Errors and biases in the diagnosis of cancer of the lung and their influence on risk estimates.

When diagnosed as primary lung cancer, metastases from the abdomen, plus false negative cases have little effect on epidemiology studies of male smokers, but may result in a severe dilution of the lung cancers among women and nonsmokers. We have attempted to quantitate this handicap for epidemiological studies using two approaches. The relative frequency of diagnosed primary lung and abdominal cancer among males, women, and nonsmokers differs substantially and is used here to calculate magnitude. The second approach postulates that the ratio of nonsmokers among persons with squamous cell lung cancer and primary adenocarcinoma of the lung would be constant by sex if there were no distortion by abdominal metastases. These two approaches indicate that the much higher ratio of metastatic disease diagnosed as primary lung cancer among nonsmoking women (factor of 15 to 20), makes it more difficult to identify an environmental carcinogen among women or nonsmokers than among male smokers in case-control studies.