Selection of the measures of exposure for epidemiology studies.
Appl Occup Environ Hyg 1991 Jun; 6(6):448-457
The implications of exposure variation within and between individual workers for selection of exposure measures for epidemiological studies were addressed. A lognormal distribution model was presented and applied to exposure data for 31 worker groups in nine industrial processes. Within worker and between worker distributions were examined to determine, respectively, importance of peak exposures and measures of exposure applicable to groups of individuals. Applications of such analyses to retrospective investigations were briefly addressed. The lognormal model described exposure distribution using any two of six possible parameters: mean and variance of lognormal distribution; geometric mean and geometric standard deviation of lognormal distribution; and mean and variance of normal distribution. Analysis of the 31 worker group data indicated large within and between worker variation, with the latter being slightly larger. Both variabilities could result in significant exposure misclassification. A toxicokinetic model was used to determine relative importance of peak versus cumulative exposures. For first order elimination and repair rates, cumulative exposure (mean times duration) was generally a valid predictor of damage. For nonlinear situations involving a threshold burden, peak exposures were considered insignificant when mean exposure was less than one fourth to one eighth that giving rise to the threshold burden and when a damage repair mechanism was involved. Retrospective assessments were particularly difficult because data on cumulative exposures were poor. Data from the 31 groups indicated that retrospective grouping was especially risky when exposures were largely intermittent and occurring outdoors. The author concludes that prospective sampling can provide necessary data for unbiased estimates of mean group exposure and within group exposure homogeneity, while this is rarely met in retrospective studies.
NIOSH-Publication; NIOSH-Contract; Contract-91-38661; Humans; Industrial-factory-workers; Epidemiology; Exposure-levels; Analytical-methods; Dose-response; Mathematical-models; Statistical-analysis; Occupational-exposure
Applied Occupational and Environmental Hygiene