Retrospective exposure assessment of perfluorooctanoic acid serum concentrations at a fluoropolymer manufacturing plant.

Perfluorooctanoic acid (PFOA) is a suspect human carcinogen, causes neonatal loss, liver enlargement, and a variety of tumors in rodents, and has been associated with increased cholesterol levels in humans. Mortality analyses of worker cohorts have not been conclusive or consistent. As part of a series of epidemiologic studies of workers in a West Virginia plant that manufactures fluoropolymers, estimates of serum PFOA for the worker cohort were developed for the period of 1950-2004. An existing database of 2125 worker biomarker measurements of serum PFOA was used to model retrospective exposures. Historical PFOA serum levels for eight job category/job group combinations were modeled using linear mixed models to account for repeated measures, along with exposure determinants such as cumulative years worked in potentially exposed jobs, the amount of C8 used or emitted by the plant over time, as well as a four-knot restricted cubic spline function to reflect the influence of process changes over calendar time on exposure. The modeled biomarker levels matched well with measured levels, including those collected independently as part of a community study of PFOA levels (Spearman correlations of 0.8 for internal data comparisons and 0.6 for external data comparisons). These annualized PFOA serum estimates will be used in a series of morbidity and mortality studies of this worker cohort.