Determining the spatial non-stationarity underlying social and natural environment in thyroid cancer in China

Background: Light at night (LAN) is a physiological environmental factor related to thyroid cancer (TC). The spatial re-lationship between the number of TC incident cases, LAN, air pollution and other macro social factors and stationarity needs to be further examined to provide evidence for regional control of TC. Methods: Spatial econometrics methods for spatial nonstationarity were used to explore the impacts of LAN, air pollut-ants, economic factors, and population size on the number of TC incident cases in 182 Chinese prefecture-level cities and the local coefficients were further tested for nonstationarity. Temporally weighted regression (TWR), geographi-cally weighted regression (GWR), and geographically and temporally weighted regression (GTWR) were compared in this study for model selection. Results: Based on the ordinary least squares (OLS), LAN, air pollutants, and urbanization all significantly affected the number of TC incident cases. GWR had the best goodness of fit, and the coefficients of all the variables passed the nonstationarity test. The strong positive impact of LAN was mainly concentrated in North China, air pollutants in Cen-tral China and neighboring regions, and urbanization in the eastern coast of China. Conclusions: The locational factors of the prefecture-level city influence the spatial pattern of the number of TC incident cases. Governments should pay attention to this influence, adhere to the Health in All Policies principle, and formulate region-specific policies based on regional characteristics, which this study provides updated evidence for.