Estimates of the association between ozone and asthma hospitalizations that account for behavioral responses to air quality information

Behavioral responses to information about forecasted air quality may introduce systematic measurement error in pollution exposure, leading to biased estimates of the impact of pollution exposure on health. This paper estimates the statistical association between ambient ozone concentrations and asthma hospitalizations in Southern California while accounting for potential avoidance behavior in response to forecasted air quality. Data on asthma hospital admissions were merged with observed and forecasted air quality and meteorological data at the daily level for the years 1989–1997. A distributed lag Poisson generalized linear model allowing for overdispersion was estimated. Accounting for potential responses to information about pollution leads to significantly larger estimates of the relationship between ozone concentrations and asthma hospital admissions, particularly for susceptible populations. Individuals take substantial actions to reduce exposure to ozone; estimates of the concentration–response function for ozone that ignore these actions are biased towards the null and may significantly understate the costs to society from ozone concentrations.