Breathing Zone Exposure Assessment - Aerosols: Measurement, Dosimetry, and Health Effects - page 34

that contribute to daily nonoccupational personal exposures, in both outdoor locations and indoor
microenvironments. They also reported that a signiicant portion of the personal aerosol exposures
occurred in locations where the participants spent only 4%-13% of their time. Ott (2007) examined
the dispersion of a gas tracer indoors in a private residence and noted a strong proximity effect
(concentration decrease with distance) at distances up to 2 m from the source. Room background
levels were typically measured within experimental error beyond 2 m from the source. They also
reported that the room air exchange rate had little effect on the proximity effect within 0.5 m of the
source. These data suggest that the room average MEM would need to be signiicantly closer to the
PEM than 2 m in order to relect the elevated BZ levels in close proximity to a source. Rodes et al.
(1995) also reported that the velocity proile near the body decreases sharply, deining a low-velocity
boundary layer adjacent to the body only a few centimeters in thickness. Since the typical BZ inlet
typically extends only a few centimeters from the body, particles will require a minimum size and
inertia to penetrate across the low streamlines to the inlet to be sampled. This bias is discussed
subsequently in more detail in Section 2.2.3.
It has become increasingly clear (e.g., Rodes et al., 2010) that the strengths of epidemiologic
investigations are strongly inluenced by the ability to identify and characterize the most exposed
in either a residential or occupational cohort. The exposure data from this segment of the cohort
deine the range of the exposure distributions and are heavily responsible for obtaining the strongest
possible epidemiologic (cause and effect) associations, as well as deining the strongest relative risks
when comparing exposed and unexposed populations. Additionally, it is now evident (Edwards
and Jantunen, 2009) that the chemical compositions of exposure samples for the most exposed
can be dramatically different from those deining the median or IQR. Rodes et al. (2010) note that
minimally biased and confounded BZ personal exposures are the most representative of the true
levels that form the “gold standard” against which to achieve the most representative epidemio-
logic indings. But technologies to conduct exposure studies applying minimally burdensome and
cost approaches have historically not been available (Weis et al., 2005). The National Institute for
Environmental Health Sciences at NIH initiated the Genes and Environment Initiative to develop
better personalized exposure methodologies to ill this gap (Schmidt, 2006). These new approaches
will tap into new sensor technologies that can be applied at the personal level to provide less biased
and confounded exposure data for both the IQR and the most exposed from which to provide the
strongest predictions of adverse biological and health responses.
2.2.2 e xPosure b iases versus c onFounding
2.2.2.1 Biases, Including Representativeness
The NRC (2004) recognized that the single most important misclassiication bias to understand
and minimize when attempting to characterize human exposures was that imposed when surrogate,
ixed-location data were used to estimate BZ exposures. This aspect was discussed and reviewed
by Sarnat et al. (2007) examining the impacts of traditional precision and accuracy type param-
eters, as well as biases from interpreting total particulate exposures, when the subset of exposures
attributed to a single source category such as ambient aerosol was more appropriate. They observed
that “…the use of (BZ) personal exposure measurements can lead to different interpretations than
those derived from the use of ambient (ixed-location) concentrations.…” They strongly recom-
mended that where possible, the contributions of source categories (e.g., aerosol of ambient origin)
be taken into account to help minimize the biases between surrogate and BZ exposure assessments.
Rabinovitch et al. (2005) noted that personal BZ exposures to the known inlammatory agent endo-
toxin contained in residential particulate matter were found to be statistically associated with sever-
ity levels in asthmatic children, while concurrent indoor and outdoor levels showed no associations.
They attributed the inding to the stronger representativeness of personal exposures conducted in
the BZ during personal activities when the children resuspended dust.