MARIA was a brief visitor to Puerto Rico. The category-4 hurricane made landfall at 6am on September 20th last year and 11 hours later she was gone. She left a trail of destruction. Some 300,000 people were displaced; and the death toll? No one knows for sure. 

The official estimate of 64 deaths seemed measly by contrast. That number includes only those directly killed by the hurricane, from flying debris and the like. Importantly it excludes indirect deaths: disruptions to medical care, for example. A back-of-the envelope calculation by The Economist of excess mortality above that expected by deaths in previous years puts the toll at about 1,200. 

A paper published this week in the New England Journal of Medicine provides perhaps the best estimate to date, albeit an uncertain one. Using clever sampling techniques of some 3,000 Puerto Rican households, on a door-by-door basis, the researchers attribute some 26 deaths to Maria in 2017. Applying that number to the population as a whole gives an estimate of 793 to 8,498 deaths. (The confidence intervals are large to account for the tiny number of actual deaths reported.) That range, however, is likely to be an underestimate: single-occupancy households that were interviewed, by definition, survived the hurricane. De-biasing those households for implicit survivorship provides a final estimate of some 1,506 to 9,889 deaths. 

That is a large range indeed. But counting deaths in the wake of disasters is not a problem unique to Puerto Rico. The death toll from Hurricane Katrina in 2005 is still disputed.