Abstract
Objectives. We examined the association of military service history with past-year suicidal ideation and past-30-days mental distress in a probability-based sample of adults.
Methods. We gathered 2010 Behavioral Risk Factor Surveillance System data from 5 states that asked about past-year suicidal ideation. Military service was defined as current or former active-duty service or National Guard or Reserves service. We stratified analyses into 18 to 39 years, 40 to 64 years, and 65 years and older age groups and used multiple logistic regression analyses, adjusted for demographic confounders, to discern the association of military service history with past-year suicidal ideation and past-30-days mental distress.
Results. Among the 26 736 respondents, 13.1% indicated military service history. After adjusting for several confounders, we found military history status among those aged 40 to 64 years was associated with both past-year suicidal ideation and past-30-days mental distress. We found no significant associations among the younger or older age groups.
Conclusions. Differences in suicidal ideation between military and nonmilitary individuals may occur in midlife. Future research should examine the possibility of cohort effects, service era effects, or both.
In the United States, suicide has over the past decade become the 10th leading cause of death.1 Since 2005, suicide rates among US active-duty service members have seen unprecedented increases,2 and US veterans are estimated to make up about 20% of individuals who die from suicide.3 Identifying, assessing, and engaging US service members and veterans who may have risk for suicide are major priorities for the clinicians, public health professionals, and organizations serving veterans and their families.2,4
In examining suicidality among service members and veterans, a preponderance of studies have focused on those already engaged in health care settings within the US Department of Defense or Veterans Health Administration environments (i.e., clinical populations).5–8 Population-based investigations of suicide risk among general samples of service members and veterans are scant, predominately reporting overall rates of suicide mortality.9–16 Research about whether US military service itself is associated independently with risk for suicide remains equivocal.12,17
Additionally, it is unclear whether phenomena related to yet distinct from suicide, such as suicidal ideation, may be elevated among service members and veterans. Studies among service members and veterans have identified varying prevalence of suicidal ideation, ranging from 6.5% to 45.9% in clinical samples18–22 and from 2.3% to 21.2% in nonclinical samples.23–28 However, most of this research has used non–probability-based sampling methods and lacked comparison groups of people without military experience. For example, using data from the 2010 Behavioral Risk Factor Surveillance System (BRFSS), Bossarte et al. (2012) found that 3.8% of veterans reported suicidal ideation in the past 12 months.28 Although this estimate resembles the national estimate for suicidal ideation among adults (3.7%),29 the data were from only 2 states (Nebraska and Tennessee) and lacked a comparison group without military experience.
Perhaps the clearest current investigation of suicidal ideation among a nonclinical, probability-based sample of people with and without military service history comes from the National Survey of Drug Use and Health (NSDUH). White et al.30 analyzed data for men from the 2008 NSDUH data set and stratified the analysis by age groups of 18 to 25 years, 26 to 35 years, and 36 years and older. They found no differences in prevalence of past-12-month suicidal ideation between men who had ever served on active duty (3.0%) and men who had never served on active duty (3.7%). Although they found no difference in suicidal ideation between military service history status within age groups, the age groupings used by White et al. were constructed in a way that may not have been sensitive to a key shift in suicide risk in the general US population: the increase of suicide among people in midlife (approximately aged 40–64 years).31–33
With the increasing focus on suicide among current and former military populations, population-based data for examining differences in suicidal ideation between people with and without military service history are remarkably scarce. The few studies that have managed to make these comparisons have limitations, and further research is needed regarding the prevalence of suicidal behaviors among those who have served in the military and the associations between suicide risk and characteristics that may be unique to members of this group. Using data from the BRFSS, which gathers population-based health risk information from each US state, we compared the prevalence of past-year suicidal ideation and past-30-days mental distress among a large probability-based sample of US adults by history of military service. Furthermore, because suicide risk varies by age in the general US population,29,34 we also examined whether prevalence of suicidal ideation and mental distress differed on the basis of the age categories (18–39 years, 40–64 years, ≥ 65 years) that were used in the detection of the midlife increase in suicide among US adults.31
Data are from the Centers for Disease Control and Prevention’s 2010 BRFSS, which used computer-assisted telephone interviews with probability-based samples of noninstitutionalized adults from all US states, territories, and the District of Columbia.35 In 2010, 5 US states chose to ask about past-year suicidal ideation on their surveys (Alaska, Montana, Nevada, New Mexico, and Ohio), but these state-added data were not gathered by the Centers for Disease Control and Prevention for use in the national data sets.
To use the Centers for Disease Control and Prevention–created weights with the data, we added the state-level data into the national dataset following a series of steps to ensure accurate merging. First, we gathered individual BRFSS datasets from each state that asked about suicidal ideation. Second, we matched deidentified observations from the state datasets to their deidentified observations in the national dataset by matching on the unique combination of respondent sequence number and Federal Information Processing Standards code for each state. Once matched, a suicidal ideation variable was created in the national dataset. We then input the data for the state-added suicidal ideation question into the national BRFSS for only those observations for which the question was received. Responses from all other states that did not administer a suicidal ideation item were set to missing for this variable. Four of the 5 states administered the question to their entire survey sample; Ohio administered the question to respondents in 2 of 3 equivalent sample splits.
The first dependent variable of interest was past-year suicidal ideation. There was slight variation across the 5 states’ suicidal ideation survey items. Alaska, Nevada, and Ohio used the same item: “During the past 12 months, did you ever seriously consider attempting suicide?” Montana’s item was worded, “During the past 12 months, have you seriously considered attempting suicide, even if you would not really do it?” Last, New Mexico’s survey contained the question “In the past year, have you felt so low at times that you thought about committing suicide?” The response categories were the same across all 5 states: yes, no, don’t know, and refused. The responses were recoded into a dichotomy of yes or no, with the other responses coded as missing.
The second dependent variable was past-30-days mental distress, operationalized with a single item in the core survey. This item is worded consistently for all BRFSS surveys: “Now thinking about your mental health, which includes stress, depression, and problems with emotions, for how many days during the past 30 days was your mental health not good?” We recoded responses into a dichotomy using a cutoff of 6 or more days as an indicator of frequent mental distress.36
History of military service was assessed with 1 item: Have you ever served on active duty in the United States Armed Forces, either in the regular military or in a National Guard or military reserve unit? Active duty does not include training for the Reserves or National Guard, but DOES include activation, for example, for the Persian Gulf War.
Response options included yes, now on active duty; yes, on active duty during the last 12 months, but not now; yes, on active duty in the past, but not during the last 12 months; no, training for Reserves or National Guard only; and no, never served in the military.
Because of the low prevalence of active-duty personnel (n = 114), among whom only 1 indicated suicidal ideation, we coded military service history to include categories 1 to 4. Respondents who indicated category 5 were considered the comparison group without military service.
Additional demographic information included age, sex, race (non-Hispanic White, non-Hispanic Black or African American, non-Hispanic other racial group, and Hispanic), and educational attainment (high school diploma, GED, or less; some college; and college degree or higher). We coded current marital status into a 3-category variable of married or coupled, formerly married (separated, divorced, widowed), and never married. We also coded employment status into 3 categories: employed (employed for wages or self-employed), unemployed (out of work for > 1 year or out of work for < 1 year), and out of the workforce (student, homemaker, retired, unable to work).
We stratified all analyses into age groups—18 to 39 years, 40 to 64 years, and 65 years and older—because our a priori research question was formulated on age groups that would take into account the increased risk for suicide in midlife.31 In addition, we found a significant veteran × age interaction that surfaced in the multivariable analyses (data not shown). Finally, Gibbons et al. suggested that the near-constant skew of age in analyses that involve veterans should not be adjusted as merely a main effects covariate.9
For each of the 3 age categories, we compared group differences between people with a military service history and people without a military service history using the χ2 test. We used multiple logistic regression models, also stratified by age group, to assess the association between military service history with past-year suicidal ideation and military service history with past-30-days mental distress (as separate outcomes) while adjusting for demographic information. All analyses were weighted using the final person weights supplied by the Centers for Disease Control and Prevention in the national data set to adjust for the complex sampling design and noncoverage and nonresponse. Unweighted frequencies are presented with weighted percentages, all means and standard deviations are weighted, and weighted adjusted odds ratios (AORs) are reported with 95% confidence intervals (CIs). Missing data in the multivariable analyses were handled with listwise deletion. We conducted the analyses using Stata SE version 12 (StataCorp, College Station, TX).
The total analytic sample across the 5 states was 26 685, with 13.1% (95% CI = 12.4, 13.8) of the sample indicating military service history. In aggregate, the group with military service history had a significantly larger proportion of men and was older and less racially diverse than the group without military service history (Table 1). The group with a military service history was also more likely to be married, have at least some college education, and be out of the workforce.
TABLE 1— Demographics and Prevalence of Past-30-Day Mental Distress and Past-Year Suicidal Ideation, by Age Group and Military Service History: Behavioral Risk Factor Surveillance System; Alaska, Montana, Nevada, New Mexico, and Ohio; 2010
| All Ages | Aged 18–39 Years | Aged 40–64 Years | Aged ≥ 65 Years | |||||
| Characteristic | Military Service (n = 4250), No. (%) or Mean ±SD | No Military Service (n = 22 435), No. (%) or Mean ±SD | Military Service (n = 300), No. (%) or Mean ±SD | No Military Service (n = 4194), No. (%) or Mean ±SD | Military Service (n = 1848), No. (%) or Mean ±SD | No Military Service (n = 11 837), No. (%) or Mean ±SD | Military Service (n = 2102), No. (%) or Mean ±SD | No Military Service (n = 6404), No. (%) or Mean ±SD |
| Sex | ||||||||
| Men | 3887 (90.7)* | 6781 (42.5) | 231 (79.7)* | 1587 (48.8) | 1640 (90.5)* | 3972 (43.2) | 2016 (96.6)* | 1222 (22.9) |
| Women | 363 (9.3) | 15 654 (57.5) | 69 (20.3) | 2607 (51.2) | 208 (9.5) | 7865 (56.8) | 86 (3.4) | 5182 (77.1) |
| Race/ethnicity | ||||||||
| Non-Hispanic White | 3412 (79.7)* | 16 870 (76.2) | 192 (72.3) | 2639 (69.3) | 1430 (76.3)* | 9060 (80.0) | 1790 (87.7) | 5171 (84.0) |
| Non-Hispanic Black or African American | 135 (7.4) | 705 (6.3) | 16 (11.4) | 175 (7.0) | 74 (8.6) | 348 (5.9) | 45 (4.0) | 182 (5.5) |
| Non-Hispanic other | 328 (7.0) | 2118 (6.9) | 50 (6.7) | 652 (8.7) | 178 (9.9) | 1077 (6.1) | 100 (3.6) | 389 (4.3) |
| Hispanic | 299 (5.8) | 2490 (10.6) | 40 (9.6) | 697 (15.0) | 131 (5.1) | 1225 (8.0) | 128 (4.7) | 568 (6.2) |
| Employment status | ||||||||
| Employed | 1610 (47.0)* | 11 627 (58.8) | 227 (74.2) | 2757 (64.4) | 1057 (61.6)* | 7909 (68.3) | 326 (15.0)* | 961 (13.3) |
| Unemployed | 209 (6.4) | 1487 (9.6) | 27 (10.3) | 471 (12.8) | 153 (9.2) | 902 (9.1) | 29 (0.8) | 114 (1.9) |
| Not in workforce | 2414 (46.6) | 9239 (31.7) | 44 (15.5) | 953 (22.8) | 634 (29.2) | 2989 (22.7) | 1736 (84.1) | 5297 (84.7) |
| Education | ||||||||
| ≤ high school diploma | 1487 (35.0)* | 8918 (39.5) | 91 (30.8) | 1655 (38.0) | 554 (31.2)* | 4135 (36.6) | 842 (41.7)* | 3128 (52.5) |
| Some college | 1307 (31.8) | 6280 (27.4) | 128 (39.9) | 1208 (28.8) | 646 (34.1) | 3391 (26.9) | 533 (24.9) | 1681 (24.9) |
| ≥ college degree | 1447 (33.2) | 7181 (33.1) | 81 (29.3) | 1326 (33.2) | 645 (34.7) | 4287 (36.5) | 721 (33.3) | 1568 (22.6) |
| Marital status | ||||||||
| Married or coupled | 2707 (72.9)* | 12 748 (64.7) | 194 (71.4)* | 2547 (59.5) | 1219 (73.8) | 7473 (72.8) | 1294 (72.5)* | 2728 (54.1) |
| Formerly married | 1221 (19.5) | 6958 (16.7) | 48 (12.6) | 406 (5.5) | 463 (18.6) | 3107 (18.5) | 710 (24.2) | 3445 (43.0) |
| Never married | 303 (7.6) | 2626 (18.6) | 55 (16.0) | 1226 (35.0) | 155 (7.6) | 1198 (8.7) | 93 (3.4) | 202 (2.9) |
| Age, y | 57.1 ±0.58* | 45.6 ±0.23 | 30.7 ±0.46* | 29.3 ±0.16 | 53.5 ±0.32* | 51.2 ±0.12 | 75.1 ±0.23* | 74.3 ±0.14 |
| Mental health | ||||||||
| Suicidal ideationa | 154 (3.3) | 749 (3.1) | 9 (4.7) | 179 (3.7) | 104 (4.7)* | 451 (3.1) | 41 (0.9) | 119 (1.5) |
| Mental distressb | 566 (14.9) | 3620 (16.9) | 52 (18.3) | 781 (18.2) | 350 (19.4) | 2097 (17.5) | 164 (7.5)* | 742 (11.5) |
Note. Frequencies are unweighted; percentages are weighted.
a Past year.
b Past 30 days.
*P < .05 for comparisons between military service vs no military service within age group.
When examined within age categories, the differences were less consistent. For example, in the 18 to 39 years category, we observed no differences in race, employment, or educational attainment between the groups with and without military service history. However, in the 40 to 64 years category, individuals with military service history differed on all demographic characteristics except marital status.
In aggregate analyses, we found no significant differences by military service history in crude prevalence of past-30-days mental distress (Table 1). When stratified by age group, however, people with military service history in the 65 years and older group had lower prevalence of mental distress than their contemporaries without military experience (7.5% vs 11.5%, respectively; P = .001), but this difference was accounted for by demographic characteristics when tested in a multivariable model (Table 2).
TABLE 2— Adjusted Odds of Past-30-Days Mental Distress, Overall and by Age Group: Behavioral Risk Factor Surveillance System; Alaska, Montana, Nevada, New Mexico, and Ohio; 2010
| Characteristic | All Ages (n = 25 724), AOR (95% CI) | Aged 18–39 Years (n = 4378), AOR (95% CI) | Aged 40–64 Years (n = 13 255), AOR (95% CI) | Aged ≥ 65 Years (n = 8091), AOR (95% CI) |
| Age | 0.98* (0.98, 0.99) | 1.00 (0.98, 1.03) | 0.97* (0.96, 0.98) | 0.99 (0.98, 1.00) |
| Sex | ||||
| Men | 0.61* (0.52, 0.71) | 0.58* (0.43, 0.78) | 0.62* (0.51, 0.74) | 0.84 (0.60, 1.17) |
| Women (Ref) | 1.00 | 1.00 | 1.00 | 1.00 |
| Race/ethnicity | ||||
| Non-Hispanic Black or African American | 0.98 (0.75, 1.29) | 1.12 (0.69, 1.81) | 0.66* (0.46, 0.94) | 1.44 (0.87, 2.38) |
| Non-Hispanic other | 1.06 (0.80, 1.40) | 1.10 (0.67, 1.80) | 0.93 (0.67, 1.29) | 1.12 (0.74, 1.69) |
| Hispanic | 0.89 (0.72, 1.08) | 0.88 (0.63, 1.23) | 0.92 (0.74, 1.15) | 1.20 (0.85, 1.70) |
| Non-Hispanic White (Ref) | 1.00 | 1.00 | 1.00 | 1.00 |
| Employment status | ||||
| Unemployed | 2.41* (1.93, 3.03) | 2.22* (1.52, 3.24) | 2.72* (2.11, 3.52) | 2.99* (1.44, 6.20) |
| Not in workforce | 1.70* (1.45, 2.00) | 1.67* (1.20, 2.32) | 2.76* (2.33, 3.29) | 1.30 (0.88, 1.91) |
| Employed (Ref) | 1.00 | 1.00 | 1.00 | 1.00 |
| Education | ||||
| ≤ high school diploma | 1.87* (1.58, 2.21) | 1.82* (1.28, 2.60) | 2.09* (1.73, 2.53) | 1.81* (1.35, 2.43) |
| Some college | 1.51* (1.27, 1.81) | 1.39 (0.98, 1.97) | 1.73* (1.42, 2.13) | 1.31 (0.92, 1.88) |
| ≥ college degree (Ref) | 1.00 | 1.00 | 1.00 | 1.00 |
| Marital status | ||||
| Never married | 1.07 (0.86, 1.33) | 1.24 (0.89, 1.72) | 1.55* (1.22, 1.97) | 1.22 (0.71, 2.08) |
| Formerly married | 1.37* (1.19, 1.58) | 1.63* (1.09, 2.43) | 1.46* (1.23, 1.74) | 1.20 (0.93, 1.55) |
| Married or coupled (Ref) | 1.00 | 1.00 | 1.00 | 1.00 |
| Military service history | ||||
| Yes | 1.30* (1.05, 1.60) | 1.25 (0.73, 2.13) | 1.49* (1.16, 1.92) | 0.84 (0.58, 1.22) |
| No (Ref) | 1.00 | 1.00 | 1.00 | 1.00 |
Note. AOR = adjusted odds ratio; CI = confidence interval.
*P < .05
In overall multivariable models adjusted for demographic characteristics, people with military service history had higher odds of reporting past-30-days mental distress than those without a military service history (AOR = 1.30; 95% CI = 1.05, 1.60). Analyses stratified by age categories revealed that this difference in past-30-days mental distress by military service history group was largely driven by the 40 to 64 years category (Table 2), which was the only category with significantly elevated odds of past-30-days mental distress among those with military service history (AOR = 1.49; 95% CI = 1.16, 1.92).
We also found no differences in crude prevalence of past-year suicidal ideation by military service history in the aggregate analyses. When prevalence was stratified by age group, a higher proportion of people aged 40 to 64 years with military service history than people without military service history reported past-year suicidal ideation (4.7% vs 3.1%, respectively; P = .01).
Even after adjusting for sex, age, race, marital status, education, and employment status, military service history was associated with an 82% increased odds of reporting past-year suicidal ideation among people aged 40 to 64 years (Table 3). No other age category showed significantly greater odds of reporting past-year suicidal ideation.
TABLE 3— Adjusted Odds of Past-Year Suicidal Ideation, Overall and by Age Group: Behavioral Risk Factor Surveillance System; Alaska, Montana, Nevada, New Mexico, and Ohio; 2010
| Characteristic | All Ages (n = 24 150), AOR (95% CI) | Aged 18–39 Years (n = 4037), AOR (95% CI) | Aged 40–64 Years (n = 12 422), AOR (95% CI) | Aged ≥ 65 Years (n = 7500), AOR (95% CI) |
| Age | 0.98* (0.97, 0.99) | 1.00 (0.98, 1.03) | 0.98 (0.96, 1.00) | 0.94 (0.88, 1.00) |
| Sex | ||||
| Men | 0.79 (0.57, 1.10) | 0.83 (0.47, 1.47) | 0.70* (0.50, 0.98) | 1.72 (0.85, 3.50) |
| Women | 1.00 | 1.00 | 1.00 | 1.00 |
| Race/ethnicity | ||||
| Non-Hispanic Black or African American | 0.47 (0.22, 1.00) | 0.64 (0.20, 1.98) | 0.30* (0.13, 0.69) | . . . |
| Non-Hispanic other | 1.80 (0.98, 3.34) | 2.42 (0.95, 6.17) | 1.09 (0.59, 2.00) | 1.34 (0.63, 2.82) |
| Hispanic | 1.02 (0.67, 1.56) | 1.11 (0.56, 2.20) | 1.05 (0.73, 1.51) | 1.33 (0.70, 2.55) |
| Non-Hispanic White | 1.00 | 1.00 | 1.00 | 1.00 |
| Employment status | ||||
| Unemployed | 2.30* (1.42, 3.70) | 2.07 (0.97, 4.44) | 2.53* (1.51, 4.24) | 4.35* (1.26, 15.02) |
| Not in workforce | 2.53* (1.76, 3.65) | 2.75* (1.41, 5.38) | 4.24* (3.00, 5.99) | 1.75 (0.95, 3.23) |
| Employed | 1.00 | 1.00 | 1.00 | 1.00 |
| Education | ||||
| ≤ high school diploma | 1.39 (0.95, 2.04) | 1.45 (0.69, 3.01) | 1.61* (1.09, 2.36) | 1.46 (0.73, 2.93) |
| Some college | 1.27 (0.88, 1.84) | 0.99 (0.51, 1.92) | 1.56* (1.07, 2.28) | 2.48* (1.29, 4.78) |
| ≥ college degree | 1.00 | 1.00 | 1.00 | 1.00 |
| Marital status | ||||
| Never married | 1.34 (0.87, 2.06) | 1.93* (1.08, 3.44) | 2.10* (1.38, 3.18) | 1.15 (0.34, 3.92) |
| Formerly married | 2.19* (1.56, 3.07) | 2.81* (1.29, 6.10) | 2.21* (1.58, 3.10) | 1.88* (1.06, 3.34) |
| Married or coupled | 1.00 | 1.00 | 1.00 | 1.00 |
| Military service history | ||||
| Yes | 1.47 (0.96, 2.26) | 1.54 (0.58, 4.10) | 1.82* (1.21, 2.75) | 0.51 (0.23, 1.12) |
| No | 1.00 | 1.00 | 1.00 | 1.00 |
Note. AOR = adjusted odds ratio; CI = confidence interval.
*P < .05.
To our knowledge, this is only the second report to investigate the association of US military service history and past-year suicidal ideation among a large, representative sample of adults. When stratified by age, however, we found that, among people aged 40 to 64 years, military service history was positively associated with both past-year suicidal ideation and past-30-days mental distress, even after adjusting for individual characteristics related to suicidal ideation (e.g., employment, marital status). In aggregate analyses, we found a prevalence of past-year suicidal ideation among individuals with military service history (3.3%) similar to the estimate of past-year suicidal ideation among veterans documented by Bossarte et al. (3.7%).28 Our findings also corroborate the results of White et al.30 from the 2008 NSDUH data set, showing that military service history, at least in overall analysis, is not independently associated with past-year suicidal ideation.
Although White et al. did not find any differences in past-year suicidal ideation among their age groups,30 direct comparison between studies is not possible because of differences in the covariates between the BRFSS and NSDUH. More specifically, White et al. were able to adjust for nuanced mental health outcomes that are comorbid with suicidal ideation (e.g., Diagnostic and Statistical Manual of Mental Disorders37 criteria for depression, substance abuse, and substance dependence). Unfortunately, the BRFSS does not collect this detailed mental health information. Thus, it is unclear whether the discordant findings by age group between the 2 studies represent truly disparate results or whether adjustment for similar covariates would have replicated the findings reported by White et al. Model specification and covariate adjustment are important implications because they bear directly on the currently contested question of whether military experience itself is an independent indicator of suicide risk.9
For instance, on one hand, specific mental health conditions may mediate the association between military service history and suicidal risk, suggesting (1) that military service is a proxy variable and (2) that there are not true differences in suicide risk between populations with and without military experience. On the other hand, military service history may itself be a significant independent moderator of the association between specific mental health conditions and suicide risk, which suggests that military service can be a significant risk indicator that warrants specific focus in future research. The present results lend cursory support for the latter in that mental distress in this sample was also elevated among the group with a history of military service in the 40 to 64 years category. The conceptualization of these associations bears vitally on the narratives that result from these studies and guide future research in this area.
One potential explanation for the positive association between military service history and past-year suicidal ideation among those in the 40 to 64 years group may be a recency effect of severance from service. Research has suggested that the period during which risk for suicide is highest is within the first 5 years after separation from active-duty military service.38 Increased prevalence of psychological distress and suicide risk have been documented after discharge from inpatient hospitalization,39 release from prison,40 or a major life event such as divorce41 or job loss.42 It is possible that increases in suicidal behavior after separation from active-duty military service are related to disruptions in social networks and availability of other supportive resources. Research is needed to confirm the increased risk after separation from active-duty military service and investigate the relationship between transition and nonfatal suicide behaviors.
Conversely, an alternate explanation may be qualitative differences surrounding the conflicts experienced by each age group. Although the BRFSS does not collect information about period of service, the 40 to 64 years group maps, at least in part, onto the cohort of US veterans who served in the Vietnam War. It is possible that Vietnam veterans endured unique combat experiences in that war.43 Furthermore, US society’s negative reception to that war and to the returning soldiers could have interacted with combat trauma or possibly have been a form of additive stress or social disconnection unique to that particular conflict. Vietnam veterans recall enduring intense stigma and shame in the wake of that war,44–46 and research with samples of Vietnam veterans has shown persistent, robust associations of perceived negative homecoming with poor mental health outcomes, such as social anxiety47 and posttraumatic stress disorder.48,49 We were limited in making any causal inferences that our findings were related to Vietnam era veterans because the BRFSS does not assess the specific theaters in which veterans served. Research is needed to examine whether unique social, environmental, and historical contexts of veterans’ experiences may play roles in suicidal ideation, self-directed violence, or mental distress.
It is important to note that suicidal ideation, although associated with suicide mortality, should be interpreted as a phenomenon distinct from suicide attempt and suicide.50 The majority of people who experience suicidal ideation may not ever attempt suicide and do not ultimately die by suicide.51–53 Consequently, it is unclear how the association of military service history may change when examined with a different outcome, such as suicide attempt. Unfortunately, national surveillance of suicidal outcomes is limited in practice, scope, and design, that is, surveys that include nuanced items about suicidal phenomena in a longitudinal manner are scant.54 Recent suicide surveillance systems implemented by the US Department of Defense and Department of Veterans Affairs offer promising opportunities for detailed, trajectory-based analyses.4
We must note several limitations. These data are cross-sectional, so causes of past-year suicidal ideation cannot be addressed. Second, although the data are population-based samples, they came from only 5 states, 4 of which are in the western United States, a region known for high rates of suicide.33 Thus, our results may not generalize to the entire United States. Third, military service history was self-reported and could not be verified with official records of service, thus inviting potential misclassification bias. Additionally, the BRFSS does not include information about military service, such as combat exposure, theaters served, or duration of service, which may be associated with mental distress and suicidal ideation. Fourth, although mental health and suicide risk differ significantly by gender,55 we were unable to explore specific comparisons of suicidal ideation among women because of the small sample size of women with military service history across the age groups (e.g., 69 women veterans aged 18–39 years). Fifth, although the past-year suicidal ideation items were mostly similar across states, a slight variation in wording may have introduced systematic response bias. For illustrative purposes, the overall state-level past-year suicidal ideation for each 2010 BRFSS survey in this project is presented with state-level past-year suicidal ideation from the 2009 to 2010 NSDUH56 in Table A (available as a supplement to this article at http://www.ajph.org). Although direct comparison between these estimates is not possible because of differences in wording (e.g., BRFSS items use the word suicide, whereas NSDUH uses the term kill yourself) and differences in survey year, this table gives some sense of the variation between the 2 surveys in regard to past-year suicidal ideation. Finally, omitted-variable bias may have stemmed from the lack of detailed mental health questions in the BRFSS that could explain the association between military service history and suicidal ideation.
The results of this study confirm some previous reports from population-based research about past-year suicidal ideation and military service history28 but are discordant in other specific aspects, namely age-related prevalence of suicidal ideation.30 Reasons for an increased prevalence of past-year suicidal ideation among individuals with military experience in midlife (ages 40–64 years) are unclear. It may possibly be a phenomenon linked with a cohort effect, a service era effect, or both. Recently, several initiatives aimed at suicide prevention among veterans have been implemented,4 but it is unclear whether specific cohorts of veterans may be in differential need or use services at differential rates given specific characteristics, such as age, time since severance, or theater of combat.
Acknowledgments
This work was partially supported by the VISN 2 Center of Excellence for Suicide Prevention and a postdoctoral fellowship to J. R. Blosnich through the US Department of Veterans Affairs Office of Academic Affiliations and the Center for Health Equity Research and Promotion, VA Pittsburgh Healthcare System.
Results of this study were delivered as a podium presentation at the 141st Annual Meeting of the American Public Health Association; November 2–6, 2013; Boston, MA.
We thank the state Behavioral Risk Factor Surveillance System coordinators from Alaska, Montana, Nevada, New Mexico, and Ohio for sharing their 2010 datasets. We also thank Brady Stephens at the VISN 2 Center of Excellence for Suicide Prevention for his assistance with data management.
Note. The opinions expressed in this work are those of the authors and do not necessarily represent the policies of the funders, institutions, the US Department of Veterans Affairs, or the US government.
Human Participant Protection
This secondary analysis of deidentified, publicly available data was approved as exempt by the institutional review board of the VA Pittsburgh Healthcare System.
