A Methodological Investigation of Fatal Disease Risks in a Large Industrial Cohort

Abstract

Proportionate mortality analyses of deaths identified in a cohort of workers employed at a single large industrial plant were performed to examine variations in estimates of fatal disease risks between the group most likely known to the employer (i.e., identifiable from company records) and the remaining set. The former group consisted of deceased retirees and workers who died within one year of last employment at the plant (“active deaths”). Risk estimates were computed by forming the ratio (PMR) of observed deaths to age and calendar year adjusted expected numbers of deaths derived from general population mortality. For most major disease categories, risk estimates were similar in the two decedent groups. However, a very low risk of tuberculosis was derived from deaths known to the employer (PMR = 0.35) with no unusual risk in the other set of deaths (PMR = 0.95). Respiratory cancer was associated with increased mortality among active and retired workers (PMR= 1.58), but not among other decedents (PMR = 1.08). When analyses were confined to the group of workers dying within one year of last employment, risk estimates were surprisingly similar to those obtained from total cohort deaths, except for a further lowering of the PMR for tuberculosis (PMR = 0.29). These results may provide a baseline for interpreting results of proportionate mortality analyses of limited data in other industrial groups exposed to potentially hazardous substances.