Geographic and epidemiologic analysis of the proximity to petroleum refinery waste dumps and the incidence of acute myeloid leukemia and multiple myeloma in Orange County, California

Benzene has been implicated as a leukemogen in occupational cohort follow-up studies. A national environmental sampling survey has concluded that the most important environmental source of benzene is cigarette smoking. A preliminary geographic analysis indicates that there may be pockets of Acute Myeloid Leukemia (AML) and Multiple Myeloma (MM) within the county. Some of these pockets can be explained by concentrations of high-incidence age groups such as the Seal Beach and Laguna Hills Leisure Worlds. Other high-incidence census tracts must be examined using Poisson regression spatial analysis.; The proposed study presented hypotheses that relate to both individual and aggregate risk models for AML and MM. The aggregate risk model tested the hypothesis that there is a significant association between the number of cases of AML and MM and the distance from the centroid of the census tract of residence to a petroleum refinery dump. The individual risk model tested the hypothesis that residence in proximity to a hazardous waste dump is a significant risk factor for AML and MM. Confounders in the individual risk model included personal exposure risk factors (occupation, industry, smoking) stratified on gender and age. The significance of the aggregate risk model was tested using the Poisson regression method, and generalized linear regression, while the individual risk model employed logistic regression.; The Poisson regression yielded statistically significant models for both AML or MM. The model chi-squares indicated some deviation from expected contingency values but coefficients of determination were between 1 and 2%. The initial Poisson model for MM indicated that the coefficient for distance from the case to the Dow Chemical dumpsite was.12 cases/mile-person year. This was statistically significant at p {dollar}<{dollar}.05.; The individual-level case-control study resulted in a significant logistic regression models for AML only. The AML model was constructed using 69 cases and 869 unmatched controls. The overall model {dollar}chisp2{dollar} was 16.603 (p {dollar}<{dollar}.0009). The AML model included significant exposure variables for current smokers, employment in a manufacturing industry, and employment in a transportation industry.