The sample of Regular Army enlisted soldiers (9650 cases [153 528 control person-months]) is a subset of the total Regular Army sample (193 617 person-months) from the Army Study to Assess Risk and Resilience in Servicemembers (Army STARRS) Historical Administrative Data Study. Control person-months were assigned a weight of 200 to adjust for undersampling. The standardized risk estimate (SRE) (number of soldiers with a suicide attempt per 100 000 person-years) was calculated assuming that other predictors were at their samplewide means. The SREs were calculated using a logistic regression analysis that adjusted for sociodemographics (sex, current age, race/ethnicity, education, and marital status) and service-related variables, including age at Army entry, time in service (1, 2, 3-4, 5-10, or >10 years), deployment status (never, currently, or previously deployed), delayed promotion, demotion, and military occupation (combat arms, special forces, combat medic, or other). The model also included a dummy predictor variable for calendar month and year to control for secular trends.
eTable 1. List and Brief Descriptions of Administrative Data Systems Included in the 2004-2009 Army STARRS Historical Administrative Data Study (HADS)
eTable 2. List of Military Occupational Specialties (MOS) in the 2004-2009 Army STARRS Historical Administrative Data Study (HADS) That Were Categorized as Combat Arms, Special Forces, and Combat Medic
eTable 3. International Classification of Diseases, Ninth Revision–Clinical Modification (ICD-9-CM) Codes Used to Identify Mental Disorders
eTable 4. The National Corrections Reporting Program (NCRP) Offense Codes Used to Identify Categories of Criminal Perpetration and Victimization
eTable 5. List of Family Violence-Related Data Sources Included in the 2004-2009 Army STARRS Historical Administrative Data Study (HADS)
eTable 6. The National Corrections Reporting Program (NCRP) Offense Codes Used to Identify Family Violence
eTable 7. Multivariable Associations of Sociodemographic, Service-Related, and Mental Health Variables With Suicide Attempt in the Total Sample of Regular Army Enlisted Soldiers
eTable 8. Counts and Rates of Suicide Attempts by Sociodemographic and Service-Related Variables Among Regular Army Enlisted Soldiers With and Without a History of Mental Health Diagnosis
eTable 9. Multivariable Associations of Marriage Start/End Recency With Suicide Attempt in the Total Sample of Regular Army Enlisted Soldiers
eTable 10. Counts and Rates of Suicide Attempts by Marriage Start/End Recency Among Regular Army Enlisted Soldiers With and Without a History of Mental Health Diagnosis
eTable 11. Counts and Rates of Suicide Attempts by Marriage Start/End Recency Among Regular Army Enlisted Soldiers With and Without a History of Mental Health Diagnosis
eTable 12. Counts and Rates of Suicide Attempts by Outpatient Physical and Injury-Related Healthcare Visits Among Regular Army Enlisted Soldiers With and Without a History of Mental Health Diagnosis
eTable 13. Multivariable Associations of Crime Victimization/Perpetration and Family Violence With Suicide Attempt in the Total Sample of Regular Army Enlisted Soldiers
eTable 14. Counts and Rates of Suicide Attempts by Crime Victimization/Perpetration and Family Violence Among Regular Army Enlisted Soldiers With and Without a History of Mental Health Diagnosis
Customize your JAMA Network experience by selecting one or more topics from the list below.
Ursano RJ, Kessler RC, Naifeh JA, et al. Risk Factors Associated With Attempted Suicide Among US Army Soldiers Without a History of Mental Health Diagnosis. JAMA Psychiatry. 2018;75(10):1022–1032. doi:10.1001/jamapsychiatry.2018.2069
What are risk factors for suicide attempt among US Army soldiers with no history of mental health diagnosis?
This longitudinal cohort study of 9650 enlisted soldiers with a documented suicide attempt and 153 528 control person-months found no history of mental health diagnosis in more than one-third of those who attempted suicide. Risk factors for attempt (sociodemographic, service related, physical health care, injury, subjection to crime, crime perpetration, and family violence) were similar regardless of previous diagnosis, although the strength of associations differed.
This study suggests that personnel, medical, legal, and family services records may assist in identifying suicide attempt risk among soldiers with unrecognized mental health problems.
The US Army suicide attempt rate increased sharply during the wars in Afghanistan and Iraq. Although soldiers with a prior mental health diagnosis (MH-Dx) are known to be at risk, little is known about risk among those with no history of diagnosis.
To examine risk factors for suicide attempt among soldiers without a previous MH-Dx.
Design, Setting, and Participants
In this retrospective longitudinal cohort study using administrative data from the Army Study to Assess Risk and Resilience in Servicemembers (Army STARRS), person-month records were identified for all active-duty Regular Army enlisted soldiers who had a medically documented suicide attempt from January 1, 2004, through December 31, 2009 (n = 9650), and an equal-probability sample of control person-months (n = 153 528). Data analysis in our study was from September 16, 2017, to June 6, 2018. In a stratified sample, it was examined whether risk factors for suicide attempt varied by history of MH-Dx.
Main Outcomes and Measures
Suicide attempts were identified using Department of Defense Suicide Event Report records and International Classification of Diseases, Ninth Revision, Clinical Modification E95 × diagnostic codes. Mental health diagnoses and related codes, as well as sociodemographic, service-related, physical health care, injury, subjection to crime, crime perpetration, and family violence variables, were constructed from Army personnel, medical, legal, and family services records.
Among 9650 enlisted soldiers with a documented suicide attempt (74.8% male), 3507 (36.3%) did not have a previous MH-Dx. Among soldiers with no previous diagnosis, the highest adjusted odds of suicide attempt were for the following: female sex (odds ratio [OR], 2.6; 95% CI, 2.4-2.8), less than high school education (OR, 1.9; 95% CI, 1.8-2.0), first year of service (OR, 6.0; 95% CI, 4.7-7.7), previously deployed (OR, 2.4; 95% CI, 2.1-2.8), promotion delayed 2 months or less (OR, 2.1; 95% CI, 1.7-2.6), past-year demotion (OR, 1.6; 95% CI, 1.3-1.8), 8 or more outpatient physical health care visits in the past 2 months (OR, 3.3; 95% CI, 2.9-3.8), past-month injury-related outpatient (OR, 3.0; 95% CI, 2.8-3.3) and inpatient (OR, 3.8; 95% CI, 2.3-6.3) health care visits, previous combat injury (OR, 1.6; 95% CI, 1.0-2.4), subjection to minor violent crime (OR, 1.6; 95% CI, 1.1-2.4), major violent crime perpetration (OR, 2.0; 95% CI, 1.3-3.0), and family violence (OR, 2.9; 95% CI, 1.9-4.4). Most of these variables were also associated with suicide attempts among soldiers with a previous MH-Dx, although the strength of associations differed.
Conclusions and Relevance
Suicide attempt risk among soldiers with unrecognized mental health problems is a significant and important challenge. Administrative records from personnel, medical, legal, and family services systems can assist in identifying soldiers at risk.
Rates of suicidal behavior in the US Army increased during the Afghanistan and Iraq wars.1-3 Among soldiers, as in the general population,4-8 mental disorders are consistent predictors of suicide attempt9-13 and death.3,14 However, only 60% of enlisted soldiers with a documented suicide attempt10 and less than 50% of soldiers who died by suicide3 actually received a prior mental health diagnosis (MH-Dx). Understanding risk in those outside the mental health care system15,16 is important for addressing the 40% of suicide attempts that occur among enlisted soldiers with no MH-Dx history.10 Risk factors for veteran suicide differ for those with and without documented psychiatric symptoms,15 but it is not known whether suicide attempt predictors in active-duty soldiers9,10 differ for those with and without an MH-Dx. Suicide attempts are more likely among enlisted soldiers who are female, younger, non-Hispanic white, less educated, in their first tour of duty (particularly the first 2 years of service), and never or previously deployed (vs currently deployed).10 Combat arms and combat medic occupations have elevated risk, whereas special forces soldiers have lower risk.16 Although findings are mixed regarding the association of marital status with military suicidal behaviors,2,10,11 recency of marriage beginning or ending may provide information about risk during marital transitions.17-19
Medical, legal, and family services records might help identify those at risk. Most civilians who die by suicide have contact with primary care providers in the year before death.20 Similarly, service members show high rates of outpatient health care use, frequently primary care, in the month before suicide or intentionally self-inflicted injury.21 Almost 97% of soldiers who died by suicide during 2004 to 2009 had a past-year non–mental health encounter, and almost 38% had a past-month encounter.22 Treatment for injuries may be particularly important.23 Recent injury-related hospitalization, including unintentional injury, predicts suicide among soldiers.24 Legal and family services encounters also provide opportunities because being subjected to crime, perpetrating crime,25-29 and family violence are associated with suicide risk.29-34 Using administrative data from the Army Study to Assess Risk and Resilience in Servicemembers (Army STARRS),35 we examined a wide range of suicide attempt predictors among active-duty Regular Army enlisted soldiers with and without previous MH-Dx and the extent to which predictors differed by MH-Dx history.
This retrospective longitudinal cohort study used data from the Army STARRS Historical Administrative Data Study, which integrates 38 Army and Department of Defense (DoD) administrative data systems, including every system in which suicidal events are documented. The Historical Administrative Data Study includes individual-level person-month records for all soldiers on active duty from January 1, 2004, through December 31, 2009 (1.66 million soldiers).36 Data analysis in our study was from September 16, 2017, to June 6, 2018. This component of Army STARRS was approved by the institutional review boards of the Uniformed Services University of the Health Sciences (Bethesda, Maryland), Harvard Medical School (Boston, Massachusetts), University of Michigan Institute for Social Research (Ann Arbor), and University of California San Diego (La Jolla), which determined that the study did not constitute human participant research because it relied on deidentified secondary data; therefore, informed consent was not obtained.
The analytic sample was 9650 Regular Army enlisted soldiers who attempted suicide from 2004 through 2009 (excluding Army National Guard and Reserve), plus an equal-probability sample of 153 528 control person-months. Data were analyzed using a discrete-time survival framework with person-month as the unit of analysis.37 We reduced computational intensity by selecting an equal-probability 1:200 sample of control person-months stratified by sex, rank, time in service, deployment status (never, currently, or previously deployed), and historical time. Control person-months excluded all soldiers with a documented suicide attempt or other nonfatal suicidal event (eg, suicide ideation)1 and person-months in which a soldier died. Each control person-month was assigned a weight of 200 to adjust for undersampling.
Soldiers who attempted suicide were identified using administrative records from the Department of Defense Suicide Event Report,38 a DoD-wide surveillance mechanism aggregating suicidal behavior information via a standardized form completed by medical providers, and from International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) E950 to E958 codes (self-inflicted poisoning or injury with suicidal intent) associated with health care encounters from military and civilian treatment facilities, combat operations, and aeromedical evacuations (eTable 1 in the Supplement). For soldiers with multiple suicide attempts, we selected the first attempt using a hierarchical classification scheme.1
Administrative records were used for sociodemographic variables (sex, current age, race/ethnicity, education, and marital status) and service-related variables (age at Army entry, time in service, deployment status [never, currently, or previously], delayed promotion, demotion, and military occupation [combat arms, special forces, combat medic, or other]) (eTable 2 in the Supplement). We also calculated recency of marriage starting or ending (ie, divorced or widowed).
Administrative medical records identified previous MH-Dx during Army service. The indicator variable combined all ICD-9-CM mental health diagnostic codes and mental health–related V codes (eg, stressors/adversities and marital problems), excluding postconcussion syndrome and tobacco use disorder (eTable 3 in the Supplement). This definition of MH-Dx detected all soldiers with and without documented, clinically significant mental health difficulties. In most cases, absence of MH-Dx indicates that significant symptoms were not reported or detected during routine medical care. It does not signify thorough assessment of mental disorders.
Administrative medical records identified number of days with an outpatient physical health care (OP-physical) visit in the previous 2 months, presence and recency of outpatient and inpatient visits for previous injury (OP-injury and IP-injury), and previous combat injury–related visits. Injury-related health care visits were identified using ICD-9-CM codes 800 to 999, codes in the 700s indicating musculoskeletal injuries,39 supplemental E codes, and injury and trauma codes from the 2050 North Atlantic Treaty Organization (NATO) Standardization Agreement.40 Administrative records suggest that these previous injuries were nonsuicidal.
We examined legal records indicating soldiers who were subjected to or perpetrated founded crimes for which the Army had sufficient evidence to warrant investigation (eTable 4 in the Supplement). Subjects of crime and perpetrators were hierarchically categorized by crime severity (major violent, minor violent, or nonviolent). We constructed a variable indicating whether soldiers were subjected to and/or perpetrated documented family violence (eg, physical, sexual, or emotional abuse of spouse or child) using legal records, the Army Central Registry (a family services data system capturing family violence–related events), and medical (ICD-9-CM) records (eTable 5 and eTable 6 in the Supplement), with details reported elsewhere.31 Consistent with previous work,31 the family violence indicator included founded or unfounded legal events and substantiated or unsubstantiated Army Central Registry events.
Analyses were conducted using a statistical software program (SAS, version 9.4; SAS Institute Inc).41 After examining the association of MH-Dx history with suicide attempt in univariable and multivariable logistic regression models, the sample was stratified by soldiers with and without previous MH-Dx. Within each group, logistic regression analyses examined multivariable associations of sociodemographic and service-related characteristics with suicide attempt, followed by separate models evaluating incremental predictive associations of marriage start/end recency (replacing marital status in those models), physical health care use, injury, subjection to crime, crime perpetration, and family violence. Interactions were examined in separate multivariable models (adjusting for sociodemographic and service-related variables) to identify whether associations differed by MH-Dx history. Logistic regression coefficients were exponentiated to obtain odds ratios (ORs) and 95% CIs. Final model coefficients were used to generate a standardized risk estimate42 (ie, number of soldiers with an suicide attempt per 100 000 person-years) for each predictor category under the model assuming that other predictors were at their samplewide means. Because the suicide attempt rate increased during the Afghanistan and Iraq wars,1 each logistic regression equation controlled for calendar month and year. Coefficients of other predictors can consequently be interpreted as averaged within-month associations based on the assumption that associations of other predictors do not vary over time. Significance was evaluated using 2-sided tests and set at P < .05.
Most of the weighted sample was male (86.3% [26 507 814 person-months]), younger than 30 years (68.4% [20 975 370 person-months]), of non-Hispanic white race/ethnicity (59.8% [18 358 008 person-months]), high school educated (76.5% [23 503 980 person-months]), and currently married (54.7% [16 814 774 person-months]) (eTable 7 in the Supplement). Among 9650 enlisted soldiers with a documented suicide attempt (74.8% male), more than one-third (3507 [36.3%]) had no previous MH-Dx. An MH-Dx history was associated with higher odds of suicide attempt (OR, 3.2; 95% CI, 3.1-3.3), which persisted after adjusting for sociodemographic and service-related variables (OR, 6.2; 95% CI, 5.9-6.5) (eTable 7 in the Supplement). When stratified by MH-Dx history, significant sociodemographic and service-related predictors were generally similar. In both groups, odds of suicide attempt were higher among soldiers who were female, younger, non-Hispanic white, less educated, in their first 4 years of service, never or previously deployed, and those with a delayed promotion, demotion, and combat arms or combat medic occupation. Associations of sex, education, age at Army entry, time in service, delayed promotion, demotion, and military occupation with suicide attempt differed by MH-Dx status (Table 1). Pairwise analyses found that ORs were generally larger among those with no previous MH-Dx. Female sex, less than high school education, delayed promotion, and past-year demotion increased odds of suicide attempt in both groups but significantly more so in soldiers without previous MH-Dx. Entering the Army before age 21 years was associated with suicide attempt in those without but not with MH-Dx history. While ORs were generally higher among soldiers with no previous MH-Dx, their standardized risk of suicide attempt was consistently lower in each predictor category (Table 1).
Marriage recency was associated with suicide attempt in both groups. Soldiers in the first month after marriage had lower odds of suicide attempt than those not currently married (OR, 0.6; 95% CI, 0.4-0.9 for no MH-Dx and OR, 0.4; 95% CI, 0.3-0.6 for any MH-Dx), and those 4 to 12 months after marriage had higher odds (OR, 1.2; 95% CI, 1.0-1.3 for no MH-Dx and OR, 1.2; 95% CI, 1.1-1.3 for any MH-Dx). Odds were higher 2 to 3 months after marriage only among soldiers with no previous MH-Dx (OR, 1.3; 95% CI, 1.1-1.6). The association of marriage recency with suicide attempt differed by MH-Dx status because the OR for 2 to 3 months was larger among soldiers without previous MH-Dx (χ21 = 8.9; P = .003). Recency of marriage ending was not associated with suicide attempt in either MH-Dx group (Table 2).
We examined physical health care visits in a series of separate adjusted models (Table 3 and Table 4). Among soldiers who attempted suicide, more than 72% of those without MH-Dx history and almost 87% of those with MH-Dx history had at least 1 OP-physical visit in the 2 months before their attempt. In both groups, odds of suicide attempt increased monotonically as visits increased. Among those with no previous MH-Dx, odds of attempt increased from 1 visit (OR, 1.2; 95% CI, 1.0-1.3) to 8 or more visits (OR, 3.3; 95% CI, 2.9-3.8). Among those with a previous MH-Dx, odds of attempt increased from 1 visit (OR, 1.5; 95% CI, 1.4-1.7) to 8 or more visits (OR, 5.1; 95% CI, 4.6-5.6). The association of OP-physical visits with suicide attempt differed by MH-Dx status (Table 3) because all ORs were significantly larger among soldiers with MH-Dx history (χ21 = 8.4-16.6; P ≤ .004 for all). The Figure shows the increase in standardized risk associated with OP-physical visits.
Among those with no previous MH-Dx, elevated odds of suicide attempt were associated with previous IP-injury (OR, 2.7; 95% CI, 2.1-3.4) and OP-injury (OR, 1.7; 95% CI, 1.6-1.8). Soldiers with a previous MH-Dx also had higher odds if they ever had an IP-injury (OR, 3.1; 95% CI, 2.8-3.4) or OP-injury (OR, 1.4; 95% CI, 1.3-1.5). The association of injury-related health care visits with suicide attempt differed by MH-Dx status (Table 3) because the OR for OP-injury visit was larger among soldiers with an MH-Dx history (χ21 = 29.9; P < .001).
Odds of suicide attempt increased with recency of OP-injury and IP-injury in both MH-Dx groups. Among soldiers with no MH-Dx history, odds of attempt were highest for past-month OP-injury visit (OR, 3.0; 95% CI, 2.8-3.3) and lowest for a visit at least 4 months prior (OR, 1.6; 95% CI, 1.5-1.8). This pattern was the same for soldiers with previous MH-Dx. Odds of attempt were similarly highest for past-month IP-injury among those without (OR, 3.8; 95% CI, 2.3-6.3) and with (OR, 6.7; 95% CI, 5.7-7.8) MH-Dx history. The association of IP-injury recency with suicide attempt differed by MH-Dx status (Table 3) because the ORs for 1 month (χ21 = 4.0; P = .047) and 3 months (χ21 = 5.3; P = .02) were larger among soldiers with previous MH-Dx. Combat injury was associated with increased odds of suicide attempt only among those with no MH-Dx history (OR, 1.6; 95% CI, 1.0-2.4).
Subjection to crime, crime perpetration, and family violence were associated with suicide attempt in both MH-Dx groups. Relative to those who were not subjected to crime, soldiers without previous MH-Dx had higher odds of suicide attempt if they were subjected to a minor violent crime (OR, 1.6; 95% CI, 1.1-2.4) but not other crimes. Soldiers with previous MH-Dx who were subjected to any category of crime had elevated odds of attempt ranging from major violent crime (OR, 1.7; 95% CI, 1.5-2.0) to nonviolent crime (OR, 1.1; 95% CI, 1.0-1.2). In both groups, perpetration in any crime category was associated with higher odds of suicide attempt, particularly major violent crime (OR, 2.0; 95% CI, 1.3-3.0 for no MH-Dx and OR, 1.9; 95% CI, 1.7-2.2 for any MH-Dx). Family violence history was also associated with increased odds of suicide attempt in both groups (OR, 2.9; 95% CI, 1.9-4.4 for no MH-Dx and OR, 1.5; 95% CI, 1.4-1.7 for any MH-Dx). Associations of crime perpetration and family violence with suicide attempt differed by MH-Dx history, but being subjected to crime did not (Table 4). The ORs for nonviolent crime perpetration (χ21 = 33.9; P < .001) and family violence (χ21 = 12.9; P < .001) were higher among soldiers with no MH-Dx history. Additional descriptive data and nonstratified results are listed in eTables 8 through 14 in the Supplement.
More than one-third of enlisted soldiers who attempted suicide had no MH-Dx history. It is likely that many of these soldiers had undetected mental disorders.5,43,44 Mental health problems may go undiagnosed because a soldier (1) did not perceive a need for treatment or did not seek help45; (2) was not assessed for mental health problems during medical evaluations; (3) did not report symptoms during postdeployment screening46; (4) screened positive but did not follow up on the referral47; or (5) screened positive and followed up, but the evaluation did not result in MH-Dx. Ensuring adequate functioning of programs to facilitate detection and treatment is essential and may be aided by improving mental health knowledge among soldiers48 and maximizing a culture of honest symptom reporting.49 Detection may be enhanced through expanded mental health screening in primary care, a universal access point to Army health care. Screening before basic training would benefit new soldiers because many enter the Army with a history of mental disorders and suicide ideation.50,51 In the present study, almost 60% of undiagnosed soldiers who attempted suicide were in their first year of service (vs about 21% of those with previous MH-Dx), suggesting that there was less time for problems to be detected and treated. This factor is noteworthy because, as in the general population,4 most transitions from ideation to attempt among soldiers occur within 1 year of ideation onset.11,50,52
It is possible that some of those with no previous MH-Dx who attempted suicide did not have a mental disorder. For those soldiers, it might be expected that acute stressors have a significant role in their suicide attempt. Some results support this idea: for example, delayed promotion, demotion, and family violence had stronger associations with suicide attempt among those without previous MH-Dx. However, this interpretation is complicated by the fact that previous injuries, particularly recent injuries, had a stronger association with suicide attempt among those with MH-Dx history. This may support a vulnerability-stress model of suicidal behavior,53 in which suicide attempt results from the interaction of stressors (eg, injuries) with the preexisting vulnerability/diathesis (eg, MH-Dx).
Sociodemographic and service-related predictors of suicide attempt were similar among soldiers in both MH-Dx groups, although ORs were generally larger in those with no MH-Dx history. In particular, being female increased odds of suicide attempt in both groups, but more so among soldiers without previous MH-Dx. Women with no MH-Dx history may be more likely to have undetected mental health problems or unreported exposure to violence or discrimination. Given the large proportion of first-year soldiers in the group without previous MH-Dx, it is possible that women face additional stressors and challenges in the initial months of service. Perhaps surprisingly, all soldiers had decreased suicide attempt risk in the first month after marriage and increased risk 4 to 12 months after marriage, highlighting how risk may change during transitions. The lack of findings associated with end of marriage may reflect the discrepancy between when the breakup began vs when it was administratively recorded.
Regardless of MH-Dx status, soldiers who attempted suicide were higher users of OP-physical health care than controls and more likely to have sustained an injury. More than 72% of suicide attempt cases with no previous MH-Dx had at least 1 OP-physical visit in the past 2 months. However, risk detection remains a challenge because soldiers in general access medical care at a high rate. Consistent with previous findings,24 suicide attempt risk was higher in both MH-Dx groups shortly after treatment for injuries, particularly inpatient-related injuries. Injuries can result in significant mental health morbidity, increasing risk of suicide attempts over time.54,55 Future research should examine whether injury type, repeated injuries, and other physical health problems56-58 are predictive of suicide attempts.
Increased suicide attempt risk associated with being subjected to crime, crime perpetration, and family violence is consistent with previous studies.25-28,30-34,59 Risk was particularly elevated among perpetrators in both MH-Dx groups and among soldiers with no previous MH-Dx who had documented family violence. These community access points offer opportunities for suicide risk assessment. Future research should examine number and recency of specific crimes, as well as interactions with sociodemographic and service-related characteristics (eg, sex and deployment history).26,27,29,31
This study has noteworthy limitations. First, administrative records are limited to events that come to the Army’s attention and are subject to classification and coding errors. Unreported suicide attempts, mental disorders, and crimes/family violence (eg, sexual assault and spousal abuse) were not captured. Future analyses of Army STARRS survey data that are linked to respondents’ administrative records will allow for examination of undiagnosed mental disorders, subthreshold mental disorders, and undocumented stressful events among soldiers with no MH-Dx. Second, our injury variable may have captured injuries that were self-inflicted but unrecognized as such. Third, although the interaction tests adjusted for the main associations of other covariates, they did not adjust for other potential interactions. Fourth, our findings may not generalize to other periods or populations.
Interactions with medical, legal, and family services systems create assessment and prevention opportunities for soldiers with previously undetected suicide attempt risk. Machine learning models incorporating variables from disparate Army data systems can provide decision support tools to assist community service providers and clinicians in targeting interventions and deploying resources.59,60 Army STARRS has produced such algorithms to detect concentration of risk for suicide and other adverse outcomes within the health care61,62 and legal63-65 systems. The accuracy of these models can be further improved through the addition of self-report screening data.66 Findings also highlight the importance of strengthening mental health care embedded within primary care. Although one-third of soldiers who attempted suicide had no previous MH-Dx, most had an OP-physical visit in the weeks before their attempt. The Army has implemented mental screening and collaborative care programs in primary care settings to improve recognition and clinical management of mental disorders.67,68 Ensuring routine assessment of psychological distress and suicide risk during all encounters can help identify at-risk soldiers who are unknown to the mental health care system.
Accepted for Publication: June 10, 2018.
Corresponding Author: Robert J. Ursano, MD, Center for the Study of Traumatic Stress, Department of Psychiatry, Uniformed Services University of the Health Sciences, 4301 Jones Bridge Rd, Bethesda, MD 20814 (email@example.com).
Published Online: August 29, 2018. doi:10.1001/jamapsychiatry.2018.2069
Author Contributions: Dr Ursano had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Ursano, Kessler, Naifeh, Herberman Mash, Fullerton, Heeringa.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Ursano, Naifeh, Herberman Mash, Aliaga.
Critical revision of the manuscript for important intellectual content: Ursano, Kessler, Naifeh, Herberman Mash, Nock, Fullerton, Wynn, Ng, Dinh, Sampson, Kao, Heeringa, Stein.
Statistical analysis: Ursano, Kessler, Aliaga, Ng, Kao.
Obtained funding: Ursano, Kessler, Nock, Fullerton, Heeringa.
Administrative, technical, or material support: Ursano, Kessler, Naifeh, Nock, Fullerton, Wynn, Dinh, Sampson, Heeringa.
Supervision: Ursano, Nock.
Conflict of Interest Disclosures: In the past 3 years, Dr Kessler reported receiving support for his epidemiological studies from Sanofi; reported being a consultant for Johnson & Johnson Wellness and Prevention, Shire, and Takeda; reported serving on an advisory board for the Johnson & Johnson Services Inc Lake Nona Life Project; and reported being a co-owner of DataStat, Inc, a market research firm that carries out health care research. Dr Stein reported being a consultant for Actelion Pharmaceuticals, Healthcare Management Technologies, Janssen Pharmaceuticals, Pfizer, Remedy Therapeutics, Oxeia Biopharmaceuticals, and Tonix Pharmaceuticals. No other disclosures were reported.
Funding/Support: The Army Study to Assess Risk and Resilience in Servicemembers (Army STARRS) was sponsored by the Department of the Army and funded under cooperative agreement U01MH087981 (2009-2015) with the US Department of Health and Human Services, National Institutes of Health, and National Institute of Mental Health. Subsequently, STARRS–Longitudinal Study was sponsored and funded by the Department of Defense (Uniformed Services University of the Health Sciences grant HU0001-15-2-0004).
Role of the Funder/Sponsor: As a cooperative agreement, scientists employed by the National Institute of Mental Health (NIMH) (Lisa J. Colpe, PhD, MPH, and Michael Schoenbaum, PhD) and Army liaisons/consultants (Steven Cersovsky, MD, MPH, USAPHC, and Kenneth Cox, MD, MPH, USAPHC) collaborated to develop the study protocol and data collection instruments, supervise data collection, interpret results, and prepare reports. Although a draft of the manuscript was submitted to the Army and NIMH for review and comment before submission, this process occurred with the understanding that comments would be no more than advisory.
Disclaimer: The contents are solely the responsibility of the authors and do not necessarily represent the views of the US Department of Health and Human Services, the National Institute of Mental Health, the US Department of the Army, or the US Department of Defense.
Create a personal account or sign in to: