Friday, October 05, 2012

Working the data behind Uncounted Casualties

For the past several months, the Statesman investigative team has been hammering away at a huge data-driven series, Uncounted Casualties, where we researched the deaths of Texas veterans of the Iraq and Afghanistan wars, to startling result.

While this story started with data, it was shoe-leather reporting and story-telling that really made it stand out. Reporters Brenda Bell, Eric Dexheimer, Dave Harmon, Tony Plohetski and Jeremy Schwartz spent six months using scant information to identify and track down the cause of death of our vets, and then talked to the families of many so they could tell the story of these fallen service members. Jay Janner and Kelly West brought it all to life through photos and video. I worked the numbers behind the scene.

The conclusions we formed -- that overdoses are as much a problem with our returning vets as suicide, and that accidents claimed a disproportionate number of our pool compared to the Texas population -- could not have been made without good record keeping, analysis and comparison to state-wide data gathered through public information requests. One reason we spent the time to do this study was because no one else has; Veteran’s Affairs does not track causes of death among all their veterans, though they are trying to take steps to do so.

We started with a list of 72,000+ veterans who had died since 2003 that were in the VA system and receiving benefits. This Beneficiary Identification Records Locator Subsystem (or BIRLS) list had only a few fields: date of death, age at death, primary disability code, branch of service and the ZIP code of residence registered with the VA. Since we wanted to focus on Iraq/Afghanistan vets only, we had to go back to the VA to get that service added to the files, narrowing our list to 345 veterans who were from Texas and had served in Operation Enduring Freedom or Operation Iraqi Freedom.

Jeremy Schwartz outlines in a blog post the strategy to identify more than 300 of these veterans, and determine the cause of death for 266 of those. My part in this was to create a database (using Caspio) that the group could use to track the information they found, so we could later do analysis on those records. We had to code our information consistently so it could be queried and counted correctly.

We also wanted to compare the causes of death among our cohort with Texas as a whole, where we could, so I handled this as well. There are death statistics available online with the Texas Department of State Health Services, but the query tool only went up to 2009, and the age breakdown was in 10-year increments. It’s important to look at death rates by age group because wide ranges skew the numbers. (Infants don’t commit suicide and older people are much less likely to die of overdose.) So we requested the most recent data we could, broken into 5-year increments so we could look more accurately at “combat-age” groups that didn’t include Texans too young to serve in the military services, or older than our comparison cohort.

Once we had all our information in hand, we were able to do comparisons, by percentage, using the age-groupings of our sample. Because we knew our group had so many more men than women, compared to the general population, we also sliced the data by gender, although we found the trends generally the same.

Once I could query data (I used Microsoft Access and Excel,) we found some alarming trends. It was no surprise that suicides cases were elevated compared to the general population, just as has been widely reported with active duty service members, but we also found that overdoses were equally high, and the vast majority of those involved prescription medications.

When we looked at the veterans who had been diagnosed with Post-Traumatic Stress Disorder (PTSD), the causes of death were even more startling. When looking at the group as a whole (the 266 where we were able to determine the cause of death), over 100 died of disease or sickness and other natural causes. But when we looked at just those veterans with PTSD, only two in 46 died of these more natural causes. Suicides, overdoses and accidents accounted for 80 percent of the deaths.

Another trend I didn’t see until we lay it out on paper as a graphic, was that almost all of our Marine veterans died young. Of the 42 Marines, only four were 35-years-old or older.

You can slice and dice this data set yourself in this wonderful interactive graphic designed by Rob Villialpando. Rob also handled the excellent online presentation, and I’d also like to wave at Scott Ladd, who brilliantly handled the print design.

We made the comparisons with a caveat. We know the group of veterans we were looking at were already seeking medical care and receiving benefits from Veterans Affairs, so they were “sick” in some fashion. While some of the older age groupings were small, the bulk of our cohort was in younger age groups where we had adequate numbers to see trends and build comparisons. We still find, and a former Texas state epidemiologist concurred, that our comparison is valid.

It was an enlightening project and one that I’m proud to be a part of. Hopefully by revealing these startling trends about suicides, overdoses and accidents among our veterans, we can better understand the scope of the problem and policy and laws can be shaped to help more veterans as they return from the battlefield and integrate with our communities.


  © Blogger template 'A Click Apart' by 2008

Back to TOP