A national population-based study of incidence of colorectal cancer and age. Implications for screening in older Americans

BACKGROUND: Population-based screening programs with flexible sigmoidoscopy have been advocated as a means to reduce the death rate from colorectal cancer. Because other studies have suggested a greater prevalence of lesions inaccessible to sigmoidoscopy in older patients, the expected yield of this procedure may differ in these subgroups. METHODS: A 100% sample of Medicare beneficiaries 65 years or older with a first known diagnosis of colorectal cancer in 1987 was studied. Tumor site was divided into rectum, distal colon (distal to splenic flexure) and proximal colon. The analysis was also stratified by sex, race, and presence or absence metastatic disease, and incidence rates at each site by 5-year age group were calculated. RESULTS: Among the 75,266 patients studied, the incidence of colorectal cancer increased from 1.59 patients/1000 in patients age 65-69 years to 3.87 patients/1000 in patients age 85 years and older. Although the incidence rates at all three sites increased, the increment was greatest for proximal tumors. The incidence trends with age also persisted in an analysis of only metastatic lesions. Moreover, incidence rates were consistently higher in men than in women and higher in whites than in blacks at all sites, though the age-related increase in incidence was consistent among all four groups. CONCLUSIONS: The proportion of tumors beyond the reach of sigmoidoscopy increased with age, as did the actual incidence of accessible lesions. These patterns were also consistent in subgroup analyses. As the age-related increase in incidence was observed for metastatic tumors at all sites in the colon, age-related differences in screening and diagnostic evaluation alone do not adequately explain the findings. These data underscore the need for further studies of the relative benefits of cancer screening and pathogenic factors in tumor development in different subgroups of the older population.