Relationships between personal, indoor, and outdoor exposures to trace elements in PM(2.5).

Twenty-four hour average fine particle concentrations of 23 trace elements (TEs) were measured concurrently in (a) ambient air in three urban neighborhoods (Battle Creek-BCK; East St. Paul-ESP; and Phillips-PHI), (b) air inside residences of participants, and (c) personal air near the breathing zone of healthy, non-smoking adults. The outdoor (O), indoor (I), and personal (P) samples were collected in the Minneapolis/St. Paul metropolitan area over three seasons (Spring, Summer, Fall) using either the federal reference (O) or inertial impactor (I,P) inlets to collect PM(2.5). In addition to descriptive statistics, a hierarchical, mixed-effects statistical model was used to estimate the mutually adjusted effects of monitor location, community, and season on mean differences between monitoring locations while accounting for within-subject and within-monitoring period correlation. The relationships among P, I, and O concentrations varied across TEs. The O concentrations were usually higher than P or I for elements like Ca and Al that originate mainly from entrained crustal material, while P concentrations were often highest for other elements with non-crustal sources. Unadjusted mixed model results demonstrated that O monitors more frequently underestimated than overestimated P TE exposures for elements associated with non-crustal sources. This finding was true even though the O TE measurements were taken in the same neighborhoods as the P and I measurements. Further adjustment for community or season effects in the mixed models reduced the number of significant O-P and O-I differences compared to unadjusted models, but still indicated a tendency for underestimation of personal and indoor TE exposures by central site monitors, particularly in the PHI community. These results indicate that community and season are important covariates for developing long term TE exposure estimates, and that personal exposure to trace elements in PM(2.5) is likely to be underestimated by outdoor central site monitors.