An Empirical Assessment of Exposure Measurement Error and Effect Attenuation in Bipollutant Epidemiologic Models

Background: Using multipollutant models to understand combined health effects of exposure to multiple pollutants is becoming more common. However, complex relationships between pollutants and differing degrees of exposure error across pollutants can make health effect estimates from multipollutant models difficult to interpret. Objectives: We aimed to quantify relationships between multiple pollutants and their associated exposure errors across metrics of exposure and to use empirical values to evaluate potential attenuation of coefficients in epidemiologic models. Methods: We used three daily exposure metrics (central-site measurements, air quality model estimates, and population exposure model estimates) for 193 ZIP codes in the Atlanta, Georgia, metropolitan area from 1999 through 2002 for PM2.5 and its components (EC and SO4), as well as O3, CO, and NOx, to construct three types of exposure error: δspatial (comparing air quality model estimates to central-site measurements), δpopulation (comparing population exposure model estimates to air quality model estimates), and δtotal (comparing population exposure model estimates to central-site measurements). We compared exposure metrics and exposure errors within and across pollutants and derived attenuation factors (ratio of observed to true coefficient for pollutant of interest) for single- and bipollutant model coefficients. Results: Pollutant concentrations and their exposure errors were moderately to highly correlated (typically, > 0.5), especially for CO, NOx, and EC (i.e., “local” pollutants); correlations differed across exposure metrics and types of exposure error. Spatial variability was evident, with variance of exposure error for local pollutants ranging from 0.25 to 0.83 for δspatial and δtotal. The attenuation of model coefficients in single- and bipollutant epidemiologic models relative to the true value differed across types of exposure error, pollutants, and space. Conclusions: Under a classical exposure-error framework, attenuation may be substantial for local pollutants as a result of δspatial and δtotal with true coefficients reduced by a factor typically < 0.6 (results varied for δpopulation and regional pollutants). Citation: Dionisio KL, Baxter LK, Chang HH. 2014. An empirical assessment of exposure measurement error and effect attenuation in bipollutant epidemiologic models. Environ Health Perspect 122:1216–1224; http://dx.doi.org/10.1289/ehp.1307772

. Input parameters for the SHEDS model: PM 2.5 , SO 4 , and EC 3  . Attenuation of model coefficients in a classical error, single pollutant framework, and in bi-pollutant models, assuming one pollutant has an effect, and one pollutant has no effect 11

Population exposure metric
Ambient pollutant concentrations supplied at ZIP code centroids (as described in the CS metric section above), were used to represent ambient outdoor concentrations for calculation of exposure at the home, outside the home, and the home garage for groups of simulated individuals. Individuals were sampled by census tract as is normally done in SHEDS Air. The target was for 100 simulated individuals to be assigned to each of the 193 ZIP codes of interest.
To obtain ZIP code level exposure estimates, individuals were also assigned to a ZIP code polygon. Because it is common for census tracts to overlay two or more ZIP codes, an area weighted scheme was used to define the likelihood of an individual in a specific census tract being assigned to a specific ZIP code.
To account for exposure during commuting and travel, each simulated individual's work location was selected stochastically based on probabilities derived from the 2000 U.S. Census Tract-toTract commuting data. Individuals can commute to any census tract overlapping the ZIP codes of interest. Workers commuting out of the sampling area (~3% of workers) are removed from the population and replaced with another worker that commutes within the sampling area. Travel other than work is limited to the closest 20 census tracts within the sampling area, and within 10 miles of an individual's home census tract. Each individual is assigned to 3 nearby census tracts to which they may travel for a variety of reasons (e.g. shopping, church, other facilities related to microenvironments in the CHAD database). Some facilities are always in the same census tract for an individual (e.g. church), and others are assigned randomly each trip (e.g. retail stores). Air quality during travel is based on the ZIP code associated with the origin of the travel.
Penetration and decay parameters can be found in Tables S1 and S2.       one pollutant has an effect, and one pollutant has no effect; δ spatial for local-regional pollutant pairs; n=193 for each box. The bottom and top of the box represent 25 th and 75 th percentiles, the band near the middle of the box is the median, and the ends of the whiskers are the 5 th and 95 th percentiles.
12 Figure S5b. Attenuation of model coefficients in a classical error, single pollutant framework, and in bi-pollutant models, assuming one pollutant has an effect, and one pollutant has no effect; δ population for local-regional pollutant pairs; n=193 for each box. The bottom and top of the box represent 25 th and 75 th percentiles, the band near the middle of the box is the median, and the ends of the whiskers th th are the 5 and 95 percentiles.
13 Figure S5c. Attenuation of model coefficients in a classical error, single pollutant framework, and in bi-pollutant models, assuming one pollutant has an effect, and one pollutant has no effect; δ total for local-regional pollutant pairs; n=193 for each box. The bottom and top of the box represent 25 th and 75 th percentiles, the band near the middle of the box is the median, and the ends of the whiskers are the 5 th and 95 th percentiles.