Approaches for estimating the distribution of short-term exposure concentrations for different averaging times

Because the distribution of short-term exposure concentrations is approximately log-normal within 1 day, the geometric mean (GM) and geometric standard deviation (GSD) are critical parameters for obtaining the overall view of exposure, but when the averaging time changes, so too do the GM and GSD, though the arithmetic mean (AM) should be fixed. For industrial hygienists the ability to transform the GM and GSD from one averaging time to another is very useful. Three approaches for estimating the GM and GSD for different averaging times were applied to 10 workers (16 worker-chemical combinations). The random approach is based on the assumption that exposure concentrations vary randomly over time. On the other hand, two autocorrelation approaches are based on an assumption that the exposure time series are autocorrelated. This study showed that the random approach tended to lead to positive-biased and negative-biased estimates for the GM and GSD, respectively. The two types of the autocorrelation approach generally gave unbiased estimates, but the dispersions of the estimates became large when the ratio of the new averaging time to the original one was large.