ABSTRACT
In a retrospective assessment of employee exposure to acrylonitrile (AN) for an epidemiological study, investigators from the National Cancer Institute (NCI) and the National Institute for Occupational Safety and Health (NIOSH) evaluated the feasibility of using historic acrylonitrile air samples without modification. The evaluation discussed here was to determine whether the air sampling results across plants were comparable. During site visits to each plant conducted between 1984 and 1986, study investigators collected personal air samples for four days on approximately ten jobs per day. During these visits, IHs at seven of the eight plants also collected personal samples to compare their sample values to the study-collected sample values. Each plant's IH collected these concurrent measurements for their own use and independent of the IHs at the other plants. The plant IHs had no common sampling protocol but, rather, used professional judgment in deciding sampling logistics for their concurrent measurement. In addition, each plant IH used a different laboratory to analyze samples (the study industrial hygienists used one laboratory). Three sampling methods were used by plant industrial hygienists to collect concurrent measurements: charcoal tubes, passive monitors, and porous polymer tubes. The study investigators only used charcoal tubes. Two hundred and sixty four (264) pairs of concurrent measurements were collected. To assess the +/- comparability of the data sets, paired-observation tests were used. The two sets of charcoal tubes were found to compare favorably with each other. The study's charcoal tubes were 1.2 times higher than results from plant passive monitors. No correlation was found between the study's charcoal tube results and plant porous polymer tube results, although the means for 34 pairs of samples were equivalent. As a result of this evaluation, the investigators decided that no adjustments would be made to the plant measurements. This type of evaluation should be considered when using measurement data in multisite epidemiological studies.
