The “wave of veteran suicides,” in the words of The New York Times editors last year, seems to cap the traumas that the vets have borne in service to the nation. It turns out, however, that actually establishing that there is a connection between military service and suicide is difficult. It may take years more research to fully understand its personal toll. The “Forever War” of an earlier generation, Vietnam, produced a particularly strong debate about serving and suicide. While the tragic consequences seemed clear to some, the data have been much more opaque. The veteran-suicide connection was, as a recent article describes, also opaque a century ago when the veterans in question had served in the Civil War.
Vietnam and Since
Advocates for Vietnam vets are convinced that their suicide rate has been abnormally high and, moreover, that the Defense Department has suppressed the true count (see, e.g., here; here). Scholars have argued over the evidence, some finding that vets were no likelier than other men of their age and background to commit suicide. (I refer to men because women took on roles closer to combat only in recent years. Note, also, that the attention here is to suicides by veterans, not to suicides by on-duty servicemen.)
At least two technical issues complicate getting a satisfying answer. One is establishing an accurate count of suicides. Official statistics underestimate suicides because authorities often label ambiguous cases such as car crashes or drug overdoses as accidents, sometimes to spare the feelings of the family. It is hard to know whether such undercounting is more or less common for veterans than for others.
The other issue is the “selection” effect. The men who ended up in the military–even when there was a draft–and the men who, once in the military, ended up in Vietnam, much less those who ended up in combat, were not a random draw of the population. Even before entering the military, they differed in systematic ways from non-vets. So, if the raw suicide rates are higher for vets, that alone cannot establish cause and effect.
How, one might ask, could war service not increase the propensity to suicide? It may be that people generally recover from the trauma of combat. Or, while service may be traumatizing for some people, perhaps for just as many others the military experience builds strength and confidence. Indeed, there is some evidence that the military tends to select and keep those who are “healthy warriors,” and that servicemen generally fare better, at least barring combat, than those who never served.
One effort to assess the veteran-suicide connection was a study reported in a community health journal in 2007. It drew on a national health survey conducted in the 1980s and 1990s of hundreds of thousands of Americans. The researchers looked up the respondents in the National Death Index to see which of the interviewees were later recorded as having died by suicide. After adjusting for all sorts of other differences among the men (age, race, marital status, education, health, etc.), the researchers found that those who had reported serving on active duty were about twice as likely as otherwise similar non-vets to have later committed suicide. The vets were not, the researchers noted, more likely than to die than the non-vets for other reasons.
However, in 2008, a “blue ribbon” Defense Department commission (pdf) concluded that various studies on the veteran-suicide connection were, in the end, mixed. Its report pointed to several technical problems in correctly ascertaining that relationship.
A May, 2012 paper published in the American Journal of Public Health challenged the 2007 analysis. These authors used the same health survey and national death data as the previous ones had, extended a few years. With a slight adjustment in procedures, they found that veterans were likelier to commit suicide by firearm than were non-veterans–veterans are likelier to own firearms–but that overall the difference between vets and non-vets was statistically marginal, perhaps a 10 percent elevated rate. Although they concluded that there was no causal effect on suicide of being a veteran, there remained reason to suspect that there was such an effect. Another analysis, reported in the same issue of the journal, using somewhat different data, reported especially elevated suicide rates among young veterans. Controversy ensued in this issue and then in the May, 2014 issue of the AJPH. It seems a muddle.
Recognizing that there are prior differences between veterans and non-veterans that may not be adjustable statistically–in the Vietnam era, for example: feelings about the war, or access to a doctor who could provide a health deferment, or being a criminal defendant who was offered the option of enlisting rather than going to jail–some researchers have sought to use a proxy for having served that is uncontaminated by a selection effect: men’s Vietnam War draft lottery number, which was randomly based on birthdays. The lower the lottery number for his birth date, the higher the chance a man had of being drafted and of serving in Vietnam, whatever his personal characteristics or inclination. (Draft number serves as an “instrumental variable” in econometrics lingo.)
For example, a 2102 article in Demography assesses whether men born between 1950 and 1952 with draft-eligible birthdays (low lottery numbers) ended up years later with higher than expected suicide death rates. The answer is no. But it is not clear whether lottery number is really a good proxy for military service during the Vietnam War. In an earlier study, two economists showed that the connection between draft lottery number and actual service, though real, is probably not strong enough to use the number as a solid proxy for being in the military.
Cannot this question be settled? A few studies of Vietnam vets suggest a clarification: that the key is not being a veteran but being a combat veteran. A 1996 paper reported that Vietnam army vets who had been wounded more than once were likelier than other vets to have later died by suicide. A 2002 article compared thousands of pairs of twins who had both served during the Vietnam era. The one who had experienced serious combat was likelier to say that he had considered suicide than the twin who had not seen major combat. Finally, a 2012 report found that Vietnam vets who reported having killed someone in Vietnam more often reported having considering suicide than ones who had not killed.
It would appear that combat experience, not military experience, did raise the risk of suicide, at least for those who served in Vietnam. That it took time to see this price of service clearly is not new. This research forms a context for a 2015 study of suicide among veterans of the Civil War.
Historian Larry Logue, in the Journal of Social History, uncovers and explores the records of suicides in Massachusetts from 1860 to 1900. One discovery is that the suicide rate of (white) Civil War veterans was about 50 percent higher than that of non-veterans. Another was that, early on, the news stories about suicides–and suicides were popular grist for the press–tended to ignore that part of the decedents’ situations. They typically failed to mention that the victim had been a veteran and typically explained their suicides as the consequences of insanity, depression, economic misfortune, and the like. Toward the end of the nineteenth century, however, press coverage more often mentioned military service and more often ascribed the suicides to the physical or mental toll of the Civil War. It was only later, Logue argues, that the media and the public were willing to revisit the war experience with more open eyes, willingness to acknowledge its tragedy, and empathy for its long-term victims.
The story of the veteran-suicide connection for Vietnam sounds similar.
– – – – – – – – – – – – – – – – – – – –
 Disclosure: I had a student deferment for a few years of the war and then drew a high draft number. I passed an Army physical in Boston but was never called.
 The study did not take into account that, if the way military experience eventually leads to suicide is through a series of difficulties (e.g., divorce, unemployment, ill health), then controlling for those factors would underestimate the extent to which service promotes suicide.
 For example: Some men report themselves as veterans even if they had washed out of basic training; a higher proportion of veterans’ suicides may be accurately reported because veterans are likelier than other suicides use firearms, rather than the more ambiguous means, to take their lives.
 Overall, unadjusted, the odds ratio for risk of suicide between self-reported veterans and other men was 1.36. This study adjusted only for year, age, and race, and came up with an adjusted OR of 1.11 (95% CI: 0.96, 1.29). But when the researchers applied similar procedures to the same years as the 2007 study, they found a significant veterans effect. The claim of nil results is not that definitive.
 For men born from 1950 to 1952, about 30 percent of those who were draft-eligible by lottery ended up serving versus about 15 to 20 percent of those with high lottery numbers (Figure 2). That’s a notable difference, but by far most men in both groups did not serve, which makes inference from birthday to service quite fuzzy.
 Two methods points: Actual reported suicide attempts were so few in the samples for the last two studies that effects on attempts could not be reliably estimated, though the trends were consistent. And: These studies also control for all sorts of presumably intermediate conditions such as PTSD and substance abuse. Properly done, the estimates of combat’s effects should have included indirect effects through those conditions and would–presumably–result in stronger findings.