Send letter to journal:
Re: Using proportion mortality ratio to evaluate the risk of cancer

On the basis of a retrospective mortality study Mastrangelo and his co-workers (Occup Environ Med 2008; 65: 697-700) concluded that “a high and prolonged exposure to cotton dust and other endotoxin-containing organic dusts was related to a lower risk of lung cancer”. The paper looked at the cancer risk, especially on lung cancer, among cotton mill workers by using the Standard Mortality Ratio (SMR). We believe that methodological limitation inherent in this study makes that conclusion unpersuasive.

The proportion mortality study is utilized well to analyze the effect of drug on the prognosis of some disease, for example, the preventive effect of aspirin on CVD death by comparison between aspirin users and non -users among patients. However, the mortality study is not suitable for causal analyses of diseases because the proportion of dead cases due to some disease, such as lung cancer in Mastrangelo’s study, has been thought to be a deceiving measure of occurrence for the disease, when it is not an acute fatal illness. Giving a cancer due to a causal exposure, the proportion of survivals among all cancer cases in index cohort will depend on cancer stage at diagnosis, the general health of patients, etc. and could be bigger or smaller than that of dead cases, and so may be among reference cohorts. This situation will be the key variant that determines the deviation of results. However, it is difficult to speculate in which direction the study results might have been biased because the magnitude of proportion of survivors is virtually unknown.

Based on the results of Mastrangelo’s study, We think, the conclusion should be “a high and prolonged exposure to cotton dust and other endotoxin-containing organic dusts was associated with a lower risk of lung cancer mortality, and due to uncontrolled confounding other risk factors for lung cancer death could not be ruled out as a contributor to the association.”