The U.S. flag, next to a damaged Puerto Rican flag, flies in the municipality of Yabucoa.Carlos Giusti / AP

Just about nobody believes Puerto Rico’s official death toll for Hurricane Maria. Researchers and journalists alike generally accept that the island’s tally of 64 people killed by the storm last September is a massive undercount, so obviously inaccurate that the Puerto Rican government has agreed to review and revise its figures. But with Puerto Rico still in disarray—from the storm’s casualties, population changes due to migration, and the absence of basic services—information on the complete human cost of the catastrophe is still woefully incomplete.

What little is known, however, portends a grim conclusion: that Hurricane Maria is one of the most significant and destructive natural disasters in recent American history.

A new study in The New England Journal of Medicine, conducted in part by researchers at Harvard, sheds new light on what’s really happened on the island. The team found that more than 4,600 deaths were potentially attributable to the hurricane, a 70-fold increase over official estimates. The survey also measured high rates of migration among people displaced by the storm and, after it passed, long periods when residents faced a loss of basic services.

As I spent time reporting from Puerto Rico three weeks after Maria, two things became clear: The storm had had a staggering impact on the island, and it was almost impossible to translate that impact to observers on the mainland. People are used to gauging the scale of far-off events by relying on official estimates of death tolls, dollar amounts of damages, and the like. But in the immediate chaos following the storm, the “official” story was clearly inadequate. Some residents just went missing. Some got swept away in floods. Entire branches of extended families went silent. Mudslides and floods essentially turned remote places in the island’s mountainous interior into islands in their own right. In my attempts to assess the human burden of the hurricane, I asked everyone I interviewed—more than two dozen people—whether they knew someone who had disappeared, died, or fled to the mainland. Each person told me “yes.”

Official counts are obviously more difficult to perform than my anecdotal one, and not just because of scale: Further complicating the picture are mismatched systems in hospitals and morgues that might double-count some victims or misidentify others, as well as tough decision making over just what counts as a hurricane-related death. In its survey of more than 3,200 Puerto Rican households, the team behind the new study tried to get around those difficulties by asking families directly about the deaths of loved ones.

As The New York Times explains, the respondents reported that “38 people living in their households had died between Sept. 20, when Hurricane Maria struck, and the end of 2017.” When extrapolated to the island’s population of 3 million people and compared with deaths from the previous year, the researchers found 4,645 so-called excess deaths over that roughly three-month period. The researchers believe even that number might be low because of various biases in their survey, including their inability to measure any single-person households whose occupant died.

Of the deaths linked by household members directly or indirectly to Hurricane Maria, the largest mortality category was people who died from interruption of necessary medical services—about one-third of the recorded deaths. The survey also found a significant degree of storm-related migration; 3 percent of households had a member leave after the storm. Most of the migrants were young, with an average age of 25, and many were destined for the mainland United States. Additionally, many households reported major barriers to basic services, such as water, electricity, phone service, and medical care.

As The Times notes, these estimates have a rather high margin of error, owing to methodological constraints; the low end of the range reaches down to 800 deaths and the high end reaches as many as 8,000. Two other analyses of estimated mortality following the hurricane found just over 1,000 deaths.

But while this survey isn’t the definitive count of exactly how many people were killed by Hurricane Maria, what it does show with clarity is that the storm was on par with other recent natural disasters that have shaken the American populace and still reverberate today. For example, the official death toll for Hurricane Katrina in 2005 was just over 1,800 people, and it also kicked off a mass migration of displaced residents. Immediately after that storm, New Orleans lost half its population, and it appears that somewhere north of 100,000 people from the city and its surrounding areas have permanently resettled in the years since. Katrina also fundamentally shaped public policy, sparking conversations on climate change, disaster risk management, environmental justice, racial equality, and class.

Hurricane Maria looks increasingly like Katrina in terms of its effects. At the lower end of the Harvard researchers’ death range, it could match Katrina’s toll, and at the higher end, it could eclipse it. If the true death count is closer to 8,000, the September storm would be the single most devastating natural disaster to hit the United States since the Galveston hurricane in 1900.

Beyond raw death counts, the woes inflicted—and uncovered—by Maria are comparable to those revealed after Katrina. As I reported in October, many of the lasting effects of flooding, contamination, and ill health in Puerto Rico compounded along lines of race and class, just as they did after Katrina hit New Orleans. Maria has also shined a spotlight on the federal government’s relationship with its largest territory, further exacerbating one of the most consequential domestic migrations since the Dust Bowl and exposing the future difficulties of austerity on a debt-riddled island.

With another strong hurricane season expected to menace the Atlantic Ocean starting in three days, many residents of Puerto Rico still live under tarps instead of roofs. Electricity is still tenuous. And many residents still face barriers to health care left over from last season’s storm. As the effort to measure the fallout from Hurricane Maria continues, perhaps the most alarming implication of that work is that there’s potentially much more damage to come.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.