The number of deaths in Puerto Rico that can be attributed to Hurricane María has become a politicized subject, so much so that the media and the public it serves seems to have lost sight of a crucially important aspect of the number of fatalities being "reported." I put that word in quotations because now that the George Washington University (GW) study is finally out, it's important for everyone to understand what is being measured here.
First of all, the GW analysis, as well as the Penn State and the Harvard studies that preceded it, all look at “excess deaths” or excess mortality. The World Health Organization defines excess mortality as "mortality above what would be expected based on the non-crisis mortality rate in the population.” The scientists at these institutions, therefore, have all made estimates – emphasis on *estimates* – of how many people might still be alive, figuratively speaking, today had María not hit, and, arguably, Irma not sideswiped the island ten days earlier.
The GW study looks at the 6 month period from September 2017 to February 2018 and concludes that (at a 95% confidence interval CI) between 2,658 and 3,290 excess deaths occurred due to the impact of the hurricane(s). The midpoint of that estimate is the figure widely disseminated by media today: 2,975 deaths. A Penn State study looked at just September and October 2017, and estimated 1,085 excess deaths (95% CI) during that timeframe. The Harvard study, which has been criticized for its sampling methodology, show’s a very broad range of 793 to 8,498 (95% CI) excess deaths from between September 20th, the day María made landfall, and December 31st 2017. The large range denotes a high degree of uncertainty. But only the midpoint of that range was widely disseminated on traditional and social media: 4,645. I would argue that to the layperson, 4,645 was seen as a deterministic, very accurate figure of how many died as a consequence of Hurricane María in Puerto Rico. And now with the GW study 2,975 is seen as a “the exact number”. Neither is true.
Here’s the rub: hurricane fatalities are not customarily counted this way. The National Weather Service and National Hurricane Center count only direct deaths – those that can be attributed to the effects of the weather like flood drownings or flying debris, for example. These agencies also look at and separately list indirect deaths, like automobile accidents, electrocutions, and carbon monoxide poisonings from power generators, to name a few. Emergency management agencies follow the same model, and their officials are normally the ones briefing the politicians. So the politicians are used to counting deaths just like the National Weather Service does.
Is this the right way to count the dead? I’m not arguing for or against that point in this post. But, for better or for worse, it’s the way we’ve historically counted tropical cyclone deaths. Excess mortality studies are *not* done for all disasters, much less all hurricanes. You can see from my summary of the María studies that they can vary greatly in the time-frame selected to count the excess deaths, as well as methodology. Excess mortality requires that the investigators look at deaths that are ‘possibly attributable to hurricanes’, as stated in the GW study. These may include fatalities from people not able to reach a doctor, or those whose critical medical equipment failed due to lack of power. It may also count those that in the post-traumatic stress might have suffered a heart attack, or a plethora of other causes of death.
Because excess mortality studies are not available for most hurricane disasters, there’s no way for us to compare what happened in regards to excess deaths in Puerto Rico to any other past disaster. It can’t be compared to other American landfalling hurricanes either because deaths in those were counted based on the customary methods used by the National Hurricane Center, or with excess mortality studies that used drastically different methods, like this one for Katrina. Therefore the comparison between the initial official Puerto Rico government-reported María death toll of 64 and the huge numbers estimated by these studies is not an apples-to-apples comparison.
One excess death is one too many. It is tragic that very likely thousands of our (yes, OUR) brothers and sisters in Puerto Rico died, directly, indirectly, or tangentially, as a consequence of Hurricanes Irma and María. But reporting of these widely contrasting figures – 64 versus 2,975 – has lacked the above-described context in just about every on-air and social media report on the subject. When looking at the results of this GW study – the one now being used by the government of Puerto Rico as the “official” death toll – remember that it’s an estimate, and not a hard count. And remember how very differently they counted.