Rural representation of the surveillance, epidemiology, and end results database

摘要

Purpose SEER data are widely used to study rural-urban disparities in cancer. However, no studies have directly assessed how well the rural areas covered by SEER represent the broader rural United States. Methods Public data sources were used to calculate county level measures of sociodemographics, health behaviors, health access and all cause cancer incidence. Driving time from each census tract to nearest Commission on Cancer certified facility was calculated and analyzed in rural SEER and non-SEER areas. Results Rural SEER and non-SEER counties were similar with respect to the distribution of age, race, sex, poverty, health behaviors, provider density, and cancer screening. Overall cancer incidence was similar in rural SEER vs non-SEER counties. However, incidence for White, Hispanic, and Asian patients was higher in rural SEER vs non-SEER counties. Unadjusted median travel time was 53 min (IQR 34-82) in rural SEER tracts and 54 min (IQR 35-82) in rural non-SEER census tracts. Linear modeling showed shorter travel times across all levels of rurality in SEER vs non-SEER census tracts when controlling for region (Large Rural: 13.4 min shorter in SEER areas 95% CI 9.1;17.6; Small Rural: 16.3 min shorter 95% CI 9.1;23.6; Isolated Rural: 15.7 min shorter 95% CI 9.9;21.6). Conclusions The rural population covered by SEER data is comparable to the rural population in non-SEER areas. However, patients in rural SEER regions have shorter travel times to care than rural patients in non-SEER regions. This needs to be considered when using SEER-Medicare to study access to cancer care.