When 4/2 does not equal 2: approaches to extending a categorical exposure rank to a quantitative exposure range.
Rice-C; Hornung-R; Moser-A; Brewer-D; Ho-M; Tollerud-DJ
Ramazzini Days 2007-Jubilee Celebration 25th Anniversary of the Collegium Ramazzini, October 25-28, 2007, Carpi, Italy. Bologna, Italy: Institute Ramazzini, 2007 Oct; :1-16
Occupational epidemiology studies frequently rely on ranked exposures indicating which jobs are associated with the lowest exposures, the highest exposures, and those at several cut points between lowest and highest. When expressed as categories, these are often designated with a unit value, such as 0 (no exposure), 1 (some exposure), 2 (less than average), 3 (about average), 4 (higher than average) and 5 (highest). In this scale, a rank of 2 cannot be considered two times the exposure of rank 1; similarly, rank 4 is not two times rank 2. Two approaches are presented based on historical exposure guidelines and utilizing scaling from the log-normal distribution to associate ranks with quantitative ranges of exposure. The following is one approach: 0 is actually unexposed; a rank of 1 has a range that includes the limit of detection for an analytical method used at the time; a rank of 4 has a range that includes the published exposure guideline (e.g., from the industry, a TLV from ACGIH or PEL from OSHA) in effect at a point in time; a rank of 5 would be limited to values that allowed continued work. Using these three benchmarks, ranges can be set for the remaining ranks, using the general guideline that the upper and lower bounds should differ by a factor of 3 or 4 to be meaningfully different. Using this approach for a compound with an exposure guideline of 50 ppm and a limit of detection of 2 ppm, the following ranges might be associated with each rank: 0=none; 1=>0-5; 2= >5-15; 3= >15-45; 4= >45-135; 5, >135. Alternatively, in a situation where exposures are generally known to be low, ranges might be set such that the highest rank is associated the published guideline; in this instance, and depending upon the limit of detection and guideline values, three or four ranks might be sufficient. The use of air sampling data to adjust the ranges developed from ranks using these two approaches is illustrated with data from an ongoing exposure reconstruction to chemical exposures at a gaseous diffusion plant.
Epidemiology; Exposure-levels; Exposure-limits; Statistical-analysis; Work-environment; Workers; Air-samples; Humans; Men; Women;
Author Keywords: Exposure assessment; categorical exposure; occupational epidemiology; quantitiatve exposure reconstruction
Carol Rice, Ph.D., CIH, University of Cincinnati, PO Box 670056, Cincinnati OH 45267-0056
Ramazzini Days 2007-Jubilee Celebration 25th Anniversary of the Collegium Ramazzini, October 25-28, 2007, Carpi, Italy
University of Louisville