Death toll in Puerto Rico


New and shocking numbers about the actual death toll from hurricane Maria in Puerto Rico.
How can the official figure be 64 and the estimate by Harvard University 4600??:


Fake News. The study referenced in the article is only an estimate based on a random survey of death rates from Sept - Dec 2017 compared to a year earlier.

Not to minimize the tragedy of the whole thing, it is ridiculous to assign every “excess” death during those months to the storm itself. It always makes me sigh when we get a snow storm here and some old out-of-shape person has a heart attack while shoveling snow off his sidewalk. These deaths are invariably attributed to the snow storm even though had it not been snowing the same individual could have suffered the same fate walking up a tall flight of stairs.

In order for the PR study to be taken seriously, you’d have to monitor the deaths of the population for far longer than those specific months. If more deaths occurred in Sept-Dec causing less to take place in the four months afterwards, it all balances out.


The difference between a hard count (or, if you’re cynical, casualties the authorities couldn’t avoid counting) and a statistical estimate of casualties resulting from the hurricane and its aftermath. Here’s the Harvard methodology:

Researchers in the United States and Puerto Rico, led by scientists at the Harvard T.H. Chan School of Public Health and Beth Israel Deaconess Medical Center, calculated the number of deaths by surveying nearly 3,300 randomly chosen households across the island and comparing the estimated post-hurricane death rate to the mortality rate for the year before. Their surveys indicated that the mortality rate was 14.3 deaths per 1,000 residents from Sept. 20 through Dec. 31, 2017, a 62 percent increase in the mortality rate compared to 2016, or 4,645 “excess deaths.”

That is manifestly iffy, because it assumes that all “excess deaths” in the period under study are attributable directly or indirectly to the hurricane (e.g. people who died because they couldn’t obtain their normal treatment.) It also assumes the baseline mortality rate was developed using a method which was statistically similar to the survey done by the Harvard group. In other words, if the baseline mortality rate underestimated mortality, then the number of “excess deaths” would appear larger. It is not implausible that some unknown number of people die in far rural areas without the paperwork being submitted.

So I would suggest that a) the number of true casualties clearly exceeds 64 and b) we’ll never know by how much.


Edit: portofdc beat me to it :slight_smile:


But you had a good point about the baseline mortality rate. And your conclusion is a fairer way to summarize the flaw in the study.


Here’s a good, sober summary of the issues:



My statistician wife tells me the analysis in the paper was sound but the confidence interval was very wide, and the authors took the middle value, which was picked up by our ever-superficial media as a hard number.

A later, and better analysis of the paper is here:

The kerfuffle caused the PR health department to cough up their counts, which show an increase in deaths over the four month period of 1,397 and a claim that there are no unaccounted bodies.

So our Administration’s self-serving BS number of 64 is just that, and the best guess is that about 1000 people were killed.