Spatio-temporal Analysis of Breast Cancer Incidence and Its Associated Risk Factors in Nigeria, 2021–2025: An Ecological Study Using Routine Health Information System Data

Abstract Background Reported breast cancer incidence in Nigeria is shaped by differences in diagnostic capacity and reporting between states. This examined the spatial and temporal pattern of DHIS2-reported breast cancer incidence across Nigerian states and its association with behavioural, socioeconomic and environmental factors. Methods In this ecological study, annual breast cancer case counts for the 36 states and the Federal Capital Territory (FCT) from 2021 to 2025 were obtained from the District Health Information Software 2 (DHIS2) and converted to incidence rates per 100,000 women. Trends were assessed with the Mann-Kendall trend test and Sen’s slope, and spatial patterns with global and local Moran’s I. Associations between 2025 incidence and 15 risk-factor variables were examined with Spearman correlation and bivariate Moran’s I. Principal component analysis (PCA) reduced the variables to four components, used in ordinary least squares (OLS) and geographically weighted regression (GWR). Results Incidence varied widely, with extreme values in Sokoto (255.4 per 100,000) and Gombe (69.1) in 2022 that were probably reporting errors. No state showed a significant trend (national Sen’s slope 0.13 per 100,000 per year; p = 1.00). Global Moran’s I was not significant in any year (− 0.055 to 0.058; p = 0.39–0.97). Local Moran’s I identified only isolated clusters and outliers. Only formaldehyde (r s = − 0.38) and vegetation (NDVI; r s = − 0.40) correlated significantly with 2025 incidence, both negatively. Four components explained 82.7% of variance. In the OLS model (R 2 = 0.264; adjusted R 2 = 0.172), only the artificial light at night (ALAN) component was significant (coefficient 3.04; p = 0.012). GWR fitted better (AICc 250.9 versus 257.1; R 2 = 0.344), with local ALAN coefficients of 1.60 to 6.50, largest in the north-west. Conclusions Reported incidence showed no stable spatial clustering or temporal trend and probably reflects diagnostic access and reporting as much as disease occurrence. The spatially varying association with ALAN is ecological and exploratory. Wider cancer registration, including private facilities, is needed before spatial risk patterns can be interpreted confidently.