Estimation of the prevalence of occult cancer and its application to the epidemiology of multiple primary cancer.

This paper focused on methodological issues in the epidemiology of multiple primary cancer, especially on methods to evaluate the frequency of a second cancer. The incidence density is the most widely used measure in cancer epidemiology. However, the measure of incidence might be biased to some extent, because it might be affected by extraneous factors which are related to the detection of the second cancer. A better alternative is the measure of the prevalence of occult cancer. The authors have formulated the functional relationship between the prevalence of occult cancer and some variables which can be obtained from ordinary epidemiologic observations, such as the incidence density, the growth function, and the distribution of the diameters of the cancers which were clinically diagnosed. The formula was then applied to an epidemiologic study of multiple primary cancer in the digestive system. All the patients with cancer of the digestive system except the stomach treated in the National Cancer Center Hospital between 1962 and 1981 constituted the study group. Among 1,561 male and 714 female patients, gastric cancers were found as the second primary cancer in nine males and one female within 2 months after the first cancers were detected. All the first cancers in these 10 cases were esophageal cancers. The expected numbers of second gastric cancers were calculated as 4.55 in males and 1.20 in females by using the formula. The observed number exceeded significantly the expected number in males (p value = 0.042), which suggested that esophageal and gastric cancers might share some risk factor.