Abstracts: ISEE 20th Annual Conference, Pasadena, California, October 12–16, 2008: Contributed Abstracts
Development of Historical Exposure Estimates for an Epidemiologic Study of Beryllium Sensitization and Chronic Beryllium Disease at a Beryllium Production Facility
A number of epidemiologic studies have reported elevated prevalence of beryllium sensitization and chronic beryllium disease among employees in specific work processes. However, exposure-response relationships have been inconsistent, perhaps due to exposure misclassification from a lack of accurate, precise and/or biologically relevant estimates of historical exposures. In 1999, an epidemiologic study surveyed 264 workers hired after 1/1/1994 at a beryllium production facility.
Personal (full-shift) beryllium exposure data from a 1999 exposure survey (n = 3,906) were used to obtain mean baseline exposure estimates (BEE) for 272 jobs in a job-exposure matrix (JEM). We used historical general-area air samples (n = 77,183) collected between 1994 and 1999 to estimate the fractional annual change in exposure for 29 different work-process areas. The data were modeled using tobit regression to account for the left censoring of the air samples (17%–98% <detection limit). Historical job-level exposure estimates were calculated by applying the fractional annual change to the BEEs of jobs in the respective work-process areas. Workers were assigned exposure estimates based on their reported work for a given year and the locations of their jobs.
The mean beryllium time weighted average BEEs for all jobs ranged from 0.01 μg/m3 for Administration to 28.01 μg/m3 for Atomizer Operator. Changes in exposure over time between 1999 and 1994 were observed in many of the work-process areas; some of these changes were non-linear and differed among work-areas. The overall median cumulative and average beryllium exposures were 1.49 μg/m3-years and 0.62 μg/m3 respectively for participants over their work histories. The historical exposure estimates will be validated using limited personal exposure data collected between 1994 and 1998.
Using our JEM, exposure-response analyses can be explored over a range of exposure metrics, including summary measures such as cumulative, annual, or peak exposures, with the ultimate objective of elucidating an exposure-response relationship.© 2008 Lippincott Williams & Wilkins, Inc.