Understanding population-based site-specific cancer incidence rates in the USA.

As compared with conventionally reported national population-based incidence rates, incidence rates better represent the "burden" of disease if they remove prevalent cases from the denominator. In order to reflect the "risk" in a disease-free population, rates should both exclude prevalent cases from the denominator and second or later diagnosed cases at the same site from the numerator. Five common cancers were evaluated through a correction method using 2005-2007 Surveillance, Epidemiology, and End Results Program data to determine the extent of difference between conventional and corrected incidence rates. These corrections lowered the incidence rates 4.0-5.8% for female breast cancer, 4.6-7.6% for melanoma, 3.0-4.0% for colorectal cancer, and 2.1-2.5% for lung and bronchus cancer. Corrected incidence rates for prostate cancer were 9.9-13.7% higher. In cancers with either high prevalence and/or high occurrence of multiple primaries at the same site, corrected population-based incidence rates are warranted.