Use of rank-order analysis of ordinal exposure data: application to vinyl chloride exposure

Abstract

The University of Louisville currently maintains a database of work histories and exposure rankings for chemicals used in the production of polyvinyl chloride from the plant where the original angiosarcoma cases were identified. A 6-point ranking scale was used to characterize worker exposure to vinyl chloride by year, building, and job. Since this is ordinal data, non-parametric statistics have been used to evaluate exposure factors in case-control studies. Rank order analysis of vinyl chloride exposure and angiosarcoma shows a strong statistical association. In contrast, the method shows no association between vinyl chloride exposure and brain cancer. Exposure ranking is a reasonable alternative to estimation of actual dose. Rank order analysis allows for a rapid assessment of exposure-disease associations for case control studies.