Mortality at an automotive stamping and assembly facility.

Abstract

Mortality among employees with two or more years of employment at an automotive stamping and assembly facility was analyzed using standardized mortality ratio (SMR), proportional mortality ratio (PMR), and standardized mortality odds ratio (SMOR) methods. A total of 633 deaths were identified from 1978-1991. The all-cause SMR at the stamping plant was considerably less than expected (SMR = 0.64, 95% confidence interval (95% CI) = 0.55, 0.75 for white males and SMR = 0.71, 95% CI = 0.47, 1.03 for black males), consistent with a strong healthy worker effect. For six stomach cancer deaths among white males, both the SMR and PMR were statistically significant (SMR = 2.98, 95% CI = 1.09, 6.50; PMR = 4.58, 95% CI = 1.68, 9.98). Based on internal exposure comparisons using logistic regression and incorporating expected proportional mortality odds derived from U.S. mortality tables, SMORS were calculated, adjusting for gender, race, age, calendar time, and select latency weighted cumulative exposure categories. Stomach cancer risk was associated with employment in stamping departments (OR = 4.18, 95% CI = 1.30, 13.4) and evaluated at mean duration of the cases, although the number of cases was small. Lung cancer mortality at the stamping plant was elevated among production welders, and more so when combined with similar exposures in prior employment (OR = 1.69, 95% CI = 0.98, 2.9). All-cause mortality at the assembly plants produced SMRs and PMRs similar to those at the stamping plant. Excess lung cancer risk using logistic regression modeling was found in the paint department at the passenger assembly plant (OR = 1.87, 95% CI = 1.02, 3.42), but not at the corresponding van plant. The database first was analyzed with 390 deaths for the period 1978-1988 and reanalyzed with 633 deaths for the period 1978-1991. With three additional years of follow-up, adjusted SMORs changed rather substantially, where previous significant risk estimates at the assembly plants decreased to nonsignificance and paint department exposures emerged as the only predictor of lung cancer mortality. Stomach cancer at the stamping plant remained statistically significant but the magnitude of the excess decreased. This study reaffirmed the observation that results in occupational epidemiological investigations which are conducted on an urgent basis with short periods of follow-up, small number of deaths, imprecise measures of exposure, multiple comparisons, and higher regional rates should be interpreted with caution.