In 2008, the Department of Defense recognized that soldier suicides were growing to alarming proportions. That year, the Centers for Disease Control and Prevention (CDC) reported that the average rate of suicides in the United States was 11.9 deaths per 100,000 people. The U.S. Army, an organization of approximately 500,000 members, in contrast, saw a rate of 20.2 deaths per 100,000 people that same year — nearly double the civilian rate. The Army realized that the issue required immediate and decisive action.
The problem, however, was that no two leaders agreed on what was causing such an increase in suicidal deaths, leaving the armed forces at a loss for how to address the problem.
To better understand the factors driving so many soldiers to take their own lives, the Army partnered with the National Institute of Mental Health (NIMH) to create the largest military-based mental health study ever undertaken, the Army Study to Assess Risk and Resilience in Servicemembers, or Army STARRS projectopens in new window, in 2009.
"The Army wanted to understand what was driving these numbers. What were the specific risks? What was the tipping point that pushed some of these soldiers past the brink?" says Terri Tanielian, a senior research analyst for the RAND Corporation and co-director of the Invisible Wounds of War project, an independent, large-scale assessment of the psychological and cognitive effects of the wars in Iraq and Afghanistan. "They hoped to see patterns across different cases that would help them identify important risk factors. And if they could do that, they had hope of targeting interventions and reducing the overall number of suicides."
Now, six years later, the Army STARRS partnership has yielded important data on the problem of soldier suicides. The initial results may have the power to inform how the military can better prevent both attempted and completed suicidal death in the future.
Michael Schoenbaum, PhD, an NIMH scientific principal for STARRS, says that trying to look at suicide risk factors is a challenge — especially since suicide is not your typical epidemiological problem.
"Suicide remains a rare event," he explains. "The vast majority of people with any given risk factor or factors will never reach that outcome."
Also, he points out, people who do attempt suicide don't always have much in common. "It's clear that there's no single answer here. We need an approach that, if we're lucky, will help us find a lot of little answers."
That's where big data, or the analysis of extremely large data sets to identify patterns and trends, come into play. The Army collects and keeps an astonishing amount of data about each soldier, says STARRS administrative principal investigator Ronald Kessler, PhD, professor of health care policy at Harvard Medical School.
"A single soldier has a single ID number that can be used to link medical data with criminal justice data, human services data, school data and employee data," he says. "These data have an enormous richness that we can use to help target soldiers at high suicide risk."
The STARRS research team has now published a series of studies in the Journal of the American Medical Association for Psychiatry over the past year about the various, and somewhat surprising, factors that did rise to the top in data collected from active duty soldiers between 2004 and 2009.
Initial STARRS studies looked solely at those individuals who had died by suicide — and found that factors including a recent demotion and family problems were common. Those who died by suicide were also more likely to be white and male — and while a previous or current combat deployment was a slight risk factor, it was more so for women than men (JAMA Psychiatry, 2014).
Additional research by the STARRS consortium revealed that approximately one-third of Army suicide attempts were linked to mental disorders that existed prior to a soldier's enlistment (JAMA Psychiatry, 2014).
Now, in its first attempt to predict which soldiers are most at-risk for suicide, the STARRS group has published a study analyzing approximately 40,000 soldiers who had been hospitalized at least once for a mental health issue. This group is at a particularly high risk for suicide. Sixty-eight soldiers took their own lives within one year of hospital discharge, making up approximately 12 percent of all Army suicides.
"They show an extremely high risk and we've known that for a long time," says Kessler. "The question was whether we could find any common risk factors beyond hospitalization that could help us predict who would go on to die by suicide once released."
The STARRS team analyzed the records of patients with a history of hospitalization for mental health disorders, focusing on more than 300 different variables thought to be related to an increased risk of suicide. Using this approach, they discovered that about 20 of those variables showed strong predictive quality. These included a hearing loss, prior criminal offenses, an initial enlistment after the age of 26, previous suicide attempts and a higher IQ score (JAMA Psychiatry, 2015).
Kessler says that the results are illuminating, but one has to be careful not to over-attribute causality. "These are markers that often represent something more subtle and complex that is underlying them. They are, at best, shorthand summaries for lots of things that are going on," he says.
But, he says, the studies do confirm that there is no one factor even in this high-risk group that predicts suicide. "Rather, there are lots and lots of things that add up over time."
The factors are useful as a checklist that might help clinicians determine if a particular patient may need more help once he or she is released from the hospital. "It's information doctors then have at their fingertips to target patients who could be assigned to enhanced post-hospital aftercare services," Kessler says.
After analyzing risk factors among psychiatric inpatient soldiers, the STARRS team is now working on a similar risk-prediction model for soldiers who visited the hospital for psychiatric outpatient therapy, using an even larger data set, says Lisa Colpe, PhD, another member of the STARRS research team. The goal is to continuously fine-tune the factors that could lead to suicide. "Army STARRS hoped to show that it is possible to use data to identify the 5,000 people who are at higher risk out of the 500,000 total people in the Army so that we can offer them something more intensive before the real catastrophe occurs," says Schoenbaum. "And we are demonstrating that we can."
Kayt Sukel is a journalist in Houston.
