Geospatial heterogeneity of hotspots for incidence and late-stage diagnosis of breast, colorectal, and lung cancer

Abstract Purpose Cancer control relies on the identification of populations at risk (hotspots) of new or late-stage cancer diagnoses. However, the extent to which hotspots differ between cancer sites or between outcome measures has been poorly characterized. We sought to determine the geospatial heterogeneity of hotspots of breast, colorectal, and lung cancer incidence and late-stage diagnoses. Methods We identified adult patients diagnosed with female breast, colorectal, and lung cancer between 2010 and 2019 in Indiana. To identify hotspots for incidence and late-stage diagnoses, we disaggregated the patient residential location information from the Census block group level to the approximated individual point level. Statistically significant hotspots were identified with kernel ratio estimation. The total areas of hotspots and the overlap between hotspots were compared. Results 133,773 patients diagnosed with breast (n = 54,903), colorectal (n = 28,594), and lung (n = 50,276) cancer were included. Geospatial visualization demonstrated marked spatial deviation, with little overlapping area between incidence and late-stage hotspots for all three cancer sites (1–266 km 2 ). However, there was slightly greater overlap in late-stage hotspots between the different cancer sites, with total overlapping hotspot areas ranging from 24 to 516 km 2 . Conclusions Our results demonstrate considerable geospatial heterogeneity of hotspots between different outcome measures and different sites of cancer. The use of disaggregated spatial data enables a more granular, precise comparison of cancer hotspots. The greatest overlap was seen between incidence hotspots of breast and colorectal cancer, which suggests similar spatial drivers and the potential for coordinated cancer control strategies.