Suicide without warning: Results from the Army Study to Assess Risk and Resilience in Servicemembers (STARRS)

Objectives

The rate of suicide increased in members of the United States Army since 2011 after the Iraq and Afghanistan wars and continues to be a major concern. In order to reverse this disturbing trend, it is vital to understand the risk and protective factors for suicide death in servicemembers.

Methods

Data were obtained from a case-control psychological autopsy study, which compared U.S. Army suicide decedent cases (n = 135) to a probability sample of living controls (n = 255) who are also service members weighted to be representative of the Army. Interviews were conducted with next-of-kin (NOK) and supervisor (SUP) informants. Multivariable logistic regressions models were constructed using predictors significant after controlling for multiple comparisons.

Results

The most parsimonious multivariable model controlling for deployment status, as reported by SUP predicting suicide death consisted of four significant variables: a spouse or partner left him or her in the past month (OR = 28.5 [95% CI = 1.8, 442.7] χ² = 5.72, p = .0168); a smaller social network (OR = 4.2 [95% CI = 1.0, 17.3] χ² = 3.97, p = .0462), less likely to seek help from a mental health counselor (OR = 3.4 [95% CI = 1.2, 9.7] χ² = 5.35, p = .0207) and more likely to be described as incautious (OR = 3.8 [95% CI = 1.2, 11.7] χ² = 5.42, p = .0199). The AUC = .88 [95%CI = 0.82, 0.94] for this regression model suggests strong prediction.

Conclusions

Our findings suggest that recent relationship problems, especially in soldiers who are less likely to seek out support from others, may be warning signs for detection and prevention of imminent risk of suicide and according to supervisor informant surveys, had neither evidence of a mental health disorder, nor disclosed suicidal ideation or self-harm. Implications for suicide prevention are discussed.