In 1988, cancer was made a mandatory reportable disease in California. The California Cancer Registry (CCR) became the repository for the data. New cases are reported by physicians, health facilities, laboratories, and death certificates, and there are penalties for failure to report cases. The actual abstracting of cases is done by certified Cancer Abstracters. Ten Regional Cancer Registries were set up to act as the first-level receiving facility and to provide quality assurance, coordination of reports from multiple sources for the same patient, provide technical help to those reporting cases, and then to report the data to the central registry in Sacramento. At this level, there is further quality assurance, elimination of duplicate entries, clearance with death certificates, and mutual referral of reports of patients actually residing in other states and countries. The CCR is certified by the North American Association of Central Cancer Registries (NAACCR) and currently the entire state is part of NCI's Surveillance, Epidemiology, and End Results Program (SEER). California has the largest population of any single total population-based registry in the world.
With California's very large population (36 million in 2003), CCR is the total population-based registry encompassing the largest population in the world, and probably, the most diverse. Sixteen years of data (2,387,316 individual cancers), with all patient, physician, and institutional identification removed, are available to researchers.
The data set includes demographics of the patient, features of the tumor such as stage and histology, and follow-up information. The data are made available to qualified researchers on a CD. Intercensal population data developed by the California Department of Finance have been supplied by the CCR.
There are no uniformly accepted definitions of "rural" and "urban" for registry data. We have chosen the one used by the California Rural Health Commission [16], using counties as the areas of interest. "Rural" counties are those with a population density of 250 people or less per square mile and no incorporated communities of more than 50,000 people. Using this definition, we identified 28 "rural" counties (population density range 1.56 to 98.2 people/sq. mi.) having a total population of 1,842,118. For an "urban" comparison group, we selected the two counties with the highest population densities (3,488.58 and 17,213.49) people per square mile and which contain 4,476,251 people), rather than those with the largest populations, leaving 28 other counties in a "mixed" category and not further studied. Since we do not have the patients' actual addresses, socioeconomic data and because these, as well as facilities, transportation, effects of educational programs, the role of mobile mammographic facilities, etc., often change with time, we cannot evaluate what part may be played by any of these. Moreover, one must be aware that even in the two groups, there may be county-to-county differences.
Using the above definitions of "rural" and "urban", all of the cases which had the years of diagnosis and Summary Stage of breast cancer, colorectal cancer, and cutaneous malignant melanoma from the years 1988 through 2003 were studied. For each type of cancer, data were tabulated by years of diagnosis and "early" versus "late" stage, using the "EpiInfo6" (CDC and WHO) software. As noted above, there are multiple levels of quality assurance, and the standards of the North American Association of Central Cancer Registries (NAACCR) are all met or exceeded. The tissue diagnoses of the original pathologists are accepted because central review would be a fiscal and operational impossibility. It should be remembered that the patient's therapy is based on the diagnosis by the local pathologist.
Early" stage is defined as a Summary Stage of carcinoma-in-situ or localized tumor, and "late" stage is a Summary Stage of regional or distant disease. Summary Stages are rigorously defined by the NCI's Surveillance, Epidemiology, and End Results Program (SEER), the American Joint Commission on Cancer, and the American College of Surgeons. The definitions of stages differ from one primary site to another. While for clinical purposes it might be preferable to use the TNM System, there are no TNM staging data for the earlier years of CCR for all sites and thus not possible to use the TNM system. Moreover, to allow comparison with earlier registry data as well as data from other registries, the Summary Stage is more useful. When the data exist, cross- tabulations of Summary Stage "localized" against "T1N0M0" show a correlation of over 90%. We do not have data on Breslow stage. Staging data are not available for every case. Finally, while the CCR has information about ethnicity, there are not enough numbers in the non-white rural population to draw any meaningful conclusions to include in this manuscript. Because this study used the above data, it was exempt from the usual methods of Ethical Approval.