I guess Bart has knowledge of the GWU model used to assess Puerto Rico.
I'd like to hear more.
Thank you for asking.
My Master's was in Information Science, and I did some statistical analysis and database design for a couple of government agencies back in the 80s. Not sure what I can add to my earlier post. Probably the key fact is that this GWU analysis is about indirect causes from a storm and not just direct (a tree fell on the person, a tidal surge carried them away, etc.) cause. Most deaths happen later when the most vulnerable segments of the population are unable to get vital medicines and/or treatments, become ill from contaminated water supplies, get heat stroke when AC systems don't work, and so on. The coroners, when writing death certificates for such deaths, tend to give what's called the proximate cause, e.g. diabetic shock, heart failure from heat stroke, septicemia, etc.
Basically, when modern technological civilization falters, lots of people who were dependent on technology will die.
So that's why GWU uses a comparative method of statistical analysis, rather than relying on a list. And their method is sound, and gives a good estimate of storm-related deaths. 3000 is not precise, but it's very much in the ballpark. No one questioned this SAME ANALYSIS when it was used after Katrina.