Construction of a Job Exposure Matrix to Dust, Fluoride, and Polycyclic Aromatic Hydrocarbons in the Norwegian Aluminum Industry using Prediction Models
Published Date:Sep 25 2015
Source:Ann Occup Hyg. 59(9):1106-1121.
Exposure Assessment-mixed Models
Polycyclic Aromatic Hydrocarbons
Polycyclic Hydrocarbons, Aromatic
Pubmed Central ID:PMC4739354
Funding:CC999999/Intramural CDC HHS/United States
The Norwegian aluminum industry developed and implemented a protocol for prospective monitoring of employees’ exposure using personal samplers. We analyzed these data to develop prediction lines to construct a job exposure matrix (JEM) for the period 1986–1995.
The protocol for personal monitoring of exposure was implemented in all seven Norwegian aluminum plants in 1986 and continued until 1995. Personal samplers were used to collect total dust, fluorides, and total polycyclic aromatic hydrocarbons (PAH). In addition, exposure could be categorized according to process, i.e. prebake, Søderberg, and ‘other’. We constructed four-dimensional JEMs characterized by: Plant, Job descriptor, Process, and Year. Totally 8074, 6734, and 3524 measurements were available for dust, fluorides, and PAH, respectively. The data were analyzed using linear mixed models with two-way interactions. The models were assessed using the Akaike criterion (AIC) and unadjusted R2. The significance level was set to 10% (two-sided) for retaining variables in the model.
In 1986, the geometric mean (95% confidence interval in parentheses) for total dust, total fluorides, and PAH were 3.18 (0.46–22.2) mg m−3, 0.58 (0.085–4.00) mg m−3, and 33.9 (2.3–504) μg m−3, respectively. During 10 years of follow-up, the exposure to total dust, fluorides, and PAH decreased by 9.2, 11.7, and 14.9% per year, respectively. Each model encompassed from 49 to 72 significant components of the interaction terms. The interaction components were at least as important as the main effects, and 65 to 91% of the significant components of the interaction terms were time-dependent.
Our prediction models indicated that exposures were highly time-dependent. We expect that the time-dependent changes in exposure are of major importance for longitudinal studies of health effects in the aluminum industry.
application/pdf application/octet-stream image/gif image/jpeg image/gif image/jpeg
You May Also Like: