The Relationship between Cross-Sectional and Time Series Studies

Two classes of observational studies have provided quantitative estimates of the influence of air pollution on mortality. These are cross-sectional and time series studies. Cross-sectional studies examine geographical variations in community mortality rates and air pollution levels and typically involve fitting the parameters of proportional exposure-mortality functions by least squares regression analysis. Control for health-related covariates of pollution, such as socioeconomic status and smoking, is accomplished by inclusion of additional variables in the regression equation. Typically the averaging time for these analyses is one year; and therefore, many assume that these studies reflect chronic effects. Time series studies examine temporal variatons in death rates and air pollution levels in a single community. Using days as the unit of observation and averaging over the community, the parameters of the models are estimated by least squares techniques. Control for covariates is accomplished by reexpressing variables as deviations from trend lines and by including weather variables in the regression equation. Studies using these techniques reflect only the acute effects of air pollution. The similarity in the disease-specific findings of time series studies and cross-sectional studies which were analyzed is discussed. It is possible that the cross-sectional studies which have been conducted to datemore » have reflected the impacts of acute rather than chronic effects of air pollution.« less