Age-period-cohort modelling of large-bowel-cancer incidence by anatomic sub-site and sex in Denmark.

In a previous investigation, statistical modelling was used to examine the relationship between large-bowel-cancer incidence and age, time period and birth cohort by anatomic sub-site and sex, using data from the Connecticut Tumor Registry (CTR) for the period 1950 to 1984. This analysis revealed differences in age-period-cohort patterns that suggested etiologic distinctions among sub-site groupings and between the sexes. To test the generalizability of the Connecticut findings, we have conducted a similar age-period-cohort analysis using data from the Danish Cancer Registry (DCR) for the period 1953 to 1987. Cancers of the large bowel were classified into 6 anatomic sub-sites: cecum, ascending colon, transverse colon, descending colon, sigmoid colon and rectum. Data were fitted to log-linear age-period-cohort models. If we interpret differences in age-period-cohort patterns as reflecting etiologic distinctions, the Denmark analysis, in conjunction with the Connecticut findings, was consistent with there being etiologic distinctions between cancers of the colon vs. the rectum in both males and females, between cancers of the cecum and the ascending colon vs. the remainder of the colon in females and between males vs. females for cancers of the sigmoid colon and rectum. Cancers of the cecum and the ascending colon were the most similar between males and females. Due to the ambiguities of age-period-cohort modelling, these should be considered only tentative conclusions that can be tested by analytical epidemiologic studies.