A study published in the New England Journal of Medicine on Tuesday attempts to answer a question that has swirled since shortly after Hurricane Maria made landfall on Puerto Rico: How many people died?

This is a surprisingly tricky question to answer in even the most hospitable of circumstances. We spoke with researchers in October who pointed out that problems with reporting and the occasional vagueness of causes of death meant that a hard count was nearly impossible.

Overlay politics and the tally becomes even more challenging. President Trump, for example, lauded the low death toll when he visited the island shortly after the storm made landfall. At the time, there were only 16 official deaths, which he compared favorably to Hurricane Katrina in 2005.

“Sixteen versus literally thousands of people,” he said. “You can be very proud.”

Trump, of course, was under fire for his administration’s slow response to Hurricane Maria and hoped to compare himself favorably with President George W. Bush’s response to Katrina, after which an estimated 1,833 people died.

The newly released study, from researchers led by teams at the Harvard T.H. Chan School of Public Health and Beth Israel Deaconess Medical Center, estimates that 4,645 people died as a result of Maria — two and a half times as many as died from Katrina.

A natural follow-up question to that estimate is how many of those deaths might have been preventable. This, too, is nearly impossible to say with certainty, but the new research does hint at an answer.

Researchers conducted a survey of 3,299 households across Puerto Rico. The two most important questions for our purposes were about changes in household residency during 2017 and questions about the broader neighborhood.

In the first case, people were asked how many people had lived in the household from the beginning of 2017 until the point at which the survey was taken and why people were no longer there. If the resident had died, the person answering the survey was asked how the death occurred, including an option for delays in medical care. That idea was broken out further, with survey respondents also being asked how many days residents may have been affected by an unusual inability to get to a hospital, obtain medicine and the like.

Respondents were also asked how many of their neighbors had died since the hurricane. It’s that question that allowed researchers to estimate the mortality rate. By comparing that rate with the average mortality rate for the same period in 2016, the number of excess deaths was estimated.

(The researchers grouped households by “remoteness,” a function of how long it would take to travel to a city of 50,000 or more.)

How all of those people died isn’t clear and wasn’t asked. (That makes sense; the odds that you could accurately report the cause of death for random neighbor seems low.) Among those households surveyed, 38 deaths occurred after the hurricane. About a third of those deaths were identified as not being related to the hurricane. Another third were identified as being a result of the interruption of medical services. About a tenth were attributed directly to the storm.

This is a small sample size from which to extrapolate outward. Among those deaths linked to the hurricane by survey respondents, the number linked to issues with receiving medical care outnumbered direct deaths by a 4-to-1 margin. If that ratio held for all 4,645 reported deaths, it would suggest that more than 3,700 people died as a result of a lack of access.

Such problems apparently were widespread. Setting aside the rampant issues with access to electricity and water, nearly 15 percent of households indicated that for at least some period of time, they’d been unable to access medicine. About 10 percent said they were unable to use respiration equipment. Slightly fewer cited damaged roads and closed facilities as impediments to getting medical care for at least some part of the period between the storm and the end of the year.

We don’t know precisely how many people died as a result of the storm, although the new study gives us some hint. We don’t know how many of those deaths were a function of the aftermath of the storm, although the study gives us some insight on that as well. We further don’t know how many, if any, of those deaths caused by a lack of access to medical care were themselves preventable — the question at the heart of Trump’s defensive comments about Katrina.

The researchers were deliberate in how they asked about the aftermath of the storm, asking about delays in access to medical care “that didn’t exist before the hurricane.” In theory, then, these were outages due to the storm and not to problems on the island that may have existed before Maria’s landfall.

Had the federal government responded more quickly and broadly, could it have cut down on the number of damaged roads or closed hospitals? Could it have improved cell service and electrical access?

Such failures were cited in The Post’s story about the new study in describing the death of a woman named Ivette Leon. She had been treated for an infection and sent home, where she took a turn for the worse. It took 20 minutes to get a cellphone signal and another 10 minutes for the ambulance to reach her, in part due to failing traffic lights. By the time it arrived, she was dead.

The researchers we spoke with last October noted that, in addition to probably never knowing the precise toll of the storm, we’d probably never know how and why those deaths occurred.