Epidemiological studies of acute ozone exposures and mortality

Many, but not all, observational epidemiological studies of ozone (O-3) air pollution have yielded significant associations between variations in daily ambient concentrations of this pollutant and a wide range of adverse health outcomes. We evaluate some past epidemiological studies that have assessed the short-term association Of O-3 with mortality, and investigate one possible reason for variations in their O-3 effect estimate, i.e., differences in their approaches to the modeling of weather influences on mortality. For all of the total mortality-air pollution time-series studies considered, the combined analysis yielded a relative risk, RR=1.036 per 100-ppb increase in daily 1-h maximum O-3 (95% Cl: 1.023-1.050). However, the subset of studies that specified the nonlinear nature of the temperature-mortality association yielded a combined estimate of RR=1.056 per 100 ppb (95% CI: 1.032-1.081). This indicates that past time-series studies using linear temperature-mortality specifications have underpredicted the premature mortality effects of O-3 air pollution. For Detroit, Ml, an illustrative analysis of daily total mortality during 1985-1990 also indicated that the model weather specification choice can influence the O-3 health effects estimate. Results were intercompared for alternative weather specifications. Nonlinear specifications of temperature and relative humidity ( RH) yielded lower intercorrelations; with the O-3 coefficient, and larger O-3 RR estimates, than a base model employing a simple linear spline of hot and cold temperature. We conclude that, unlike for particulate matter (PM) mass, the mortality effect estimates derived by time-series analyses for O-3 can be sensitive to the way that weather is addressed in the model. The same may well also be true for other pollutants with largely temperature-dependent formation mechanisms, such as secondary aerosols. Generally, we find that the O-3-mortality effect estimate increases in size and statistical significance when the nonlinearity and the humidity interaction of the temperature-health effect association are incorporated into the model weather specification. We recommend that a minimization of the intercorrelations of model coefficients be considered (along with other critical factors such as goodness of fit, autocorrelation, and overdispersion) when specifying such a model, especially when individual coefficients are to be interpreted for risk estimation.