BACKGROUND: Public health has recently discovered the importance of including “place” as a data element in surveillance systems. As of late, central cancer registries throughout North America have begun to geocode patient-level data using a standardized geocoding tool. Frontier states, like Montana, have had trepidations about geocoding personal health information for security and utility concerns. The aim of this study was to describe the quality of geocode results and to describe the utility of reporting cancer incidence by Census tract in a frontier state.
METHODS: Montana Central Tumor Registry cases diagnosed in 2008-2013 were geocoded using the North American Association of Central Cancer Registries’ geocoder, housed at Texas A&M University, in September 2015 to assign Census tract, county, latitude and longitude. Certainty codes were assigned based on quality of Census tract and GIS coordinate assignments.
RESULTS: Geocode results were produced for 99.9% of cases (39,048/39,094). 79.6% of cases were geocoded based on street address while 15.5% of cases were geocoded based on PO Box ZIP codes. Geocoded county did not match reported county for 342 cases (0.9%). For all analyses, geocoded county derived from street address was used rather than reported county. County reconciliation resulted in a case count percent change that ranged from +22.5% to -5.7%; this did not result in a statistical change for all-site age-adjusted incidence rate for any county. Reporting of 5-year all-site incidence data by Census tract would allow for 95.9% of tracts to report an age-adjusted rate and suppress 1.5% of tracts due to small numbers (<5 cases). Census tract reporting of prostate, breast, lung, or colorectal cancer would require suppression of 11.8% -21.4% of tracts.
CONCLUSIONS: Geocoded results show that reported county of residence was misclassified for a small percentage of cancer cases in Montana but this has negligible effects on incidence estimates at the county-level. Sub-county level cancer incidence estimation can be achieved for the most common types of cancer which would be beneficial to state and local public health officials. Surveillance systems in frontier states should consider geocoding their data.