P.R. gov’t’s low post-María death toll pushed Harvard to continue study
By Jeniffer Wiscovitch-Padilla, Center for Investigative Journalism
Four days before the results of the “Mortality in Puerto Rico” study were published after Hurricane María, Harvard University reported the findings to the government of Puerto Rico, but was not acknowledged.
It was not until “The New York Times” contacted the government to talk about the results that La Fortaleza requested a meeting with the researchers, Rafael Irizarry-Quintero the professor at Harvard’s University School of Public Health told the Center for Investigative Journalism (CPI, for its initials in Spanish.)
“They answered when ‘The New York Times’ called them. They answered and asked to speak with us …,” said the professor who worked on the statistical aspect of the study.
During that meeting at La Fortaleza, they discussed the results and how difficult it had been for Harvard researchers to obtain data from the Demographic Registry, the professor explained.
At the meeting “there was a group that said yes” they were providing the required information, when it was not that way, he recalled.
Although he could not specify who was in the meeting, he did say that there were government statisticians and that the governor or the secretary of Public Security, Héctor Pesquera, who was the one who offered information about the deaths at that time, were present.
“The conversation [at the meeting] was focused, [on] us explaining what we had found in detail [in the investigation], and at that point we were saying that we had not gotten the data from the [Demographic] Registry,” Irizarry-Quintero explained to the CPI.
According to the professor, the academics explained to the officials that it would not have been necessary to make the estimate of deaths based on surveys if the Demographic Registry’s data had been available.
The Harvard study, published in May 2018, suggested that
between 793 and 8,498 people died directly or indirectly from the hurricane and that the deaths continued to happen until the end of 2017.
The professor emphasized the importance of always referring to the confidence interval. In this case, the confidence interval was 95 percent from 793 to 8,498 (deaths), where 4,645 fell in the middle of that amount. This means that the statistical method implemented in the study generated a confidence interval that has a 95 percent probability of including the actual death count.
Lack of access led to Harvard study
The lack of access to the government data on mortality for 2017 was what led the Harvard group to do their study through random surveys in which they estimated 4,645 deaths after the hurricane.
After the government published the mortality data from 2015 to 2017 in June 2018, the researchers adjusted the estimated figure to 3,000 deaths, “a number very similar to George Washington University,” Irizarry-Quintero said.
The GWU study, commissioned by the government of Puerto Rico, estimated 2,975 deaths after the storm.
Irizarry-Quintero explained that the adjusted statistical model took into account changes in the mortality rate, the effect of the season (more people die in winter than in summer) and the number of people who had left Puerto Rico after the hurricane.
“We were not sure at one point if we were going to continue the investigation because you (the CPI), ‘The New York Times’ and researchers at Penn State had presented the numbers, and it was clear that the government was wrong,” he said.
“We made the decision to continue because the government, despite seeing that clear evidence, continued insisting on the low numbers. Second, we were worried that the effect would continue after October [of 2017]. Because everything that had come out at that time was until October,” he said.
The professor said that Pesquera’s comments comparing the mortality of September 2017 with that of December and January motivated them to continue with their research, since there are usually more deaths in winter than in the summer.
“When you read that and you are a statistician, you worry, because you are comparing apples with oranges,” said the professor during a conversation at the architecture amphitheater of the University of Puerto Rico.
Little money for a big effort
The Harvard study had six times less funds than those allocated by the government to GWU, which cost $305,000.
“We had a little funding from the School (Harvard Public Health), but very little, $50,000, and it was all going to the people who did the surveys. A larger survey would have been more informative,” he said.
Carlos Albizu University was commissioned to do the 3,299 surveys. The interviewers went to houses with the coordinates given by the researchers and then the information gathered quickly reached the researchers electronically, Irizarry-Quintero said.
New findings in a second study with data at hand
After having all the official information on deaths in Puerto Rico from 2015 to 2017, the Harvard researchers compared Maria’s mortality with other hurricanes on the island such as Georges and Hugo. They also compared Maria’s data with those of Katrina (Louisiana) and Irma (Florida).
This new information was published six months ago in bioRxiv, a portal that allows to expose studies that are in the process of approval, the professor said.
The statistical work done by Irizarry-Quintero with doctoral student Rolando Acosta, found a prolonged increase in mortality after Katrina and María, with a duration of at least 125 and 207 days respectively, which means that events related to María caused longer-term damage.
They also found that in 1998, Hurricane Georges had an impact comparable to that of Katrina with a prolonged increase in fatalities of 106 days, resulting in 1,400 excess deaths.
The causes of death that increased, the professor said, were heart attacks, dialysis and failures of the nervous and respiratory system, which are consistent with a collapsed health system.
The CPI had already reported since December 2017 that the deaths after María soared mainly in hospitals and elderly homes, with conditions such as diabetes, Alzheimer’s, cardiac arrest, kidney failure, hypertension, pneumonia and other respiratory diseases.
In the study conducted after the government made the mortality data public, the researchers concluded that at least since 1998, Puerto Rico’s health system is in a precarious state and that almost a year after Hurricane María, the island was unprepared for the onslaught of another hurricane.
Irizarry-Quintero recommended that the government address the issue of electricity, especially in hospitals, and ensure that people have access to the medical services they need.