(Bloomberg) – Hurricane Maria probably killed about 5,000people in Puerto Rico last year even though the official countremains at just 64, according to a Harvard University study released Tuesday.

Such mortality would far outstrip the 1,833 who died inHurricane Katrina in 2005, belying President Donald Trump's boastsabout the low death toll of the storm that occurred on his watch.Tuesday's study, published in the New England Journal ofMedicine, said delayed or interrupted health care accountedfor a third of the deaths in Maria, which hit the commonwealth inSeptember 2017.

Residents without basic services for months

The research, funded by the Harvard T.H. Chan School of PublicHealth and others, is likely to prompt renewed scrutiny of both thefederal government and the commonwealth's handling of the recovery.Large swaths of the island were without electricity and other basic services formonths after the disaster, in some cases leaving dialysispatients and the elderly without care. The White House has been theobject of criticism for its allocation of federal resources, butthe commonwealth generated the death count.

The toll has shaped public discourse around Hurricane Maria.When Trump visited the island about two weeks after thestorm, he seemed to pat himself on the back for the lownumbers, comparing them favorably to “a real catastrophe likeKatrina.”

At the time, the government had only certified 16 fatalities,despite what was plainly a major unfolding disaster. The Harvardsurvey conducted among 3,299 households suggests 4,645 more deathsoccurred between the storm's Sept. 20 landfall and Dec. 31 thanwould have been expected under normal conditions. The researcherssaid the figure was a conservative estimate; when they tweaked thedata to adjust for possible shortcomings in the sample, theyproduced a result of more than 5,000 deaths.

Deadliest storm since 1900?

That would make it the deadliest storm since the Great GalvestonHurricane of 1900, which killed an estimated 8,000 people.

For weeks after Maria, officials from Governor RicardoRossello's administration, led by public safety Secretary HectorPesquera, appeared at press conferences to defend their count, andtook umbrage at the suggestion that the real number was muchhigher. CNN and the New York Times, among others, subsequentlyproduced studies showing an undercount, but neither arrived at atotal as high as the Harvard study.

'No reason to want to conceal the truth'

Speaking to reporters Tuesday in San Juan, Rossello said hewelcomed the Harvard study and other resources to help betterprepare for the hurricane season starting June 1. Rossello, whoaligns himself with Democrats in mainland politics, said he had noreason to want to conceal the truth, and said the main objectivewas saving more lives.

Rossello's government has tapped researchers at GeorgeWashington University to conduct a review of the count, but thosefindings have not yet been released.

Related: Irma and Maria leave a wake of heartbreak anddevastation

“We had a protocol that really was subpar, and we recognize it,”the governor said.

At the same press conference, however, Pesquera drew adistinction between Harvard's survey and a tally “based onscientific data,” as he said the George Washington study would be.But he was subsequently asked whether he was minimizing theresults, and he gave a succinct response: “No, I'm not.”

San Juan mayor vindication?

San Juan Mayor Carmen Yulin Cruz, who has blasted the TrumpAdministration's response and drew fire from Secretary Pesquera forher questioning of the count last year, tweeted a news story aboutthe findings and hinted at vindication.

“Harvard agrees with what I've said since October 2017,” shetold her 243,000 followers. “Many deaths were caused by poormanagement of the crisis.”

Related: Global insured losses from disasters in 2017 werehighest ever, Swiss Re finds

Copyright 2018 Bloomberg. All rightsreserved. This material may not be published, broadcast, rewritten,or redistributed.