Making Sense of Hurricane Katrina’s Death Toll
How many people died because of the Hurricane Katrina?
I’ve gotten that question a lot over the last few days, with reporters being specifically interested in Louisiana and New Orleans. It is a question that I spent 7 years focused on, which culminated in my dissertation in 2011 which was published as a scientific monograph in 2015. I can tell you that it is a difficult question the answer, and that the final number is both dependent on how you ask the question and subject to some uncertainty.
First, the easy part:
– Mississippi: 238
– Florida: 14
– Alabama: 2
– Georgia: 2
– Ohio: 2
– Kentucky: 1
These are the number of deaths attributed directly to Hurricane Katrina’s impacts in those states. These numbers are largely undisputed and they total to 259 U.S. Hurricane Katrina related deaths outside Louisiana. While the storm caused minor impacts in the Bahamas and Cuba, I am not aware of any storm related deaths outside the U.S.
When examining Hurricane Katrina’s impacts in Louisiana, arriving at a consistent death toll becomes complicated and nuanced. First, consider that are a few different ways to count and classify disaster related deaths. The most common way is to classify deaths as either directly or indirectly related to the disaster. Usually, the direct deaths are presented as the disaster’s official toll, with the indirect deaths mentioned as an aside. This scheme has a number of shortcomings, most notably “indirectly” is ambiguous and open ended.
Another method is to look at the cause of death and then refer to the International Classification of Diseases coding system to see which ones fall under the category for “victims of a cataclysmic storm”. A third way would be to gather reports from the families of those who are believed to have died from the storm. A fourth way would be to use obituary listings. Finally, the approach which I adopted for my dissertation is to look at the types of circumstances of the recorded deceased victims.
Various deaths tolls have been reported for Hurricane Katrina’s impacts in Louisiana. The official figure for Louisiana comes from the State Medical Examiner’s Office, which was formed by an executive decision by the governor in the immediate aftermath of the storm. They were tasked with overseeing the recovery, examination, identification, and eventual release to the families for the victims that were recovered from areas where local capabilities were decimated by the flood, meaning Orleans and St. Bernard Parishes. When they closed on August 2, 2006, their final report stated that “there have been 1,464 victims of Hurricane Katrina from Louisiana.”
Screenshot of an archived version of the August 2, 2006 final report on Hurricane Katrina Deceased Victims from the Louisiana State Medical Examiners Office. The original site has been scrubbed from the web.
However, this estimate reflects some uncertainties and bias that resulted from the operational environment of the Medical Examiner’s Office. For example, this count reflects two arbitrary dates: October 1, 2005 and August 2, 2006. The first date is the date that was chosen as the cutoff date for considering victims that die while displaced due to the storm. A person who died on September 30 because of disruptions in routine care of chronic conditions while in a shelter in Colorado would count, but not a person who died on October 2 of identical circumstances. The second date reflects the date the office closed down. Victims whose bodies were not found until after that date are not counted, even if they met all the other criteria. Finally, this count likely reflects a bias resulting from the Medical Examiner’s goal to ensuring that families of the disaster victims obtained burial related disaster assistance, and thus should be classified as related to Hurricane Katrina.
In an effort to better understand the toll of the disaster, a group of epidemiologists with Centers for Disease Control and Louisiana Department of Health and Hospital completed an independent review of the State Medical Examiner’s Office records. Brunkart, Namuland, and Ratard (2008) compared victim’s cause-of-death to criteria in the International Classification of Diseases. They concluded that 986 victims fit the criteria of “victims of a cataclysmic storm”, of which 971 occurred in Louisiana and 15 occurred outside the state. They also noted that due to limitations in the records, the true number is likely higher. Still given their utilization of internationally accepted standards, this would be the best number (along the above listed figures from other states) to use when comparing the toll of Hurricane Katrina to other disasters.
For my dissertation research, I created what I call the “Hurricane Katrina Victim Database.” This was my effort to compile as complete and accurate a record as possible. In doing so, I worked closely with the State Medical Examiner, and many of my primary sources came from them. I did not stop there though, and included other victims with reliable documentation. These include 20 victims from Jefferson Parish listed in The Times-Picayune, 6 victims described to me by the Plaquemines Parish emergency manager, and 40 victims found in Orleans Parish after the Medical Examiner ceased operations. In the end, my database listed 1572 victims of Hurricane Katrina’s impacts in Louisiana.
One thing about this disaster that I think we have taken for granted is the level of detailed data that it produced. In doing this research, I worked with high resolution satellite imagery, detailed elevation and flood depth datasets, the results state-of-the-art computer models including ADCIRC and SOBEK, and many other datasets.
Deceased victim recovery locations along with depth of flood waters for Greater New Orleans area.
These datasets gave me important hazard characteristics, such as wind speed and flood depth, at the locations where bodies were found. The SOBEK model even provided velocity of the flood waters and its rate-of-rise. My research team and I also completed the grim task of visiting over 400 locations where victims had been found after the disaster. During these field surveys, we measured water marks and first floor elevations. We also looked for signs of structural damage, indications of attempted escape and rescue, and evidence of disability, such as handicap license plates and decals indicating things like oxygen tanks.
With evacuees scattered across the country, Hurricane Katrina related deaths among Louisiana residents were also scattered across the country.
Using all this data and trying to make sense of these trends, I borrowed a concept from a paper written by two of my colleagues. Jonkman and Kelman in 2005 published a paper called “An Analysis of the Causes and Circumstances of Flood Disaster Deaths.” I did not have access to the medical cause of death for each victim, but I did have enough information to draw inferences regarding the causes of death.
Specifically, of the 1572 victims in my database, I found that nearly all of them could be put into one of three categories: i) direct flood deaths, ii) emergency circumstances deaths, and iii) evacuation/displacement deaths. A fourth category included up to 16 possible wind related deaths, and then a separate study described below covers long-term deaths. One main strength of using this framework is that a major disaster response is organized around the circumstances of the victims more than the specific medical conditions of the victims.
Inference tree used to classified victims in terms of the circumstances of death
Within the direct flood deaths category, there are 557 known victims. When the victims recovered from unknown recovery locations and a portion of the missing are added, the most likely number of flood deaths is between 600 and 700. Among the known flood deaths, the mean age was 69, 52% were black, and 54% were male.
In the emergency circumstances death category, there are 284 known victims. Of these, 152 were recovered from hospitals, 19 from nursing homes, and 42 from a shelter, school, or temporary medical clinic. The mean age was 70, and they were 57% black and 53% female. This category also includes 10 victims whose listed age was zero and are assumed to be still-born babies from flooded and powerless hospitals.
The final category was evacuation/displacement deaths, and it includes 631 known victims. This total includes 364 victims that died outside Louisiana, 243 victims reported from Louisiana parish that received evacuees, and 24 reported by the Medical Examiner. With a mean age of 71, these victims were 61% female and 57% Caucasian.
By the way, if you add the 1572 victims in the database to the 259 victims in other states, you get something close to the most cited total of 1,833.
Screenshot (taken August 28, 2015) of Google search results for “Katrina Death Tool” The 1,833 figures is based on a 2005 NOAA report.
What didn’t I include?
Like the other estimates, my database has limitations. Specifically, I did not second guess the judgement of Medical Examiner on whether or not individual victims should be included. I merely sought to complement their numbers with additional victims that would apply under their criteria. So their October 1 cutoff date applies to my total, even though I know that the disaster continued to kill after that date. Two sources shed some light on those numbers.
Professor John Mutter is a professor with the Earth Institute at Columbia University. Following the disaster, he compiled his own list of victims based on reports from the families. He even created a web based reporting system. While his list had been published on the internet for a while, it was not available as of the time of this writing. I was however able to find a July 2008 archived version of the site that lists 1294 total deceased along with 595 missing. Among his deceased, he lists 914 as residents of New Orleans. In my database, 853 victims were residents of New Orleans, and any additional ones listed by Mutter are likely attributable long term impacts.
Screedshot of July 2008 archive summarizing the list of victims compiled by Professor John Mutter.
Finally, Dr. Kevin Stephens and his co-authors took a different approach to obtain an estimate of long term deaths attributed to the disaster. Comparing obituaries listed in the Times-Picayune during the first six months of 2006 with the same period in 2004, they estimated 1,600 excess deaths attributable to the long term impacts of the storm. If you apply this value at a uniform rate over the 2 year initial recovery period, you get an estimate of 6,400 long terms deaths due to Hurricane Katrina’s impacts in Louisiana.
Yes, it is complicated, but let me try to summarize things by answering two key, but distinct questions.
First, what is number to use when comparing the death toll from Hurricane Katrina to other disasters? This would be 1,245, based on the 259 deaths outside of Louisiana and the 986 Louisiana victims that were classified as “Victims of a Cataclysmic Storm.”
Second, what is the best estimate of the true number of lives lost in Louisiana due to Hurricane Katrina and the catastrophic levee failures around Louisiana? This number would be around 8,000, based on the 1,572 victims documented in my database plus the estimated 6,400 long term deaths plus up to 130 people reported missing.