Also a good point. My main point is when looking at drug deaths, the total should reflect (deaths caused-deaths prevented), rather than a raw count of total deaths. At least when looking at it from a macro, pure optimization perspective.