Suicide Prevention in Veterans

This article was originally published in our Fall 2020 print issue.

Veterans have a unique job description: many have put their lives at risk every day doing field work, handling firearms and engaging in combat. One task not listed in the description? Wrestling with suicidal thoughts. 

Suicide is one of the leading causes of death in the United States. According to NIMH,  85% of adults who attempted suicide were found to have suicide ideation. Suicide ideation is defined as thinking about or planning suicide. Because of their past experiences as military personnel, veterans are especially at risk for suicide ideation, causing them to be more likely to commit suicide than other demographics. With the average number of suicides per day by veterans on the rise from 15.9 in 2005 to 16.8 in 2017, it is more important than ever to ask why veterans are such an at-risk population, and what we can do to prevent such high rates of suicide among this vulnerable group.

Matthew Nock, professor of psychology at Harvard University, has been working to answer these questions. When asked what factors make veterans such a vulnerable population, Nock replied, “there is no silver bullet, but a constellation of factors. It is difficult to know which factors are causal and have the biggest impact.” According to Nock, a sense of disconnection, loss of camaraderie, mismatch of mentality, disconnection when coming back into society, as well as a lack of connection with services contribute to higher risk for suicide ideation in veterans. Although it is easy to assume PTSD or memory of combat takes enough of a toll on every veteran to directly cause suicide ideation, risk factors in veterans are in fact very similar to risk factors in civilians and service members. Legal problems, relationship problems, and substance abuse, which veterans are more susceptible to, are all common risk factors for suicide ideation in veterans. It is the problems associated with the transition from a military lifestyle to a civilian lifestyle that contribute most to suicide ideation in veterans. 

The Veterans’ Association clearly recognizes the need for programs to support veterans and lower risk for suicide attempts. It lists a number of mental health programs meant to support military personnel after their time of service; offerings include an online self-help portal under the Veteran Training program, a smartphone app to track daily habits, a “telemental health” program, and the option to speak to a fellow veteran who has struggled with mental health in the past. Veterans have the option to seek help online, over the phone, or in person. Some services, such as the telemental health program, are available 24 hours a day, 7 days a week. So why is it that ⅔ of veterans who die of suicide are not connected with Veterans’ Association services, which are so seemingly accessible?

One common trait among these programs stands out above all others: the fact that they are optional. A clear absence of mandatory mental health screening or treatment after service leaves many veterans to fend for themselves when experiencing suicide ideation. Programs set up by groups like the Veterans Association may never reach the individuals they are meant to help, even though they could be vital in preventing suicide ideation as well as suicide attempts. Veterans are in desperate need of mandatory, rather than optional, post-service programs that will screen or predict their likelihood of falling into suicide ideation. 

Nock recognizes this need, and he has a solution in mind. Machine learning, he says, will be the answer to predicting risk for suicide in veterans. “A clinician will ask questions about depression, suicide, and substance abuse, but the human brain is not equipped to assess the risk factors just because there is a combination of so many different risk factors. Machine learning helps combine all the different factors.” Now that health information is largely on an electronic record, it is possible to use machine learning to build statistical models, which can predict whether someone will attempt suicide. These statistical models can handle up to 3 dozen or more factors, making predictions more accurate than any human could. Although these models started out by using numeric code to create prediction models, they now use natural language processing to comb through doctor’s notes for certain keywords. These flagged keywords may lead to indicators of suicide ideation. 

In addition to machine learning, clinicians rely on behavioral tests to detect suicide ideation in an individual. For example, a patient may be given an implicit association test, where they are asked to associate concepts with each other in a timed environment. By measuring and analyzing reaction time to each of the questions, clinicians can look at patterns to see if the individual may be at risk for suicide ideation.

With the need for mandatory screening growing as the suicide rate in veterans increases, incorporating machine learning-based statistical models and implicit bias tests into the discharge is more important than ever. The Veterans’ Association must recognize this need and work with the US Armed Forces to take more upstream approaches to suicide prevention and mental health before we lose more of this vulnerable population.