Spatially varying effect estimates of mediation in the presence of spatial auto-correlation

A GWR/SVC framework with the product method accurately and precisely estimates spatially heterogeneous natural direct and indirect effects while handling spatial confounding in grid simulations.

• Why it matters: Mediation analysis in epidemiology often assumes homogeneous effects, yet environmental and social settings can show spatial dependence and heterogeneous mediation roles.
• What they did: The paper compares spatial mediation approaches and runs extensive simulations, then applies the method to PM2.5’s spatially varying mediating role between income and cardiovascular diseases.
• The result: The combined geographically weighted regression and spatially varying coefficient approach is accurate and precise for spatially heterogeneous effect estimates in grid settings, and it can extend to other mediation and confounding structures.

AI summary of the abstract · may contain errors · not medical advice · read the paper.