United States Environmental Protection Agency Atmospheric Research and Exposure Assessment Laboratory Research Triangle Park, NC 27711 Research and Development EPA/600/SR-93/050 May 1993 &EPA Project Summary Particle Total Exposure Assessment Methodology (PTEAM): Riverside, California Pilot Study - Volume I E. D. Pellizzari, K. W. Thomas, C. A. Clayton, R. W. Whitmore, R. C. Shores, H. S. Zelon, and R. L. Peritt EPA's Atmospheric Research and Exposure Assessment Laboratory (AREAL) and the California Air Re- sources Board sponsored a study of human exposure to inhalable particles in the Los Angeles Basin. A total of 178 residents of Riverside, CA, wore specially designed personal monitors for a day, and allowed their homes and back yards to be monitored concur- rently, in the fall of 1990. Personal ex- posures averaged 150 ng/m3 during the day, compared to indoor and outdoor concentrations of 94-95 |ig/m3. Daytime personal exposures to 14 of 15 ele- ments were also significantly increased compared to indoor and outdoor con- centrations. Housework (vacuuming, dusting, cooking) and sharing a home with a smoker were two activities asso- ciated with significantly increased ex- posures to particles and metals. This Project Summary was developed by EPA's Atmospheric Research and Exposure Assessment Laboratory, Re- search Triangle Park, NC, to announce key findings of the research project that is fully documented in a separate report of the same title (see Project Report ordering information at back). Introduction In 1986, Congress mandated that the USEPA undertake a study of exposure to particles. EPA's Atmospheric Research and Exposure Assessment Laboratory (AREAL) joined with California's Air Re- sources Board to sponsor a study in the Los Angeles Basin. Small portable per- sonal monitors were designed to measure inhalable particles (aerodynamic diameter less than 10 |im or PM10) In addition, stationary microenvironmehtal monitors were designed to sample both PM10 and PM . (fine particles <2.5 jim in diameter). Following a 9-home study to test the mea- surement methods in the Azusa, CA, area, a study of 178 residents of Riverside, CA, was carried out in the fall of 1990. Procedure Measurement Methods A personal exposure monitor (PEM) was designed to collect PM10 using a sharp-cut impactor with a circular set of holes 1.9 mm in diameter. Particles are collected at a flow rate of 4 Lpm on a 37-mm Teflon filter mounted below a greased impactor plate. The PEM consists of a soft canvas bag containing the pump and battery pack that can be worn on the hip, stomach, lower back, or over the shoulder. A broad shoulder strap supports the sampling head, which can be moved to a comfortable position near the collarbone using a Velcro fastener. When worn on the body, the pump/battery pack slides freely on a belt, allowing it to be shifted to the most com- fortable position depending on people's activities or changes of posture. A leather backing for the sampling head was added to prevent it from being accidentally turned toward the body, and a 2-inch guard shaped like a hooded traffic light was added to the top of the sampling head to protect against skin flakes, hairs, or fibers from clothes. A small quiet monitor for concurrent in- door and outdoor sampling was also cre- ated. This monitor is called the Stationary Ambient Monitor (SAM) when used out- Printed on Recycled Paper ------- doors and the Stationary Indoor Monitor (SIM) when used indoors. The monitor employs identical sampling heads and flow rates as the PEM to collect PM10, but operates off line current instead of batter- ies. The sampling head can be replaced with one having holes 1.4 mm in diameter to collect fine particles (PM ). Laboratory studies indicate that the PEM and the SAM10 have a sharp outpoint at about 11 urn, while the SAM25has a sharp outpoint at 2.5 |im. Study Design The City of Riverside, CA, was selected for study because it is known to have highly variable outdoor PM10 concentra- tions and because the socioeconomic char- acteristics of the community appeared to provide a reasonably representative mi- crocosm of the southern California popu- lation. A wide range of outdoor concentra- tions offers the best chance of determin- ing the contribution of outdoor levels to indoor levels and personal exposures. The fall season was selected since Santa Ana winds occur then; such winds can have strong effects on the outdoor concentra- tions of particles. The main goal of the study was to esti- mate the frequency distribution of expo- sures to PM10 particles for all nonsmoking Riverside residents aged 10 and above, based on a probability sample of 178 resi- dents. A second major objective was to estimate the frequency distribution of con- centrations of PM10 and PM25 in residences and nearby outdoor air (e.g., back yards). Other objectives included determining the effect of outdoor air on indoor concentra- tions, and the contribution of personal ac- tivities to exposure. A three-stage probability sampling pro- cedure was adopted. Thirty-six areas within Riverside were selected for study follow- ing socioeconomic stratification. Several homes from each area were sent letters explaining the study. Interviewers then col- lected information about each household and invited eligible residents to partici- pate. Respondents represented 139,000 ± 16,000 (S.E.) nonsmoking Riverside resi- dents aged 10 and above. Smokers were excluded from participat- ing, but nonsmoking members of their fam- ily were not. Employed persons were slightly oversampled, since employment was thought to be a possible risk factor for exposure to particles. Each participant wore the PEM for two consecutive 12-hour periods. Concurrent PM10 and PM samples were collected by the indoor SIM and outdoor SAM at each home. This resulted in 10 samples per household (day and night samples from the PEM10, SIM10, SIM25, SAM10, and SAM25). Air exchange rates were also cal- culated for each 12-hour period, using the perfluorotracer technique. Participants were asked to note activi- ties that might involve increased particle levels (nearby smoking, cooking, garden- ing, etc.). Following each of the two 12- hour monitoring periods, they answered an interviewer-administered recall ques- tionnaire concerning their activities and locations during that time. Up to four participants per day could be monitored, requiring 48 days in the field. A central outdoor site was maintained over the entire period (Sept. 22-Nov. 9, 1990). The site had two high-volume samplers (Wedding & Assoc.) with 10-|j.m inlets (ac- tual outpoint about 9.0 urn), two dichoto- mous PM10 and PM25 samplers (Sierra- Andersen) (actual outpoint about 9.5 |im), one PEM and one SAM. All filters were weighed on-site and then analyzed for elements by x-ray fluores- cence (XRF). An additional set of about 600 citric-acid treated filters from personal and indoor samplers was analyzed for nicotine. Filters were weighed before use and again within 48 hours of collection at an on-site weighing facility with controlled tem- perature and humidity. Replicate weighings were required to be within 4 ^g/filter. Blank filters were weighed, sent out with field samples, and reweighed along with the field samples. Duplicate indoor and out- door samples were collected at 10% of the homes. Duplicate SAM and PEM samples were also collected at the central . site. Duplicate PEM samples were also collected by EPA, RTI, and Harvard sci- entists while on site. Results Of 632 permanent residences contacted, 443 (70%) completed the screening inter- view. Of these, 257 were asked to partici- pate and 178 (69%) agreed. More than 2750 particle samples were collected, about 96% of those attempted. Quality of the Data Blank PEM and SIM/SAM filters (N = 51) showed consistent small increases in mass of 5-10 jig. Blanks (N = 9) placed near ungreased impactor plates had simi- lar increases of 7 jig. Blank dichot filters (N = 41) showed increases averaging 4 p.g. Blank SSI filters had increases of about 170 |ig. XRF analyses indicated that the increase was not due to aerosol; possibili- ties include water vapor or electric charge, although stringent efforts were made to control humidity and static charge in the on-site weigh room. The effect of the in- crease is small (0.4-4 jig/m3) and was corrected for by subtracting the mean blank value from all samples. Limits of detection (LODs), based on three times the stan- dard deviation of the blanks, were on the order of 10 ng/m3. All field samples ex- ceeded the LOD. Duplicate samples (N = 363) showed excellent precision for all types of sam- plers at all locations, with median relative standard deviations ranging from 2-4%. The collocated samplers at the central site showed good agreement, with corre- lations ranging from 0.96 to 0.99. As had been noted in the pilot study, the PEM and SAM collected about 12% more mass than the dichotomous samplers (Figure 1), perhaps due to their higher outpoint (11 jam compared to 9.5 |j.m) or to a par- ticle "bounce" effect, measured in the labo- ratory at less than 9%. The Wedding sam- plers collected about 13% less mass than the dichots at night, but about the same level during the day, reflecting a possible temperature dependency on the part of the Wedding. Although these small differ- ences were significant, they do not affect the main conclusions. All PEM, SIM, SAM, and dichotomous sampler filters (about 2500) were ana- lyzed by XRF for a suite of 42 elements. The analysis was carried out at EPA's Atmospheric Research and Exposure As- sessment Laboratory in Research Triangle Park, NC. Some filters were analyzed twice under blind conditions. A subset of about 100 filters was analyzed by the Lawrence Berkeley Laboratory (LBL) for quality as- surance purposes. Background levels on laboratory and field blanks were very low for 19 of 20 elements. Blank levels for iron were slightly higher but were 4 to 100 times lower than observed concentrations. Analyses of stan- dard reference materials (SRM 1832 and 1833) were within 7% of the correct val- ues for all 12 elements contained. Median relative standard deviations (RSD) for du- plicates analyzed blindly by the principal laboratory were less than 15% for all 15 prevalent (more than 30% of samples with measurable quantities) elements. Median RSDs for duplicates analyzed by the two laboratories were less than 21% for all elements except manganese (76%) and copper (27%). The LBL laboratory reported 10-20% higher average values for 13 of 14 elements. All filters analyzed by LBL had been first analyzed by EPA. ------- Comparison of Methods PEM-SAM vs. Dichot 250 100 Mean of Dichots (\ig/m 3) 150 200 Figure 1. The collocated PEM and SAM showed good precision but a positive bias with respect to the dichotomous samplers. Concentrations Outdoor 12-h PM10 concentrations at the central site ranged from 20-200 |ig/ m3, with the fine particles accounting for most of the variation (Figure 2). On the six windiest (16-20 mph) days, the coarse particles accounted for most of the PM10 mass. Population-weighted daytime personal PM10 concentrations averaged about 150 ng/m3, compared to concurrent indoor and outdoor mean concentrations of about 95 p.g/m3 (Table 1; Figure 3). The overnight personal PM10 mean was much lower (77 |j.g/m3) and more similar to the indoor (63 |ig/m3) and outdoor (86 jig/m3) means (Fig- ure 4). Approximately 25% of the popula- tion was estimated to have exceeded the 24-h National Ambient Air Quality Stan- dard for PM10 of 150 jig/m3. Over 90% of the population exceeded the California Ambient Air Quality Standard of 50 ng/m3. Fine (PM25) particles accounted for about 50% of the total PM10 mass both indoors and outdoors. The measurements at the central site showed good agreement with the outdoor measurements at homes throughout the City of Riverside (Figure 5), indicating that a single central-site PM10 monitor can char- acterize a large urban area adequately. Although the correlations of indoor air con- centrations with outdoor air are lower, there is evidence (Figure 6) that outdoor air PM10 concentrations can affect indoor air concentrations. Population-weighted mean elemental concentrations for 15 prevalent elements are provided in Table 2. As with the par- ticle mass, daytime personal exposures were consistently higher than either in- door or outdoor concentrations of all the elements save sulfur. At night, levels were similar in all three types of samples. The weighted mean element/particle mass ra- tios are provided in Table 3. The personal and indoor PM10 samples are depleted in the crustal elements (Si, Al, Fe) compared to the outdoor samples, by amounts rang- ing from 15 to 25%. The indoor PM25 samples show no depletion in any ele- ments and may be slightly enriched in Ca, K, Cl, and (night only) S. Models of Exposure Questionnaires were analyzed to detect activities associated with increased expo- sure. Housework (dusting, vacuuming, cooking) was associated with significantly increased personal exposures and indoor air concentrations during the day (Table 4). Sharing a home with one or more smokers also led to increased personal exposures and indoor air concentrations during the night. Persons who commuted to work had significantly lower exposures than those who stayed at home, perhaps due to the housework activities of the lat- ter group. Discussion Source of Excess Personal Exposure The source or sources of the roughly 50% increase in daytime personal expo- sure compared to the indoor and outdoor air concentrations remain unclear. Sev- eral possibilities include 1) The apparent increase is due to differ- ent sampling characteristics of the per- sonal monitor. 2) The increase is due to skin flakes or clothes fibers accumulating on the per- sonal monitor. 3) The increase is due to increased expo- sures encountered while participants are out of the house. 4) The increase is due to generation or reentrainment of particles during per- sonal activities. The first possibility has been tested in several ways. The only difference between the PEM and the SIM is the pump (Casella vs. Medo). Laboratory tests of the two pumps failed to show any difference in sampling characteristics on a test aerosol. Wind speed and direction were also tested and had little effect on either the PEM or the SIM. Particle bounce should affect both the PEM and SIM equally, since the sam- pling heads are identical. It remains pos- sible that the constant motion of the PEM may somehow affect its sampling charac- teristics compared to the fixed SIM. The second possibility was tested by scanning electron microscopy (SEM) on three sets of personal, indoor, and out- door filters. Although skin flakes were found in large numbers on one personal filter, their mass seemed insufficient to explain the mass difference. Also, if most of the increased mass were due to skin flakes or fibers, increases in elements other than carbon would not be expected; how- ever, 14 of 15 elements were also el- evated in the personal samples. The third possibility has been partially tested by comparing persons who went to work on the day of monitoring with those who did not. Even though their daytime exposures included round-trip commutes in Los Angeles County traffic, their expo- sures were significantly lower than those of participants who stayed at home. The fourth possibility seems likely. Per- sons engaging in activities such as vacu- uming, dusting, and cooking had signifi- cantly higher exposures than the other participants. House dust is a mixture of airborne outdoor aerosols, tracked-in soil and road dust, and aerosols produced by ------- Central Site: PM-10 and Coarse Particles 200 20 40 60 80 12-Hour period beginning Sept. 22, 1990 100 Figure 2. During the 48-day sampling period, two extended peaks characterized by elevated fine particles (PM-2.5) occurred. Coarse particles were elevated on days with high wind speeds. Table 1. Population-Weighted' Concentrations and Standard Errors (iig/m3) Sample Type Daytime PM10 Personal Indoor Outdoor Overnight PMW Personal Indoor Outdoor Daytime PM25 Indoor Outdoor Overnight PM2S Indoor Outdoor N 171 169 165 168 163 162 173 167 166 161 Median 130 ±8 82 ±8 83 ±5 66±4 52 ±4 74 ±4 34 ±4 36±4 26 ±2 35 ±2 Arithmetic Mean 150 ±9 95 ±6 94 ±6 77 ±4 63 ±3 87 ±4 48 ±4 49 ±3 36 ±2 51 ±4 Percentile 90% 260 ± 12 180 ±11 160 ± 7 140 ± 10 120 ± 5 170 ± 5 100 ± 7 100 ±6 83 ±6 120 ±5 98% 380 240 240 190 160 210 170 170 120 160 'Personal samples weighted to represent nonsmoking population of 139,000 Riverside residents aged 10 or above. Indoor-outdoor samples weighted to represent 61,500 homes with at least one nonsmoker aged 10 or above. indoor sources. As such, it should contain crustal elements from soil, lead and bro- mine from automobiles, and other elements from combustion sources. This would be consistent with the observation that nearly all elements were elevated in personal samples. The fact that personal overnight samples showed smaller mass increases than the personal daytime samples is also consistent with the fact that the partici- pants were sleeping for much of the 12- hour overnight monitoring period and were thus not engaging in these particle-gener- ating or reentraining activities. There re- mains the problem of sulfur, which showed no increase in personal samples compared to indoor or outdoor samples. This may be because sulfate ions have a much smaller mass median diameter and a lower deposition velocity than other ionic con- stituents of fine particles. Thus, sulfur would not tend to accumulate in house dust as much as other elements. Also, smaller particles may be harder to dis- lodge from surfaces, due to electrostatic or Van der Waals forces. Conclusions and Recommendations The personal and microenvironmental monitors designed especially for this study performed well. About 96% of all samples attempted were collected and median pre- cision was 2-4%. A positive bias of about 12% was noted with respect to the refer- ence dichotomous sampler method. The major finding of the study was the 50% increase in daytime personal expo- sures to PM10 compared to indoor and outdoor concentrations. The increase ap- pears to be due to personal activities such as dusting, vacuuming, cooking, and shar- ing a home with a smoker. This suggests that reduction of dust levels in the home could decrease exposure to airborne par- ticles. Future Publications Volume II of this three-volume series presents the results of measurements of polyaromatic hydrocarbons (PAH) and phthalates in 120 of the 178 homes in this study. Entitled "PTEAM: Monitoring of Phthalates and PAHs in Indoor and Out- door Air Samples in Riverside, Califor- nia—Volume II," it is available from the California Air Resources Board, Sacra- mento, CA. Volume III will present the results of additional statistical analyses and physi- cal modeling. It will also contain the re- sults of the nicotine analyses and the air exchange rate measurements. Volume III will be available in 1993 from the National Technical Information Service (NTIS). ------- Particle Levels: Daytime PTEAM Study: Riverside, CA 1000 co 100 10 _L 1000 Figure 3. Population- or household-weighted frequency distributions of 12-h average concentrations of PM-10 and PM-2.5 show a nearly log-normal shape for personal, indoor, and outdoor air. Daytime personal levels are 50% higher than concurrent indoor/outdoor levels. 100 25% 50 10 75 90 95 98 99% In 2.5 Pers 10 -O-- Out2.5 —•— In 10 —- Out 10 Particle Levels: Overnight PTEAM Study: Riverside, CA WOO 100 10 Figure 4. At night, personal concentrations sink to levels comparable with outdoor air. Indoor levels of both PM-10 and PM-2.5 also fall to 60-70% of outdoor levels. 25% 1000 100 50 Pers 10 In 2.5 75 —•— In 10 --O- Out 2.5 90 95 98 99% --- Out 10 10 ------- Outdoor PM-10: Homes vs. Central Site 250 200 150 100 50 0 50 The line is the 1:1 line. 100 150 SAM-10(\ig/m3) 200 250 Figure 5. Outdoor PM-10 levels near homes were also well characterized by the identical monitor at the central site. Indoor vs. Outdoor PM-10: Overnight 200 100 150 Outdoor (\ig/m 3) 200 250 Figure 6. Although there is considerable scatter due to indoor sources and activities, outdoor concentrations near the home appear to have considerable impact on indoor concentrations. ------- Table 2. Element Si Al Ca Fe Mn K Br Pb S Zn Cl Ti Cu Sr P Mean Elemental Concentrations (ng/m3) in Personal, Daytime PJ SAM 740. b 330. 400. 12. 230. 8.8 20. 1500. 41. 83. - - - - SIM 700. - 380. 340. 9.8 260. 9.1 17. 1300. 42. 130. - 11. - - SAM 7700. 3100. 2300. 2300. 51. 1100. 10. 30. 1800. 65. 230. 210. 15. 18. - PM10 SIM 6300. 2300. 2300. 1800. 38. 1100. 13. 27. 1700. 86. 410. 190. 22. 15. - Indoor, and Outdoor Samples' PEM 12000. 4700. 4300. 3400. 69. 1900. 25. 40. 1800. 150. 840. 390. 41. 25. 230. P/l SAM 380. - 170. 260. 9.9 150. 11. 23. 1600. 38. 170. - 9.6 - - J25 SIM 360. . 200. 200. 7.5 200. 8.6 20. 1300. 34. 100. . 8.9 . - Nighttime SAM 5000. 2000. 1500. 1700. 37. 800. 13. 32. 1900. 56. 500. 140. 17. 14. - PMW SIM 3300. 1200. 1200. 990. 22. 650. 11. 27. 1500. 60. 290. 100. 15. 9.8 - PEM 4200. 1400. 1700. 1200. 24. 800. 14. 26. 1500. 67. 440. 130. 19. 11. - "Results are weighted to reflect the target population of individuals (PEM samples) or households (SIM and SAM samples). "Fewer than 30% of samples with concentrations greater than the uncertainty limit. Table 3. Mean Ratios of Element/Particle Masses (%) in Personal, Indoor, and Outdoor Samples' Element Si Al Ca Fe Mn K Br Pb S Zn Cl Ti Cu Sr P P/l SAM 2.16 b 0.94 1.17 0.04 0.60 0.02 0.06 3.30 0.13 0.16 - - - - SIM 2.37 - 1.19 1.09 0.03 0.69 0.02 0.05 3.22 0.13 0.27 - 0.04 - - Daytime SAM 9.04 3.63 2.70 2.72 0.06 1.27 0.01 0.03 1.92 0.08 0.28 0.25 0.02 0.02 - PMW SIM 7.15 2.56 2.57 1.99 0.04 1.18 0.01 0.03 2.01 0.11 0.44 0.21 0.03 0.02 - PEM 7.57 2.88 2.85 2.14 0.04 1.22 0.02 0.03 1.36 0.11 0.58 0.25 0.03 0.02 0.15 Nighttime PM25 PM,0 SAM 1.39 - 0.58 0.83 0.03 0.47 0.02 0.06 3.62 0.12 0.40 - 0.03 - SIM 1.74 - 0.90 0.85 0.03 0.72 0.03 0.07 4.28 0.15 0.34 - 0.04 - SAM 7.02 2.78 2.12 2.25 0.05 1.09 0.02 0.04 2.23 0.07 0.68 0.19 0.02 0.02 SIM 5.97 2.08 2.14 1.71 0.04 1.11 0.02 0.04 2.73 0.11 0.52 0.18 0.03 0.02 PEM 5.72 1.87 2.31 1.56 0.03 1.09 0.02 0.03 2.16 0.10 0.61 0.18 0.03 0.02 "Results are weighted to reflect the target population of individuals (PEM samples) or households (SIM and SAM samples). Estimated means < 0 are reported as 0. "Fewer than 30% of samples with concentrations greater than the uncertainty limit. 'ftV.S. GOVERNMENT PUNTING OFFICE: 1993 • 7S047I/IMM07 ------- Table 4. Effects of Activities on Mean Personal and Indoor Air PM10 Concentrations Activity and Sample Type Housework (Day) Personal Indoor Smoking (Night) Personal Indoor Work (Day) Personal Homes/Persons With Activity N 110 111 29 30 59 Mean 162' 106' 104' 93' 127 (SE) (11) (8) (8) (9) (12) Homes/Persons Without Activity N 61 58 139 131 111 Mean 125 71 71 55 162' (SE) (11) (6) (3) (3) (10) "Both arithmetic mean (shown) and geometric mean significantly (p<0.05) higher than correspond- ing value for other group. E. D. Pellizzari, K. W. Thomas, C. A. Clayton, P. W. Whitmore, R. C. Shores, H. S. Zelon, and R. L. Peritt are with Research Triangle Institute, Research Triangle Park, NC 27709-2194 Lance A. Wallace is the EPA Task Manager (see below). The complete report, entitled "Particle Total Exposure Assessment Methodology (PTEAM): Riverside, California Pilot Study-Volume I" (Order No. PB93-166 957/ AS; Cost: $44.50, subject to change) will be available only from: National Technical Information Service 5285 Port Royal Road Springfield, VA 22161 [ Telephone: 703-487-4650 The EPA Task Manager can be contacted at: Atmospheric Research and Exposure Assessment Laboratory U.S. Environmental Protection Agency Research Triangle Park, NC 27701 United States Environmental Protection Agency Center for Environmental Research Information Cincinnati, OH 45268 BULK RATE POSTAGE & FEES PAID EPA PERMIT No. G-35 Official Business Penalty for Private Use $300 EPA/600/SR-93/050 ------- |