An approach to adjust standardized mortality ratios for competing cause of death in cohort studies

The calculation of standardized mortality ratios (SMRs) is a standard tool for the estimation of health risks in occupational epidemiology. An increasing number of studies deal with the analysis of the mortality in employees suffering from an occupational disease like silicosis or coal-worker pneumoconiosis (CWP). Their focus lies not on the mortality risk due to the occupational disease itself, but on other diseases such as lung cancer or heart diseases. Using population-based reference rates in these studies can cause misleading results because mortality rates of the general population do not reflect the elevated mortality due to the occupational disease investigated. Hence, the purpose of the present paper is to develop an approach to adjust the risk estimates for other causes of death with respect to the effect of an occupational disease as a competing cause of death in occupational mortality cohort studies.

To overcome the problems associated with SMRs, the paper makes use of proportional mortality ratios (PMR), which are a further approach for the estimation of health risks in occupational epidemiology. The cause-specific SMR can be rewritten as a product of PMR times the overall SMR. The PMR can be adjusted by ignoring the competing cause of death. Hence, an adjusted cause-specific SMR can be derived by multiplying this adjusted PMR with the overall SMR. This approach is applied to studies concerning lung cancer risk in coal miners suffering from CWP.

The usual approach for calculating SMRs leads to an underestimation of the real lung cancer risk in subgroups of miners suffering from CWP. The same effect can be observed in workers exposed to respirable silica already suffering from silicosis. The presented approach results in more realistic risk estimation in mortality cohort studies of employees suffering from an occupational disease. It is easily calculable on the basis of usually published values of observed cases and the corresponding cause-specific SMR.