Average Kernel Density as Dependent Variable in Regression

I have point pattern dataset and I estimated a kernel surface from it. I want to use a census tracts average kernel density as a dependent variable in regression analysis. Is OLS appropriate or should I also fit a spatial regression model? The dependent variable would inherently be autocorrelated because of how the kernel surface is estimated, but if IV’s are autocorrelated I would still need a spatial model, correct?

Forgive me if this is an obvious question as I am fairly new to GIS and working with spatial data.