Relative power of alternative ambient air pollution metrics for detecting chronic health effects in epidemiological studies

A number of alternative metrics for characterizing ambient air pollution concentrations were compared with regards to their relative power for detecting health effects as associated with long- term ambient concentrations of ozone and total suspended particulates (TSP). The health effects studied included development of overall airway obstructive disease (AOD), chronic bronchitis and asthma, as well as increasing severity for AOD, chronic bronchitis, and asthma. These health effects were studied in a cohort of 3914 California non-smoking Seventh-day Adventists who had 20 years of residence and work location history. The alternative metrics for ozone included the 24 hour mean, an 8 hour average (9:00 a.m. to 5:00 p.m.) as well as exceedance frequencies (hours above) and excess concentrations (the integrated sum of time x concentration above) for cutoffs of 6, 8, 10, 12, and 15 parts per hundred million (pphm). For TSP the alternative metrics were the 24 hour mean, as well as exceedance frequencies and excess concentrations for cutoffs of 60, 100, 150 and 200 micrograms per cubic metre (mu g/m(3)) The relative power for detecting health effects was assessed using the ratio of the estimated pollutant regression coefficient to its estimated standard error in multiple logistic regression models for development of disease and multiple linear regression models for increasing severity of disease. For ozone the metric with the greatest power for the most health outcomes was the 8 hour average. For TSP the metric with the greatest power for the most health outcomes was excess concentration above 200 mu g/m(3) of TSP.