Does one size fit all? The suitability of standard ozone exposure metric conversion ratios and implications for epidemiology.

Several exposure metrics have been applied in health research and policy settings to represent ozone exposure, such as the 24 h average and daily 8 h maximum. Frequently, results calculated using one exposure metric are converted using a simple ratio to compare or combine findings with results using a different metric. This conversion, however, assumes that such a ratio is constant across locations and time periods. We investigated the appropriateness of this conversion method by examining the relationships among various forms of ozone concentrations (24 h average, daily 1 h maximum, and daily 8 h maximum) within and between communities for 78 US communities from 2000 to 2004 and compared results to commonly used conversion ratios. We explored whether the relationships between ozone exposure metrics differ by region, weather, season, and city-specific characteristics. Analysis revealed variation in the relationship among ozone metrics, both across communities and across time within individual communities, indicating that conversion of ozone exposure metrics with a standard ratio introduces uncertainty. For example, the average ratio of the daily 8 h maximum to the daily concentration ranged from 1.23 to 1.83. Within a community, days with higher ozone levels had lower ratios. Relationships among metrics within a community were associated with daily temperature. The community-average exposure metric ratios were lower for communities with higher long-term ozone levels. Ozone metric ratios differed by season because of the different rate of change of ozone metrics throughout the year. We recommend that health effects studies present results from multiple ozone exposure metrics, if possible. When conversions are necessary, more accurate estimates can be obtained using summaries of data for a given location and time period if available, or by basing conversion ratios on data from a similar city and season, such as the results provided in this study.