Passive Smoking and Lung Cancer: A Reanalysis of Hirayama’s Data
The study was not designed to test the hypothesis, whether passive smoking is associated with lung cancer or not. It can therefore only generate this hypothesis, not prove it.
The cohort was not representative for the population of Japan. A selection bias is possible.
The exposure indicator — the fact of being married to a man who smokes — is not reliable, not valid and not specific.
The event indicator — dying on lung cancer as noted on death certificates — is neither reliable nor valid.
Various confounding factors — for instance exposure at the working place, indoor air pollution, overall air pollution, type of medical care — were not accounted for.
Bias in registering the fact, that a woman is a nonsmoker, was not controlled. Resulting differential misclassifications of the cases, who were smokers and had to be excluded, have not been considered.
Almost nothing is known about the 200 cases. No case reports are available, autopsy and histology arc available in only 11.5%.
Unable to display preview. Download preview PDF.
- 1.Bishop YV, Fienberg StE, Holland PW (1980) Discrete multivariate analysis: theory and practice. MIT Press, Cambridge, pp 97Google Scholar
- 2.Garfinkel L (1984) Time trends in lung cancer mortality among nonsmokers and a note on passive smoking. J Natl Canc Inst (USA) 66:1061–1066Google Scholar
- 3.Hartung J, Elpelt B, Klösener KH (1985) Statistik. Lehr-und Handbuch der angewandten Statistik. München Wien Oldenburg, p 501–503Google Scholar
- 4.Hirayama T (1984) Lung cancer in Japan: effects of nutrition and passive smoking. In: Mizell M, Correa P (eds) Lung cancer: causes and prevention. Chemie, Weinheim, pp 175–195Google Scholar