Environ. Sci. Technol. 2009, 43, 4641–4646
The Effects of Human Activities on Exposure to Particulate Matter and Bioaerosols in Residential Homes QING CHEN† AND LYNN M. HILDEMANN* Stanford University, Civil and Environmental Engineering Dept., Stanford, California 94305
Received August 19, 2008. Revised manuscript received February 26, 2009. Accepted March 9, 2009.
Indoor and outdoor airborne particle mass, protein, endotoxin and (1f3)-β-D-glucan in three size fractions (PM2.5, PM10, and TSP) were measured in ten single-family homes, along with quantifying household activities in the sampling room. Correlations between human activity levels and elevations in the indoor concentrations of particles and biomarkers were evaluated using four approaches for distinguishing activity levels: diurnal differences, the number of occupants, selfestimated occupancy, and activity strength. The concentrations of particles, protein, endotoxin and (1f3)-β-D-glucan in all three size fractions (PM 10 µm) were found, in most cases, to be significantly elevated during the day, and with higher activity levels in the room. The coarser fractions of particle mass and bioaerosols were more strongly correlated with human activity levels. Activity strength was the most statistically robust measure for relating human activities to indoor bioaerosol levels. While selfestimated activity and analysis of diurnal differences both offer reasonable (but not perfect) alternatives to activity strength, the number of occupants appears to be a weaker indicator for homes.
Introduction Airborne particulate matter (PM) in indoor environments includes constituents which are biological. These bioaerosols are of concern because they have the potential to cause allergies and other human health impacts (e.g., ref 1). Many previous studies have compared bioaerosol levels indoors and outdoors (e.g., refs 2 and 3), and a few studies have evaluated associations of indoor bioaerosols with factors such as pets, dampness, relative humidity and temperature (3-6). Some studies have evaluated the effects of occupant activities on PM emission rates, indoor concentrations, and size distributions (7-11). Only coarser (>1 µm) PM mass concentrations were found to measurably increase with the time occupants spent indoors (9, 12, 13). However, the influence of human activity on bioaerosol exposures has not received much attention. Still, there is evidence that human activity influences indoor bioaerosol levels. Lehtonen et al. (14) found that some residential activities (e.g., sweeping floors, bed making) had a pronounced effect on culturable fungi counts, while other activities (e.g., vacuuming, baking) did not. * Corresponding author phone: 650-723-0819; Fax: 650-725-3164; e-mail:
[email protected]. † Current address: Washington State Department of Ecology, Air Quality Program, Olympia, WA. 10.1021/es802296j CCC: $40.75
Published on Web 04/06/2009
2009 American Chemical Society
A few researchers have evaluated the contributions of human activities to indoor culturable bioaerosol concentrations. A study (15) of two hospitals reported that bacterial counts tracked closely with foot traffic and with extent of physical activity. In contrast, a study of a middle school (16) found little correlation between person-minutes of activity (during 20 min before sampling) and subsequent 2-min airborne bacterial and fungal samples. One study (17, 18) used a more detailed method of quantifying human activity, the hours per day spent in a living room by up to four persons, quantified as personhours per week (person-hrs/wk). These researchers found higher occupancy levels were associated with higher endotoxin and (1f3)-β-D-glucan concentrations in house dust. No previous studies, to our knowledge, have compared person-hour estimates with simultaneous airborne microbial levels indoors. While studies of indoor bioaerosols have reported number of occupants (3, 15, 16, 19-22), none evaluated contributions from human activities. No published studies have compared human activity levels with nonculture-based bioaerosol measurements, or with bioaerosol size distributions. Traditional culturing methods allow sampling within minutes, but replicate samples show high variability; in addition, culturing requires substantial time for growth and counting, and results in selective underestimation of bioaerosols (23-25). Due to these concerns, our study instead utilized highly sensitive, reproducible chemical markers for measuring bioaerosols, with samples collected over a period of hours. This study measured different sizes (PM2.5, PM10, and TSP) of airborne protein (existing in all biologic materials), endotoxin (the outer membrane portion of gram-negative bacteria), and (1f3)-β-D-glucan (glucose polymers in the cell walls of most fungi) in 10 single-family homes (26). We compared two common approaches for measuring human activity, the number of occupants and the diurnal differences, with two more sophisticated approaches achievable with our data, self-estimated occupancy and activity strengths. Finally, we assessed the effects of human activity levels on the sizes and concentrations of indoor PM and bioaerosols.
Materials and Methods Study Homes. The methodology for selection of study homes has been described in detail elsewhere (26). Briefly, living room dust samples and surveys were taken in 20 homes in the Bay Area of northern California. The more intensive sampling (between Oct. 2005 and May 2006) involved a subset of 10 homes chosen to create a balanced sample set for living room dust loading, house characteristics, and occupant lifestyles (26). Occupant density ranged from 3 to 8 per household, with the majority at 4 to 5 per household. Activity and Household Information. The living or family room where residents spent the most time served as the sampling room. Neither the ventilation nor the daily activities were scripted. Instead, normal activities in the sampling room during each sampling period were logged by the occupants, including the type (e.g., walking, reading) and duration, and the number of people involved. This was converted to a quantitative “activity strength”: the sum over all activities occurring during the sampling period of (duration × number of persons involved) was divided by the total sampling time period. Households reported times when windows were opened. In most cases, sampling room windows were kept closed during visits. There were three samples collected (from two VOL. 43, NO. 13, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4641
different homes) that included a small portion (∼1 h) of the total sampling time with a sampling room window open. No tobacco smoking occurred in any of the households; cooking activities were not logged since they did not take place in the sampling room. Collection of Particulate Samples. The sampling system has been described elsewhere (26). Briefly, airborne TSP, PM10, and PM2.5 samples were collected in duplicate from each home, using aluminum filter holders (Pall Corp., East Hills, NY), with size selective cyclones. Two identical setups, one with prebaked (300 °C overnight) tissue quartz fiber filters (47 mm, Pall Corp., East Hills, NY) for endotoxin and (1f3)β-D-glucan, and another with Teflon filters with polypropylene rings (2 µm, 47 mm, Pall Corp., East Hills, NY) for protein and gravimetric assays, were placed ∼1.2 m above ground, in the middle of the room. The first measurement series consisted of one night (typically ∼9 p.m. to ∼8 a.m.) and two daytime samples (typically, ∼10 a.m. to ∼7 p.m.) with a second set of two daytime samples ∼3 weeks later. The duration for each sample (9-12 h) was recorded. Sample Analysis. Teflon filters were weighed with a microbalance (MT-3, Mettler Instrument Corp., Highstown, NJ) before and after sampling. Before weighing, each filter was equilibrated at 72 ( 3 °F and 55 ( 5% relative humidity for g24 h, and exposed to 210Po R-sources to minimize static charge. Proteins were extracted from Teflon filters in a lysis buffer (2% sodium dodecyl sulfate (SDS), 0.4 mM ethylenediamine tetraacetic acid (EDTA), 10 mM Tris-HCl, pH 7.5) and analyzed by a detergent compatible micro bicinchoninic acid (BCA) method (Micro BCA kit, Pierce Biotechnology, Rockford, IL). Endotoxin was extracted from quartz filters in 10 mL pyrogen-free water; a small aliquot of supernatant was analyzed for endotoxin. Enough NaOH was added into the remaining extract to achieve 0.3 N for (1f3)-β-D-glucan extraction at 4 °C. Endotoxin and (1f3)-β-D-glucan were quantified using a kinetic chromogenic limulus Amebocyte lysate (LAL) method (Pyrochrome Kit, Associates of Cape Cod, East Falmouth, MA) and a modified LAL method (Glucatell Kit, Associates of Cape Cod, East Falmouth, MA), respectively, with a VersaMax microplate reader (Molecular Device Corp., Sunnyvale, CA). The limits of detection (LOD), propagated errors, and estimated precision for these analyses are described in detail elsewhere (26). Briefly, out of a total of 480 samples per biomarker, levels were below detection for 14 protein analyses. When calculating PMTSP-10 and PM10-2.5, levels were less than half the propagated errors for 27 protein, 6 endotoxin, and 3 (1f3)-β-D-glucan values. These values were replaced with values corresponding to either half the LOD or half the propagated error, as appropriate. The biological measurement methods we chose offer rapid, inexpensive, and sensitive analyses, enhanced stability of the biomarkers allowing for longer sampling times, and the ability to measure total (viable and nonviable) bioaerosol levels. However, there are also some limitations. First, these methods do not allow genus or species identification and they do not differentiate between viable and nonviable bioaerosols. Second, not all species of microorganisms will have the same amount of biomarker per cell, or per unit mass. Thus, higher levels of biomarkers do not necessarily correlate with higher counts or mass concentrations of microorganisms. Third, the biomarker specificity varies. Protein is not limited to microorganisms, but also is found in pollen, animal dander, and other organisms. Besides being in most fungi, (1f3)-β-D-glucan is also present in some bacteria and most plants. Endotoxin is only present in gramnegative bacteria (GNB); it does not include gram-positive bacteria. However, GNB are of greater health concern. Moreover, exposure to endotoxin and (1f3)-β-D-glucan may 4642
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 13, 2009
be associated with health effects such as respiratory symptoms and inflammatory responses (23, 28, 29). Statistical Analysis. Statistical analyses were performed using S-Plus statistical software (Student version 7.0; Insightful Corp., Seattle, WA). Distributions of the PM, protein, endotoxin, and (1f3)-β-D-glucan size fractions were much closer to log-normal than normal (Wilks-Shapiro normality test), so geometric means (GMs) and geometric standard deviations (GSDs) were used to summarize the data. However, a few of the data subsets were not log-normal, and our numbers per group were small, so nonparametric analyses (e.g., Wilcoxon rank-sum test, Spearman correlation) were utilized for evaluating correlations. Comparing between biomarkers (e.g., using multiple repeated measures analysis of variance) was judged to be beyond the scope of this paper. The between-home GSD of activity strengths was calculated by averaging log-transformed measurements for all sampling periods for each home, and computing the antilog of the standard deviation (SD) of the average for the 10 homes. The within-home GSD of activity strengths was calculated by taking the SD for log-transformed measurements for all sampling periods for each home and computing the antilog of the average SD for the 10 homes. This mathematical approach is derived from approaches reported by Chew et al. (30). For diurnal variability, differences in concentrations across daytime and nighttime for the first 24 h were evaluated via the Wilcoxon signed-rank test. Associations between daytime geometric means (GMs) and number of occupants were examined by Spearman correlation coefficients and also by classifying houses into two groups. The self-estimated use rate was also compared with GM levels using Spearman correlation coefficients. Both the correlations between activity strength and bioaerosol levels and those between activity strength and the indoor-to-outdoor (I/O) ratios of bioaerosols were evaluated by Spearman correlation tests. The data were also grouped into higher vs lower activity levels, with differences assessed by the Wilcoxon rank-sum test.
Results and Discussion Diurnal Variability. Details regarding the ranges of indoor and outdoor mass and biomarker concentrations measured in this 10-home cohort can be found elsewhere (26). One simple approach for assessing the effects of human activities on indoor pollutants is to compare daytime concentrations, when activity levels should generally be higher, with nighttime concentrations. We examined differences in indoor concentrations between daytime and nighttime for the first 24 h, as well as simultaneous outdoor differences. Indoor diurnal variability was further evaluated using I/O ratios to normalize for diurnal variations in outdoor contributions. GMs are reported in Table 1, along with p-values (Wilcoxon signed-rank test). The most notable diurnal differences (Table 1) were for the indoor particle components between 2.5 and 10 µm in diameter (PM10-2.5): for particle mass and the three biomarkers, daytime indoor concentrations were significantly higher (p < 0.05) than at night. This was true even though outdoor levels of PM10-2.5 endotoxin and (1f3)-β-D-glucan were significantly higher at night (p ) 0.06 and 0.05, respectively). For all four components, the PM10-2.5 I/O ratios were higher in the daytime than at night; whether day or night, most (7/8) ratios were g1.0. Together, these observations indicate there were substantial indoor sources of PM10-2.5 mass and biomarkers, and that indoor sources were stronger in the day. Three other indoor fractions showed significant diurnal differences in Table 1 (p < 0.05): the coarse particle and (1f3)β-D-glucan concentrations (>10 µm), and the fine protein
TABLE 1. Nonparametric Statistical Comparison of Daytime and Nighttime Airborne Concentrations, Indoors and Outdoors, Showing Geometric Mean Concentrations sizes Cindoor (n ) 10)
10 µm
Coutdoor (n ) 9)
a
10 µm
I/O (n ) 9)a
10 µm
periods
particle (µg/m3)
protein (µg/m3)
endotoxin (Eu/m3)
glucan (ng/m3)
night day night day night day night day night day night day night day night day night day
6.7 8.6b 4.8 7.8b 4.3 7.7b 8.5 7.4 4.4 6.4b 4.2 9.7b 0.8 1.1b 1.1 1.2 1.0 0.8
0.9b 0.7 0.3 0.6b 0.3 0.4 1.4b 0.6 0.2 0.2 0.2 0.3 0.7 1.2 1.5 2.5 1.4 1.1
0.4 0.5 0.9 1.4b 0.5 0.9 0.4b 0.2 0.8b 0.3 0.3 0.7 1.2 1.9 1.0 3.9b 1.4 1.3
0.4 0.5 1.0 1.4b 0.8 1.5b 0.6 0.5 1.8b 0.7 1.5 4.2b 0.7 0.8 0.5 1.9b 0.5 0.4
a Comparisons between night and day samples for first 24 h; one nonpaired outdoor data set was removed. b Marks each group (day or night) of 9-10 home measurements found to be significantly higher than the group for the other time period, based on the Wilcoxon rank sum test: bold-faced values indicate a significance level of p < 0.05, and underlined values indicate 0.05 e p < 0.10. Geometric mean values for each group are provided for context.
TABLE 2. Housing Characteristics and Activity Levels of Study Cohort
home ID
home age (yrs)
S04 S05 S07 S11 S14 S17 S19 S21 S22 S24 geometric mean
49 15 16 49 49 5 58 20 51 35
a
number of residents
self-estimated occupancy (person-hrs/wk)b
daytime activity strength (n ) 4 days), GM (Range) (person-hrs/hr)
nighttime activity strength (n ) 1 night) (person-hrs/hr)
4 8 5 4 5 3 6 4 4 4 4.5
106 213 63 165 208 44 124 193 40 126 110
0.22 (0.15-0.30) 0.53 (0.40-0.86) 0.07 (0.03-0.15) 0.81 (0.71-1.13) 1.39 (0.94-2.63) 0.07 (0.03-0.16) 0.55 (0.46-0.65) 0.22 (0.18-0.28)c 0.17 (0.02-0.54) 1.49 (1.04-1.96) 0.37
0.16 0.43 0.11 0.23 0.30 0.17 0.53 N/Ac 0.47 0.48 0.28
a Age in years as of 2005. b Person-hrs/wk ) 5 × ∑i n) 1 hri + 2 × ∑j n) hours in living or family room. c 1 night and 2 day records were lost.
( 0.60, Table 3) with the GMs for PM2.5 mass, and for PM10-2.5 endotoxin and (1f3)-β-D-glucan. For nighttime, no significant correlations for any measures were found (r ) -0.31-0.49, p > 0.10, data not shown). Activity Strength and Indoor Levels. Strong, significant (r > 0.6, p < 0.05) correlations were found between GM activity strengths and daytime GM PM10-2.5 and PMTSP-10 levels for all 4644
9
TABLE 4. Spearman Correlation Coefficientsa For Airborne Particle and Biomarker Levels (Day and Night Combined) in Various Size Fractions Compared with Two Activity Measures: Activity Strength and Weighted Activity Strength
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 13, 2009
sizes
10 µm weighted activity 10 µm activity strength
particle protein endotoxin glucan 0.52 0.65 0.81 0.54 0.66 0.84
0.44 0.65 0.67 0.41 0.64 0.67
0.52 0.70 0.54 0.54 0.73 0.56
0.26b 0.50 0.47 0.29 0.54 0.48
a Bold-face: strong association (r g 0.6); underlined: moderate association (0.4 e r < 0.6); four sets of data were removed (three due to lost activity strength data logs, and one due to a microwave fire). b All p-values are p < 0.05 except for activity strength vs glucan 0.05), except for particles in the PM10-2.5 (r ) 0.88, p ) 0.01), and PMTSP-10 (r ) 0.53, p ) 0.14) fractions, and fine protein (r ) 0.53, p ) 0.14). The latter two correlations, while stronger, were not statistically significant. Activity strengths captured more than twice as many strong daytime correlations with PM mass and bioaerosols as were found using self-estimated occupancy (Table 3). The poor correlation of airborne levels with human activity measures at night indicates that other sources obscure the limited nighttime contributions from human activities. As shown in the upper half of Table 4, activity strength was at least moderately (r > 0.4) and significantly (p < 0.05) correlated with all size-segregated measures except fine (1f3)-β-D-glucan. The strongest correlations (r > 0.6) were for PM10-2.5 and PMTSP-10 mass and protein, and for PM10-2.5 endotoxin. A limitation in our calculation of activity strength is that types of activities are not weighted to reflect differences in vigor - some activities should have more pronounced effects on indoor bioaerosol levels than others. Self-reporting of activities does not allow an accurate assessment of this; however, as a preliminary test, we separated nonsitting activities (e.g., walking, folding laundry) from sitting activities (e.g., watching television). A weighting factor of unity was assigned to nonsitting activities, while one-half was arbitrarily chosen for sitting activities. As shown in the lower half of Table 4, correlation coefficients increased (albeit only slightly) for all size fractions of PM mass, endotoxin, and (1f3)-β-D-glucan (but not for protein) when weighted activity strengths were used. It may prove valuable, in the future,
FIGURE 1. Indoor to outdoor ratio for size fractionated PM mass, protein, endotoxin and glucan versus activity strength (n ) 24, four points were omitted due to nonparallel samples and two points due to missing activity strengths); Spearman correlations (r) between the ratios and activity strengths for each size fraction represent all nighttime and daytime data pooled together: **p < 0.05, significant correlated; * 0.05 e p < 0.10, marginally significantly correlated. Red dashed lines mark an indoor to outdoor ratio of 1.0.
TABLE 6. Geometric Means of Indoor/Outdoor Ratios (I/O) for Particle Mass, Protein, Endotoxin and (1f3)-β-D-Glucan in Three Size Fractions, Sorted by Two Activity Strength Levels (In Person Hrs/Hr) PM mass a
10 µm
protein b
person-hrs/hr
I/O
p
0.40 (n ) 13) 0.40 (n ) 13) 0.40 (n ) 13)
0.71 1.33 0.80 1.76 0.64 1.32
0.02 0.003 0.02
endotoxin b
I/O
p
0.76 1.76 1.47 2.98 1.21 2.49
0.05 0.02 0.09
glucan b
I/O
p
1.37 2.54 1.22 5.99 1.36 1.35
0.04 0.001 0.59
I/O
pb
0.88 1.38 0.67 1.62 0.44 0.55
0.21 0.04 0.41
a There were no data points between 0.30 and 0.40 person-hrs/hr. b p-value for difference between groups, based on Wilcoxon rank-sum test; Values are bold-faced when p < 0.05, and underlined when 0.05 e p < 0.10.
to investigate how to more accurately weight the contributions from each type of activity. We also grouped homes based on nonweighted activity strengths, using an arbitrary dividing line of 0.30 personhrs/hr. The GMs of PM mass, protein, endotoxin and (1f3)β-D-glucan (shown in Table 5) are significantly less across all sizes for the lower activity strength group (p < 0.05, Wilcoxon rank-sum test), except for PM2.5 (1f3)-β-D-glucan (p ) 0.18). For the higher activity strength group, the GMs of mass, protein, endotoxin and (1f3)-β-D-glucan were ∼2.5-3.5 × as high in the two coarser fractions, but only ∼1.2-2.4 × as high in the fine fraction. An important effect of human activity is particle resuspension from surfaces, which has been reported to be much greater for particles >1 µm (8, 13). Thus, the effect of activity strengths on both the levels and the sizes of PM mass and bioaerosols in residences is pronounced, even though these data average over a 9-12 h period. Additional sources associated with occupants include shedding of skin cells and clothing fibers, outdoor particles transported in with people, and release from surfaces due to mechanical disturbances. For example, Gorny et al. (33) found that vibration, at a level typical for children jumping or a door closing, increased the release of fungal particles. All these sources would be expected to increase with activity strength. Activity Strengths and I/O Ratios. Pooling day and night
data together, I/O ratios for particle, protein, endotoxin and (1f3)-β-D-glucan in three fractions were plotted against activity strengths, and Spearman correlations were assessed (Figure 1). Six data sets were removed because they were nonparallel ratios, or lacked activity log information. Most correlations with activity strength decreased in strength and/ or significance when using I/O ratios (Figure 1) instead of indoor concentrations (Table 4). Overall, a sizable portion of the I/O ratios were >1.0, especially for larger activity strengths. Another home study (34) found lower I/O ratios for PM2.5 and PM10 mass when inhabitants were absent (means: 0.54 and 0.67, respectively) than present (means: 1.23 and 1.40). The GM I/O ratios for PM2.5 and PM10 mass in our lower vs higher activity strength groups (0.71 and 0.75 vs 1.33 and 1.52, respectively) were similar. And the GM I/O ratios were significantly less in the lower activity strength group (p < 0.05, Wilcoxon rank-sum test, Table 6) for all size fractions of particle mass, PM2.5 and PM10-2.5 protein and endotoxin, and PM10-2.5 (1f3)-β-D-glucan. For the lower activity group, for all size fractions, GM I/O ratios for mass and for (1f3)-β-D-glucan were 1.0. One possible explanation for the higher indoor endotoxin I/O ratios is that humans are believed to be an important bacterial source. VOL. 43, NO. 13, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4645
Thus, the presence of occupants indoors, active or not, will increase the endotoxin levels. The biomarker I/O ratios tended to cluster together (Figure 1) when the activity strength is low (0.4 person-hrs/hr, the range of biomarker I/O ratios greatly increased. This suggests that only above a certain critical value can the activity strength become a dominant factor influencing the I/O ratio. Implications for Exposure. Activity strength, which accounts for the duration of activities as well as the number of occupants, is strongly correlated with the coarser fractions (PM10-2.5 and PMTSP-10) of daytime PM mass and bioaerosols inside homes. However, the daytime biomarker I/O ratios were 2-4 times as high for PM10-2.5 as for PMTSP-10. A small portion of this elevation in the I/O ratios for PM10-2.5 can be attributed to greater penetration efficiencies from outdoors for particles 3)-glucan in house dust of German homes: Housing characteristics, occupant behavior, and relations with endotoxins, allergens, and molds. Environ. Health Perspect. 2001, 109, 139–144. (18) Bischof, W.; Koch, A.; Gehring, U.; Fahlbusch, B.; Wichmann, H. E.; Heinrich, J. Predictors of high endotoxin concentrations in the settled dust of German homes. Indoor Air 2002, 12, 2–9. (19) Goh, I.; Obbard, J. P.; Viswanathan, S.; Huang, Y. Airborne bacteria and fungal spores in the indoor environmentsA case study in Singapore. Acta Biotechnol. 2000, 20, 67–73. (20) Wu, P. C.; Li, Y. Y.; Chiang, C. M.; Huang, C. Y.; Lee, C. C.; Li, F. C.; Su, H. J. Changing microbial concentrations are associated with ventilation performance in Taiwan’s air-conditioned office buildings. Indoor Air 2005, 15, 19–26. (21) Kalogerakis, N.; Paschali, D.; Lekaditis, V.; Pantidou, A.; Eleftheriadis, K.; Lazaridis, M. Indoor air qualitysbioaerosol measurements in domestic and office premises. J. Aerosol Sci. 2005, 36, 751–761. (22) Fox, A.; Harley, W.; Feigley, C.; Salzberg, D.; Sebastian, A.; Larsson, L. Increased levels of bacterial markers and CO2 in occupied school rooms. J. Environ. Monit. 2003, 5, 246–252. (23) Douwes, J.; Thorne, P.; Pearce, N.; Heederik, D. Bioaerosol health effects and exposure assessment: Progress and prospects. Ann. Occup. Hyg. 2003, 47, 187–200. (24) Palmgren, U.; Strom, G.; Malmberg, P.; Blomquist, G. The Nuclepore filter methodsa technique for enumeration of viable and nonviable airborne microorganisms. Am. J. Ind. Med. 1986, 10, 325–327. (25) Niemeier, R. T.; Sivasubramani, S. K.; Reponen, T.; Grinshpun, S. A. Assessment of fungal contamination in moldy homes: Comparison of different methods. J. Occup. Environ. Hyg. 2006, 3, 262–273. (26) Chen, Q.; Hildemann, L. M. Size-Resolved Concentrations of Particulate Matter and Bioaerosols Inside versus Outside of Homes. Aerosol Sci. Technol., 2009, in press. (27) Chen, Q. Biological Composition of House Dust: Relation to Home Characteristics, Occupants, and Airborne Levels. Chapter 6, Ph.D Thesis, Stanford University, 2007. (28) Douwes, J. (1->3)-beta-D-glucans and respiratory health: a review of the scientific evidence. Indoor Air 2005, 15, 160–169. (29) Rylander, R.; Norrhall, M.; Engdahl, U.; Tunsater, A.; Holt, P. G. Airways inflammation, atopy, and (1->3)-beta-D-glucan exposures in two schools. Am. J. Resp. Crit. Care 1998, 158, 1685–1687. (30) Chew, G. L.; Douwes, J.; Doekes, G.; Higgins, K. M.; van Strien, R.; Spithoven, J.; Brunekreef, B. Fungal extracellular polysaccharides, beta (1->3)-glucans and culturable fungi in repeated sampling of house dust. Indoor Air 2001, 11, 171–178. (31) Koutrakis, P.; Briggs, S. L. K.; Leaderer, B. P. Source apportionment of indoor aerosols in Suffolk and Onondaga counties, New-York. Environ. Sci. Technol. 1992, 26, 521–527. (32) Ozkaynak, H.; Xue, J.; Spengler, J.; Wallace, L.; Pellizzari, E.; Jenkins, P. Personal exposure to airborne particles and metals: Results from the particle team study in Riverside, California. J. Expos. Anal. Environ. Epidemiol. 1996, 6, 57–78. (33) Go´rny, R. L.; Reponen, T.; Willeke, K.; Schmechel, D.; Robine, E.; Boissier, M.; Grinshpun, S. A. Fungal fragments as indoor air biocontaminants. Appl. Environ. Microbiol. 2002, 68, 3522–3531. (34) Monn, C.; Fuchs, A.; Hogger, D.; Junker, M.; Kogelschatz, D.; Roth, N.; Wanner, H. U. Particulate matter less than 10 µm (PM10) and fine particles less than 2.5 µm (PM2.5): relationships between indoor, outdoor and personal concentrations. Sci. Total Environ. 1997, 208, 15–21.
ES802296J
