A Record Hurricane Disaster in Puerto Rico
The Puerto Rican government's estimate of 64 deaths from Hurricanes Irma and Maria struck researchers from Harvard T.H. Chan School of Public Health as too low, so they decided to investigate. The team’s door to door survey revealed that the death toll was on the scale of thousands. At least 800, and as many as 8,000 if not more, perished as a result of the hurricanes' destruction and interruption of the power grid, water infrastructure and medical services, making it the worst natural disaster in the US in more than a century. Host Steve Curwood sat down with senior authors emergency medicine physician Satchit Balsari and epidemiologist Caroline Buckee to learn why their survey is such an important tool, and how it can spur preparedness in the future.
CURWOOD: It’s Living on Earth, I’m Steve Curwood. Hurricanes Irma and Maria devastated Puerto Rico when they hit the island last September, and as hurricane season begins again, power is still not fully restored and infrastructure remains fragile. But official figures claimed that the death toll was modest – that a mere 64 people had died as a result of the disaster. Well, now a team led by the Harvard T.H. Chan School of Public Health has made a new calculation of mortality related to the hurricanes, and the new figure shows the official estimates to be a startling undercount. We’re joined now by two of the senior authors – team leader and epidemiologist Caroline Buckee, and emergency room physician Satchit Balsari – welcome to Living on Earth!
BALSARI: Thank you for having us.
BUCKEE: Thank you.
CURWOOD: Caroline, what was the genesis of this? What was the idea that generated this study?
BUCKEE: After Hurricane Maria, it was clear and, in fact, the government acknowledged that there should be some reevaluation of the death count in the wake of the hurricane because the official estimate was 64 deaths due to the hurricane. And in the aftermath of Hurricane Maria it was clear that there was a ton of infrastructure damage and that, in fact, this was probably an underestimate. In the wake of disasters it’s very hard actually to estimate mortality because when infrastructure is so damaged issuing death certificates is difficult and normal health systems are not functioning well.
And so one of the ways that you can provide an independent estimate of mortality in the wake of disasters is to go and do a household survey study like the one that we did. And what that means is that you are providing an estimate that is not dependent on the death registry system or on death certificate issuance. And so you can capture deaths as well as well as other nuances like infrastructure damage, loss of utilities like electricity and water, as well as other types of suffering. So, what the reported causes of death were according to family members.
CURWOOD: Satchit, why did you get involved in this project?
BALSARI: Well here at the FXB Center for Health and Human Rights, we have routinely worked with vulnerable populations to leverage science, big data, for human rights advocacy. So, these kinds of situations are situations that our scientists are very familiar with and very comfortable working in. I've been a student of disaster medicine and humanitarian crises for the past 20 years and I work closely with Professor Jennifer Leaning, and we felt when Caroline reached out that this was another excellent opportunity for us to leverage the experience that we had elsewhere to advocate for the people in Puerto Rico.
CURWOOD: Caroline, as I understand it your study came up with some 4,600 excess deaths from the storm – actually the storms I should say, right? Because this covers a longer period of time. Just what was your method there to come up with the numbers?
BUCKEE: So, before I describe the method, I think it's very, very important, and we've seen in the last few days, the importance of emphasizing that this is not a precise estimate. So, there are very large confidence intervals around this estimate. And so the range of the 95 percent confidence intervals is from about 800 to over 8,000, and the estimate as it stands came out as 4,645, but it's misleading to quote that as a number because it’s, by definition, it’s not a precise number. And that's because of the methodology.
So, we used a survey approach where we randomized households across Puerto Rico and then we went door to door and we are asked members of that household to tell us about their families. What the composition of the household was, how old the people were, the genders, if people had died, what they had died of and when they died, as well as asking about displacement, so who had moved into the house, who had left the house during that time period. And then we asked about utilities, so how long had you been without electricity or water or cellular telephone coverage.
So we surveyed 3,299 households which covers a population of approximately nine and a half thousand people, and from that we could calculate a mortality rate. And we compared that mortality rate to the same period of time in Puerto Rico from 2016. So over that period of time from September 20 to December 31, we compared the mortality rates and the rate difference tells you about how many excess deaths occurred. And then from that rate difference we multiplied it up to cover the whole population. And so there's a large amount of uncertainty associated with that calculation because there are many different ways to estimate the population size of Puerto Rico in the aftermath of the hurricane, and we used a fairly low number to account for the fact that we know people left the island leading up to the hurricane, and so we wanted to make sure that we were conservative with our estimate.
Nevertheless, it is very uncertain. And the key point here is that this approach does not rely on infrastructure. It's very quick, it's very cheap, and if the government wanted to employ this type of approach in the future when their death registry system was not working well, it would be straightforward to implement a much larger survey and then they could make the estimate much more precise. So, I think the key takeaway here is that the precise number is less important than the fact that the mortality rate was increased and it stayed high throughout the rest of the year.
CURWOOD: By the way, for people who aren't scientists, they might say "excess deaths? I mean, any death feels very personal.” What's the scientific meaning of "excess death?"
BUCKEE: So, essentially we always expect some level of frequency of deaths in a population. People die. And so we're talking about deaths over and above what we would normally expect in that period of time. So, that's what we're trying to calculate: the difference between what would normally happen and what we observed due to the hurricane.
CURWOOD: It’s interesting, though, that your numbers, if it's a range between 800 and, say, 8,000, still the government was talking about 64. So, even the low end of what you're talking about is quite a bit larger than what has been officially acknowledged.
BUCKEE: Absolutely, and interestingly our results are consistent with several reports – a few academic reports and in the press – that came out after the hurricane. And so out of those handful of studies, the only outlier is the government estimate of 64. And in fact, the government does acknowledge that that is an underestimate, and that's why they have issued this reevaluation of the death registry data.
CURWOOD: Satchit, so how has the government of Puerto Rico responded to your research?
BALSARI: The Governor of Puerto Rico in a press conference hours after the paper was published welcomed the research. He said he was looking forward to learning more. We have reached out to his offices and offered to be available to sit down and explain our findings to the officers.
CURWOOD: Now, you're an emergency physician. I know that your research didn't look at death certificates. So, you don't have causes of death per se, but what do you think happened to these people who perished?
BALSARI: Well, while we did not look at death certificates, we did ask people what their family members died from. We also asked people what they had continued to suffer from, the interruption in services. And what we find are a few different things. As Caroline explained, the persistence of the elevated mortality rate is what is of concern. When we think about disasters, disaster response planning, we think of an acute event, a whole bunch of people die, and then life goes on.
Well, that is not what we're observing in disasters now and certainly not in Puerto Rico. There was interruption of basic utilities for a long period of time. Large groups of populations did not have access to water, cellular phone coverage, and electricity for a good amount of time. One third of all those that had died in the households that we had surveyed attributed those deaths to interruption of medical care. There's a variety of reasons that people provided including broken down roads: disruption in the health care facilities themselves, and availability of doctors, difficulty accessing medications. Then not being able to use their respiratory machines, for example; nebulizers or CPAPs that people used at home, or dialysis machines that they went to their clinics for.
This phenomenon of interrupted medical care is important for responders in the United States to pay attention to. We have seen that all the way back in Hurricane Katrina. We continued to see that in Sandy. Nursing homes are affected. The elderly are dependent on life-sustaining equipment that is electricity-dependent. Even the evacuation of hospitals in Sandy was because the generators flooded and failed and patients had to move. It is that same phenomenon that we're seeing play out in Puerto Rico, but not just on the days of the event, but for months thereafter.
CURWOOD: So, take a step back for a moment. We're in a period of time of a disrupted climate and there are likely to be more very serious storms that hit Puerto Rico. I mean just this past year we saw Houston, we saw Florida, we saw Puerto Rico … oh, and out west we saw a whole bunch of fires. What does your research tell us we should be thinking about, acting about, going forward as a society as we deal with the pressures of climate disruption?
BALSARI: We should map our vulnerabilities better. That is not just the role of the government, but community preparedness is really the key to all our societies faring better with these disasters. It is important to know where the uncles and aunts that are on dialysis, who is on breathing machines, what equipment is your neighbor dependent on. Even high resource populations that think they're otherwise insulated from disruptions and climate change are very much vulnerable. Think about the elderly that lived in high rises in downtown Manhattan after Hurricane Sandy. This was a rich population, normally used to all kinds of services, that were trapped without food, water, electricity, and access to medicines on the 30th story and the 40th story of high buildings. Again, the California bushfires and the demographic that was affected. So no one, no one is spared from the impact of these adverse climate events that we are observing.
CURWOOD: I gather that this research was conducted over a very short period of time. It just took you a couple of months to do the interviews and then crunch the numbers. How do you feel about that, and how do you feel about the response that the public has had to this?
BUCKEE: So, the quick turnaround of this study and the ability to do it within – we completed the surveys within three or four weeks, and the analysis took another couple of weeks. So, this was a very quick assessment, and we had a small budget and we were constrained in the sample size as a result. So, 3,299 households was a lot, but since deaths are rare events, even after the hurricane, that means that our estimate is imprecise. So, although overwhelmingly the response has been positive and people have welcomed some insights, and I think they’ve welcomed the fact that we've made the data freely available online for anyone who wants to analyze it, there is some pushback about the uncertainty of our estimate. And I think one of the challenges here is communicating scientific uncertainty to the public. This is a really important part of our job, and I think we probably don't do it very well. It's very important that we continue to state that this was a small survey, the uncertainty was large, it doesn't invalidate the fact that we find significantly higher mortality.
BALSARI: Household-based surveys are fairly standard, so the innovation here was not that we did a household-based survey. For decades, the scientific literature has deliberated on how best to do these household surveys and how they complement death registry data and other forms of accounting for deaths after disasters. So, while the response has been overwhelming, we hope that it complements any future endeavors that the government undertakes to establish both accuracy of the death toll, but also to support closure for these families.
Going into the hurricane season, we are hoping – and the Governor said so – that they will relook at how deaths are counted in the aftermath of disasters. And what Puerto Rico comes up with could potentially help the rest of the country as well. In the United States, currently, a death is attributed to a disaster event if it is noted on the death certificate, but the criteria for noting death on a death certificate is not known to most. There are very few people that are trained in understanding that even in the days following a disaster there are several indirect causes of death that can be attributed to the event and that this has to be noted on the death certificate. So, we do need expanded training and awareness of how these deaths have to be recorded appropriately so that they can be counted.
CURWOOD: In other words, you're saying if somebody couldn't get to their, say, cardiac medication for several days and they had a heart attack – if they write “heart attack” on the death certificate, and they don't attribute it to the fact that the streets were flooded, that they're missing something.
CURWOOD: How does your study help with closure from such a disaster?
BALSARI: Well, I think it is important for deaths from disasters to be acknowledged. It is certainly important to families. They want to know that society knows why they lost their beloved. It also has financial ramifications. People have access to aid assistance from FEMA, from other government agencies, if they have lost someone in a disaster. It is important that that be recognized and that the families be given access to the aid that they deserve.
CURWOOD: Why do you suppose that the government got this so wrong, the scale of the loss of life in Puerto Rico?
BALSARI: The numbers that the government declared were numbers that they were working with. It is how deaths were being counted in the early days after the disaster. I think the failure was in recognizing that the official methodology for counting death was highly flawed. And had they looked around and gathered data from multiple sources outside of just the death registry data, they might have been working with very different information.
CURWOOD: I mean, you guys took – what? – less than two months to do this. Why didn't Puerto Rico, why didn't the Federal Emergency Management Agency? Why didn't they trot out this more sophisticated math themselves?
BUCKEE: Well, currently this type of approach has been largely an academic exercise in the wake of disasters, and in fact, this has been occurring in studies since the 1970s to try and estimate the impact on the population. And as it stands now, the way that the government counts deaths is through death registries, and it does not use these additional sources of information. So, what we're advocating is to have a shift where we do start to see these types of studies being done by the government as well. And we hope that it will become more standard. At the moment these are largely scientific exercises undertaken by academics rather than by the government itself.
CURWOOD: Academic exercises – and yet your study’s on the front pages of many newspapers and many national broadcasts.
BUCKEE: That's true. The fact that the government has acknowledged from very early on that they needed to reevaluate the number of deaths attributed to the hurricane – and the fact that they have commissioned George Washington University to go through all of the death registry data again which will take some time – it shows that they are aware that there are limitations. The infrastructure damage in Puerto Rico after the hurricane was enormous, and they were dealing with all kinds of other priorities at the same time as trying to keep track of mortality. And so, you know, that is a huge challenge and it remains a huge challenge. And so moving forward, I think the most important thing is that we can complement activities when infrastructure is destroyed with these quite simple approaches, which are quick and can be done in a simple and transparent way.
CURWOOD: You know famously, Heisenberg the researcher said anything you measure you're going to affect. How did doing this research in Puerto Rico affect you?
BALSARI: Profoundly … I think these kinds of projects or spending time in these environments is a constant reminder of how lucky the rest of us are.
BUCKEE: Yeah, it was very affecting, and of course my heart goes out to all of the people who suffered in the wake of Hurricane Maria. But I was also quite hopeful in some ways because the students that we worked with who conducted the surveys, these were graduate students in psychology from Carlos Albizu University and Ponce Health Sciences University. They were utterly dedicated, very bright, they worked so hard round the clock to get these surveys done, and they have been working on their own time with the communities, supporting communities who have suffered so much from day one after the hurricane, and their energy and dedication, I felt was really a hopeful thing. And so while, you know, the devastation has been enormous to the island, I do think that there is hope because the Puerto Ricans we worked with were just so resilient and hardworking and determined, and so that gave me some small hope.
CURWOOD: Caroline Buckee and Satchit Balsari are with Harvard University's T.H. Chan School of Public Health and authors of the study of Puerto Rican mortality. Thank you both for taking the time with us today.
BALSARI: Thank you for having us.
BUCKEE: Thank you so much.
Living on Earth wants to hear from you!
P.O. Box 990007
Boston, MA, USA 02199
Newsletter [Click here]
Donate to Living on Earth!
Living on Earth is an independent media program and relies entirely on contributions from listeners and institutions supporting public service. Please donate now to preserve an independent environmental voice.
Sailors For The Sea: Be the change you want to sea.
Innovating to make the world a better, more sustainable place to live. Listen to the race to 9 billion
The Grantham Foundation for the Protection of the Environment: Committed to protecting and improving the health of the global environment.
Energy Foundation: Serving the public interest by helping to build a strong, clean energy economy.
Contribute to Living on Earth and receive, as our gift to you, an archival print of one of Mark Seth Lender's extraordinary wildlife photographs. Follow the link to see Mark's current collection of photographs.
Buy a signed copy of Mark Seth Lender's book Smeagull the Seagull & support Living on Earth