I don't think you should be expecting people per square mile to scale 1:1 with mortality rate anyhow. I would guess putting twice as many people within a square would result in the virus spreading an order of magnitude faster. Similar to how increasing infection radius in an SIR model increases spread by much more.