Defining
Intake Fraction This concept simplifies discussions of emissions-to-intake relationships, enabling easy intercomparison of the results of many risk investigations.
ach day, thousands of chemicals, many of them toxic, are released into the environment. Addressing the challenges that these substances pose requires consistent, reliable, and easily communicated information about their potential adverse effects. This need is driving measurement and modeling efforts aimed at linking emissions from a range of human products and activities to resulting human exposures and subsequent health effects. For many pollutants, a preliminary estimate of the risk posed by environmental releases can be determined from knowledge of the quantity released, the incremental intake per unit release, and the likelihood and severity of adverse effects per unit intake. This article considers the emissions-to-intake relationship.
E
Concept evolution
D E B O R A H H . B E N N E T T, THOMAS E. MCKONE, JO HN S. EVANS, W I L L I A M W . N A Z A R O F F,
CYRIL CABRY
MANUE L E D. MARG NI , O L I V I E R J O L L I E T, A N D KIRK R. SMITH © 2002 American Chemical Society
Several research groups have described approaches for relating source emissions of pollutants to human intake (1); multiple terms, definitions, and units exist for what appears to be a single, yet multifaceted concept (2−11). The first known articulation of the source-to-intake relationship, named committed dose, was introduced to describe the fraction of released radioactive elements entering a defined population through multiple exposure pathways (12). Subsequently, radiation dosimetry researchers proposed that this type of scheme could be extended to other materials. For crustal elements, Cohen calculated the probability that an element released to soil, surface water, oceans, or air reaches human populations (13). Bennett extended the concept along a series of exposure compartments by systematically assessing multipathway sourceto-intake relationships using the term exposure commitment (2). MAY 1, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
I
207 A
In the past two decades, the literature has also provided many examples of calculations relating the amount of a substance inhaled to the amount emitted to air, often using varying terminologysee historical terminology in Table 1, below. Establishing this relationship was an important step beyond the concentration−source ratio introduced in the early 1970s for interpreting the significance of atmospheric dispersion (14). Recently, there have been efforts to estimate multimedia intake by a population or an individual relative to a specified release, also resulting in varying terminology and units in the cited studies. This inconsistency in terminology and definitions quantifying emissions-to-intake relationships creates difficulties when comparing the results of different research groups and a lack of transparency in communicated information. Thus, more consistent terms, definitions, and units could improve the coordination among research groups and communicated results. Such intergroup consistency might ultimately emerge regardless of whether an effort to unify the language is undertaken, but this could take a long time to happen. To encourage development of more consistent terminology, we have formed, along with others, the Intake Fraction Working Group (IFWG), which seeks to derive a set of terms and associated definitions that are descriptive, simple, accurate, consistent, flexible, and that reflect consensus. On the basis of IFWG’s deliberations, and as discussed later, intake fraction (iF), is proposed as the primary descriptor for quantifying emissions-to-intake relationships. This concept is gaining momentum in various applications, and now is the time to gain consensus on terminology.
Terms and definitions In Equation 1, iF is the integrated incremental intake of a pollutant released from a source or source category (such as mobile sources, power plants, or refineries) and summed over all exposed individuals during a given exposure time, per unit of emitted pollutant.
iF =
Σ mass intake of pollutant by an individual (mass)
people, time
mass released into the environment (mass)
Although the pollutant intake is summed over population and time, in actuality, when a pollutant is released, there is a distribution of individual exposures within the exposed population. Individual exposure can be quantified in terms of the individual intake fraction, iFi. The total intake fraction comprises iFis summed over all members of a potentially exposed population. The iF can be calculated over different time horizons, depending on usefor example, a particular policy issue. Depending on the horizon chosen, the exposed population may include current and future generations. An infinite time horizon specified as a default value causes no problem for those compounds that decay rapidly in the environment, because in this case, a short-term iF essentially equals the infinite iF. To calculate an infinite iF for persistent pollutants, such as metals, requires predicting the fate of the pollutant in the environment, as well as potential human exposures over a time horizon of hundreds or even thousands of years, adding uncer-
TA B L E 1
Historical terminology
208 A
I
Term
Definition
Exposure efficiency (3)
The fraction of total emissions likely to reach people or the ratio of human intake to the amount emitted
Exposure factor (4)
The ratio of total population exposure (µg/m3 person-year) to total emissions (tons)
Exposure effectiveness (5)
The fraction of released material that actually enters someone’s breathing zone as measured in exposure units (µg/m3 person-year)
Exposure efficiency (1, 6)
The fraction of material released from a source that is eventually inhaled or ingested
Inhalation transfer factor (7)
The pollutant mass inhaled by an exposed individual per unit pollutant mass emitted from an air pollution source; the population inhalation transfer factor is defined as the sum of the inhalation transfer factors over all members of the exposed population
Exposure constant (8)
The absorbed individual intake resulting from a unit release as calculated by the Uniform System for the Evaluation of Substances—the multimedia, multiple-pathway exposure model (15)
Potential intake (9)
A multimedia source-to-intake factor based on the CalTOX multimedia model (16).
Population-based potential dose (10)
The exposure to a population for a unit emission source to a population based on the CalTOX multimedia model (16)
Fate factor (11)
A parameter for converting emission flow into its related concentration increase for transfers of an air-released substance to farm crops, then to agricultural products, and finally into humans
ENVIRONMENTAL SCIENCE & TECHNOLOGY / MAY 1, 2002
(1)
tainty to the calculation. In such cases, a shorter time horizon could be chosen, but differences in calculated outcomes would be expected. This has been an issue in other fields—global warming potential is usually determined for several time horizons, leaving the choice of which to use to the analyst (17). Another consequence of Equation 1 is that it assumes a linear relationship between emissions and population intake. If there are nonlinearities— for example, because of environmental chemistry—an incremental iF can be defined as the first derivative of the relationship between intake and emissions evaluated at current conditions. What follows is a synopsis of the options and the criteria we considered in determining terms, which build on the work of Zartarian et al. (18). The term should be consistent with common usage. Most previous expressions of the emissionsto-intake concept used either “exposure” or “dose” to characterize human contact and/or uptake of a pollutant. Although definitions vary, most investigators accept that “exposure” represents “the contact between an agent and a target” (18). In contrast, “dose” is “the amount of pollutant that is absorbed by a target” (18). Rather than recommending either term, we propose “intake” because it refers to “the amount of pollutant that enters a target after crossing a contact boundary” (18), where the contact boundaries for inhalation, ingestion, and dermal exposure are defined as passage through the nose (or mouth for oral breathing), mouth, and skin, respectively. This approach necessitates specifying intake rates, such as breathing and ingestion rates, which vary temporally and among individuals. Generally, population average intake values or nominal values consistent with regulatory risk assessment practices can be used. Structuring iF as a dimensionless parameter provides numerical constancy, independent of the system of units used for intake or emission. It also permits clearer communication of information. For transient release scenarios, iF is made dimensionless by dividing the time-integrated intake by the total quantity released. For steady-state release and exposure conditions, iF is the rate of intake divided by the rate of release, both using units expressed as mass per time. In both cases, obtaining iF requires measuring or calculating the fraction of a released substance entering the receptor population; hence, the term “fraction” is used in the proposed terminology. For an idea to be accepted and widely used, it
should be unique and expressed concisely. Unlike some proposed terms (see Table 1), which have alternative definitions and thus might cause confusion, we are unaware of any other definitions for iF. Furthermore, iF should be broadly applicable, independent of species, environmental media, and exposure pathways. By basing the measure on intake, the analyst can incorporate information on any exposure routes. The concept, moreover, is not limited to a single evaluation method—iF can be estimated using simple methods or sophisticated modeling tools simulating environmental fate and transport and human activities. It can also be determined experimentally or through measurement. For example, a pollutant tracer released at a constant rate can be measured as the steadystate tracer concentration in a medium such as air, water, or food. The iF for the tracer is then the concentration times the rate of intake of the medium divided by the rate of release.
Attributes of iF Note that iF depends on several factors, including chemical properties of the contaminant, emission locations, environmental conditions, exposure pathways, receptor locations and activities, and population characteristics. Hence, iF is an extrinsic property of a pollutant—the emission scenario and the exposure conditions are essential attributes that must be communicated in an iF calculation. For example, if a volatile pollutant were released into a crowded auditorium, a greater fraction would be inhaled than if the same pollutant were released outdoors in a sparsely populated region or into an empty auditorium. This idea is further demonstrated in the sidebar on the next page with example iF calculations for two benzene exposure scenarios. Chemicals’ intrinsic properties can also produce differences in iF for the same release scenario. For example, if two reactive, volatile chemicals are released into the same urban airshed, the one with a shorter environmental lifetime may have a lower iF. The iF differences are more extreme where there is significant pollutant partitioning in the food chain. For example, based on measurements from a recent EPA study, the iF of 2,3,7,8-tetrachloro-p-dioxin equivalents was ~0.002 for the United States in 1995 (19, 20). This is up to 5 orders of magnitude greater than calculated iF values for a volatile compound with a short atmospheric half-life, for example, 1,3-butadiene, demonstrating a clear distinction in iF between various compounds (19). When cumulative pollutant intake effects are represented by a linear dose–response function, this inMAY 1, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
I
209 A
formation can be easily combined with iF to yield an overall measure of risk. If the health effects depend on exposure rather than cumulative dose, iF must be modified to quantify the toxic effect. For example, if the toxic effect is based on the average exposure concentration, this concentration must be calculated from iF by backing out the breathing rate and number of exposed individuals. In cases where the toxic response depends on the intake rate rather than the average or cumulative intake or if there is a dose–response function threshold, iF must be disaggregated spatially and temporally to make risk estimates. For some iF calculations, the place and time where pollutants are released can be as important in determining health impact as their relative toxicities. In some cases, only the aggregate iF of an entire exposed population is of interest, and in other situations, it becomes important to disaggregate iF into its component parts in one or more dimensions, such as across individuals, time periods, exposure routes, and exposure pathways as discussed below. The box on the next page suggests a notation for specifying such disaggregation. Although iF is the sum of the iFis over all exposed individuals, quantifying each individual exposure may be impractical. A clear statement of the population under study should be included in any calculation along with a comment on the anticipated relationship between the studied population and the actual population. In a typical assessment, an iF can be constructed by adding up average individual intakes within population subgroups, then summing across all potentially exposed subgroups. For some applications, it may be sufficient to use an average intake for the entire population. In other cases, it may be necessary to quantify the intake for numerous popula-
Two sources of benzene exposure Motor vehicles are a major source of benzene emissions to urban air. In contrast, environmental tobacco smoke (ETS) contributes negligibly to urban-air benzene concentrations, but it contributes significantly to population dose because the iF for indoor air pollutant emissions is much larger than for outdoor emissions. Consider California’s 16,000 km2 South Coast air basin (SoCAB), which is home to 14 million people, who daily drive private vehicles ~0.5 billion km (21), and ~1.9 million smokers, who consume 42 million cigarettes daily (22, 23). Sales tax records indicate that 59 million L of gasoline are consumed daily by on-road vehicles (24). Remote sensing and tunnel studies of vehicle tailpipe emissions indicate that ~280 mg of benzene is emitted per liter of gasoline consumed (24, 25). Thus, total daily SoCAB benzene emissions are ~17 metric tons from this source. Depending on meteorology, size, and population density, iFs for distributed urban air pollution sources lie in the range (1–500) × 10−6 (7). A recent analysis suggests that the average inhalation iF for nonreactive, primary pollutants emitted from motor vehicles in the SoCAB is ~ 60 × 10−6 (26). Cigarette smoking in public buildings is not permitted in
210 A
I
ENVIRONMENTAL SCIENCE & TECHNOLOGY / MAY 1, 2002
tion subsets to capture exposure variability, such as where steep gradients in the concentration and population density occur around a release location. Disaggregation among routes of inhalation, ingestion, or dermal intake may yield an understanding of the underlying fate and transport of the pollutant or be important for identifying control options. Additionally, if the pollutant reaches different target organs, is absorbed, or metabolized differently depending on whether it is inhaled, ingested, or taken through the skin, it is important to quantify the potential impact via each intake route. In this case, the toxicity would vary among routes of exposure, and the health risk would be proportional to a weighted sum of the disaggregated iFs. Determining intake through the ingestion route requires considering both regional food production rates and sources of regional food consumption, because pollutants may partition into the local foods but be consumed by an individual living outside the affected region due to natural pollutant transport. In some cases, quantifying source-to-intake relationships from different exposure pathways for various exposed populations may be of interest. An exposure pathway defines the path from the source to exposure. For example, Evans et al. recently considered perchloroethylene exposure from dry cleaning operations (6). They described subpopulations of workers, dry cleaning consumers, and residents. Although all were exposed through the inhalation route, the pathway for each sub-population was different—workers were exposed at the facility, whereas consumers were exposed at home. Component iFs for these subpopulations can be summed to determine a total iF. Altogether, the nondimensional iF is a simple, transparent, and potentially comprehensive measure
California. Assume that 50% of cigarettes consumed in the SoCAB area are smoked in private residences. Average benzene emission factors for ETS are reported to be 280–610 µg per cigarette (27). Using the midpoint of this range, total estimated residential emissions of benzene from ETS in SoCAB are ~9 kg/day, or ~0.05% of the emissions from motor vehicles. The iF for indoor emissions of a nonreactive pollutant is estimated as the ratio of the occupant’s breathing rate to the building’s ventilation rate. For average conditions in California residences, this iF is ~7 × 10−3, more than 100 times as high as for outdoor emissions. Combining results, the total estimated inhalation intake of benzene for SoCAB residents is 1.0 kg/day, owing to motor vehicle tailpipe emissions and 60 g/day from ETS. The former intake is calculated as the product of the iF for distributed sources of nonreactive, primary pollutants in SoCAB times the total daily SoCAB benzene emissions. The latter intake is calculated as the product of the iF for indoor emissions of a nonreactive pollutant times the total estimated residential emissions of benzene from ETS in SoCAB. Thus, although the contribution of ETS to total benzene emissions into the air basin is negligible, the contribution to inhalation intake is not.
Notation for disaggregated iFs We recommend that a qualifier describing a modeled scenario be placed in parentheses after iF. Authors should define any such qualifiers in the text of their articles in cases where disaggregation is useful. For example, when defining different iFs based on the exposure route, release media, and particular subpopulation, specify the intake fraction as follows: iF (route, media, subpopulation). This describes iF for a population with a given route (inhalation, ingestion, dermal, total), media (release to air, water, soil), and subpopulation (workers, residents, all exposed). Some of the qualifiers in the parentheses are components of the total iF, such that they can be summed. For example, the iF by each exposure route can be summed to obtain a total iF, if multiple routes simultaneously expose the same population. Equation 2 shows the mathematical expression for exposure through all pathways: iF (total) = iF (inhalation) + iF (ingestion) + iF (dermal) (2)
of the relationship between emissions and human exposure. Although there is inherent complexity in calculating all exposed individuals’ intake, expressing the result as an iF compresses this complexity into an easily understood measure, which facilitates comparisons among investigators. Over time, a compendium of relevant methods, case studies, and results will be assembled, including various release scenarios and pollutants. Such information will provide an important resource for researchers and policy makers.
Acknowledgment D. H. Bennett, T. E. McKone, and M. D. Margni were supported in part by EPA’s National Exposure Research Laboratory through Interagency Agreement DW-988-38190-01-0 with the U.S. Department of Energy (DOE) under grant no. DE-AC03-76SF00098 at Lawrence Berkeley National Laboratory. Bennett also acknowledges general support under Kresge Center for Environmental Health, National Institute of Environmental Health Sciences (NIEHS) ES000002. The views expressed here are those of the authors and do not necessarily state or reflect those of EPA, DOE, or NIEHS. The IFWG consists of the authors of this paper, as well as J. I. Levy, D. Hattis, E. G. Hertwich, D. W. Pennington, and W. J. Riley.
References (1) Evans, J.; Wolff, S.; Phonboon, K.; Levy, J.; Smith, K. Chemosphere 2002, in press. (2) Bennett, B. Ecotoxicol. Environ. Saf. 1982, 6, 363–368. (3) Harrison, K.; Hattis, D.; Abbet, K. Implications of Chemical Use for Exposure Assessment: Development of an ExposureEstimation Methodology for Application in a Use-Clustered Priority Setting System; MIT Center for Technology Policy and Industrial Development; Cambridge, MA, 1986. (4) Smith, K. R. Environment 1988, 30, 10−15; 33−38. (5) Smith, K. Annu. Rev. Energy Environ. 1993, 18, 529–566. (6) Evans, J.; Thompson, K.; Hattis, D. J. Air Waste Manage. Assoc. 2000, 50, 1700–1703.
(7) Lai, A. C. K.; Thatcher, T. L.; Nazaroff, W. W. J. Air Waste Manage. Assoc. 2000, 50, 1688–1699. (8) Guinee, J.; Heijungs, R. Chemosphere 1993, 26, 1925–1944. (9) Hertwich, E. G.; Mateles, S.; Pease, W.; McKone, T. Environ. Toxicol. Chem. 2001, 20, 928–939. (10) Bennett, D. H.; McKone, T. E.; Kastenberg, W. E. In Human and Ecological Risk Assessment: Theory and Practice; Paustenbach, D. J., Ed.; John Wiley and Sons: New York, 2002; 619–644. (11) Jolliet, O.; Crettaz, P. Int. J. Life Cycle Assess. 1997, 2, 104–110. (12) Limits for the Intake of Radionuclides by Workers, Part 1; ICRP Publication 30; Annals of the International Commission on Radiological Protection, 1979, 2 (3/4). (13) Cohen, B. Health Phys. 1984, 47, 281–292. (14) Gifford, F.; Hanna, S. Atmos. Environ. 1973, 7, 131–136. (15) National Institute of Public Health and the Environment (RIVM). Uniform System for the Evaluation of Substances 3.0 (USES 3.0); RIVM 60145004; RIVM, Ministry of Housing, Spatial Planning and the Environment, Ministry of Health, Welfare, and Sport: Bilthoven, The Netherlands, 1999. (16) McKone, T. E. CalTOX, A Multimedia Total-Exposure Model for Hazardous-Wastes Sites Part I: Executive Summary; prepared for the State of California. Department of Toxic Substances Control; UCRL-CR-111456 Pt 1; Lawrence Livermore National Laboratory: Livermore, CA 1993. (17) Intergovernmental Panel on Climate Change. Climate Change 1994: Radiative Forcing of Climate Change; Cambridge University Press: United Kingdom, 1995. (18) Zartarian, V. G.; Ott, W. R.; Duan, N. A. J. Exposure Anal. Environ. Epidemiol. 1997, 7, 411–437. (19) Bennett, D.; Margni, M.; McKone, T.; Jolliet, O. Risk Anal. 2002, in press. (20) U.S. EPA. Exposure and Human Health Reassessment of 2,3,7,8-Tetrachlordibenzo-p-dioxin (TCDD) and Related Compounds; EPA/600/P-00/001B; U.S. Government Printing Office: Washington, DC, 2001. (21) California Air Resources Board (CARB). The 1999 California Air Quality and Emissions Almanac; CARB: Sacramento, CA, 1999, Chap. 4. (22) Reese, S.; et al. Morbidity and Mortality Weekly Report 2000, 49, 978–982. (23) U.S. Census Bureau. Census 2000 Data for the State of California; Department of Commerce, U.S. Government Printing Office: Washington, DC, 2001. (24) Singer, B.; Harley, R. Atmos. Environ. 2000, 34, 1783–1795. (25) Kirchstetter, T. W.; Singer, B. C.; Harley, R. A.; Kendall, G. R.; Hesson, J. M. Environ. Sci. Technol. 1999, 33, 329–336. (26) Marshall, J. D.; Riley, W. J.; McKone, T. E.; Nazaroff, W. W. Estimating exposure to motor vehicle emissions: A dosefraction approach. Presented at the 11th Annual Meeting of the International Society of Exposure Analysis, Charleston, SC, Nov. 4–8, 2001; paper to be submitted to Atmos. Environ. 2002. (27) Singer, B. C.; Hodgson, A. T.; Guevarra, K. S.; Hawley, E. L.; Nazaroff, W. W. Environ. Sci. Technol. 2002, 36, 846–853.
Deborah H. Bennett is an assistant professor and John S. Evans is a senior lecturer in the School of Public Health, Harvard University. Thomas E. McKone is an adjunct professor in the School of Public Health, University of California at Berkeley, and is a senior scientist in the Environmental Energy Technologies Division, Lawrence Berkeley National Laboratory. William W. Nazaroff is a professor in the Department of Civil and Environmental Engineering, University of California at Berkeley, and is a faculty scientist in the Environmental Energy Technologies Division, Lawrence Berkeley National Laboratory. Manuele D. Margni is a doctoral candidate and Olivier Jolliet is a professor in the Institute of Environmental Science and Technology, Swiss Federal Institute of Technology, Lausanne. Kirk R. Smith is a professor in the School of Public Health, University of California at Berkeley. Direct correspondence to Deborah H. Bennett at E-mail:
[email protected]. MAY 1, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
I
211 A
