Non-Occupational Exposures to Pesticides for Residents of Two U.S. Cities

Transcription

1 Arch. Environ. Contain. Toxicol. 26, (1994) AROHIVES OF: Environmental Contamination a n d Toxicology 1994 Springer-Verlag New York Inc. Non-Occupational Exposures to Pesticides for Residents of Two U.S. Cities R. W. Whitmore 1, F. W. Immerman 2, D. E. Camann 3, A. E. Bond 4, R. G. Lewis 4, J. L. Schaum 5 Research Triangle Institute, P.O. Box 12194, Research Triangle Park, North Carolina , USA 2 American Cyanamid, Pearl River, New York 10965, USA 3 Southwest Research Institute, San Antonio, Texas 78228, USA 4 U.S. Environmental Protection Agency, Research Triangle Park, North Carolina 27711, USA 5 U.S. Environmental Protection Agency, Washington, DC 20460, USA Received: 30 April 1993/Revised: 27 July 1993 Abstract. The Non-Occupational Pesticide Exposure Study, funded by the U.S. Environmental Protection Agency, was designed to assess total human exposures to 32 pesticides and pesticide degradation products in the non-occupational environment; however, the study focused primarily on inhalation exposures. Two sites--, Florida (USA) and Springfield/ Chicopee, Massachusetts (USA)--were studied during three seasons: Summer 1986 ( only), Spring 1987, and Winter Probability samples of 49 to 72 persons participated in individual site/seasons. The primary environmental monitoring consisted of 24-hr indoor, personal, and outdoor air samples analyzed by gas chromtography/mass spectrometry and gas chromatography/electron capture detection. Indoor and personal air concentrations tended to be higher in than in Springfield/Chicopee. Concentrations tended to be highest in summer, lower in spring, and lowest in winter. Indoor and personal air concentrations were generally comparable and were usually much higher than outdoor air concentrations. Inhalation exposure exceeded dietary exposure for cyclodiene termiticides and for pesticides used mainly in the home. Dietary exposures were greater for many of the other pesticides. Inhalation risks were uncertain for termiticides (depending on rates of degradation) but were negligible for other pesticides. The data were insufficient to support risk assessments for food, dermal contact, or house dust exposures. The U.S. Environmental Protection Agency (EPA) sponsored the "Non-Occupational Pesticide Exposure Study" (NOPES) from initial design work in 1985, through field data collection in , and analysis and reporting in The design of the NOPES was based on the EPA's Total Exposure Assessment Methodology (TEAM) approach to monitoring human exposures, which uses probability-based survey sampling Correspondence to: R. W, Whitmore procedures combined with questionnaire data collection and modem personal exposure monitoring equipment to obtain statistically defensible estimates of exposure levels in the population (Wallace 1987). The NOPES had both methodological and analytical objectives. The primary methodological objective was to develop the monitoring instrumentation, laboratory procedures, and survey questionnaires needed for a study of nonoccupational exposures to pesticides commonly used in and around the home. The primary analytical objective was to estimate the distribution of non-occupational exposures through air, drinking water, dermal contact, and food in two geographic regions. The EPA selected a set of 32 pesticides and pesticide degradation products for development of monitoring methodology based on regulatory interest, potential for occurrence in household environments, and analytical feasibility. Table 1 displays the target compounds and the analytical method(s) used to quantify the concentration of each compound in air and water samples. Development of the analytical methods is discussed by Hsu et al. (1988). Given the study objective of monitoring total non-occupational human exposure to the 32 selected pesticides and pesticide degradation products, monitoring was performed for air, water, food, and dermal routes of exposure, although air monitoring was the primary focus of the study. Air samples were collected over 24 h to monitor personal exposures as well as air concentrations inside and outside the residence of each participant. The sample homes were stratified by high, moderate, and low potential for indoor air pesticide residues based on questionnaire responses. Prior monitoring data indicated that drinking water was not an important route of exposure for the selected compounds, and, hence, sampling of tap water was very limited. Developing methodology for collecting samples of the food actually consumed and testing them for pesticide residues was considered beyond the scope of the study. Instead, each participant was asked to complete a 24-h dietary recall questionnaire. These data were combined with information from the U.S. Department of Agriculture's Total Diet Survey regarding pesticide residues on raw agricultural commodities to estimate

2 48 R.W. Whitmore et al. Table 1. Non,occupational pesticide exposure study (NOPES) target compounds and estimated limits of detection in air by analytical method (ng/m 3) Analytical method a Analyte GC/ECD GC/MS Insecticides * Aldrin X * alpha-bhc X Bendiocarb (Ficam ) 7-38 Carbaryl (Sevin ) 8-42 * Chlordane (technical) Chl0rpyrifos (Dursban ) X * 4,4'-DDT b X ** 4,4'-DDD b X ** 4,4'-DDE b X Diazinon Dichlorvos (DDVP) X Dicofol X * Dieldrin X gamma-bhc (lindane ) X * Heptachlor X ** Heptachlor epoxide X Malathion Methoxychlor X ** Oxychlordane X cis-permethrin X trans-permethrin X Propoxur (Baygon ) 3-16 * Ronnel X Resmethrin 8-48 Disinfectant ortho-phenylphenol 5-22 Fungicides Captan X Chlorothalonil (Bravo ) X Folpet X Hexachlorobenzene X Herbicides Atrazine ,4-D methyl ester (Summer 1986) 0.5~3.8 X 2,4-D butoxyethyl ester (other seasons) X Dacthal X * Use banned or discontinued in the United States ** Breakdown product of pesticide banned or discontinued agc/ecd = gas chromatography/electron capture detection, GC/ MS = gas chromatography/mass spectometry. Compounds analyzed by both GC/ECD and GC/MS (indicated by an "X" in the GC/MS column) were normally quantified by using the GC/ECD result bnot measured in Summer 1986 and not analyzed by GC/MS in Spring 1987 food route exposures. Pre-extracted cotton gloves were used to collect data on dermal exposure to the applicator during pesticide application events. The EPA chose to demonstrate the NOPES methodology by studying sites with little agricultural pesticide use in two regions of the United States----one with relatively high household pesticide use and another with lower, but not negligible, use. Northern Florida and New England were purposively selected as the high-use and low-use regions, respectively. Following discussions with regional, state, and local officials,, Florida (USA) was selected as the relatively high-use Table 2. Number of participants by site and season Season Springfield/Chicopee Total Summer Spring a a Winter a 52 a 123 a Total 208 a 101 a 309 a a Multiple-season participants (about one-third of the first-season sample) are counted for each season in which they participated study site, and the neighboring cities of Springfield and Chicopee, Massachusetts (USA) were chosen as the low-use site. A pilot test of the study methodology was conducted in in August of 1985 with a purposively selected sample of nine homes. Draft questionnaires were administered. Air, water, and glove samples were collected and analyzed. Questionnaires, sample collection protocols, and analytical laboratory procedures were revised based on the pilot test experience (Lewis et al. 1988). Methods and Materials Study Design To permit assessing regional and seasonal variations in exposures, samples were collected for the NOPES main study in three seasons: (1) Summer 1986:, FL site only (August 21-September 18). (2) Spring 1987:, FL (March 20-April 13) and Springfield/Chicopee, MA (May 29-June 17). (3) Winter 1988:, FL (January 30-February 17) and Springfield/Chicopee, MA (March 11-28). The numbers of participants in each study site are summarized by season in Table 2. The following data collection activities were performed for each sample member who agreed to participate: (1) A questionnaire was used to collect data on personal demographic characteristics, characteristics of the residence, dietary intake for a 24-h recall period, and the pesticide products in storage at the residence. (2) The participant was given a personal air monitor mounted on a vest to wear or keep in close proximity for 24 h. (3) Two or more fixed-site air samplers were set up inside and outside the participant's residence in frequently used locations and run for 24 h. (4) A questionnaire was administered at the end of the 24-h monitoring period regarding the participant's activities while being monitored. Subsamples of homes were also selected for duplicate or triplicate air sampling for quality assurance. In each study site, some sample members were asked to participate in all seasons of the study and others were recruited for only a single season. Monitoring some people in more than one season permitted assessing between- and within-household components of seasonal variation. Short-term temporal variation was also addressed by monitoring a subsample of households twice in the same season, 3-10 days apart. Small numbers of homes were purposively selected for tap water sampling, carpet dust collection, and dermal and personal air monitoring during pesticide application events. Tap water was sampled in six

3 Non-Occupational Exposure to Pesticides 49 homes for each site and season. Cotton gloves were worn by 22 people to monitor dermal exposure during pesticide application events. Carpet dust samples were collected in nine homes in during the Winter 1988 season using a high volume surface sampler (Roberts et al. 1991). Statistical Sampling Design In each study area, the target population consisted of the civilian, non-institutionalized population aged 16 or older who were residing, at the time of data collection ( ), in households in which no household member was employed in a position in which the primary activity involved the handling or use of pesticides. The age restriction was imposed because of the physical requirements and level of responsibility required to participate in the personal exposure monitoring and questionnaire data collection. Occupationally exposed persons were excluded because of the difficulty of differentiating their occupational and non-occupational exposures. The target population was estimated to consist of approximately 290,000 people residing in 150,000 households (in the 10 central Census County Divisions of Duval County). The Springfield/Chicopee population was estimated to consist of about 135,000 people living in 73,000 households. Standard area household survey sampling procedures were used to select participants using a three-stage probability sampling design (Kish 1965). At the first stage, a stratified sample of areas defined by 1980 Census blocks was selected for each study site. Within each sample area, all housing units were listed and an equal probability sample of housing units was selected. A short screening interview was attempted at each sample housing unit to obtain a household roster and to classify the housing unit with respect to its potential for having detectable pesticide residues in the indoor air. At the third stage of sampling, a stratified sample of individuals was selected for personal exposure monitoring. No more than one participant was selected from any household. The purpose of stratification of the first-stage sample of areas within each study site was to ensure that the sample was representative of a variety of different types of households. Thus, the sample size was proportionally allocated to the strata, and the strata were all sampled at the same rate. The samples were stratified by average housing unit value and by percent single-family residences, both of which were computed from 1980 Census block data for all areas in the study sites. The purpose of stratification at the third stage of sampling was to improve the precision of estimates of upper percentiles of exposure distributions by sampling people with high potential for exposure more heavily than those with a low potential for exposure. Thus, the strata were defined to represent high, moderate, and low potential for indoor pesticide residues based on responses regarding the following household characteristics: (1) use of pesticides on indoor plants (2) use of insecticides (e.g., flea powder, spray, or shampoo) on household pets (3) use oftermiticides. (4) use of insecticides to control household insect pests. The goal for each season was to sample 50% high, 30% medium, and 20% low potential for exposure households. When the first-season monitoring sample was selected for each study site, one-third of the sample subjects were randomly designated to be monitored during all subsequent study seasons. In addition, for each site and season five households were randomly selected for duplicate sampling, three for triplicate sampling, ten for replicate sampling 3 to 10 days apart within the same season, and six for sampling tap water. Environmental Sampling and Analytical Methods Air was drawn through a glass cartridge containing a 22-mm OD x 7.5-cm long precleaned polyurethane foam (PUF) plug (density g/cm 3) at a flow rate of 3.8 L/rain for 24 hours to collect each air sample, following the procedure of Lewis and MacLeod (1982). PUF plugs and other samples were Soxhlet-extracted in 6% diethyl ether in hexane for 16 h within 7 days of collection. Split sample extracts were analyzed by dual-column gas chromatography/electron capture detection and by gas chromatography/mass spectrometry/multiple ion detection as shown in Table 1 (Hsu etal. 1988; ASTM 1991a, 1991b). Detection limits for each NOPES target analyte were estimated from minimum peak heights based on the judgment of the analyst that the signal-to-noise ratio exceeded three. The actual limits of detection varied between analytical batches and sampling seasons, being higher in batches in which the instrument gave less response to the standard. Limits of detection were also higher for matrices that produced larger analytical interferences than for ambient air samples. Ranges of estimated limits of detection in the air samples are presented in Table 1. Several types of quality assurance/quality control (QA/QC) measures were employed throughout the study. Blank sampling matrices equivalent to a minimum of 3% of the field samples were transported to and from the field as field controls. Matrix spikes were used to assess analyte recoveries. An additional 10% of the field air samples were duplicates or triplicates. Spiked matrices prepared by an independent laboratory were used to evaluate accuracy. Field audits of sampler performance and field laboratory systems audits were conducted by U.S. EPA quality assurance teams. The QA/QC results for this study were generally acceptable, with good agreement usually obtained between simultaneous duplicates and triplicates. Overall, recoveries of laboratory spikes were acceptable (~>75%), although recovery was lower for the more polar pesticides such as propoxur and o-phenylphenol. External field audits by the U.S. EPA of the air samplers showed satisfactory performance. Blind spike PUF catridges provide by the U.S. EPA and a reference laboratory afforded good recoveries for most of the 12 analytes teslled (avg. 76%). None of the results presented in this paper are corrected for recovery efficiency. Results All sample surveys of the human population must deal with the fact that not all sample subjects will participate in the survey. Response rates for the NOPES are presented in Table 3. The overall NOPES response rate for each site and season is the product of the response rate for the household screening interviews conducted at the second stage of sampling and the response rate for the third-stage monitoring phase of the study. Only first-time participants are considered in computing the response rates in Table 3 because those sample members who participated in a previous season were more likely to agree to an additional season of participation. The overall response rates were 40% for Springfield/Chicopee and 45% for. The NOPES response rates are low compared 1Lo those often achieved in traditional surveys but are similar to the response rates achieved in comparable exposure assessment studies (Wallace 1987; Cox et al. 1988). Weight adjustments based on sample-based weight adjustment cells were used to reduce the potential for nonresponse bias (Kalton and Maligalig 1991). Selected characteristics of sample participants are presented in Table 4 by study site. The participants in and Springfield/Chicopee were similar with respect to the distributions of sex, age, employment status, and occurrence of occupational exposure during the 24 h of monitoring. In each site, approximately 50% of the participants reported that they recalled that the housing unit had been treated for termites. The two study sites differed somewhat in ethnic composition, age and type of housing stock, and number of pesticide prod-

4 50 R.W. Whitmore et al. Table 3. Response rates Springfield/Chicopee Summer Spring Winter Spring Winter '86 '87 '88 Total '87 '88 Total Seeond Stage Sample size ,501 1,422 1,050 2,472 Eligible ,372 1, ,339 Respondents , ,774 Response rate 74% 66% 81% 73% 70% 84% 76% Third Stage First time sample: Selected Eligible t Respondents Response rate 54% 73% 61% 61% 55% 51% 53% Overall response rate a 40% 48% 49% 45% 39% 43% 40% Follow-up sample: Selected Eligible Respondents Response rate -- 66% 84% 73% -- 75% 75% Total: Selected Eligible Respondents aoverall response rate = (second-stage response rate) * (third-stage response rate) for first time members of the sample Note: The time available for second-stage contacts was limited because of the seasonal nature of the study and to limit field data collection costs ucts in storage. The percentage of respondents who were either non-white or Hispanic was higher in than in Springfield/Chicopee (28% versus 14%). The Springfield/ Chicopee sample contained a higher percentage of older homes (48% over 35 years old versus 31%). The Springfield/Chicopee sample contained more attached single-family dwellings (12% versus 3%) and contained more mobile homes (10% versus 2%). In addition, the average number of pesticide products in storage was slightly higher in Springfield/Chicopee housing units (5.3 pesticide products per Springfield/Chicopee household versus 4.2 for ). Air Sampling Results Tables 5 and 6 present the estimated percentages of detectable levels of pesticide residues in the population of air samples that could have been collected from the and Springfield/Chicopee households, respectively, by season for indoor, outdoor, and personal air samples. The tables are restricted to the 22 pesticides for which the estimated percentage detectable was at least 10% in at least one of the media in one site/season. Estimated mean concentrations are presented for these same pesticides in Tables 7 and 8 for and Springfield/Chicopee, respectively. The means were calculated using zero as the concentration of each analyte that was not detected, which may bias the means downward, especially for analytes with relatively high limits of detection (see Table 1). Levels in were generally higher than in Springfield/Chicopee~ as expected. For the most frequently detected compounds, levels tended to be highest in summer, lower in spring, and lowest in winter. This pattern is illustrated for heptachlor and diazinon, in Figures 1 and 2, respectively. Indoor air concentrations and personal exposures were usually comparable, with indoor air levels slightly higher. Both tended to be considerably higher than outdoor air levels. The Spearman rank-order correlation between indoor and personal levels was 0.75 or greater for several compounds (5-14) for each site/season. The findings with regard to the relationship between indoor, outdoor, and personal exposure levels are very similar to those of the EPA's TEAM studies of exposures to volatile organic compounds (Wallace 1987). They also confirm the findings of other studies that indoor air levels of toxic substances are often considerably greater than outdoor air levels (Lewis and MacLeod 1982; Lewis and Lee 1976; Nigg et al. 1990). From Table 7, seven compounds----chlordane, chlorpyrifos, diazinon, dichlorvos, heptachlor, ortho-phenylphenol, and propoxur--had the highest mean concentrations in each season in for both indoor air and personal air. Moreover, these seven compounds generally account for about 85% of the mass of pesticides detected for each season in. In Table 8, these same compounds account for nearly all (about 98%) of the mass of pesticides detected in each season in Springfield/Chicopee. However, gamma-bhc and dicofol replace dichlorvos in last place among the top seven compounds by total mass across seasons for indoor air and for personal air, respectively, in Springfield/Chicopee. The third-stage sample was stratified by high, moderate, and low potential for indoor air pesticide residues, as defined from the household screening data. Higher sampling rates were used for strata with higher potential for residential exposures to better characterize the upper percentiles of the exposure distributions. The effectiveness of the third stage stratification can be addressed either with respect to specific analytes or with respect to general, non-specific occurrence of pesticide residues. For

5 Non-Occupational Exposure to Pesticides 51 Table 4. Third-stage respondent and household characteristics Percent of respondents Characteristic Springfield/Chicopee Sex Male Female Race/Ethnicity White, non-hispanic Nonwhite or Hispanic Age Over Employed Yes No Occupational Exposure Yes 6 7 No Type of housing unit Unattached single family Attached single family 3 12 Multiunit (apartment) Mobile home 10 2 Age of housing unit Less than 6 years old More than Any termiticide treatment of housing unit Yes No Don't know specific analytes, the strata used in this study were not consistently predictive of relative exposure levels, as shown in Table 9. Therefore, using higher sampling rates in the strata representing higher potential for exposure generally led to unnecessary loss of precision. However, Table 10 shows that the stratification was effective in a more genera/ sense. The mean number of analytes detected in indoor air was generally largest for the high exposure stratum, less for the moderate exposure stratum, and smallest for the low exposure stratum. Thus, the exposure strata may have led to some gains in precision if equal sampling rates, or proportional allocation, had been employed. Dietary Exposure Results The NOPES data were used for limited comparisons of the relative contributions of air and dietary exposure pathways for the target pesticides. Mean daily air exposures were estimated by multiplying the average daily personal concentration (ng/ m 3) in Tables 7 and 8 by 20 m 3 of air respired per day. Mean daily dietary and air exposure estimates are shown in Table 11 for and Springfield/Chicopee. Three estimates of mean daily dietary exposure were computed for each analyte: (1) One estimate was computed from (a) national estimates of mean dietary intake of pesticide residues per kilogram of body weight by age and sex based on eight "market baskets" collected for the FDA's Total Diet Study (TDS) from April 1982 through April 1984; (b) national estimates of weight by age and sex from the Statistical Abstract of the United States (U.S. Bureau of the Census 1983a); and (c) age and sex distributions for and Springfield from the Census Bureau's County and City Data Book (U.S. Bureau of the Census 1983b). (2) A second estimate was computed from (a) four "market baskets" of the TDS collected between July 1986 and May 1987 and (b) the NOPES dietary recall questionnaires for the 65 respondents in the Summer 1986 sample. (3) A third estimate was computed from (a) one "market basket" of the TDS collected in April and May 1987 in the South (Brownsville, TX; Birmingham, AL; and Baton Rouge, LA) and (b) the NOPES dietary recall questionnaires for the 65 respondents in the Summer 1986 sample. Because these estimates of exposures via the food route are more indirect than the estimates of the exposures via the air route, they are subject to considerably greater uncertainty. However, as seen in Table 11, the various estimates of the food route exposures are generally comparable. The comparability of the first set of estimates based on national statistics for residue intake by age, weight, and sex with the other two sets of estimates that utilized the NOPES dietary recall questionnaires suggests that the sample of 65 respondents was sufficient to produce estimates that reliably indicate the order of magnitude of exposure. Because of budget limitations, estimates utilizing the Spring 1987 and Winter 1988 NOPES data were not computed. The mean dietary and air exposures are presented in Table 11. The analytes are categorized by whether (a) the air exposures appear to be highest, (b) the dietary exposures appear to be highest, or (c) the data are inconclusive regarding the relative contributions of these two routes of exposure. Of the seven target pesticides for which no dietary data were reported, three--dichlorvos, ortho-phenylphenol, and propoxur--are used mainly in the home and occurred at relatively high levels in the personal air samples. The air route of exposure is likely to be most important for these analytes, and they are listed as such in Table 11. The dietary route of exposure appears to be more important for many of the analytes studied, except for cyclodiene termiticides and pesticides used mainly in the home. Additional research is needed to better quantify the relalfive contribution of diet to total exposure. Application Event Exposure Results A small component of the overall study was designed to address dermal and respiratory exposure during household application of pesticide products. Monitored application events included the use of commercial aerosol spray cans, hand-pumped broadcast sprayers for ready-mixed and user-mixed formulations, mechanical spreaders for granular formulations, and insecticide-containing pet shampoos. Of 22 events monitored, 12 involved one or more of the targeted pesticides, i.e., carbaryl,

6 52 R.W. Whitmore et al. Table 5. Estimated percent of population with detectable levels in air a Indoor Outdoor Personal Analyte Summer Spring Winter Summer Spring Winter Summer Spring Winter Dichlorvos alpha-bhc Hexachlorobenzene gamma-bhc Chlorothalonil < 1 19 Heptachlor Chlorpyrifos Aldrin Dacthal Heptachlor epoxide ,4-D ester b Dieldrin Methoxychlor Dicofol Chlordane ,4'-DDT ,4'-DDE ortho-phenylphenol Propoxur Bendiocarb Diazinon Carbaryl Malathion atarget pesticides not listed were estimated to have less than 10% detectable levels for all media, all seasons, and both sites bmethyl ester in summer, butoxyethyl ester in spring and winter Table 6. Estimated percent of Springfield/Chicopee population with detectable levels in air a Indoor Outdoor Personal Analyte Spring Winter Spring Winter Spring Winter Dichlorvos alpha-bhc Hexachlorobenzene gamma-bhc Chlorothalonil < Heptachlor Chlorpyrifos < Aldrin Dacthal Heptachlor epoxide ,4-D butoxyethyl ester Dieldrin Methoxychlor Dicofol Chlordane ,4'-DDT < ,4'-DDE ortho-phenylphenol Propoxur Bendiocarb Diazinon Carbaryl Malathion ~Target pesticides not listed were estimated to have less than 10% detectable levels for all media, all seasons, and both sites

7 Non-Occupational Exposure to Pesticides 53 Table 7. Estimated mean air concerntrations for residents a'b (ng/m 3) Indoor Outdoor Personal Analyte Summer Spring Winter Summer Spring Winter Summer Spring Winter Dichlorvos alpha-bhc < < Hexachlorobenzene gamma-bhc Chlorothalonil < Heptachlor Chlorpyrifos Aldrin Dacthal Heptachlor epoxide ,4-D ester c < Dieldrin < Methoxychlor Dicofol Chlordane ,4'-DDT ,4'-DDE ortho-phenylphenol < Propoxur Bendiocarb Diazinon Carbaryl Malathion atarget pesticides not listed were estimated to have less than 10% detectable levels for all media, ban estimate of "0" means no detectable levels were observed Methyl ester in summer, butoxyethyl ester in spring and winter all seasons, and both sites ab Table 8. Estimated mean air concentrations for Springfield/Chicopee residents (ng/m 3) Indoor Outdoor Personal Analyte Spring Winter Spring Winter Spring Winter Dichlorvos alpha-bhc < Hexachlorobenzene <0.05 gamma-bhc Chlorothalonil Heptachlor Chlorpyrifos < Aldrin Dacthal Heptachlor epoxide ,4-D butoxyethyl ester Dieldrin Methoxychlor Dicofol Chlordane ,4'-DDT < ,4'-DDE ortho-phenylphenol Propoxur Bendiocarb Diazinon Carbaryl Malathion atarget pesticides not listed were estimated to have less than 10% detectable levels for all media, all seasons, and both sites b An estimate of "0" means no detectable levels were observed

8 54 R.W. Whitmore et al. 180, Mean Concentration (ng/m 3) mlndoor air ~Outdoor air ii ~iiiiiiiii Personal air Summer Spring Winter Spring Winter Jackeonville Springfield/Chicopee Fig. 1. Heptachlor mean concentrations for indoor, outdoor, and personal air 5oo O Mean Concentration (ng/m 3) Summer /~iiiiii~ii i~i~ i!~!ii!j!i~!ii!i mlndoor air ~Outdoor air ~Pereonal air Spring Winter Spring Winter Spring field/chicopee Fig. 2. Diazinon mean concentrations for indoor, outdoor, and personal air Table 10. Overall effectiveness of the exposure stratification model: Mean number of analytes detected Potential indoor air Exposure stratum High Medium (Standard errors in parentheses) Low Indoor air Summer 9.0 (1.5) 8.0 (0.5) 8.7 (0.5) Spring 8.0 (0.4) 7.3 (0.5) 6.7 (0.8) Winter 9.3 (0.5) 8.4 (0.8) 7.8 (0.6) Springfield Spring 4.0 (0.4) 3.9 (0.7) 3.4 (0.6) Winter 4.1 (0.1) 3.7 (0.3) 4.3 (4.1) Personal air Summer 7.7 (1.2) 8.4 (0.5) 7.8 (1.1) Spring 7.1 (0.6) 6.3 (0.5) 5.7 (1.3) Winter 8.0 (0.3) 8.9 (0.9) 7.9 (0.4) Springfield Spring 3.7 (0.4) 3.0 (1.2) 4.1 (0.7) Winter 4.5 (0.1) 3.0 (0.5) 4.5 (0.6) chlorpyrifos, diazinon, dichlorvos, dicofol, malathion, methoxychlor, and resmethrin. Participants were provided with preextracted cotton gloves and a portable PUF sampler, which were used for the duration of the application event only. In all events involving the application of one or more target compounds, the compounds were measured on the sample gloves, usually at high concentrations. In the majority of these events, detectable levels of the applied target compound were also measured in the personal air samples collected during the events. The five cases in which nothing was detected in the air samples may have been due to the high limits of detection inherent in short-duration sampling. In 18 of the 22 monitored events, analytes other than those being applied were also detected on the gloves. These other analytes were usually, but not always, at low concentrations. They may have been present as residues on the application equipment or on the respondent from a previous application. The residues found on the cotton gloves were used as a measure of dermal exposure to the hands. Participants were Table 9. Ranks of exposure strata mean indoor air concentrations for commonly detected analytes a Springfield/Chicopee Summer Spring Winter Spring Winter Rank b Rank b Rank b Rank b Rank b Analyte Dichlorvos L M H L M H H M L M H L M H L Heptachlor M H L M L H L H M L H M H M L Chlorpyrifos H L M H M L H L M H M L H M L Dieldrin H M L L H M H L M L H = M L M H Chlordane H M L H L M L H M L H M H M L ortho-phenylphenol H M L H L M H L M M L H H L M Propoxur L H M L M H M L H H M L H L M Diazinon H M L H M L L M H M H L L H M ah = high exposure stratum; M = medium exposure stratum; L = low exposure stratum b For each analyte, the stratum with the highest mean indoor air concentration was ranked "1," next highest ranked "2," and lowest ranked "3"

9 Non-Occupational Exposure to Pesticides 55 Table 11. Mean dietary and air exposure estimates (ng/day) Dietary exposure Relative exposure level Air exposure a analyte Springfield '82-'84 b '86-'87 c Air Higher than Dietary Aldrin <7 NR Chlordane ~ 3,942 3, Heptachlor f 2, Dichlorvos 1, NR NR ortho-phenylphenol 1, NR NR Propoxur 3, NR NR Dietary Higher than Air alpha-bhc Captan 2 2 1,375 1,590 Carbaryl ,682 3,170 ~DDT g ,954 1,530 Dacthal Dicofol ND ,540 Dieldrin Folpet NR 640 Hexachlorobenzene 10 ND Malathion ,701 5,560 Methoxychlor 6 ND cis-permethrin 18 ND 168 1,030 trans-permethrin 6 ND Relative Level Unclear Chlorothalonil NR 20 Chlorpyrifos 3, Diazinon 1, ,140 Gamma-BHC Ronnel ND 2 <7 NR Springfield '87 d,82_,846 NR < NR NR NR NR NR NR ,468 2,000 1, , NR ,510 4, O NR NR <7 ND = not detected in NOPES personal air samples NR = not reported in the applicable "market basket(s)" of the Total Diet Study (TDS) aassuming 20 m 3 air respired per day bbased on eight TDS "market baskets" collected from April 1982 through April 1984 and the city's population distribution by age and sex from the Census Bureau's County and City Data Book (U.S. Bureau of the Census, 1983b) CBased on four TDS "market baskets" collected from July 1986 through May 1987 and the NOPES dietary recall questionnaires for Summer 1986 dbased on one TDS "market basket" collected in the South in April and May 1987 and the NOPES dietary recall questionnaires for Summer 1986 ~Includes oxychlordane elncludes heptachlor epoxide gtotal of 4,4'-DDT; 4,4'-DDD; and 4,4'-DDE asked to wear their usual rubber or work gloves over the cotton gloves, but few did so. The provided cotton gloves may have overestimated the dermal exposure, because of their absorptivity, particularly in the case of liquid formulations. In contrast, the air monitors may have underestimated exposure due to lack of sensitivity. Because of these limitations and the small sample size, it is not possible to accurately estimate the relative importance of dermal and respiratory exposure during residential pesticide application events. However, in the worst-case situation, a participant applied a granular chlorpyrifos formulation by hand, leaving a residue of 50 mg of the insecticide on the gloves. Personal air monitoring during the short (ca. 5-min) application indicated that he was exposed to average air levels of 13.4 p~g/m 3. Therefore, the participant would have been exposed to a 2 Ixg inhalation dose (assuming a lightwork respiratory rate of 30 L/min) and a 500 p~g dermal dose (assuming 1% absorption). The average resident was exposed to 24-h mean chlorpyrifos air concentrations (as determined by personal air sampling) of 0.3 p~g/m 3 at the time of the application event (August 1986), resulting in an average daily inhala- tion exposure of 13 ~g/day. The estimated dermal dosage in this worst case was 250 times the estimated air route exposure during the application and nearly 40 times the average daily respiratory dose. However, such applications would not be expected to occur frequently in most households. It appears, therefore, that chronic respiratory exposure to pesticides may result in greater annual cumulative doses than del~al contact during pesticide applications for most pesticides. Using disinfectants and treating pets may be exceptions to the assumption of infrequent application events, which could alter the relative importance of dermal exposure in these situations. Carpet Dust Sampling Results Carpet dust samples were collected from nine purposively selected houses during the Winter 1988 season in. In two of these homes, samples were taken from two different carpets. In one case, the carpets were the same age, while in the other the carpets were one and eighteen years old. The collected

10 56 R.W. Whitmore et al. Table 12. Levels of pesticides detected most frequently in carpet dust ~ Analyte No. of times Median Mean Mean detected in dusff in dust e in air in dust b (tzg/g) (p~g/g) (p~g/m 3) Heptachlor Chlorpyrifos Aldrin <0.01 Dieldrin Chlordane ~DDT d <0.01 ortho-phenylphenol Propoxur Diazinon Carbaryl ND Atrazine ND ND = not detected aln nine homes sampled in the Winter 1988 season bbased on the 11 carpets tested c Mean and median of the detectable levels in dust dtotal of 4,4'-DDT; 4,4'-DDD; and 4,4'-DDE dust was sieved to exclude particles larger than 150 I~m in diameter (e.g., fibers), weighed and subjected to the same extraction and analysis procedures used for the air samples. The average number of the targeted pesticides found in the carpet dust samples was 12, compared to 7.5 in the air samples collected in the same nine residences. Thirteen pesticides were found in the dust that were not detected in the air. These tended to be less volatile compounds. All but three of the target analytes were found in at least one carpet dust sample. Mean concentrations of the pesticides detected in carpet dust ranged from less than 0.01 Ixg/g to 15.4 p~g/g. Chlordane, detected in 10 of the 11 carpets, showed the highest concentrations (mean = 14.9 txg/g; median = 6.3 txg/g). The mean and median levels of the most frequently detected pesticides are shown in Table 12 with the corresponding mean air concentrations in these same nine homes. The mean values were elevated by residues found in the 18-year-old carpet in one home, which had chlordane at 98.6 Ixg/g, chlorpyrifos at 21.9 p~g/g, dieldrin at 18.2 ixg/g, and ~DDT at 6.3 ~g/g. The 1-year-old carpet in this same home also had relatively high pesticide loadings (although nearly an order-of-magnitude lower), presumably from cross-contamination. The populations at greatest potential risk of exposure to pesticides found in carpet dust are infants and toddlers (ages years) who may ingest dust through mouthing of hands, toys, and other objects. Literature values for soil ingestion by infants and toddlers range from 10 mg to 10 g/day (Lewis et al. 1994). These estimates are based largely on ingestion of outdoor soil rather than house dust. While the average young child spends much more time playing on the floor inside the home than outdoors, no data are available to assess the relative importance of indoor and outdoor soil ingestion. However, if we assume that a child ingests 100 mg per day of dust from contact with carpeted floors and breathes 6.3 m 3 of air per day (24 h inside Table 13. Weighted estimate of average daily air concentrations, cancer risk, and hazard index for air (pesticides other than cylodiene termiticides) Analyte Average Slope Excess Reference daily concen, factor lifetime dose Hazard (ng/cu m) (kg-day/mg) cancer risk g (mg/kg-day) index h Dichlorvos c 2E-02 alpha-bhc a 2E-06 Hexachlorobenzene c 2E-04 gamma-bhc a IE-02 Chlorothalonil b 3E ~ 2E-05 Chlorpyrifos a 2E-02 Dacthal a 1E-07 2,4-D ester b 6E a 2E-06 Methoxychlor c 2E-06 Dicofol f < b < 3E-06 f ~ < 1E-02 f 4,4'-DDT ~ a 5E a 3E-04 4,4'-DDE e b 6E-08 ortho-phenylphenol b 3E-08 Propoxur b 4E a 1E-02 Bendiocarb c 9E-04 Diazinon c 5E-01 Carbaryl a 2E-05 Malathion a 2E-04 a Source: Integrated Risk Information System (IRIS) bsource: Memorandum from Reto Engler to Health Effects Division Branch Chiefs and Selected OPP Division Directors, US EPA, October 27, 1989 c Source: Reference Dose Tracking Report, Health Effects Division, Office of Pesticides, US EPA, October 12, 1989 d Methyl ester in summer, butoxyethyl ester in spring and winter e Concentration calculated as (3 * spring + winter)/4 because this analyte was not measured in the summer season fnot found above detection limit. The concentration shown is the maximum detection limit encountered in this study. The corresponding risk estimate represents an upper bound g Risk ~< 1E-06 is generally considered "negligible" by the USEPA h Hazard ~< 1E00 is generally considered "negligible" by the USEPA

11 Non-Occupational Exposure to Pesticides 57 Table 14. Weighted estimate of average daily air concentrations, cancer risk, and hazard index for Springfield/Chicopee air (pesticides other than cylodiene termiticides) Average Slope Excess Reference daily concen, factor lifetime dose Hazard Analyte (ng/cu m) (kg-day/mg) cancer risk g (mg/kg-day) index h Dichlorvos 3.3 alpha-bhc a 4E-07 Hexachlorobenzene f <2.2 gamma-bhc 1.9 Chlorothalonil lb 2E-09 Chlorpyrifos 7.1 Dacthal 2 2,4-D ester a'f < b <2E-07 f Methoxychlor f <7.8 Dicofol f b 4,4'-DDT ~ a 9E-08 4,4' -DDE e b 4E-07 ortho-phenylphenol h 2E-08 Propoxur b 3E-08 Bendiocarb 0.3 Diazinon 7.9 Carbaryl 0.1 Malathion ~ 1E c <8E-04 f a 2E c 1E a 7E a 1E a <9E-04 f 0.05 c <5E-05 f c 2E a 5E a 1E c 2E c 3E a 3E a 6E-06 asource: Integrated Risk Information System (IRIS) bsource: Memorandum from Reto Engler to Health Effects Division Branch Chiefs and Selected OPP Division Directors, USEPA, October 27, 1989 CSource: Reference Dose Tracking Report, Health Effects Division, Office of Pesticides, USEPA, October 12, 1989 dmethyl ester in summer, butoxyethyl ester in spring and winter econcentration calculated as (3 * spring + winter)/4 because this analyte was not measured in the summer season fnot found above detection limit. The concentration shown is the maximum detection limit encountered in this study. The corresponding risk estimate represents an upper bound grisk ~< 1E-06 is generally considered "negligible" by the USEPA hhazard ~< 1E00 is generally considered "negligible" by the USEPA the house), he would potentially receive about 50% greater total exposure to the target pesticides via the air route than by ingestion of house dust based on the results of this study. However, virtually all the non-dietary exposure to EDDT would come from dust ingestion. The same is true for aldrin, atrazine and carbaryl, which were largely absent in the indoor air. In the worst case scenario involving the 18-yr-old carpet, potential dust ingestion would outweigh air route exposure by a factor of 9 for chlordane and 4 for chlorpyrifos. The data from this study suggest that dust ingestion could constitute a substantial portion of a child's exposure to pesticides in some homes. These data also suggest that a child may be exposed to a greater number of pesticides in the home by ingestion of house dust than by inhalation. Dermal absorption of pesticides from house dust may also be a potential route of exposure for small children. C = [Summer + (2*Spring) + Winter]/4 [(3*Spring) + Winter] / 4 for for Springfield/Chicopee The adult average daily exposure, E, was estimated as: E=C*I/W where I represents inhalation rate and W represents body weight. EPA "standard factors" of 20 m3/day and 70 kg were used for I and W, respectively. Given the average daily exposure estimate, the excess lifetime cancer risk, R, was calculated as R=E*Q Risk Assessment for Air Route Exposures Health risks due to airborne non-occupational exposures to pesticide residues were estimated using the seasonal average personal air exposures shown in Tables 7 and 8. The data were insufficient to support risk assessments for food, dermal contact, or house dust exposures. The average personal air concentration across seasons, C, was estimated as: where Q is the 95% upper confidence limit of the linear slope factor for cancer risk for the analyte (mg/kg-day)- 1. Likewise, the non-cancer hazard index, H, was calculated as: H=E/D where D is the reference dose for the analyte (mg/kg-day). The cancer slope factors, Q, and the reference doses, D, were obtained, whenever possible, from the EPA's Integrated Risk Information System (IRIS). Otherwise, the cancer slope factors

12 58 R.W. Whitmore et al. Table 15. Weighted estimate of average daily air concentrations, cancer risk, and hazard index (cyclodiene termiticides) Average Slope Excess lifetime cancer risk g Reference Hazard index h daily concen, factor dose Analyte (ng/cu m) (kg-day/mg) (a) (b) (mg/kg-day) (a) (b) Heptachlor a 2E-04 c 6E-06 d b 7E-02 c 3E-03 d Aldrin a 1E-04 c 5E-06 d b 3E-01 ~ 1E-02 d Dieldrin a 3E-05 ~ 1E-06 d b 4E-02 1E-03 a Chlordane a 7E-05 c 3E-06 d ~ 1E+00 c 4E-02 d Heptachlor epoxide e ~ 1E-06 c 4E-08 d b 1E-02 c 5E-04 d Springfield/Chieopee Heptachlor ~ 4E-05 ~ 1E-06 d b 2E-02 ~ 6E-04 d Aldrin a 5E-07 c 2E-08 d b 1E-03 ~ 4E-05 d Dieldrin a 4E-06 c IE-07 a b 5E-03 c 2E-04 d Chlordane a 7E-05 c 3E-06 d b 1E+00 c 4E-02 d Heptachlor epoxide ~'f < a < 1E-06 ~ <4E-08 d b <le-02 c <5E-04 d a Source: Integrated Risk Information System (IRIS) b Source: Reference Dose Tracking Report, Health Effects Division, Office of Pesticides, US EPA, October 12, 1989 CThe risk estimates presented in this column assume that the concentrations remain constant over 70 years. Since these pesticides have been cancelled or withdrawn, some reduction in risk will occur due to degradation athe risk estimates presented in this column assume that the pesticide degrades with a 2-yr half-life. No reliable degradation data are available and these estimates are included as an example of the possible reductions in risk due to degradation Breakdown product of cyclodiene termiticide fnot found above detection limit. Concentration shown is the maximum detection limit encountered in this study. The corresponding risk estimate represents an upper bound grisk is calculated for the inhalation route assuming: (a) constant average exposure and (b) a 2-yr half-life. Risk ~< 1E-06 is generally considered "negligible" by the USEPA hhazard is calculated for the inhalation route assuming: (a) constant average exposure and (b) a 2-yr half-life. Hazard ~<IE00 is generally considered "negligible" by the USEPA and reference doses were based on assessments on file with the EPA's Office of Pesticide Programs (OPP). For pesticides other than cyclodiene termiticides, risk estimates are presented in Tables 13 and 14 for and Springfield/Chicopee, respectively. As compared to thresholds commonly used by the EPA [10-6 for cancer risk and one (1) for non-cancer hazard], the risks for non-occupational airborne exposures all appear to be negligible. Of course, the risk estimates are based exclusively on estimated average concentrations via the air route of exposure, assuming exposure at this average level throughout a 70-yr lifetime. Moreover, the reference dose and slope factors are subject to considerable uncertainty. The risk estimates for cyclodiene termiticides are presented in Table 15. However, assuming a constant average exposure at current levels over a 70-yr lifetime is not realistic because these pesticide products have been banned from use in the United States. Average exposure levels will decline because of product degradation, although carpet dust levels suggest that degradation indoors may be slow (Lewis et al. 1994). Therefore, a second set of risk and hazard levels based on assuming a 2-year half-life for each analyte is also presented in Table 15. Given this assumed rate of degradation, the risks are again estimated to be negligible. Conclusion NOPES was designed to assess total human exposures to pesticide residues in the non-occupational environment. Potential routes of exposure that were assessed include: (1) air (personal exposure plus indoor and outdoor air monitoring at the residence for 24 h) (2) water (tap water samples) (3) food (dietary questionnaires linked with USDA's Total Diet Survey) (4) dermal contact (glove samples during application events) (5) dust ingestion (carpet dust in nine homes). However, monitoring of routes of exposure other than airborne exposures was minimal because water was not considered to be a significant route of exposure and because reliable sampling and analytical methods were not available for food, dermal, and dust routes of exposure. Monitoring was conducted in one site considered a priori to be a high-use area,, Florida, and another site considered a priori to be a low-use area, Springfield and Chicopee, Massachusetts. The study investigated seasonal variation by monitoring during three seasons: Summer 1986, Spring 1987, and Winter The NOPES methodological objectives were generally achieved. The air sampling instrumentation and analytical procedures reliably characterized personal, indoor, and outdoor air concentrations for the majority of the study analytes. Because the study was based on a probability sampling design, valid statistical inferences regarding population exposures via the air route were possible for the two study areas. Airborne concentrations of pesticide residues were generally higher in than in Springfield/Chicopee, as expected. Concentrations tended to be highest in summer, lower in spring, and lowest in winter. Personal air exposures and indoor air concentrations tended to be comparable, and they

13 Non-Occupational Exposure to Pesticides 59 both tended to be higher than outdoor air concentrations. Inhalation exposure exceeded dietary exposure for termiticides, while the dietary route exposures appeared to be greater for many other pesticides. The chronic risk of cancer development and the acute risk of other adverse health effects as a result of airborne exposures were estimated for each compound based on the estimated average exposure level for a person spending their lifetime at either study site. In most cases, the inhalation risks were estimated to be negligible based on EPA guidelines (<10-6 for chronic effects, < 1 for acute effects). The only possible exceptions were for cyclodiene termiticides, especially heptachlor and aldrin. The risks associated with these banned compounds are particularly uncertain since their degradation rates have not been well established. Assuming a 2-yr half-life, the risks appear to be negligible for these compounds. However, further research on degradation of cyclodienes and other pesticides under field conditions is recommended. Acknowledgments. The information in this document has been funded wholly or in part by the United States Environmental Protection Agency under Contracts and to Environmental Monitoring and Services, Inc. and Research Triangle Institute, respectively. It has been subjected to agency review and approved for publication. Mention of trade names or commercial products does not constitute endorsement or recommendation for use. References Annual Book of ASTM Standards (1991a) Standard practice for sampling and analysis of pesticides and polychlorinated biphenyls in indoor atmospheres (D ), American Society for Testing and Materials, Philadelphia, pp (1991b) Standard test method for chlordane and heptachlor residues in indoor air (D ), ASTM, Philadelphia, pp Cox BG, Mage DT, Immerman FW (1988) Sample design consider- ations for indoor air exposure surveys. J Air Poll Contr Assoc 38: Hsu JP, Wheeler HG, Camann DE, Schattenberg H J, Lewis RG, Bond AE (1988) Analytical methods for detection of nonoccupational exposures to pesticides. J Chromatogr Sci 26: Kalton G, Maligalig DS ( 1991) A comparison of methods of weighting adjustment for nonresponse. Bureau of the Census 1991 Annual Research Conference Proceedings, pp Kish L (1965) Survey sampling. Wiley, NY Lewis RG, Bond AE, Fortman RC, Camann DE (1994) Evaluation of methods for monitoring the exposure of small children to pesticides in the residential environment. Arch Environ Contam Toxicol 26:37-46 Lewis RG, Bond AE, Johnson DE, Hsu JP (1988) Measurement of atmospheric concentrations of common household pesticides: a pilot study. Environ Monit Assess 10:59-73 Lewis RG, MacLeod KE (l 982) A portable sampler for pesticides and semi-volatile industrial organic chemicals. Anal Chem 54: Lewis RG, Lee RE Jr (1976) Air pollution from pesticides: sources, occurrences and dispersion. In: Lee RE Jr (ed) Air pollution from pesticides and agricultural processes. CRC Press, Boca Raton, FL, pp 5-55 Nigg HH, Beier RC, Carter O, Chaisson C, Franklin C, Lavy T, Lewis RG, Lombardo P, McCarthy JF, Maddy KT, Moses M, Norris D, Peck C, Skinner K, Tardiff RG (1990) Exposure to pesticides. In: Baker SR, Wilkinson CS (eds) The effect of pesticides on human health: advances in modem environmental toxicology, Vol XVII. Princeton Scientific, NY, pp Roberts JW, Budd WT, Ruby MG, Bond AE, Lewis RG, Wiener RW, Camann DE (1991) Development and field testing of a high volume sampler for pesticides and toxics in dust. J Exp Anal Environ Epidem 1: Wallace LA (1987) The total exposure assessment methodology (TEAM) study: Summary and analysis: Volume I (EPA/600/6-87/ 002a). Office of Research and Development, U.S. Environmental Protection Agency, Washington, DC US Bureau of the Census (1983a) Statistical abstract of the United States: US Government Printing Office, Washington, DC --(1983b) County and city data book, US; Government Printing Office, Washington, DC