fb-pixel Skip to main content

Nearly 3,000 deaths in Puerto Rico linked to Hurricane Maria, new report says

A neighbor walked out on a piece of metal that fell on Carmen Charriez's home when Hurricane Maria hit.
A neighbor walked out on a piece of metal that fell on Carmen Charriez's home when Hurricane Maria hit. Jessica Rinaldi/Globe Staff/file 2017

NEW YORK — A long-awaited report on the devastation wrought by Hurricane Maria in Puerto Rico says that nearly 3,000 more deaths than expected occurred in the months after the storm, the first official outside evaluation of the toll in a disaster whose damage in some cases took months to unfold.

The report, made public Tuesday by researchers at George Washington University’s Milken Institute School of Public Health, was commissioned by the governor of Puerto Rico after The New York Times and other media outlets and researchers last year estimated that the death count far exceeded the government’s official toll of 64.


The new research, which was performed independently of the government, compared the actual death rate with what would have been expected had the storm not occurred — a method used in several earlier analyses. In December, The Times found that 1,052 more people than usual died in the 42 days after Hurricane Maria made landfall.

By the new tally, 22 percent more people died during the roughly six months after the storm than would normally be expected in that period.

“Hurricane Maria was a catastrophe of historic proportions, as never seen or lived before in the United States,” Carlos Mercader, executive director of the Puerto Rico Federal Affairs Administration, said in an e-mailed statement. He said the new analysis “is sobering, and its insights make clear that Congress and FEMA must work with us to establish a better system for the preparation and distribution of supplies ahead of future disasters.”

The researchers released few details of their methodology or analysis, making it difficult to assess the quality of their work, which was carried out at a cost of $305,000 and included an analysis of how the government certified deaths and communicated about them.

They wrote that “national and international experts in different fields” had reviewed their methods. The report has not yet been subjected to the more rigorous process of being published in a peer-reviewed medical journal, though it has been submitted for publication.


A study by Harvard researchers published in the New England Journal of Medicine earlier this year estimated the death toll could range from 800 to more than 8,000 people, and was based on a household survey.

The new study has two main differences from most previous analyses. Researchers looked at deaths for a longer time period, from September 2017 — when two hurricanes, Irma and Maria, hit the island in close succession — until February 2018. Deaths continued to be elevated throughout this period.

The researchers also adjusted their calculations for what they estimated to be an 8 percent drop in the population after the storm and prolonged power failures, when thousands of people fled for the mainland. Using data from the territory’s Bureau of Transportation Statistics as well as from a government survey of airline travelers, the researchers estimated that nearly 280,000 fewer people were living on the island in February 2018 as in September 2017. That out-migration made it even more significant that deaths had increased compared with previous years.

People in poorer municipalities and older men had a higher risk of death throughout the study period, the researchers found.

“The lesson from this is that efforts for assistance and recovery need to focus as much as possible on lower-income areas, on people who are older, on people who are more vulnerable, because of having fewer of their own resources,” Lynn Goldman, dean of the Milken Institute, said at a news conference Tuesday. She said a similar analysis has not been carried out in previous disasters. “I think we fail to appreciate the multitude of ways these major disasters impact people’s health and lives.”


The researchers also tried to understand why the official death count was so low, despite the overall spike in deaths.

In completing death certificates, which were the basis for the official count, doctors followed conventions that considered the disaster a factor only if it directly contributed to a death — for example, through “structural collapse, flying debris, floods and drownings.” The doctors at the time were not aware of new guidelines from the federal Centers for Disease Control and Prevention, released the month after the hurricane, which recommend that doctors also consider a natural disaster’s indirect impact — for example, through delayed access to health care.

Some doctors told the researchers in interviews that they were confused about the new guidelines, or concerned that listing the hurricane as a contributor was too subjective, or could raise questions of liability. The death certificates were otherwise highly complete, although their completion was often delayed.

“Most physicians have no formal training in completing a death certificate and thus are not aware of appropriate death certification practices, especially in a disaster setting,” the study found.

The researchers found this problem was not isolated to Puerto Rico. “The lack of training on death certificate completion is also a problem on the mainland,” they wrote. “To date, there are no formal courses on death certification in natural disasters for persons who fill out death certificates.”


The proposed next phase of the study will examine the death certificates recorded in the months after the hurricane and will try to determine, through interviews with family members and others, whether individual deaths should be attributed to the storm. Researchers said they did not know when that next phase would be completed.

The research was performed in collaboration with the University of Puerto Rico Graduate School of Public Health. The actual number of excess deaths was statistically estimated to be in the range of 2,658 to 3,290.