Spatial and temporal variability in outdoor, indoor, and personal PM2.5 exposure

Outdoor, indoor and personal PM2.5 measurements were made in a population of nonsmoking adults from three communities in the Minneapolis-St. Paul metropolitan area between April and November 1999. Thirty-two healthy adult subjects (23 females, 9 males; mean age 42 +/- 10, range: 24-64 yr) were monitored for 2-15 days during the spring, summer, and fall monitoring seasons. Twenty-four hour average gravimetric PM2.5 samples were collected using a federal reference monitor (Anderson RAAS2.5-300) located at outdoor (O) central sites in the Battle Creek (BCK), East St. Paul (ESP) and Phillips (PHI) communities. Concurrent 24-h average indoor (1) and personal (P), and a limited number of outdoor-at-home (O@H) samples were collected using inertial impactors (PEM(TM) Model 200, MSP, Inc). The O (geometric mean {GM} = 8.6; n = 271; range: 1.0-41 mug/m(3)) were lower than I concentrations (GM = 10.7; n = 294; range 1.3-131 mug/m(3)), which were lower than P concentrations (GM =19.0; n = 332; range 2.2-298 mug/m(3)). Correlation coefficients between O concentrations in the three communities were high and measured GM O levels in BCK were significantly lower than ESP, most likely because of local sources, but GM concentrations in PHI were not significantly different from BCK or ESP. On days with paired samples (n = 29), O concentrations were significantly lower (mean difference 2.9 mug/m(3); p = 0.026) than O@H measurements (GM =11.3; range: 3.5-33.8 mug/m(3)), likely due to local sources in communities. Observed I and P concentrations were more variable, probably because of residential central air conditioning and hours of household ventilation for I and P, and occupational and environmental tobacco smoke exposures outside the residence for P. Across all individuals and days the median PM2.5 personal cloud was 5.7mug/m(3), but the mean of the average for each participant was 15.7 mug/m(3), with very low values in participants who did not work outside the home and much higher values in subjects with active lifestyles. Across all households and individuals the correlation between P and O concentrations was not significant, but the overall I-O correlation (0.27) and P-I correlation (0.51) were significant (p<0.05). Relatively little spatial variability was observed in O PM2.5 concentrations across the three communities compared to the variability associated with I and P samples, and the measured O levels were relatively low compared to other large metropolitan areas in the United States. (C) 2002 Elsevier Science Ltd. All rights reserved.