In poorer areas, mortality rates remained elevated, the study found. Much of Puerto Rico continued to have no electricity for months, and the US and territorial governments struggled to carry out disaster recovery efforts.
The GWU researchers decided to look at six months of data because they assumed that, during that time period, the elevated mortality rate would return to a normal level. But in a briefing at the university today, the leaders of the research effort said that in low-income areas the mortality rate remained somewhat elevated even after six months.
They said further investigation of mortality rates after February could push the estimate even higher.
"We did not capture the entire epidemic. There may have still been a few in March," said Lynn Goldman, dean of the Milken Institute School of Public Health.
The lack of power, intermittent water service and damage to roadways and bridges cut off entire communities from basic necessities. But the GWU report does not specify why people died; it is a statistical study.
The researchers did not go household to household, though they said they hope to conduct a more detailed investigation in the future if funding permits. It is unclear who would fund the study and what role the Puerto Rico Government would play.
"We can come up with a hundred different hypotheses," Goldman said. "What we don't have is the ability today to tell you these are the factors that caused this."
She threw out a few possibilities: "Many people still did not have power restored to their homes. We do know that when there is a loss of power, that causes a number of issues, everything from having to operate medical equipment, to having to do work manually that normally a machine would do."
The GWU report estimates that 2658 to 3290 excess deaths occurred between September and February. The poorer and older the resident, the higher the risk for death, especially among men older than 65.
Eighteen municipalities spread across metropolitan San Juan, western Puerto Rico and Vieques experienced some of the highest increases in mortality. Maria's aftermath resulted in chaos for the territory's physicians and funeral directors. Morgues were overwhelmed and families were left waiting for up to 27 days for a death certificate.
In Naguabo, an eastern coastal town, the lone medical facility for kilometres around held onto eight bodies after the storm, waiting on instructions on where to send the cadavers. The majority were quickly cremated without an autopsy and mourned in hasty ceremonies inside funeral homes.
Many residents were outraged by the Government's intransigence in amending the official toll, even after far higher estimates were released, including a Harvard University survey estimate this northern spring that put the number within a range of 800 and 8000 deaths.
The midpoint, 4645, became a rallying cry for Puerto Ricans who demonstrated on and outside the island, most notably by setting thousands of pairs of shoes in front of the Capitol building in San Juan with notes bearing the names of the dead.
Hours after the report was released to news organisations under an embargo, the interim director of Puerto Rico's Bureau of Forensic Sciences, Mónica Menéndez, resigned. A news release from the governor's office cited personal reasons.
GW's report found that communication problems led to failures in certifying deaths properly during the disaster. Physicians across the island were not trained and did not know how to appropriately complete death certificates. The Government's failure to communicate proper protocols led to the initial undercount, and the lack of transparency fueled suspicions and rumours.
Planning for communication during a crisis "was very poor," said the principal investigator of the GWU study, epidemiologist Carlos Santos-Burgoa.
GW researchers interviewed nearly three dozen public officials and other leaders, some of whom said they were reluctant to attribute deaths to Hurricane Maria because they were concerned about subjectivity and liability.
The report found that the Puerto Rico Government was inadequately prepared for the disaster, operating from an outdated emergency plan for a Category 1 hurricane - one hardly adequate for Maria, which roared in with Category 4 winds.
Coordination between agencies was stifled, and researchers found that the government communications staffs were more concerned with media relations than with reaching out to first responders, municipal leaders and the general public before the storm and during the ensuing crisis.
While researchers said Governor Ricardo Rosselló was an effective communicator on the mortality issue, he tended to defer to Hector Pesquera, the public safety secretary, "instead of calling upon an expert to provide information about mortality surveillance."
