Deaths and tumours among workers grinding stainless steel: a follow up

To study cause specific mortality and cancer morbidity in workers exposed to the dust of grinding materials, grinding agents, and stainless steel, especially with regard to a possibly increased risk of respiratory, stomach, and colorectal cancer.

Retrospective cohort study, using reference cohorts of blue collar workers and population rates for comparison. The exposed cohort comprises workers with at least 12 months employment time at two plants, producing stainless steel sinks and saucepans (n = 727). Also, reference cohorts of other industrial workers (n = 3965) and fishermen (n = 8092) were analysed. The observation period began 15 years after the start of employment. Standardised mortality or incidence ratios (SMRs, SIRs; county reference rates) were calculated for cause-specific mortality between 1952 and 1993, and for cancer morbidity between 1958 and 1992.

In the exposed cohort, overall mortality, cardiovascular mortality, and all malignant mortality and morbidity were slightly lower than expected. Also, the risk estimates for cancer in the upper and lower respiratory tracts and for stomach cancer were lower than expected. There was an increase in morbidity from colon cancer, which was explained by an excess of tumours in the sigmoid part only. Here, the risk estimates were higher in workers with long employment time (1-14 years: four observed cases, SIR 1.7, 95% confidence interval (95% CI) 0.4 to 4.5; > or = 15 years: three observed cases, SIR 4.3, 95% CI 0.9 to 13) and the increased risk was especially pronounced among those first employed before 1942. A slight nominal excess of rectal cancers (nine observed cases, SIR 1.4, 95% CI 0.6 to 2.6), and a significant excess of prostate cancer morbidity (36 observed cases, SIR 1.7, 95% CI 1.2 to 2.4) were found. These risk estimates did not, however, increase with employment time.

The finding of an increased risk of cancer in the sigmoid part of the colon, which was not found in the reference cohorts, and with indication of a relation between duration of employment and response, is consistent with a causal relation. The limited size of the exposed cohort makes a detailed exposure-response analysis unstable, and the confidence limits are wide. Albeit slightly raised, the risk estimate for rectal cancer in the exposed cohort was not different from the estimate among the other industrial workers.