PURPOSE: Based on previous reports of increased serum lipid levels in workers at a U.S. polymer manufacturing facility, the study objective was to investigate ischemic heart disease (IHD) mortality as well as a broad range of mortality causes for an occupational cohort at the facility. METHODS: The cohort comprised 6,027 men and women who had worked at the facility between 1948 and 2002; these years delimit the mortality follow-up period. Standardized mortality ratios (SMR) were estimated to compare observed numbers of deaths to expected numbers derived from mortality rates for 3 reference populations: the U.S. population, the West Virginia state population, and an 8-state regional employee population from the same company. RESULTS: Most SMR estimates based on U.S. and state populations were below 100. Comparison to the employee population also resulted in many SMR estimates at or near a no-effect level. Relative to the regional worker population, a nonsignificant elevation for IHD mortality was observed (SMR = 109, 95% confidence interval [CI]: 96, 124). Mortality associated with diabetes was significantly increased compared to the regional worker population (SMR = 197, 95% CI: 123, 298). A corresponding increase in the SMR for IHD and diabetes mortality was not detected for comparisons with the two general populations. CONCLUSIONS: The results reported herein show little evidence of increased cause-specific mortality risks for workers at the plant. This study demonstrates the utility of comparing occupational cohorts with a similar worker reference population in order to reduce bias associated with the healthy worker effect.
PURPOSE: Based on previous reports of increased serum lipid levels in workers at a U.S. polymer manufacturing facility, the study objective was to investigate ischemic heart disease (IHD) mortality as well as a broad range of mortality causes for an occupational cohort at the facility. METHODS: The cohort comprised 6,027 men and women who had worked at the facility between 1948 and 2002; these years delimit the mortality follow-up period. Standardized mortality ratios (SMR) were estimated to compare observed numbers of deaths to expected numbers derived from mortality rates for 3 reference populations: the U.S. population, the West Virginia state population, and an 8-state regional employee population from the same company. RESULTS: Most SMR estimates based on U.S. and state populations were below 100. Comparison to the employee population also resulted in many SMR estimates at or near a no-effect level. Relative to the regional worker population, a nonsignificant elevation for IHD mortality was observed (SMR = 109, 95% confidence interval [CI]: 96, 124). Mortality associated with diabetes was significantly increased compared to the regional worker population (SMR = 197, 95% CI: 123, 298). A corresponding increase in the SMR for IHD and diabetes mortality was not detected for comparisons with the two general populations. CONCLUSIONS: The results reported herein show little evidence of increased cause-specific mortality risks for workers at the plant. This study demonstrates the utility of comparing occupational cohorts with a similar worker reference population in order to reduce bias associated with the healthy worker effect.