Colon Cancer in Pennsylvania: Predictors of County-Level Incidence and Geographic Clustering

Background: The incidence of colon cancer in Pennsylvania is higher than the US. County-level analysis and mapping of cancer clusters may help prioritize cancer control. Purpose: 1) Determine whether geographic clustering of colon cancer incidence occurs and whether cluster patterns differ by stage and site (left- versus right-sided); and 2) Identify county-level characteristics associated with incidence. Methods: Cases diagnosed between 1994-2002 were identified through the Pennsylvania Cancer Registry. Principal components analysis was used to create a county-level SES index based on the 2000 Census. County-level colon cancer screening and obesity prevalence were derived from the Behavioral Risk Factor Surveillance System. SaTScan was used to determine geographic clustering by county. Poisson regression estimated incidence rate ratios (IRR) and 95% confidence intervals (CI) according to the county-distribution of age, sex, race, SES index, obesity, screening, and rural residence as defined by the Beale rural-urban continuum. Results: We identified 54,617 colon cancer cases. Lower SES was associated with higher left-sided colon cancer incidence (IRR=1.11, 95% CI: 1.03-1.19, p-trend=0.003), adjusting for age, sex, race, screening and obesity. Clustering of late stage cancer persisted in two metro regions, including a three-county area in the north east (Beale=2) and two southeastern counties (Beale=1). Adjustment for SES revealed a cluster of late stage cancer incidence in rural non-metro counties in the western portion of the state. Conclusion: Geographic clustering of late stage colon cancer occurs. The association between left-sided colon cancer and low SES is worthy of further investigation.