Secondary Logo

Journal Logo

Original Articles

Gender Differences in Lifetime Prevalence and Onset Timing of Suicidal Ideation and Suicide Attempt Among Post-9/11 Veterans and Nonveterans

Hoffmire, Claire A. PhD*,†; Monteith, Lindsey L. PhD*,‡; Forster, Jeri E. PhD*,†; Bernhard, Paul A. BA§; Blosnich, John R. PhD, MPH∥,¶; Vogt, Dawne PhD#,**; Maguen, Shira PhD††,‡‡; Smith, Alexandra A. MS*; Schneiderman, Aaron I. PhD§

Author Information
doi: 10.1097/MLR.0000000000001431

Abstract

Among rising suicide rates in the United States, the veteran suicide rate appears to be increasing at an especially high rate; this is particularly notable for females.1,2 Among men, the suicide rate increased 43% in veterans compared with 7% in nonveterans, and among women, it increased 61% for veterans compared with 33% for nonveterans between 2005 and 2017. It is unknown whether these trends extend to suicidal ideation (SI) and suicide attempt (SA), which are major predictors of suicide and key points for intervention and prevention.3,4 Research on gender differences in veterans’ SI and SA has been limited to one study that reported greater odds of past-year SI among men, but no gender differences in past-year SA.5 However, this study focused on a regional sample and had no nonveteran comparison group.

In addition to assessing if there are gender differences in SI and SA prevalence, understanding when differences in SI and SA emerge is essential for suicide prevention efforts. A life-course perspective can advance understanding of both timing of events and gender differences to inform clinical care,6 as few studies have examined the onset or the trajectory of SI and SA among veterans.

Veterans and nonveterans have distinct experiences that support applying a life course perspective to understand differences in the onset and trajectory of SI and SA. Veterans are more likely than nonveterans to report adverse childhood experiences,7–9 which are associated with increased risk for SI and SA.10–12 Exposure to military sexual trauma, combat-related killing, perceived life threat, and stress-related to family separations during military service may also increase the risk for veterans.13–17 Furthermore, among veterans, women are more likely to experience sexual violence across the lifespan, whereas men are more likely to experience combat and killing during military service.18–20

These disparities may contribute to gender differences in rates and timing of suicide risk following separation from military service.21 Research on gender differences in SI and SA prevalence and timing of onset among veterans is limited; however, Monteith et al.22 recently examined SI and SA prevalence and onset relative to military service (pre-, during, and post-military); female veterans were most likely to report SI and SA following separation; however, onset was most likely to occur before military service.22 These findings suggest different periods of risk for SI and SA onset versus the highest burden (overall prevalence).

Veterans, particularly women, may be more likely than nonveterans to experience SI and SA but direct comparisons of veterans and nonveterans on SI and SA prevalence have been limited and findings mixed.23–25 To our knowledge, no studies have examined whether the onset of SI and SA differ for veterans and nonveterans nor have any studies examined the role of age and gender on the relationship between veteran status and SI or SA onset or prevalence.

To address these knowledge gaps, we analyzed data from the Comparative Health Assessment Interview (CHAI) Study, which included a national sample of post-9/11 veterans and matched nonveterans. We addressed the following aims in gender-stratified analyses: (1) describe lifetime and time-period (age; less than 18 y, 18 y and above) point prevalence of SI and SA among veterans and nonveterans; (2) describe time-period point prevalence of SI and SA in relation to military service (pre, during, post), among veterans; (3) compare at what age SI and SA were most commonly reported for veterans and nonveterans; (4) determine when veterans were most likely to report experiencing SI and SA relative to military service (pre, during, or post); and (5) determine within which time-period onset of SI and SA was most likely to occur relative to age (veterans and nonveterans) and military service (veterans).

METHODS

Participants and Survey Procedures

CHAI data collection occurred in 2018 and was approved by the Department of Veterans Affairs (VA) Central Institutional Review Board. Additional background on CHAI objectives and sampling methodology can be found in the online supplement (Supplemental Digital Content 1, https://links.lww.com/MLR/C110). Briefly, the veteran sampling frame was identified from the US Veterans Eligibility Trends and Statistics (USVETS) dataset, which included information on all separated veterans who served on active duty military service after September 11, 2001.26,27 Post-9/11 veterans were included irrespective of whether they deployed in support of the wars in Iraq and Afghanistan. Veterans were initially recruited by mail (invitation, 2 reminders) to complete the survey online or via a computer-assisted telephone interview with additional follow-up calls to nonresponders. Invitation letters included a $1 preincentive; completers received a $50 postincentive. A 40% response rate was achieved.

The nonveteran sample was selected from KnowledgePanel, the largest US probability-based, nationally representative online respondent panel (see supplement for details, Supplemental Digital Content 1, https://links.lww.com/MLR/C110).28 Eligible nonveterans were noninstitutionalized US residents 18 and older without military service, selected from both active and inactive panels to match the age and gender distributions of deployed veterans, and recruited via email using preincentives and postincentives using KnowledgePanel’s standard process. Response rates of 56.5% and 8.4% were achieved for active and inactive panelists, respectively.

Study Measures

Lifetime Suicidal Ideation and Suicide Attempt Point Prevalence

An adapted version of the Columbia-Suicide Severity Rating Scale29–31 was administered to assess the lifetime prevalence of SI (“Have you ever actually had any thoughts of killing yourself?”) and SA (“Have you ever made a suicide attempt?”).

Time-period Suicidal Ideation and Suicide Attempt Point Prevalence

All respondents who indicated lifetime SI or SA were asked whether they experienced these “before age 18” (childhood/adolescence) and/or “age 18 or older” (adulthood). Veteran respondents affirming SI or SA were also asked whether the experience occurred before, during, and/or after military service. Participants were given the option to select all that applied, allowing us to capture SI and SA point prevalence for each time period.

Suicidal Ideation and Suicide Attempt Onset

For respondents reporting lifetime SI and/or SA, onset was captured by age and military service time period. The same Columbia-Suicide Severity Rating Scale follow-up questions described above were used to assess SI and SA onset. For age-based analyses, if individuals indicated experiencing SI or SA “before age 18,” onset was set to childhood/adolescence, otherwise, onset was set to adulthood. For military service-based analyses, onset was similarly set to the earliest time period endorsed (before, during, or after military service).

Other Measures

Gender (stratification variable) was primarily defined by self-report [male, female, transgender male (female to male), transgender female (male to female), gender nonconforming, different identity] and was collapsed into a binary variable (men, women). The small number of transgender respondents (n=33) were categorized according to their gender identity [male to female=woman (n=11), female to male=man (n=22)]. Where self-reported gender was missing (n=46), the sampling frame sex variable (male, female) was imputed. For nonmissing observations, consistency between survey gender and frame sex was high (99.0% veterans, 98.4% nonveterans), supporting this approach. Those who reported gender as nonconforming or of different identity (n=41, 0.21%) were removed from the analysis, as the frame sex variable was unlikely to accurately represent gender identity. Additional self-report variables to describe the sample and determine the time at risk for SI and SA within each period included demographics and military service characteristics and dates. See Table 1 for a complete list and description. Time at risk (covariate) was computed from age at entry and separation from military service, and date of survey completion (Supplement, Supplemental Digital Content 1, https://links.lww.com/MLR/C110).

TABLE 1 - Sample Characteristics by Veteran Status and Gender
Raw N (Weighted %)
Veterans (N=15,082) Nonveterans (N=4638)
Characteristics Men (n=9544) Women (n=5538) Men (n=3225) Women (n=1413)
Age [mean (SD)] 38.74 (0.05) 36.90 (0.10) 43.96 (0.35) 44.43 (0.36)
Race/ethnicity
 White single race, non-Hispanic 6581 (68.70) 3092 (55.38) 2263 (62.62) 945 (62.30)
 Black single race, non-Hispanic 1119 (10.90) 1208 (21.16) 242 (11.98) 144 (12.11)
 Other single race, non-Hispanic 346 (3.86) 181 (3.47) 177 (8.23) 74 (6.51)
 2+ races, non-Hispanic 525 (5.65) 333 (6.18) 83 (1.60) 49 (1.99)
 Hispanic 973 (10.89) 724 (13.81) 460 (15.58) 201 (17.09)
Sexual orientation (n=19,656)
 Straight/heterosexual 9290 (97.28) 4921 (88.11) 2904 (91.33) 1300 (93.07)
 Lesbian/gay 84 (1.02) 289 (5.34) 220 (5.97) 29 (1.84)
 Bisexual 93 (1.11) 258 (5.51) 72 (1.84) 65 (3.88)
 Other 47 (0.59) 46 (1.05) 21 (0.86) 17 (1.21)
Marital status (n=19,690)
 Never married 1447 (19.57) 1013 (21.27) 1087 (34.20) 374 (28.42)
 Married/domestic partnership 6649 (65.44) 3093 (54.18) 1838 (55.38) 872 (55.48)
 Separated/divorced 1379 (14.56) 1352 (23.59) 276 (9.42) 145 (11.83)
 Widowed 55 (0.43) 71 (0.96) 18 (1.00) 21 (4.28)
Education (n=19,711)
 Less than high school/high school/GED 1175 (13.59) 294 (7.20) 632 (33.28) 398 (43.81)
 Some college credit, no degree 2683 (29.92) 1206 (24.82) 622 (20.55) 226 (17.30)
 Associate’s or bachelor’s degree 3881 (40.56) 2686 (47.15) 1309 (30.92) 534 (27.39)
 Graduate degree 1801 (15.94) 1347 (20.83) 662 (15.25) 255 (11.50)
Paid employment (n=19,673)
 Working 7854 (83.58) 4003 (72.72) 2678 (72.79) 994 (60.74)
 Not working, actively looking 588 (7.02) 476 (9.25) 196 (7.72) 112 (8.14)
 Not working, not looking 1071 (9.40) 1050 (18.03) 347 (19.50) 304 (31.13)
Current residence* (n=19,672)
 MSA 7973 (83.56) 4746 (85.17) 2870 (88.94) 1196 (82.65)
 Non-MSA 1547 (16.44) 768 (14.83) 355 (11.06) 217 (17.35)
Region (n=19,672)
 Northeast 886 (9.87) 475 (8.67) 577 (18.84) 249 (17.54)
 Midwest 1858 (20.53) 898 (17.48) 829 (22.81) 314 (20.14)
 South 4550 (46.36) 3011 (53.91) 1055 (34.56) 508 (38.03)
 West 2211 (23.13) 1121 (19.80) 764 (23.79) 342 (24.29)
 Other 15 (0.11) 9 (0.15) 0 (0.00) 0 (0.00)
Primary branch of service (n=15,033)
 Army 4868 (50.27) 2783 (52.71) NA NA
 Marine Corps 1191 (15.04) 225 (5.56) NA NA
 Navy 1523 (17.20) 1102 (18.86) NA NA
 Air Force 1912 (17.36) 1415 (22.87) NA NA
 Coast Guard 14 (0.12) 0 (0.00) NA NA
Ever Reserve/National Guard Service
 Yes 4916 (46.92) 2951 (48.69) NA NA
 No 4628 (53.08) 2587 (51.30) NA NA
Deployment—ever
 Yes 7237 (73.95) 3483 (51.76) NA NA
 No 2307 (26.05) 2055 (48.24) NA NA
Deployment—OEF/OIF/OND
 Yes 5406 (59.04) 2720 (38.70) NA NA
 No 4138 (40.96) 2818 (61.30) NA NA
OEF/OIF/OND Combat (n=8155)
 Yes 4337 (82.05) 1853 (67.80) NA NA
 No 1087 (17.95) 878 (32.20) NA NA
Percentages were calculated based on the available sample size for each variable. The sample size for each variable was based on n=19,720, unless noted otherwise. Percentages were weighted using the main study population weight.
*MSAs contains at least 1 county with a city population ≥50,000 or a Census Bureau-defined urbanized area with a total population of ≥100,000.
Northeast: Connecticut, Maine, Massachusetts, New Hampshire, New Jersey, New York, Pennsylvania, Rhode Island, Vermont; Midwest: Illinois, Indiana, Iowa, Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, and Wisconsin; South: Alabama, Arkansas, Delaware, District of Columbia, Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, West Virginia; West: Alaska, Arizona, California, Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington, Wyoming,
GED indicates General Educational Development; MSA, Metropolitan Statistical Area; NA, not available; OEF, Operation Enduring Freedom; OIF, Operation Iraqi Freedom; OND, Operation New Dawn.

Statistical Analysis

The dataset included main population weights and rescaling bootstrap replicate weights32 for all participants in addition to standardization weights created for the nonveteran sample (Supplement, Supplemental Digital Content 1, https://links.lww.com/MLR/C110).33 Standardization weights were used in models comparing veterans and nonveterans; population weights were used for all other within-group comparisons. Analyses were conducted with SAS, v.9.4 (©2016, SAS Institute Inc.).

Proportions of participants reporting having ever experienced SI or SA (lifetime point prevalence) and corresponding 95% confidence intervals (CIs) were computed, followed by time-period point prevalence by age (veterans and nonveterans) and in relation to military service (veterans only).

Next, to compare lifetime and time-period specific SI and SA prevalence between groups and time points, logistic generalized estimating equations logistic regression was used. This analytical strategy accounts for correlations between repeated measures (ie, intraindividual correlation of responses clustered within an individual, a concern given that participants who experience SI or SA in an early time period are more likely to experience it again in a future time period). We included years at risk within each time period as a covariate to control for the potential confounding effect of variations in time spent within each time period (eg, increased length of military service increases likelihood of experiencing SI during military service). SI or SA (yes/no) was modeled as a function of the categorical time period and veteran status (group comparison models only), and contrasts were constructed within each model for the comparisons of interest. No additional covariates were included because weights standardized the nonveteran group to match the age/sex/race-ethnic/education distribution of the veteran sample and all analyses were gender stratified. Odds ratios (ORs) and 95% CIs were computed. The ratio of odds ratios (ORRs) for veterans relative to nonveterans was also computed to assess the interaction between veteran status and the relationship between age and SI or SA prevalence.

The same generalized estimating equations logistic regression approach was used to determine the age and military service (veterans only) time periods during which SI and SA was most likely to first occur (onset). For each model, a repeated dichotomous onset outcome variable (yes/no) was defined for each individual/time period and modeled as a function of the categorical time period, controlling for years at risk in each period. The onset variable was equal to zero for the period(s) before onset, set to 1 for the period of onset, and set to missing for any subsequent periods (c.f. Monteith et al22). This approach ensures that the onset variable does not contribute information to the model beyond the time period where the first onset occurs.

RESULTS

Participant Characteristics

The analytic sample comprised 15,082 veterans and 4638 nonveterans. Women accounted for 36.7% (n=5538) of veterans and 30.5% (n=1413) of nonveterans. Table 1 presents demographic and military service characteristics across groups.

Suicidal Ideation and Suicide Attempt Lifetime and Period Prevalence

Suicidal Ideation

The unadjusted lifetime prevalence of SI was significantly higher for veteran men [24.0% (95% CI=23.0–25.0)] and women [28.8% (95% CI=27.3–30.3)], compared with nonveterans, and significantly higher for veteran women compared with veteran men. After accounting for years at risk and intraindividual correlation, only veteran men experienced significantly higher odds of lifetime SI compared with nonveterans [OR=1.13 (95% CI=1.01–1.27)]. For both men and women veterans, unadjusted SI prevalence increased monotonically over time.

Considering SI prevalence within age groups, findings differed relative to veteran status. After accounting for intraindividual correlation, the ORs for veterans reporting SI in childhood/adolescence relative to nonveterans were 0.42 (95% CI=0.33–0.53) for men and 0.59 (95% CI=0.48–0.72) for women, indicating significantly lower odds of childhood/adolescence SI for veterans. Conversely, adjusted ORs for veterans reporting adulthood SI relative to nonveterans indicated significantly higher odds of adulthood SI for veterans (Table 2).

TABLE 2 - Lifetime and Period Prevalence* of SI and SA, by Veteran Status and Gender
Veterans (N=15,082) Nonveterans (N=4641) Veterans Relative To Nonveterans
Men Women Men Women Men Women
Outcome % 95% CI % 95% CI % 95% CI % 95% CI OR 95% CI OR 95% CI
SI
 Lifetime 24.0 23.0–25.0 28.8 27.3–30.3 20.3 18.7–21.9 20.6 18.0–23.2 1.13 1.011.27 1.15 0.97–1.37
 <18 y 5.9 5.3–6.3 12.0 11.0–13.0 10.8 9.63–12.1 14.0 11.8–16.3 0.42 0.330.53 0.59 0.480.72
 18+ y 22.8 21.8–23.8 26.4 25.0–27.9 16.5 15.0–18.0 15.3 13.1–17.4 1.34 1.201.50 1.33 1.111.59
 Premilitary 6.5 6.0–7.1 13.3 12.3–14.4 NA NA NA NA NA NA NA NA
 During-military 13.5 12.6–14.3 18.2 17.0–19.4 NA NA NA NA NA NA NA NA
 Postmilitary 18.7 17.9–19.6 19.9 18.7–21.2 NA NA NA NA NA NA NA NA
SA
 Lifetime 5.6 5.0–6.1 12.4 11.4–13.5 4.07 3.3–4.9 7.8 6.0–9.6 1.25 0.99–1.57 1.35 1.041.77
 <18 y 1.2 0.9–1.4 5.5 4.9–6.2 1.9 1.5–2.4 5.7 4.1–7.2 0.45 0.320.64 0.74 0.54–1.01
 18+ y 4.9 4.4–5.5 9.5 8.6–10.4 2.8 2.1–3.5 3.6 2.4–4.9 1.68 1.252.24 2.12 1.463.09
 Premilitary 1.3 1.1–1.6 6.0 5.3–6.7 NA NA NA NA NA NA NA NA
 During-military 2.6 2.2–2.9 5.6 4.9–6.3 NA NA NA NA NA NA NA NA
 Postmilitary 3.4 2.9–3.8 5.8 5.0–6.5 NA NA NA NA NA NA NA NA
Missing observations were removed from the following analyses: n=1 missing observation for: male veteran SI<18 years, male veteran SI 18+ years and n=2 missing observations for female veteran SA <18 years, female veteran SA 18+ years.
*All proportions presented were weighted using the main population weight. Proportions are not adjusted for the duration of time at risk.
OR comparing veterans to nonveterans’ odds in favor of experiencing SI and SA control for the duration of time (ie, years) at risk and intraindividual correlation of responses over time. Individuals are at risk for 9 years in the <18-year timeframe, given that experiencing SI or SA before 10 years of age is uncommon.
Significant OR (α<0.05) are bolded.
CI indicates confidence interval; NA, not available; OR, odds ratio; SA, suicide attempt; SI, suicidal ideation.

Suicide Attempt

The prevalence of lifetime SA was also significantly higher for veteran men [5.6% (95% CI=5.0–6.1)] and women [12.4% (95% CI=11.4–13.5)] compared with nonveterans and significantly higher for women compared with men veterans. Counter to SI results, odds of experiencing lifetime SA for veterans relative to nonveterans was significantly higher for women only [OR=1.35 (95% CI=1.04–1.77)]. Furthermore, SA prevalence across military service periods increased for men but remained constant for women.

As observed for SI, the unadjusted prevalence of SA was significantly higher in adulthood for veterans. Accounting for a time at risk and intraindividual correlation, odds of SA in childhood/adolescence were significantly lower for veteran men relative to nonveteran men [OR=0.45 (95% CI=0.32–0.64)]. In contrast, odds of SA in adulthood were significantly higher for men [OR=1.68 (95% CI=1.25–2.24)] and women [OR=2.12 (95% CI=1.46–3.09)] veterans relative to nonveterans.

Identifying When Veterans and Nonveterans Are Most Likely to Experience Suicidal Ideation and Suicide Attempt

Suicidal Ideation

Adjusted odds of experiencing SI in adulthood relative to childhood/adolescence were significantly higher for nonveteran men [OR=1.37 (95% CI=1.18–1.59)], veteran men [OR=4.32 (95% CI=3.81–4.90)], and veteran women [OR=2.53 (95% CI=2.23–2.88)], but not for nonveteran women. In addition, ORRs for men and women veterans compared with nonveterans indicated the ORs for experiencing SI in adulthood relative to childhood/adolescence were higher for veterans (ie, a significant interaction was observed between veteran status and the relationship between age and SI prevalence; Fig. 1). Furthermore, for both men and women veterans, odds of experiencing SI was significantly associated with military time period with odds elevated across comparisons: during-military to premilitary; postmilitary to premilitary; and postmilitary to during-military (Table 3).

F1
FIGURE 1:
Ratio (veterans relative to nonveterans) of odds ratios (18+ relative to less than 18 y of age) for SI and SA. Error bars depict 95% confidence intervals for each presented ORR. The ORRs presented here can be interpreted as assessing the interaction between veteran status and the relationship between age and SI or SA prevalence. An ORR of 1 would indicate that the odds of experiencing SI or SA in adulthood as compared with childhood/adolescence is the same for veterans and nonveterans. All ORRs presented are significantly >1 indicating that the odds ratios for experiencing SI and SA in adulthood relative to childhood/adolescence are higher for veterans. ORR indicates ratio of odds ratio; SA, suicide attempt; SI, suicidal ideation.
TABLE 3 - Identifying When Veterans and Nonveterans are Most Likely to Experience Suicidal Ideation and Suicide Attempt, Accounting for Time at Risk
Suicidal Ideation Suicide Attempt
Men Women Men Women
Comparison OR 95% CI OR 95% CI OR 95% CI OR 95% CI
Nonveterans
 18+ relative to <18 1.37 1.181.59 1.13 0.97–1.31 1.02 0.71–1.46 0.56 0.360.88
Veterans
 18+ relative to <18 4.32 3.814.90 2.53 2.232.88 3.79 2.875.01 1.60 1.212.13
 During-military relative to premilitary 2.24 2.062.44 1.54 1.401.69 2.00 1.602.50 1.00 0.86–1.16
 Postmilitary relative to premilitary 3.39 3.123.68 1.70 1.551.87 2.69 2.143.37 1.02 0.87–1.19
 Postmilitary relative to during-military 1.51 1.421.61 1.11 1.021.20 1.34 1.151.58 1.02 0.87–1.20
Missing veteran observations were removed from the following analyses: suicidal ideation 18+ relative to <18, 1 man; suicide attempt 18+ relative to <18, 1 man and 2 women; suicidal ideation relative to military service, 3 men; suicide attempt relative to military service, 1 man and 5 women. The veteran-nonveteran standardization weight was used for age-based analyses to facilitate a comparison of findings for veterans and nonveterans. The main study population weight was used for veteran analyses relative to military service. All OR control for the duration of time (ie, years) at risk and intraindividual correlation of responses over time. Individuals are at risk for 9 years in the <18-year timeframe, given that experiencing SI or SA before 10 years of age is rare.
Significant OR (α<0.05) are bolded.
CI indicates confidence interval; OR, odds ratio.

Suicide Attempt

The adjusted odds of experiencing SA in adulthood relative to childhood/adolescence was significantly elevated for veterans, with larger OR observed for men [OR=3.79 (95% CI=2.87–5.01)] than women [OR=1.60 (95% CI=1.21–2.13)]. Conversely, nonveteran women had significantly reduced odds of experiencing SA in adulthood relative to childhood/adolescence [OR=0.56 (95% CI=0.36–0.88)]. As observed for SI, the veteran: nonveteran ORRs indicated significant interactions between veteran status and the relationship between age and SA prevalence (Fig. 1). In contrast to findings for SI, odds of experiencing SA were not significantly associated with the military service period for women. For veteran men, the findings for SA were similar to those for SI.

Suicidal Ideation and Suicide Attempt Onset

Age

SI and SA onset varied considerably for veterans and nonveterans and by gender within veteran groups. No differences in odds of SI or SA onset relative to age were observed for nonveteran men, whereas for nonveteran women, odds of initial onset of SI and SA in adulthood were significantly less than the odds of SI and SA onset in childhood. Veterans, both men and women, experienced significantly higher odds of initial onset of SI and SA in adulthood relative to childhood/adolescence. Among veterans, the magnitude of this effect was significantly higher for men compared with women, for both SI and SA.

Military Service

The odds of initial onset of SI were significantly elevated among women premilitary relative and during-military relative to postmilitary service; for male veterans, odds of initial onset were significantly lower for premilitary relative to both during and following military service. The observed pattern of findings was similar for SA, except, among women veterans, there were also significantly higher odds of onset premilitary relative to during-military service (Table 4).

TABLE 4 - Identifying When Veterans and Nonveterans Are Most Likely to Experience Onset of Suicidal Ideation and Suicide Attempt, Accounting for Time at Risk
Suicidal Ideation Onset Suicide Attempt Onset
Men Women Men Women
Comparison OR 95% CI OR 95% CI OR 95% CI OR 95% CI
Nonveterans
 18+ relative to <18 0.86 0.68–1.09 0.48 0.320.70 1.00 0.67–1.51 0.43 0.240.75
Veterans
 18+ relative to <18 4.61 4.095.20 1.79 1.532.09 5.74 4.427.46 1.51 1.191.91
 Premilitary relative to during-military 0.61 0.550.68 1.11 0.97–1.28 0.57 0.440.72 1.44 1.161.77
 Premilitary relative to postmilitary 0.67 0.590.68 2.06 1.732.46 0.65 0.490.85 2.22 1.702.91
 During-military relative to postmilitary 1.09 0.97–1.23 1.85 1.552.21 1.15 0.91–1.45 1.55 1.15–2.08
Missing veteran observations were removed from the following analyses: suicidal ideation 18+ relative to <18, 1 man; suicide attempt 18+ relative to <18, 1 man and 2 women; suicidal ideation relative to military service, 3 men; suicide attempt relative to military service, 1 man and 5 women. The veteran-nonveteran standardization weight was used for age-based analyses to facilitate a comparison of findings for veterans and nonveterans. The main population weight was used for veteran analyses relative to military service. All OR control for the duration of time (ie, years) at risk and intraindividual correlation of responses over time. Individuals are at risk for 9 years in the <18-year timeframe, given the assumption that experiencing SI or SA before 10 years of age is very rare.
Significant OR (α<0.05) are bolded.
CI indicates confidence interval; OR, odds ratio.

DISCUSSION

This is the first study to examine differences in prevalence and onset of SI and SA for US veterans compared with nonveterans and to assess the role of gender and age in these differences. Striking differences were observed based on both veteran status and gender, underscoring their roles in the onset and experience of SI and SA. One notable finding was the higher lifetime prevalence of SI and SA observed among veterans, compared with nonveterans. Adjusted comparisons indicated a higher lifetime prevalence of SI for veteran (vs. nonveteran) men and of SA for veteran (vs. nonveteran) women. These findings differ from prior research that suggested no differences, albeit in past-year SI and SA,30–32 and support the importance of taking a life course perspective. Although veterans were less likely than nonveterans to report SI (both genders) and SA (men only) in childhood/adolescence, a different pattern emerged in adulthood. In adulthood, veteran men and women were both more likely to report SI and SA, compared with their nonveteran peers. Nonetheless, the “black box” of elevated SI and SA for veterans in adulthood remains elusive. Clinical screening and prediction tools are currently used to identify veterans using VA health care services at increased risk for suicide,34,35 but further research is needed to determine whether such tools can accurately predict SI and SA.

The discrepant pattern of findings for veterans (relative to nonveterans) and in relation to age suggest that nonveterans are at greater risk for SI and SA in childhood/adolescence, whereas veterans, particularly men, maybe less “at risk” for SI and SA in childhood. In contrast, veterans appear to be at greater risk for SI and SA in adulthood and in relation to specific military service periods. There are several potential explanations for these findings. Selection and screening criteria for military service may exclude those with prior SI and SA. Alternatively, the “healthy soldier effect” which operates in earlier life (ie, when people typically enlist) may diminish in adulthood following challenges, traumas, and disruptions experienced during military service and civilian reintegration.13–17 For example, SI may occur after exposure to trauma during deployment and worsen as mental health conditions are exacerbated following military separation. The Interpersonal-Psychological Theory of Suicide36 may also help explain these findings. Veterans may feel misunderstood and experience difficulty connecting with others (thwarted belonging) and burdensome to friends and family (perceived burdensomeness) due to difficulties with postdeployment or postseparation readjustment and may have been habituated to the pain and fear of dying (acquired capability) through military experiences.

Observing meaningful gender differences in our findings underscores the need to consider gender when conducting suicide research and designing prevention strategies. Studies often do not include enough women to examine gender differences; our ability to do so while taking a life course approach is a major strength. By considering these less studied nuances, several novel findings were revealed. Gender differences observed for veterans (eg, increased lifetime and time-period SI and SA prevalence among women) may be explained by the fact that women are more likely to experience the dual burden of warzone exposures and military sexual trauma, which is associated with suicide.15,16 This can also couple with the cumulative burden of childhood adversity, which may reach a tipping point when additional traumas are experienced during the war, increasing women veterans’ risk for SI and SA due to cumulative lifetime exposures.37,38 Thus, for women veterans, prevention and intervention efforts at distinct military service periods which address trauma across the life course may be critical.

SI and SA are major predictors of suicide, and our findings present a picture of opportunity regarding prevention and intervention. Veterans report less early life SI and SA than nonveterans (at least among women), suggesting that Department of Defense, VA, and community practitioners may have opportunities in critical periods to prevent or identify and treat incident SI or SA for active duty service members and veterans if onset is occurring in adulthood. However, stakes are higher for veterans because incident SI or SA in adulthood also means that onset is likely occurring after (or potentially during) military training, which includes the potential for experiences that may increase capability for suicide (eg, extensive firearm training). Finally, that SI is highest postmilitary highlights the need to focus on interventions that maximize well-being and readjustment after separation. Increasing belongingness and decreasing access to lethal means are 2 additional avenues for upstream suicide prevention.

Study limitations include the lower than optimal response rate (for the nonveteran inactive panelists, see Supplement, Supplemental Digital Content 1, https://links.lww.com/MLR/C110), potential for bias due to self-report, dichotomization of gender, and selection based on survival, limited adjustment for other potential SI and SA risk factors, and the cross-sectional design. Further research is warranted to evaluate how additional SI and SA risk factors (eg, sexual orientation) relate to the timing of SI and SA onset for veterans as compared with nonveterans. To evaluate the potential for bias and mitigate concerns whenever possible, numerous factors were considered. First, reliance on a nationally representative panel of randomly recruited general US population members and weighting for nonresponse help to mitigate the possible introduction of bias in the nonveteran sample. Furthermore, although participants may have had difficulty recalling SI or SA related to age or military service, there is no reason to suspect such self-report bias would differ across study groups and the time periods in question are major milestones in life which should assist with recall. In addition, to avoid dropping observations with missing gender, we imputed sex from administrative sampling frame data when self-reported gender was missing. Although this approach could introduce gender misclassification, this concern is minimal as the number of respondents with missing data was small (n=46) and the sampling frame sex variable contained a mix of birth sex and gender. Finally, the time lag between military separation (after 2001) and data collection (2018) may introduce selection bias as only those who survived to the time of data collection could participate; longitudinal data are essential to fully address this concern in future research. Key study strengths include the life course perspective and associated data and a large, gender-balanced, nationally representative sample.

In sum, the comparison of SI and SA among veterans and nonveterans across different time periods resulted in a more comprehensive understanding of how veterans and nonveterans compare in their risk over time. Findings also suggest numerous avenues for future research. As some timeframes were quite broad (eg, above 18 y), it will be important to delineate when within these timeframes risk for SI and SA are most elevated. It will be particularly important to consider timing in relation to specific transition experiences, such as military entry or separation, or other significant life events. Identifying proximal factors that drive risk for SI and SA in each period is also essential. In addition, it would be useful to examine these findings in other veteran samples, such as those who served before 9/11. Finally, better understanding factors that buffer against SI and SA at various time periods, and delineating how these differ by gender, is critical to enhancing suicide prevention services for active duty service members, veterans, and nonveterans alike.

REFERENCES

1. Department of Veterans Affairs. 2019 National Veteran Suicide Prevention Annual Report; 2019. Available at: www.mentalhealth.va.gov/docs/data-sheets/2019/2019_National_Veteran_Suicide_Prevention_Annual_Report_508.pdf. Accessed February 15, 2020.
2. Department of Veterans Affairs. 2017 National Suicide Data Appendix; 2019. Available at: www.mentalhealth.va.gov/docs/data-sheets/2017/2005-2017_National_Data_Appendix.xlsx. Accessed February 15, 2020.
3. Bostwick M, Pabbati C, Geske JR, et al. Suicide attempt as a risk factor for completed suicide: even more lethal than we knew. Am J Psychiatry. 2016;173:1094–1100.
4. Nock MK, Borges G, Bromet EJ, et al. Suicide and suicidal behavior. Epidemiol Rev. 2008;30:133–154.
5. Blosnich JR, Brenner LA, Bossarte RM. Population mental health among US military veterans: results of the Veterans Health Module of the Behavioral Risk Factor Surveillance System, 2011–2012. Ann Epidemiol. 2016;26:592–596.
6. Bevans M, Clayton JA. Integrating the influence of sex and gender in research: keeping women veterans in focus. Womens Health Issues. 2019;29:S9–S11.
7. Blosnich JR, Dichter ME, Cerulli C, et al. Disparities in adverse childhood experiences among individuals with a history of military service. JAMA Psychiatry. 2014;71:1041–1048.
8. Katon JG, Lehavot K, Simpson TL, et al. Adverse childhood experiences, military service, and adult health. Am J Prev Med. 2015;49:573–582.
9. McCauley HL, Blosnich JR, Dichter ME. Adverse childhood experiences and adult health outcomes among veteran and nonveteran women. J Womens Health. 2015;24:723–729.
10. Blosnich JR, Montgomery AE, Dichter ME, et al. Social determinants and military Veterans’ suicide ideation and attempt: a cross-sectional analysis of electronic health record data. J Gen Intern Med. 2020;35:1759–1767.
11. Carroll TD, Currier JM, McCormick WH, et al. Adverse childhood experiences and risk for suicidal behavior in male Iraq and Afghanistan veterans seeking PTSD treatment. Psychol Trauma. 2017;9:583–586.
12. Dube SR, Anda RF, Felitti VJ, et al. Childhood abuse, household dysfunction, and the risk of attempted suicide throughout the life span: findings from the Adverse Childhood Experiences Study. JAMA. 2001;286:3089–3096.
13. Maguen S, Metzler TJ, Bosch J, et al. Killing in combat may be independently associated with suicidal ideation. Depress Anxiety. 2012;29:918–923.
14. Monteith LL, Maguen SBongar B, Sullivan G, James L. Combat-related killing and suicide through the Interpersonal-Psychological Theory of Suicide. Handbook of Military and Veteran Suicide: Assessment, Treatment, & Prevention. New York, NY: Oxford University Press; 2017:64–75.
15. Monteith LL, Holliday R, Hoyt T, et alKumar U. Military sexual trauma and suicidal self-directed violence: a narrative review and proposed agenda for future research. The Routledge International Handbook of Military Psychology and Mental Health. Oxfordshire, UK: Routledge; 2019:411–424.
16. Khan AJ, Li Y, Dinh JV, et al. Examining the impact of different types of military trauma on suicidality in women veterans. Psychiatry Res. 2019;274:7–11.
17. Kline A, Ciccone DS, Falca-Dodson M, et al. Suicidal ideation among National Guard troops deployed to Iraq: the association with postdeployment readjustment problems. J Nerv Ment Dis. 2011;199:914–920.
18. Street AE, Vogt D, Dutra L. A new generation of women veterans: Stressors faced by women deployed to Iraq and Afghanistan. Clin Psychol Rev. 2009;29:685–694.
19. Vogt D, Vaughn R, Glickman ME, et al. Gender differences in combat-related stressors and their association with postdeployment mental health in a nationally representative sample of US OEF/OIF veterans. J Abnorm Psychol. 2011;120:797–806.
20. Maguen S, Luxton DD, Skopp NA, et al. Gender differences in traumatic experiences and mental health in active duty soldiers redeployed from Iraq and Afghanistan. J Psychiatr Res. 2012;46:311–316.
21. Bullman T, Hoffmire C, Schneiderman A, et al. Time dependent gender differences in suicide risk among Operation Enduring Freedom and Operation Iraqi Freedom veterans. Ann Epidemiol. 2015;25:964–965.
22. Monteith LL, Holliday R, Miller CN, et al. Suicidal ideation, suicide attempt, and non-suicidal self-injury among female veterans: prevalence, timing, and onset. J Affective Disorders. 2020;273:350–357.
23. Logan J, Bohnert A, Spies E, et al. Suicidal ideation among young Afghanistan/Iraq War Veterans and nonveterans: individual, social, and environmental risk factors and perception of unmet mental healthcare needs, United States, 2013. Psychiatry Res. 2016;245:398–405.
24. Blosnich JR, Gordon AJ, Bossarte RM. Suicidal ideation and mental distress among adults with military service history: results from 5 US states, 2010. Am J Public Health. 2014;104(suppl 4):S595–S602.
25. Pease JL, Monteith LL, Hostetter TA, et al. Military service and suicidal thoughts and behaviors in a national sample of college students. Crisis. 2015;36:117–125.
26. US Department of Veterans Affairs. VA National Center for Veterans Analysis and Statistics. 2018. Available at: www.va.gov/vetdata/. Accessed August 13, 2020.
27. Hauser G. United States Veterans Eligibility Trends and Statistics (USVETS): A new data source with socioeconomic variables. VA HSR&D VIReC Database and Methods Seminar; 2019. Available at: www.hsrd.research.va.gov/for_researchers/cyber_seminars/archives/video_archive.cfm?SessionID=3626. Accessed August 1, 2020.
28. Ipsos. KnowledgePanel. Available at: www.ipsos.com/en-us/solutions/public-affairs/knowledgepanel. Accessed August 13, 2020.
29. Posner K, Brown GK, Stanley B, et al. The Columbia-Suicide Severity Rating Scale: initial validity and internal consistency findings from three multisite studies with adolescents and adults. Am J Psychiatry. 2011;168:1266–1277.
30. Matarazzo BB, Brown GK, Stanley B, et al. Predictive validity of the Columbia-Suicide Severity Rating Scale among a Cohort of At-risk Veterans. Suicide Life Threat Behav. 2018;49:1255–1265.
31. Katz I, Barry CN, Cooper SA, et al. Use of the Columbia-Suicide Severity Rating Scale (C-SSRS) in a large sample of Veterans receiving mental health services in the Veterans Health Administration. Suicide Life Threat Behav. 2019;50:111–121.
32. Rao JK, Wu CJ, Yue K. Some recent work on re-sampling methods for complex surveys. Surv Methodol. 1992;18:209–217.
33. Lumley T. Survey: analysis of complex survey samples. R package version; 2019;3:35–41.
34. Kessler RC, Hwang I, Hoffmire CA, et al. Developing a practical suicide risk prediction model for targeting high-risk patients in the Veterans Health Administration. Int J Methods Psychiatr Res. 2017;26:e1575.
35. McCarthy JF, Bossarte RM, Katz IR, et al. Predictive modeling and concentration of the risk of suicide: implications for preventive interventions in the US Department of Veterans Affairs. Am J Public Health. 2015;105:1935–1942.
36. Joiner TE. Why People Die by Suicide. Cambridge, MA: Harvard University Press; 2007.
37. Briere J, Godbout N, Dias C. Cumulative trauma, hyperarousal, and suicidality in the general population: a path analysis. J Trauma Dissociation. 2015;16:153–169.
38. LeBouthillier DM, McMillan KA, Thibodeau MA, et al. Types and number of traumas associated with suicidal ideation and suicide attempts in PTSD: findings from a US Nationally Representative Sample. J Trauma Stress. 2015;28:183–190.
Keywords:

veterans; deployment; suicidal ideation; suicide attempt; gender

Supplemental Digital Content

Copyright © 2021 Wolters Kluwer Health, Inc. All rights reserved.