Why Puerto Rico Can’t Nail Down Hurricane Maria’s Death Toll

It’s been almost almost a year, but there’s no consensus on how many people died. Why not?

shoes holding puerto rico flag death toll maria katrina counting death statistic demography george washington harvard university climate change global warming

Ricardo Arduengo/Getty

On Thursday, the Puerto Rican government announced that Hurricane Maria contributed to the deaths of nearly 1500 people—a significant increase from the initial death toll estimate of 16.

“According to initial reports, 64 lives were lost,” a draft of a not-yet released government report noted, according to The New York Times. “That estimate was later revised to 1,427.”

Puerto Rico’s death toll count from the hurricane has long beleaguered investigators and researchers alike, and the number remains fuzzy despite official and academic reports, ranging from as low as 16 people in the immediate aftermath of the disaster to as high as 8,498 dead, according to a June report in the New England Journal of Medicine.

The disparity is vast and illustrates the difficulty even experts have answering the simple question: How many people died from Hurricane Maria?

For months, the numbers have volleyed back and forth between official and academic accounts. Hurricane Maria struck Puerto Rico on September 20, 2017. On September 25, about a week ahead of President Trump’s visit, the Puerto Rican government announced its initial count of 16 people dead. When Trump met Governor Ricardo Rosselló a week later, he praised the island for dodging a “real catastrophe” like Hurricane Katrina, which counted deaths in the thousands. On November 2, about a month and a half after Maria struck, the government raised the official toll to 55.  

Those low numbers seemed odd to experts, who countered with new estimates, nearly all of which posited that at least ten times as many people had been killed. A few weeks after the Puerto Rican government upped its estimate to 55 people, a team at Penn State released estimates that 1,085 people were actually killed.

How many people died from Hurricane Maria?

That kicked off a series of estimates and counter-estimates between the government and academic experts. On December 9, The New York Times reviewed the daily mortality data from Puerto Rico’s vital statistics bureau and reported a similar number to Penn State, 1,052. That very same day, the Puerto Rican government also updated their death toll to 64. In June, a team led by Harvard’s T.H Chan School of Public Health reignited the debate by estimating an even higher death toll in the New England Journal of Medicine, estimating that between 793 and 8,498 had died. Their midpoint estimate was 4,645 people dead—four times the number academics were initially thinking and nearly 73 times the last official Puerto Rican estimate.

Even in the aftermath of Thursday’s announcement, the government is sticking to their December figure. Pedro Cerame, a spokesman for the Puerto Rican government’s Federal Affairs Administration, told the Times that although the governor has repeatedly acknowledged that death toll is likely higher than 64, the government is not ready to publicize 1,427 as an “official number.” The final version of the report, he added, will say that the 1,427 deaths “may or may not be” attributable to Maria.

The numbers vary so much in part because there is no one way to count what it means to die from a hurricane, said John Mutter, a professor of Earth and Environmental Sciences at Columbia who specializes in studying natural disasters.

“There’s no standards,” he told The Daily Beast. “Everybody chooses what they want to choose.”

Although the CDC provides voluntary guidelines for calculating death tolls in the wake of natural disasters, there are no mandatory requirements. Researchers, news organizations, and governments are all free to use their own methodology. But these choices—who to include, how to count them, and how long to count them for—vary drastically between studies, causing lasting confusion about the numbers they intend to clarify.

One of the most common methods is the simple vital statistics count, which requires very little data: comparing the number of deaths in a period before an event like Maria with the number in the months afterwards. A team from the University of Puerto Rico at Mayagüez, led by statistics professor Roberto Rivera and math professor Wolfgang Rolke, were one of the many to use this strategy.

Get The Beast In Your Inbox!

Daily Digest

Start and finish your day with the top stories from The Daily Beast.

Cheat Sheet

A speedy, smart summary of all the news you need to know (and nothing you don't).

By clicking “Subscribe,” you agree to have read the Terms of Use and Privacy Policy
Thank You!
You are now subscribed to the Daily Digest and Cheat Sheet. We will not share your email with anyone for any reason.

In Rivera and Rolke’s study, published in the journal Significance this February, the pair calculated the average daily number of deaths in the 19 days before Maria struck (83.3) and subtracted it from the average daily number of deaths that occurred between September 20 and October 31 (102.8). That allowed them to determine how many additional deaths occurred each day (about 19.5). They then multiplied the result by 42, to determine how many deaths occurred in the six weeks following Maria. Accounting for statistical uncertainty, they determined that between 605 and 1,039 people were killed as a result of the hurricane during that time period.

The numbers get even higher given that tens of thousands of people left the island in the weeks following Maria, Rivera told The Daily Beast. Accounting for migration, estimates go up “by about 100 on all bounds” to 726 to 1150, he said. The Penn State team (led by Alexis Santos, a demographer at Penn State), the New York Times, and Puerto Rico’s Center for Investigative Journalism all used relatively similar methods, over similar time periods, and came up with estimates in the same range: 1,085, 1,052, and 985 hurricane-related deaths, respectively. Santos’ team updated their estimate in a research letter published in the Journal of the American Medical Association on August 2nd, now positing between 1,006 and 1,272.

The Harvard team, meanwhile, took an entirely different approach. They surveyed 3,299 Puerto Rican households, and asked residents in-person if their relatives had died as a result of Maria. They found 38 deaths, calculated a mortality rate, and extrapolated that rate to the island’s entire population to get an expected additional death toll between 793 and 8,498.

If the government is putting out an unbelievable figure, like 64, as an official count, and simultaneously not sharing the vital statistics system data, then I think it is important for us to use alternatives that permit us to get something closer to the truth.
Rafael Irizarry, professor of applied statistics at Harvard

Many researchers pointed out that that’s an extremely wide range. “It’s like me telling you, ‘I caught the criminal who stole my house,’” Santos said. “‘Oh, where is he?’ Well, he’s between New York City and Seattle, which lie [on] opposite sides of the continent.’”

But it was their midpoint estimate of 4,645 additional deaths that captivated the media, largely because it so drastically exceeded all previous estimates. It fed into a preexisting narrative—that the government was massively underreporting the devastation wrought by the hurricane on an unprecedented level.

Even that 4,645 number is incredibly misleading. It might be the midpoint of the estimated range, but there’s no indication that it’s anywhere near the actual toll. As an exact measurement of Maria’s destruction, “that number is simply dead wrong,” Rolke said.  

Santos agreed. Although including a midpoint estimate is standard academic practice, he worried that the number could cripple efforts to communicate a more accurate figure. “It really hurt the discussion of this topic,” he said. “Whatever number comes out now is going to be substantially lower than 4,000. So anybody who sees an analysis of less than 4,000, they might not believe it.”

Rafael Irizarry, a professor of applied statistics at Harvard and one of the authors of the Harvard study, told The Daily Beast via email that methods like Santos’ or Rivera’s would certainly give a “more precise” accounting of Maria’s death toll. But he defended the validity of his study.

“I personally don't think estimating exactly how many people died is the most important challenge,” Irizarry said. “We do want to have a general sense about how much the death rate increased [...] In my opinion, the more important findings in our paper were that the death rate appeared to be elevated up until December and that a large proportion of those surveyed stated that lack of access to medical care as a cause of death.

“If the government is putting out an unbelievable figure, like 64, as an official count, and simultaneously not sharing the vital statistics system data,” Irizarry continued, “then I think it is important for us to use alternatives that permit us to get something closer to the truth and then inform the public, public health professionals, and policy makers to make up for the government’s unwillingness to do so.”

The government contests the claim that they were ever trying to hide data, contending that the simple vital statistics information—the raw number of people who died in the months following the hurricane—was always available, and that releasing the names and death certificates of the deceased would have violated the families’ right to privacy.

“The registry [gathers] some private information from the deceased that may also include information [about] their family members, [that] may also include information as to the causes of the death,” Cerrame told The Daily Beast. “That’s what we were trying to protect at that moment. We were trying to comply with the information requests, but also [be] responsible with the privacy of the deceased and/or the family members.” Ceramme did not return a request for comment about Thursday’s figures.

But no matter the process or the availability of data, Irizarry’s words underscore the most important, objective conclusion from these studies: The number wasn’t 64. It was nowhere near 64.

How, then, did the Puerto Rican government come up with such a low figure? Robert Anderson, chief of the Mortality Statistics Branch of the CDC’s National Center for Health Statistics, explained that the government only counted death certificates that linked the death directly to Hurricane Maria, and that this method tends to cause significant underreporting.

According to Anderson, that’s because doctors don’t often understand that’s it important to include that information on the death certificate. He noted that the CDC provides guidelines for reporting deaths, which the Puerto Rican government says it ascribes to—but that it’s unclear how well they’re followed in practice.

It’s also because in the absence of mandatory, well-communicated guidelines, deciding whether a death is “hurricane related” is an extremely subjective determination. Direct deaths, like drowning or being struck by a falling tree, are easy.

In the case of a hurricane, however, there are harder questions: Do you count someone who caught an infection from a cut sustained while escaping the hurricane? Someone whose oxygen machine shut off when they lost power? Someone who committed suicide because they were so overwhelmed by Maria’s devastation?

People need to understand that this was a seriously horrible event. [People say] ‘Well, Puerto Rico was only 64, it can’t have been that bad.' And that is obviously not true.”
Wolfgang Rolke, math professor, University of Puerto Rico at Mayagüez

The Puerto Rican government has made clear that all hurricane-related deaths—whether direct or indirect—should count in the official total. But it’s not that simple. In order for bodies to be counted, a doctor must decide that the death qualifies as hurricane-related, and report it to the Bureau of Forensic Sciences in San Juan. The body then must be examined in-person by a Bureau medical examiner.

It’s easy to see why this system likely undercounted so massively, as evidenced by a series of excellent investigations into the system’s breakdown. Funeral directors and physicians didn’t fully understand the process, or the importance of reporting hurricane-related deaths to the capital. People began cremating their dead before they could be examined, because they couldn’t afford a funeral.

Despite all of these obstacles, the deputy director of the Bureau told CNN that her team had received 843 deaths for analysis in the first month alone—but somehow, only 55 of those were counted in the official toll. Five were still pending review at the time. A second CNN investigation revealed that at least six people were listed as a victim of a “cataclysmic storm” on their death certificate, but did not make the government’s eventual list of 64. And the names on this list matter. Relatives of those whose deaths made the official list can apply for funeral assistance money from FEMA, while others may not be able to afford to bury their dead.

Even if there were only 64 hurricane-related death certificates, why would Puerto Rico’s government stick to this method for so long, and not report the conclusions of other studies? Mutter called the decision “bizarre,” noting that governments that have fewer resources generally highball, not lowball, death toll estimates to draw in more international aid.   

Rolke agreed, but noted that the government may not be entirely responsible. “Obviously, on the one hand, the reaction is unbelievable. I mean, ‘How could they? How dare they?’ But on the other hand, they don’t come up with these numbers out of thin air,” he said. “They have rules and regulations. So rather than wanting to blame the people who do this, I would think it's more of a problem with the official methodology with which bodies are counted.”

Anderson agreed. “When you have a system in place to count deaths, you generally assume that that works,” he said. “I think a lot of people didn’t understand the limitations of the system, particularly in a case like this. So they said, ‘Well, our vital records people, our official count of deaths, said 64. So that must be right’—not realizing that there are specific limitations related to these kinds of events.”

Carlos Mercader, executive director of the Puerto Rico Federal Affairs Administration, agreed with this conclusion, noting that since the end of 2017, the governor has acknowledged that their process was insufficient, and that the real death toll is almost certainly higher. He added that the government proved their commitment to accuracy by commissioning a $300,000 project from George Washington University’s Milken Institute School of Public Health to re-count the death toll and propose a new counting procedure.

The George Washington study, however, has already provided some cause for concern. The first section of the report was scheduled to be released in May. But instead, it was released to news outlets in late July—with no numbers or significant findings. The GW Hatchet reported that researchers were “still acquiring data,” and that the first section of the report would now be released sometime this summer.

The second section, in which researchers are expected to conduct an extensive review of the death toll accounting, is still scheduled for early 2019.

But George Washington University isn’t the only group that will propose a new way to take death tolls. In early June, Rep. Nydia Velázquez (D-NY) and Sen. Kamala Harris (D-CA) introduced a bill that would require FEMA to work with the National Academy of Medicine in order to redesign and implement a standard death toll protocol. The Counting Our Unexpected Natural Tragedies’ Victims Act (COUNT Act) would work to ensure that a better counting policy became standard, both so that the nation could understand the full scope of the disaster and so families could apply for FEMA’s funeral benefits.

Mutter emphasized that finding a way to measure an accurate death toll is vital. “We’re human,” he said. “We want to know this. We scale tragedy by human losses, not by financial losses. It’s a human thing. We want to know.”

Rolke agreed. “People need to understand that this was a seriously horrible event. [People say] ‘Well, Puerto Rico was only 64, it can’t have been that bad,’” he said. “And that is obviously not true.”