Confounding by dietary factors in case–control studies on the efficacy of cancer screening in Japan

Abstract

Evaluation of cancer screening using case–control studies is less valid in comparison to randomized controlled trails, due to the intrusion of a possible self-selection bias in the former. The randomized controlled trial approach, however, may be difficult in developed countries where mass cancer screening programs are already being performed nationally. Accordingly, case–control studies are often performed instead of randomized controlled trials. In case–control studies, no reports could be found in the literature using dietary habit, an important influencing factor in carcinogenesis, as an adjusting item. We surveyed nutrition and food intake status through a nutrition survey using the weighing method, and smoking prevalence and alcohol consumption with questionnaires in the general population in northern Japan, in subjects over 30 years of age. We then compared these results among non-participants and participants in cancer screening programs, and evaluated how any differences between the two groups might affect the results of case–control studies. Non-participants had a significantly lower intake of vegetables, carotene, vitamin C, and dietary fiber, which are thought to be beneficial factors in the prevention of carcinogenesis. Non-participants had a significantly higher cigarette smoking prevalence than participants. In relation to the intake of vegetables, a difference between participants and non-participants was evident, as proper adjustment for vegetable intake led to an odds ratio closer to the null value. The value of the odds ratio will probably not decrease, but it might increase when other dietary factors or smoking are taken into consideration.