Municipal-level estimation of colorectal cancer incidence in Brazil and its association with social and health system determinants: A nationwide spatial analysis.
BACKGROUND Colorectal cancer (CRC) incidence varies in relation to social determinants of health and health system capacity, particularly in settings with territorial inequalities. This study aims to estimate CRC incidence at the municipal level in Brazil and assess its association with social, demographic, and health system determinants, accounting for spatial heterogeneity. METHODS Population-based cross-sectional ecological study with spatial analysis. Municipal CRC incidence for 2024 was estimated using corrected mortality data (1980-2023) and state-level incidence-mortality ratios from Population-Based Cancer Registries (PBCR). Estimated age-standardized incidence rates (estimated ASR) were calculated using the Segi-Doll standard population. Spatial autocorrelation was assessed using Global and Local Moran's I, while associations were examined using Ordinary Least Squares (OLS), spatial autoregressive (SAR), and Multiscale Geographically Weighted Regression (MGWR) models. RESULTS Higher estimated ASRs were observed in the South, Southeast, and parts of the Central-West, while lower rates predominated in the North and Northeast. In the SAR model, positive associations were observed for the Municipal Human Development Index (β = 1.48; p < 0.001), hospital beds (β = 0.023; p = 0.002), premature mortality due to neoplasms (β = 0.002; p < 0.001), and ultra-processed food consumption (β = 0.027; p < 0.001). Negative associations were found for under-1-year mortality (β = -0.009; p = 0.004), dependency ratio (β = -0.013; p < 0.001), and hospital admission rates (β = -0.002; p < 0.001). MGWR showed spatial heterogeneity, with MHDI and ultra-processed food consumption significant in all municipalities and dependency ratio and under-1-year mortality in over 98%, with stronger effects in the North and Northeast. CONCLUSION Estimated CRC ASRs showed marked territorial heterogeneity and was associated with socioeconomic, dietary, and health-system characteristics. Because the estimates were derived from mortality and state-assigned I/M ratios, the observed patterns may reflect a combination of disease occurrence, healthcare access, data quality, and model assumptions. These findings may support territorially tailored cancer surveillance and control planning.