Materials and Methods
Colorectal cancer incidence rates aggregated over 5 years by cancer registry were obtained from IARC's Cancer Incidence in Five Continents (CI5) databases. The CI5 series aims to provide data on cancer incidence from populations all over the world for which high-quality data are available; therefore, data sources vary and include national registries (e.g., Czech Republic, New Zealand), local registries (e.g., Murcia in Spain, Miyagi in Japan), or aggregates of local registries (e.g., nine SEER registries which we used to represent the United States; refs. 9, 10). We restricted our analysis to 51 cancer registries that have incidence data beginning at least in 1983-87; for completeness, however, we also explored four additional registries in Africa with long-term data of varying time intervals. We examined the change in the incidence rates over 20 years among men and women by calculating the ratio of the incidence rates in 1998-2002 to that in 1983-87. Statistical significance was determined by calculating 95% confidence intervals for the rate ratios (11). In describing the change, the terms “increase” or “decrease” were used when the rate ratio was statistically significant; otherwise the term “stable” was used. In addition to calculating the change in incidence between the two time intervals, we present all available incidence rates since 1953-57 for select registries in Figs. 1 through 3 and in Supplemental Figs. S1 and S2 on the web. All rates were age-standardized to the 1960 world standard population in order to compare data across countries over time with different age compositions.