Age-period-cohort Analysis of Renal Cell Carcinoma in United States Adults

Abstract

To determine the extent to which the year of diagnosis, year of birth, and age at diagnosis influence the incidence trends of kidney cancer in the United States. Cancer registry data from the National Cancer Institute's Surveillance, Epidemiology, and End-Results (SEER) program were obtained for 64,041 patients with kidney cancer diagnosed between 1973 and 2008. Overall and age-specific incidence rates were calculated and adjustments were made for birth cohort and period effects. Results were stratified by race and sex. Age-period-cohort analysis was used to examine the effects of age, year of diagnosis (period), and year of birth (cohort) on incidence trends. The overall age-standardized annual incidence per 100,000 increased during the study period (1973 to 2008) by race, from 6.75 (95% confidence interval, 6.18-7.36) to 19.56 (18.85-20.20) among whites, from 5.31 (3.50-7.71) to 25.38 (23.00-27.92) among blacks, and from 5.61 (3.50-8.50) to 13.98 (12.41-15.71) among other races; and by sex, from 9.44 (8.49-10.47) to 26.48 (25.39-27.60) among men and from 4.21 (3.65-4.84) to 13.38 (12.64-14.11) among women. Age-period-cohort analysis revealed a strong influence from period and cohort effects. The 1983 birth cohort, for example, had a 2-fold increase in kidney cancer (incidence rate ratio, 1.93 [1.63-2.25]) compared with the referent 1948 cohort. From 1973 to 2008, the incidence rate of kidney cancer increased for each sex and race across all age groups. Age-period-cohort models revealed that period-related factors, although significant, cannot alone account for these unfavorable temporal trends.