What is the main culprit of uncertainty in the health effects of air pollution?

Naresh Kumar (1)

(1) The University of Iowa

     Abstract Number: 430
     Last modified: November 27, 2009

Abstract
BACKGROUND: The association between air pollution and birth outcomes is inconclusive, primarily because of uncertainty arising from the coarse resolution of air pollution data and their mismatch with the location of birth data.
OBJECTIVES: This research examines how different methods and spatial-temporal scales of air pollution data used for exposure computation influence the association between air pollution exposure and birth outcomes, namely low birth weight (LBW), pre-term delivery (PTD), and intra-uterine growth restriction (IUGR).
METHODS: A total of 400,000 live births (during 2000 to 2004) were geocoded to street address and exposure to criteria pollutants were computed using different methods, namely nearest neighbor average, inverse distance weighting and Kriging. Exposure computation was restricted for cases within different distance intervals (3 miles, 6 miles, 9 miles, 12 miles and by counties) from the monitoring stations.
RESULTS: The effect of air pollution exposure, especially that of PM$_(2.5), PM$_(10), CO and O$_3, on birth weight declines as the distance from the monitoring station used for computing exposure increases. Our analysis suggest that the intensity of relationship between air pollution and birth weight changes significantly with the change in the method of exposure computation and geographic distance between place of residence and monitoring stations. The increase in the average daily exposure during three trimesters and entire pregnancy showed a statistically significant inverse association with birth weight when controlled for confounders and exposure were computed at fine geographic scale.
CONCLUSIONS: Ad hoc methods of exposure computation are partly responsible for uncertainty in the linkages between air pollution exposure and birth weight. Time-space resolved estimates of exposure (at the place of mother’s residence) used in this study, suggests that the elevated concentration of most of the criteria pollutants were significantly associated with decline in birth weight, greater risks of LBW and IURG. We suggest the use of hybrid approaches to derive exposure at any point location (on to geographic space) and on a given day to improve our understanding of the health effects of air pollution.