and Health (1986~. Both reports concluded that exposure of nonsmokers to ETS results in acute irritation of the eyes, nose, and throat; unacceptable odor; upper-airway problems in children, including increased prevalence of respiratory symptoms (cough, sputum production, and wheezing), decreased lung function, increased lower-respiratory-tract illnesses, and increased rate of chronic ear infections; and increased risk of lung cancer. The reports also noted that other outcomes related to the growth and health of children had positive associations In studies, including low birthweight and reduced growth and development. However, the results of some of these studies continue to be debated, and other related studies are ongoing. Thus, it is unportant to examine ways to improve techniques to assess more accurately exposure to ETS. Until recently, epidemiological studies of the acute and chronic health effects of ETS have been handicapped by limitations in assessing exposures to ETS. Exposures occur at a u ide range of concentrations for highly variable periods and in numerous indoor environments. Unlike active smoking, expo- sure to ETS cannot now be easily assessed with standardized methods. Prev~- ous epidemiological studies of the chronic effects of ETS, particularly lung cancer, have determined exposure solely by questionnaires, which have not been standardized or validated. The questionnaires have usually obtained information on smoking habits of occupants of residences to permit assess- ment of ETS exposures and have not adequately addressed the Impact of occupational exposures. The use of such questionnaires might pose problems

OCR for page 207
216 ASSESSING-HUMAN EXPOSURE in misclassification of subjects by exposure status and obscure possible e~o- sure-effect relationships. In the past few years, new techniques have been developed that permit a more accurate assessment of individual exposures to ETS (Leaderer, 1990~. They are being applied to test hypotheses ~ epidemiological studies on the relationships between ETS and acute and chronic health and nuisance effects. The methods use advances in the applications of markers or profanes of ETS, air monitoring, modeling, questionnaire survey, and biological markers. Air OCR for page 207
CURRENT AND ANTICIPATED APPLICATIONS 217 ETS-generated RSP in various indoor microenvironments (Repace and Low- rey, 1980, 1982~. It is also being used to estimate ETS exposures retrospec- tively and to assess risk (Repace and Lowrey, 1990~. As input, the model uses known rates of RSP emission from tobacco combustion and data from several sources, including measured and estimated smoking densities, infiltration and ventilation rates, and deposition rates. The tapered element oscillating micro- balance could be used to continuously monitor indoor concentrations of RSP (Patashnick and Rupprecht, 1986~. Biological Marlters Physiological fluids can be analyzed for specific biological marker com- pounds indicative of exposure to ETS. Thiocyanate, carboxyhemoglobin, nicotine and cotinine, hydroxyproline, N-nitrosoproline, aromatic amines, and protein or DNA abducts have all been considered as Indicators of dose of tobacco smoke (NRC, 1986; Office of Smoking and Health, 1986~. Those biological markers indicate that exposure has taken place, but might not be directly related to the source or to the specific adverse effect under study. Furthermore, a biological marker of exposure might not be specific for the contaminant related to the effect, does not provide an exact measurement of ETS exposure in a single environment, and does not provide information on the environmental factors that affect the concentration in the environments in which people spend time. Biological markers of ETS exposure can also vary widely from person to person, because of differences in uptake, distribution, and metabolism. Some markers are not specific for ETS exposure (e.g., carboxyhemogIobin); while others (e.g., thiocyanate) might be useful for active smoke exposure, but not sensitive enough for ETS exposure. Cotinine and nicotine measurements in the blood, urine, and saliva are specific for tobacco- smoke exposure, and have been widely used as indicators of ETS exposure (NRC, 1986~; they are valuable in determining the total or integrated short- term (hours to days) dose of ETS across all locations in which a person spends time. Questionnaires Questionnaires have been used extensively in epidemiological studies for the classification of people into broad categories of ETS exposure on the basis of reported exposure. Questionnaires are also used to obtain information on the physical environments in which exposures take place, the factors affecting

OCR for page 207
246 ASSESSING HUMAN EXPOSURE Models The single compartment and multicompartment ma.cc-balance model dis- cussed in Chapter 6 can be used to good purpose in designing and conducting a study to elucidate the relationship between health effects and exposures to airborne contaminants. Those models can provide a conceptual framework for designing the sampling strategy to be used in various buildings, can help to predict exposure concentrations from various sources or infernug source strengths from concentration measurements, and can be useful In designing controlled human exposure experiments in which concentrations of indoor contaminants are varied. Empirical models constitute a second class of model that can often be used as hypothesis-generating and testing tools. Such models are typically multivar- iate. For an SBS study, some health end point would be related to env~ron- mental variables in a stepwise multiple regression. Conclusions The SBS problem has surfaced only in the last decade. Thus, the methods for understanding it have not had time or resources to be adequately devel- oped by exposure analysts. However, the database obtained from ~nvestiga- tions of BRI has suggested better approaches for the design of SBS studies and a need to develop measures to reduce exposure and the Incidence of SBS. Issues of technique include the development of more refined hypotheses; the use of a broader range of physical, chemical, and biological measurements; more complete and standardized health and activity questionnaires; and the use of more sophisticated models of total exposure. TOXICS RELEASE INVENTORY Introduction Title III of the Superfund Amendments and Reauthorization Act of 1986 (SARA), Public Law 99-499, is a free-standing statute titled "The Emergency Planning and Community Right-To-Know Act of 1986". In the development of SARA Section 313, it was acknowledged during discussions in Congress that the extent of human exposure to toxic chemicals released by industry was a major concern Congressional Record, H11205, December 5, 1985~. It was also expressed in Congress that much work is needed before human exposure

OCR for page 207
CURRENT AND ANTICIPATED APPLICATIONS 247 to toxic chemicals can be effectively managed and that acquisition of informa- tion is the next necessary step. Section 313 of SARA requires industrial facili- ties that manufacture, process, or use toxic chemicals to report annual envi- ronmental release information to the EPA. Initial requirements for submis- sion of the information are specified by EPA in the Toxic Chemical Release Reporting Final Rule (Fed. Reg., 1988a). The database resulting from the information reported to EPA is referred to as the Toxics Release Inventory (TRI). The TRI was seen as a means to gather information for three general objectives: (a) to identify the chemicals of the greatest concern; (b) to identify locations where the chemicals are manufactured, used, and released; and (c) to determine the quantities released into the environment (Congressional Record, S11772, September 19, 1985~. The initial list of toxic chemicals for TRI reporting contains 308 specific chemical compounds and 20 chemical categories and can be modified only by a rule-making, such as the deletion of titanium dioxide Fed. Reg., 1988b). Information reported to the TRI includes routine releases (e.g., emissions from stacks) and accidental releases to air, land, and water. The first reports were filed on June 30, 1988. This case study examines the issues that should be addressed in the TRI to make it useful for assessing exposure to toxic chemicals. The purpose of the TRI is to inform the public and government officials about total releases of toxic chemicals. Section 313 of SARA requires EPA to develop the TRI information into a computerized database for public ac- cess. The information is intended, among other purposes, to assist research and aid in the development of various regulations, guidelines, and standards (EPA, 1988c). There are no requirements to perform risk assessments or to regulate any TRI-listed chemicals. To minimize the burden of data-gathering on industry, Section 313 of SARA allows release reports to be based on esti- mates; monitoring data and other available information are not required, but can be reported if available. Applications to Exposure Assessment Although the TRI provides useful information on estimated mass quantities of chemical releases, it does little to assist in understanding the potential for human exposure to those releases and resulting impacts on public health. The inclusion of accidental and routine emissions in the total releases reported to the TRI makes estimation of downwind concentrations and exposures techni- cally infeasible. The separate types of releases involve different exposure issues and require different analyses for determination of exposure and expo

OCR for page 207
248 ASSESSING HUMAN EXPOSURE sure impact. The TRI database is useful In identifying chemicals of concern, which may, with further analysis, provide data needed for exposure assess- ment. The TRI requires reporting of chemical quantities released directly into the environment or transferred to off-site locations, identity of releasing facility, geographical location (latitude and longitudes, identity of all sites to which the reporting facility transports chemical wastes, how the reported chemicals are used, and types and efficiency of on-site methods to treat chemical wastes. The data, by themselves, are inappropriate for assessing either acute or chron- ic exposures, because they are not linked specifically to the potential concen- trations and locations of exposure of the general population (Levin and Spence, 1989~. TRI data are only one type of input data for air-dispersion models (see Chapter 6) used to estimate potential downwind concentrations, which are then linked with human time-activ~ty data to assess potential e~o- sures (see Chapter 5~. Even the simplest dispersion models cannot be used to estimate downwind concentrations of released toxic chemicals on the basis only of TRI data. The TRI provides some data useful in determining downwind concentrations, such as facility location, latitude and longitude (to assist in describing meteorolog~- cal transport), and categorization of releases as either point sources (e.g., stack emissions) or fugitive releases. Additional data are needed for air-dis- persion analyses. Values of various model parameters on individual sources are needed: release temperature and discharge velocity, orifice diameter and height of release; frequency and duration of releases; and nearby structure characteristics likely to affect small-scale air movements. Because the TRI does not collect those additional data, industry is not likely to obtain and store them, so they cannot be obtained simply by calling the TRI coordinator at each facility and requesting them. Some facilities have taken the initiative of estimating potential exposure concentrations of released chemicals reported to the TRI. Such information more fully prepares facilities to respond to inquiries from the public about impacts of their toxic chemical releases on public health and the environment. Acute toxicity is the primary concern for assessment of exposure to acci- dental releases. To identify possible carcinogenic impacts, analysis of lifetime exposure to routine emissions is required. Even if all the necessary model data were provided for each release source, the results would be of little use for exposure assessment, because of the combination of data in the TRI on routine emissions and accidental releases. For example, at low concentrations, hydrogen cyanide (HCN) gas, an acute- ly toxic agent, can be detoxified by the body. However, at high concentrations, HCN causes breathing loss and death. Reporting all emissions of HCN on an

OCR for page 207
CURRENT AND ANTICIPATED APPLICATI0115 249 annual basis could give a false impression of potential exposures that had acute health outcomes. The estimation of exposure on an annual basis might be acceptable for long-term effects, but even a single breath of HCN at more than 2,300 ppm~v) would result in death. Only time will tell whether the TRI database will be applied incorrectly to Closure assessment activities. It is clear today, on the information submitted to the TRI database, that industry has committed resources to reduce emis- sions (Steyer, 1988) and EPA is expected to move more rapidly to develop regulations for several specific hazardous air pollutants. Implications The TRI reporting requirement will, in all probability, provide tangible environmental benefits. Data on releases to all media are important for understanding the impact of a chemical release on total human exposure and the global environment, but releases to air warrant special attention. Releases to air probably result in the most immediate, and perhaps the most important, exposures of the public living near an industrial facility that produces or uses toxic chemicals. Exposure to airborne toxic chemicals can occur directly through inhalation of contaminated air or through ingestion of food or water that contains contaminants deposited from the air. In the future, acutely and chronically toxic chemicals should be reported separately to allow proper focus of resources on the most important exposure issues. A source and receptor database needed for the proper exposure as- sessment for both acutely and chronically toxic chemicals should be carefully considered for inclusion in the data-collection effort In any revision of SARA Section 313. Industry burden of providing the additional information is an important aspect of the considerations. Because the technology used for exposure assessment is changing rapidly, it would be appropriate to define data needs in regulations, rather than in specific laws. Regulations could then be changed as necessary to respond to advances ~ exposure assessment without the need to amend laws. RADON Introduction Exposure of the general public to radon and its decay products appears to constitute an important naturally occurring environmental health risk. Radon

OCR for page 207
250 ASSESSING HUhlAN EXPOSURE decay products have clearly produced lung cancers in exposed underground miners (NRC, 1987a). However, there are considerable uncertainties in how the risks identified in the miner studies can be extrapolated to the general public. No clearly identified lung-cancer mortality in the general population ran yet be specifically linked to exposure to radon decay products (NCRP' 1984a,b). Four relatively small case-controlled studies have suggested a possible relationship between lung cancer and building construction or residential radon exposure (Axelson et al., 1979; Edling et al., 1984; Lees et al., 1987; Svensson et al., 1987), but there are no unequivocal measurements of the Inug-cancer risk associated with indoor radon. Because the estimated risks are higher than those associated with many other environmental agents suspected of having adverse health effects, there has been considerable inter- est In looking for clear evidence of radon-related lung cancer in the general population. The problem of protecting the public health has been exacerbated by the uncertainties in the exposures and the corresponding risk estimates. Risk- management decisions of EPA have suggested radon concentrations in indoor air that should trigger mitigation action, and those action concentrations if too low would result in unnecessary expenditures and concern. EPA reported in a September 1988 press conference that radon causes 2O,000 lung-cancer deaths a year in the United States. However, that estimate is at the high end of the range estimated by the National Council on Radiation Protection and Measurements (NCRP, 1984a,b), so the risk estimates are not in good agree- ment. Major factors affecting the uncertainty in risk estimates are related to the measurement of the proportion of exposure that is environmental. There are major difficulties in assessing exposure to natural airborne radio- activity, particularly to those radionuclides of greatest health-effect potential. It is universally agreed that the short-lived decay products of radon (hippo' 2~4Pb, alibi, and 2~4po) cause the presumed health effects, but radon Is gener- ally measured as a surrogate for these other radionuclides, because it is rea- sonably easy and inexpensive to measure the indoor radon concentration. One must be careful in extrapolating short-term screening measurements made under nontypical conditions (e.g., in a basement in a closed house during winter) to annual average exposures, although these nontypical-condition measurements may represent a maximum exposure condition. Methods for measuring long-term, average exposure to radon require further development, and better communication is necessary to explain the risk uncertainties to the public. The amount of airborne radon decay products in a room depends on sever- al factors, including the amount of radon to produce them, the concentration of airborne particles to which they can become attached, and the aerodynamic

OCR for page 207
CURRENT AND ANTICIPATED APPLICATIONS 251 processes that contribute to the deposition of radioactivity on surfaces in the room (walls, ceilings, furniture, etc). Thus, the actual concentration of a~r- boruc radioactivity is a complicated function of several environmental vari- ables. . ~ The health effects of radon decay products also depend heavily on their aerodynamic behavior in the indoor atmosphere. Particularly for Typo, parti- tioning between the unattached state and the attached forms (i.e., combined with pre-existing aerosol particles) has an important impact on the calculation of the dose to the lung from a given airborne decay-product concentration. In the dose models commonly used to relate tissue dose to airborne radioac- tiv;ity concentrations (Jacob; and Eisfeld, 1980; James et al., 1980), a substan- tially increasing effective dose to the target tissue is predicted with decreasing particle size down to about 3 nm. The increase in dose is due to the increase in effective deposition through molecular diffusion as particle size approaches that of free molecules. Small changes In particle size in this range result in large changes In the diffusion coefficient and in depositional behavior, particu- larly in regard to the location of deposition in the tracheobronchial tree. These models of delivered dose of alpha radiation to Jung tissue show radon to be a reasonable surrogate for exposure to the `decay products because several of the opposing factors in the exposure cancel each other. Hypothesis and Study Design The hypothesis of interest is that increased exposure to radon decay prod- ucts In the indoor environment increases the risk of induction of lung cancer. Exposure to tobacco smoke and differential residential mobility are substantial confounding factors in the estimation of health risk. Two epidemiological studies are attempting to relate lung cancer to envi- ronmental radon and decay-product exposure through retrospective measure- ment of indoor radon concentrations. One is being conducted by the state of New Jersey and the other by Argonne National Laboratory. Both are con- cerned with obtaining better risk estimates related to exposure of the general population to radon decay products and are using cases of lung cancer in white women as the subjects of case-controlled studies The New Jersey study (Schoenberg et al., 1987) is an earlier extension of a statewide population-based case-controlled interview study of New Jersey women. The cases include all of the female residents of New Jersey whose histologically confirmed primary cancers of the lung were newly diagnosed in the period from August 1982 to September 1983. For cancer patients who were interviewed, age- and race-matched controls were chosen from New

OCR for page 207
252 ASSESSING HUMAN EXPOSURE Jersey drivers-license files and from Health Care Financing Administration files for Medicare enrollees. For next-of-kin interviews, matched controls were selected from state death-certificate files. For the 1,306 cases identified, 994 patients or next-of-kin were interviewed; of the 1,449 controls chosen, 995 were interviewed. Some 53% of the interviews were with the patients, and the rest were with next-of-kin. The study began without consideration of indoor radon, and residential housing information had been collected only on the towns in which the sum jects lived. The subjects or next-of-kin were therefore recontacted to obtain street-address information. It was assumed that there is a minimal 10-year latency period between exposure and onset of cancer. Only one house was tested per subject because of resource limitations, so the study focused on subjects who lived for at least 10 years at an address in New Jersey during the period 1953-1972, about 10-30 years before the case diagnosis or control selection. It was found that 17% of the subjects had not lived in New Jersey for at least 10 years dunng 1953-1972 and that 10~o had not lived at any address for at least 10 years during the critical period. In another 2% of the cases, it was not possible to determine specific street addresses. It was possi- ble to obtain addresses for 1,216 subjects that met the criteria. Of those addresses, 82 no longer existed or were dwellings in upper floors of apart- ments, trailers, or other situations in which radon exposure would be expected to be negligible; that left 1,134 addresses. Short-term charcoal-canister meas- urements were made for a quick screening. For a better determination of the annual average concentrations, two alpha-track detectors were deployed in the 1,134 dwellings. In 10% of the dwellings, a third track-etch detector was collocated with one of the other detectors for quality assurance. The Argonne study provides a good example of a potentially useful study design. The study population comprises white females born in Pennsylvania who lived in eastern and central Pennsylvania, excluding Philadelphia and Pittsburgh, and died of lung cancer between 1970 and 1987. Controls will be chosen from white females born in Pennsylvania in the same years as cases, selected by random-digit dialing and random selection from vital statistics. A large number of lung cancer cases are available (more than 6,000 through 19843 in an area where there are likely to be high indoor radon concentra- tions. Separate case series will be defined by histopathological type of lung cancer and by smoking status. The first case series of about 500 cases in- cludes all lung cancers and categories of smokers. For each of the dwellings that the subjects have occupied and that can be identified, both short-term charcoal and long-term track-etch measurements will be made for all levels in the dwellings. When current occupants are willing, two sets of sequential 6-month track-etch samplers will be left and picked up by project personnel,

OCR for page 207
CURRENT AND ANTICIPATED APP~C4TIONrS 253 to ensure adequate response. Charcoal-canister measurements will be used to screen the dwellings to make preliminary assignments as to radon exposure. When current occupants are not willing to allow measurements, radon concen- tration will be based on the age and construction lope of the dwelling and its geological setting. Although data on a number of dwellings already permit building of a predictive model for indoor radon, the results from the coopera- tive dwellings with occupants win improve the database on which the models are built. The researchers also plan to measure radon-decay product concentrations to assess exposures to radon progeny more directly. There is no apparent plan to measure the radioactive particle size distributions. Thus, it will not be possible to assess the potential for deposition, and the analyses will have to include estimates of the effectiveness of the measured concentrations In pro- ducing specified doses. Measurement Methods Short-term charcoal and long-term track-etch detectors will be used. In both cases, it is assumed that current radon concentrations reflect past radon. If there have not been changes In the insulation, heating system, or general nature of a dwelling, the assumption should be reasonable. However, with the extensive energy-conservation efforts many homeowners made in the late 1970s and early 1980s, many homes might have been modified. Estimation of prior concentrations would then constitute a considerable problem. Another important problem is the concentration of decay products relative to the radon concentration. If, for example, one or more occupants smoked and then quit, indoor particle concentrations might be much lower now than in the past. Higher particle concentrations result In higher decay-product concentrations, but lower the concentrations of more diffusive unattached decay products and thus result in a lower average dose per unit of airborne radioactivity. Similarly, if a gas stove were traded for an electric unit, particle concentrations resulting from cooking would be lower; this could change the effective exposure to decay products. Air cleaners can substantially increase the unattached fraction, so EPA does not recommend the use of air cleaners to mitigate the effects of radon decay products. Models The choice of radon as the measured entity suggests the possibility of an

OCR for page 207
254 ASSESSING HURON E~OSU~ implicit use of dose models that make the following prediction: the inverse relationship between the concentration of airborne particles, the total decay products, and the unattached fraction cancels out the effects of particle con- centration on dose (James, 1988~. Therefore, exposure can be adequately measured by determining the integrated, average radon concentration. Such calculations have been presented by Vanrnarcke et al. (1985~. The capability to predict indoor radon concentrations Is central to the development of radon exposure models. Considerable effort is being devoted to the development of models of radon entry into houses as a function of soil characteristics (e.g., radium content and permeability), cInnatic conditions, and house characteristics (e.g., substructure type, type of heating system, and air- leakage area). Indoor radon concentrations are predicted by combining the models of radon entry into basements (Loureiro, 1987; Mowris and Fisk, 1988), generally with steady-state, two- or three-dimensional numerical codes that model the convective (pressure-driven) entry of soil gas (containing ra- don) through openings In the substructure. These models are being upgraded at Lawrence Berkeley Laboratory to account for diffusive entry of radon, spatial variability of soil properties, simultaneous transport of soil moisture, and transient effects. A smaller effort has been devoted to the entry of radon into houses with craw! spaces (Mowris and Fisk, 19~) and to the development of simplified closed-form or statistical models (Mowris and Fisk, 1988; Rev- zan, 1989~. None of these models has been adequately validated, although a current experimental effort by Lawrence Berkeley Laboratory should provide critical data on radon entry into basements during the next few years. Advances Both of the large studies discussed here are making direct measurements in at least one of the dwellings occupied by each of many lung-cancer subjects over a long period and thus should yield a reasonable estimate of radon con- centrations to which they have been exposed. In addition, information has been obtained on smoking behavior and mobility to try to account for these strongly confounding variables. The Argonne study will be partially supple- mented by direct measurement of concentrations of radon decay products although the particle size distributions and their potential influence on dose are not explicit parts of either study. New measurement methods have recently been developed that permit the determination of both concentrations and size distributions of radon decay products. The use of single screens in nonconventional diffusion batteries (graded screen arrays) and measurement of the radioactivity that passes

OCR for page 207
CURRENT AND ANTICIPATED APPLIC4TIOli/S 255 through each screen permit one to obtain size distribution over the range of 0.5-500 rim (Reineking and Porstend~orfer, 1986; Holub and Knutson, 1987; Ramamurthi and Hopke, 1988~. A new system can provide hourly measure- ments of concentrations and size distributions of each decay product, so it is now possible to measure directly the species that are responsible for hearth effects without resorting to assumptions and models (Ramamurthi, 1989~. This system can be used soon to test the variability of concentrations in ~ffer- ent size ranges directly so that a better understanding of the dynamics of inducer radon decay products will be possible.

OCR for page 207