Longitudinal Trends in Suicidal Thoughts and Behaviors Among US Military Veterans During the COVID-19 Pandemic

Key Points

Question  Have suicidal thoughts and behaviors increased in US military veterans over the course of the first 3 years of the pandemic?

Findings  In this cohort study, the prevalence of suicidal ideation did not significantly increase among US veterans during the pandemic, and the prevalence of suicide planning and suicide attempts remained low and consistent with prepandemic levels. Greater severity of prepandemic psychiatric distress and loneliness and lower purpose in life were the strongest predictors of new-onset suicidal thoughts and behaviors.

Meaning  The prevalence of suicidal thoughts and behaviors has not increased for most US veterans during the first 3 years of the COVID-19 pandemic.

Importance  Concerns have been raised since the onset of the COVID-19 pandemic that vulnerable populations, such as military veterans, may be at increased risk of suicidal thoughts and behaviors (STBs).

Objective  To examine longitudinal trends in STBs in US military veterans during the first 3 years of the COVID-19 pandemic.

Design, Setting, and Participants  This cohort study is a population-based longitudinal study including US military veterans that used 3 surveys from the National Health and Resilience in Veterans Study. Median dates of data collection were November 21, 2019 (prepandemic); November 14, 2020; and August 18, 2022.

Main Outcomes and Measures  Lifetime and past-year suicidal ideation, suicide planning, and suicide attempt.

Results  In this longitudinal study including 2441 veterans (mean [SD] age, 63.2 years [14.0]; 2182 [92.1%] male), past-year suicidal ideation decreased from 9.3% prepandemic (95% CI, 8.2%-10.6%) to 6.8% a year later (95% CI%, 5.8-7.9%) and then slightly increased to 7.7% (95% CI, 6.7%-8.9%) 2 years later. In total, 9 veterans (0.4%) reported attempting suicide at least once during the follow-up period, while 100 (3.8%) developed new-onset suicidal ideation and 28 (1.2%) developed new-onset suicide planning. After adjusting for sociodemographic and military characteristics, factors strongly associated with new-onset suicidal ideation included higher education (odds ratio [OR], 3.27; 95% CI, 1.95-5.46), lifetime substance use disorder (OR, 2.07; 95% CI, 1.23-3.46), prepandemic loneliness (OR, 1.28; 95% CI, 1.09-1.49), and lower prepandemic purpose in life (OR, 0.92; 95% CI, 0.86-0.97). Factors associated with new-onset suicide planning included lifetime substance use disorder (OR, 3.03; 95% CI, 1.22-7.55), higher prepandemic psychiatric distress (OR, 1.52; 95% CI, 1.06-2.18), and lower prepandemic purpose in life (OR, 0.88; 95% CI, 0.81-0.95).

Conclusions and Relevance  Contrary to expectations, the prevalence of STBs did not increase for most US veterans during the COVID-19 pandemic. However, veterans with preexisting loneliness, psychiatric distress, and lower purpose in life were at heightened risk of developing new-onset suicidal ideation and suicide planning during the pandemic. Evidence-based prevention and intervention efforts that target these factors may help mitigate suicide risk in this population.

As of November 2022, complications related to COVID-19 have led to the death of more than 1 million US lives (6.6 million globally)^1,2 and contributed to considerable physical, financial, and social hardship. While measures to combat the spread of the disease, such as social distancing, have proven effective,³ there has been concern since the beginning of the pandemic that these measures may contribute to increases in loneliness, suicide, overdose, and alcohol-related diseases^4-8 that can have long-lasting ramifications. Despite these concerns, emerging evidence suggests that predictions of a parallel mental health crisis^9,10 may have been overstated or at least that the consequences have not yet emerged. A recent meta-analysis of 65 longitudinal studies found that, while many people experienced an increase in anxiety symptoms early in the pandemic, these increases generally reverted back to prepandemic levels a few months later.¹¹ Moreover, an analysis of data from 33 countries found that suicide rates decreased over the first year of the pandemic.¹² Collectively, these findings suggest most adults have adapted relatively well during the COVID-19 pandemic.¹³

In spite of this positive trend, some individuals may be at risk of experiencing a deterioration of mental health related to the pandemic. In the aforementioned meta-analysis,¹¹ individuals with preexisting physical health conditions experienced greater increases in distress than those without these conditions. Further, among all participants in that analysis, depressive symptoms increased during the pandemic, suggesting that some adverse consequences of the COVID-19 pandemic may persist or take longer to emerge. Thus, there is need to identify risk and protective factors for worsening mental health in vulnerable populations.

One group considered at risk of worsening mental health, such as increases in suicidal thoughts and behaviors (STBs), during the COVID-19 pandemic are US military veterans.^14,15 Relative to nonveterans,^16,17 veterans experience higher rates of several psychiatric disorders that are strongly linked to STBs, such as posttraumatic stress disorder, major depressive disorder, and alcohol use disorder.¹⁸ Veterans also endorse high rates of adverse childhood experiences¹⁹ and combat exposure, which are associated with emotion dysregulation²⁰ and greater stress susceptibility,²¹ as well as loneliness²² and poor physical health,^18,23 which are robust correlates of STBs.^24-27 Moreover, prior to the pandemic, the age- and sex-adjusted suicide rates for US veterans were more than 60% higher than those of nonveterans.²⁸ These findings suggest that veterans may compose a high-risk group for the development of new-onset STBs during the pandemic.

In an earlier study,²⁹ we examined longitudinal changes in STBs during the first year of the COVID-19 pandemic among US veterans. Our results revealed that, at the population level, the prevalence of suicidal ideation decreased 1 year into the pandemic (from 10.6% to 7.8%). However, they also indicated that a small portion of veterans (2.6%; approximately 450 000 individuals based on population benchmarks) experienced new-onset suicidal ideation 1 year in the pandemic, and low prepandemic social support and history of psychiatric problems were among the strongest predictors of new-onset suicidal ideation.

Here, we expanded on this study by examining the prevalence of STBs (ie, suicidal ideation, suicide planning, and suicide attempt) before the pandemic and at 1 year and almost 3 years after the prepandemic survey. We also examined the prevalence of incident (ie, new-onset) STBs in those without a history of STBs and identified empirically informed risk and protective factors¹⁸ that differentiated veterans with and without new-onset STBs.

Data were analyzed from the National Health and Resilience in Veterans Study (NHRVS), a nationally representative, longitudinal study of US veterans. All participants provided informed consent, and the ethics committee of VA Connecticut Healthcare System approved the study. The NHRVS followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) and American Association for Public Opinion Research (AAPOR) reporting guidelines.

The NHRVS sample was drawn from KnowledgePanel, a research panel of more than 50 000 households maintained by Ipsos, a survey research firm. KnowledgePanel is an online, probability-based, nonvolunteer survey panel of a nationally representative sample of US adults that covers approximately 98% of US households. A total of 7860 veterans were invited to participate in the initial survey, and 4069 (51.8%) completed it. To permit generalizability of study results to the US veteran population, Ipsos computed poststratification weights using the following benchmark distributions of US military veterans from the most recent (August 2019) Current Veteran Population Supplemental Survey of the US Census Bureau’s American Community Survey: age, sex, race and ethnicity (chosen by participants from Census categories), Census region, metropolitan status, education, household income, and military branch and years of service. An iterative proportional fitting (raking) procedure was used to produce the final poststratification weights.

A total of 4069 veterans completed a prepandemic survey before the first documented COVID-19 case in the United States (median completion date, November 21, 2019). Then, 3078 veterans completed a second survey 1 year later (hereafter referred to as the peripandemic survey) during the 2020 fall/winter surge of COVID-19 cases and before the widespread public availability of COVID-19 vaccines (median completion date, November 14, 2020). A total of 2441 veterans (60% of the prepandemic group, 79% of the peripandemic group) completed a third survey during late summer 2022 (median completion date, August 18, 2022). A comparison of veterans who did (n = 2441) and did not (n = 1628) complete all 3 surveys did not reveal any differences in rates of prepandemic STBs (eMethods in Supplement 1).

At the first, prepandemic survey, STBs were assessed using the Suicidal Behaviors Questionnaire–Revised (SBQ-R).³⁰ The STBs included past-year suicidal ideation, lifetime suicide planning, and lifetime suicide attempt. Measures are described in eTable 1 in Supplement 1.

The peripandemic survey reassessed STBs and inquired about pandemic-related stressors (eg, social restriction stress). To assess suicide planning and attempts during this second survey, the prompt for SBQ-R question 1 was slightly modified to read, “In the past year, have you ever thought about or attempted to kill yourself?” To positively endorse suicide planning, respondents had to endorse either, “I have had a plan at least once to kill myself but did not try to do it,” or “I have had a plan at least once to kill myself and really wanted to die.” To positively endorse suicide attempt, respondents had to endorse either, “I have attempted to kill myself but did not want to die,” or “I have attempted to kill myself and really hoped to die.”

Almost 3 years after the first survey was done, in August 2022, the third survey reassessed STBs. It also assessed infection-related stressors (eg, knowing someone who died from COVID-19–related complications) and measures of change in risk factors (eg, increases in psychiatric distress) between the prepandemic and peripandemic surveys (eMethods in Supplement 1). To assess suicide planning and attempts at this survey, the prompt for SBQ-R question 1 was slightly modified to read, “In the past 2 years, have you ever thought about or attempted to kill yourself?” To positively endorse suicide planning, respondents had to endorse either, “I have had a plan at least once to kill myself but did not try to do it,” or “I have had a plan at least once to kill myself and really wanted to die.” To positively endorse suicide attempt, respondents had to endorse either, “I have attempted to kill myself but did not want to die,” or “I have attempted to kill myself and really hoped to die.”

Descriptive statistics were calculated to estimate the prevalence of STBs at each survey point. McNemar χ² tests were computed to evaluate differences between prevalence estimates at each survey point. χ² and independent-samples t tests were conducted to identify significant differences (P < .05) between veterans who did and did not experience new-onset suicidal ideation or suicide planning. Veterans were identified as having new-onset suicidal ideation or planning if they did not screen positive for this outcome prepandemic but did screen positive at either of the follow-ups, which were combined to maximize statistical power.

Next, multivariable logistic regression analyses using a forward selection (Wald) method were conducted to examine predictors/correlates of new-onset suicidal ideation and suicide planning. Variables that significantly differentiated veterans with and without new-onset suicidal ideation or planning in bivariate analyses were included as independent variables. We deliberately set a liberal α level (P < .05) to maximize the number of potentially informative variables that might be included in the multivariable models.

Finally, following these analyses, relative importance analyses³¹ were conducted. These analyses partition the explained variance attributable to each significant independent variable while accounting for intercorrelations among these variables. Analyses predicting new-onset suicidal ideation and planning included only those veterans who denied lifetime suicide planning or suicide attempt and past-year suicidal ideation at the prepandemic assessment; data for 331 veterans were excluded from these analyses.

The mean (SD) age of participants in the full sample (N = 2441) was 63.2 years (14.0). The majority were male (n = 2182; 92.1%) and non-Hispanic White (n = 2032; 79.5.%), while the remainder were either non-Hispanic Black (n = 165; 10.2%) or Hispanic (n = 164; 6.0%) or reported a mixed or other race and ethnicity (n = 80; 4.3%).

Figure 1 shows the prevalence of past-year suicidal ideation at the prepandemic, peripandemic, and third assessments in both the full sample and stratified by age and sex. Table 1 shows the prevalence of all STBs at each assessment. Results revealed that the prevalence of past-year suicidal ideation decreased from prepandemic (9.3%; 95% CI, 8.2%-10.6%) to peripandemic (6.8%; 95% CI, 5.8%-7.9%) in the full sample, as well as in all sex and age subgroups except veterans 65 years and older. Results also indicated that the prevalence of suicidal ideation increased from the peripandemic survey to the third survey in the full sample (7.7%; 95% CI, 6.7%-8.9%) and among veterans aged 45 to 64 years (9.1%; 95% CI, 7.2%-11.2%, to 11.3%; 95% CI, 9.2%-13.5%). For all groups except female veterans and veterans 65 years and older, the prevalence of prepandemic suicidal ideation was higher than that observed at the third survey.

Of the 2110 veterans who did not endorse any history of STBs at the prepandemic survey, 100 veterans (3.8%; 95% CI, 2.9%-4.7%) developed new-onset suicidal ideation during the pandemic, and 28 (1.2%; 95% CI, 1.0%-1.7%) developed new-onset suicide planning. Two veterans (0.1%) reported incident suicide attempts. Given the low number of suicide attempts during the pandemic, we did not further analyze this variable. Variables that differentiated US military veterans with and without new-onset suicidal ideation and new-onset suicide planning at the bivariate level are shown in eTables 2 and 3 in Supplement 1.

Table 2 and Table 3 show results of multivariable regression analyses examining independent predictors and correlates of new-onset suicidal ideation and planning during the COVID-19 pandemic. Analyses revealed that higher education, lifetime history of alcohol use disorder and/or drug use disorder, prepandemic loneliness, increases in loneliness, and greater worsening of relationships during the pandemic were independently associated with greater odds of new-onset suicidal ideation. Higher prepandemic purpose in life was independently associated with lower odds of new-onset suicidal ideation. Analyses also indicated that a lifetime history of alcohol use disorder and/or drug use disorder and greater prepandemic psychological distress were independently associated with greater odds of new-onset suicide planning, and higher prepandemic purpose in life was associated with lower odds of this outcome.

Results of relative importance analyses (Figure 2) revealed that lower prepandemic purpose in life (30.2% relative variance explained [RVE]) and greater prepandemic loneliness (27.1% RVE) were the strongest predictors of new-onset suicidal ideation. Greater prepandemic psychological distress (57.1% RVE) and lower prepandemic purpose in life (31.6% RVE) were the strongest predictors of new-onset suicide planning.

To our knowledge, this is the first nationally representative study to examine longitudinal changes in STBs among US military veterans during the first 3 years of the COVID-19 pandemic. Results revealed that, at the population level, the prevalence of suicidal ideation decreased over the course of the pandemic and the prevalence of suicide planning remained consistently low. Suicide attempts in the total sample during the pandemic were also low (0.4%) and consistent with prepandemic estimates.³² Collectively, these findings align with emerging research^11,12 suggesting that early concerns about substantial increases in STBs related to the COVID-19 pandemic^4-8 may have been overstated. Indeed, despite grim prognostications about the effect of the pandemic on suicidal behavior, accumulating evidence from epidemiologic studies suggest that there has been a decline in suicide among adults since the onset of the pandemic.^12,33

Nevertheless, despite an overall decrease in suicidal thinking, a significant portion of veterans (3.8%) developed new-onset suicidal ideation and/or new-onset suicide planning (1.2%) during the pandemic. Based on US veteran population benchmarks of the proportion of veterans without prepandemic histories of STBs, this suggests that approximately 580 000 veterans contemplated suicide for the first time during the pandemic, and approximately 180 000 veterans made a plan to end their lives. Moreover, a 0.1% incidence of suicide attempts suggests that approximately 15 000 veterans attempted suicide during the pandemic. These findings are concerning, especially given that a history of suicidal ideation and/or a suicide plan significantly increases the likelihood that a person will attempt suicide in the future.^34,35 It will be important to identify and monitor veterans who experienced new-onset suicidal ideation and planning during the pandemic to ensure that adequate safety is maintained (eg, development of a crisis support plan, means safety counseling) and resources are appropriately allocated. Suicide-specific interventions, such as cognitive-behavioral therapy for suicide prevention,³⁶ crisis response planning,^37,38 and dialectical behavioral therapy³⁹ that help individuals to problem-solve, manage distress, reduce self-harm risk (eg, means restriction), and promote social connectedness, could help mitigate suicide risk in these veterans.

Results of the current study revealed that both prepandemic risk and protective factors, as well as COVID-19–related stressors, predicted new-onset suicidal ideation. For example, veterans whose relationships significantly worsened because of COVID-19 were more likely to develop suicidal ideation relative to those who reported lower levels of such difficulties. This finding is consistent with several conceptual models of suicide, which suggest that increased feelings of disconnection (eg, thwarted belongingness) may heighten risk of suicidal ideation.^40-43 Changes in relationship quality due to the COVID-19 pandemic could have also eroded social support, which may have helped buffer against suicidal ideation before the pandemic. This interpretation aligns with a larger body of research showing that loneliness is strongly associated with STBs.⁴⁴ Indeed, prepandemic loneliness and increases in loneliness from the prepandemic to the peripandemic survey significantly increased the likelihood that veterans in our sample developed suicidal ideation. Policies of social distancing and quarantine may have exacerbated relationship difficulties and perceptions of loneliness, which could have led to greater hopelessness and thoughts of suicide.⁴⁴ Interventions to promote social connection^45,46 may therefore help attenuate suicide risk in veterans.

Consistent with prior work,¹⁸ lifetime history of a substance use disorder (SUD) was associated with 2- and 3-fold greater odds of new-onset suicidal ideation and new-onset suicide planning, respectively. In our sample, veterans with a lifetime SUD had significantly higher levels of impulsivity, which prior research suggests could have contributed to greater risk for the emergence of suicidal thinking.^47,48 Individuals with a history of SUD may have also been less likely to engage in adaptive coping strategies during the pandemic,⁴¹ ultimately increasing the risk of developing suicidal ideation and planning. The association between lifetime SUD and new-onset suicidal ideation and suicide planning could also reflect a more general process of stress sensitization.⁴² This may explain why prepandemic psychiatric distress was also associated with incident suicide planning, as these veterans may have already been struggling to cope with various life stressors, and the onset of the pandemic may have exacerbated these symptoms and in turn increased the risk of suicide planning.

Results of the current study also illuminate factors that may help protect against the development of suicidal ideation and suicide planning. Higher prepandemic purpose in life was linked to a reduced odds of new-onset suicidal ideation and planning during the pandemic. Purpose in life refers to the extent to which individuals’ lives are directed by personally valued goals and life aims.⁴⁹ Individuals with higher purpose in life have been shown to be more effective at regulating stress,⁵⁰ more likely to engage in adaptive coping,⁵¹ and more likely to seek out new goals when existing ones become unobtainable.⁵² They have also been shown to consistently engage in healthy behaviors, such as physical exercise,⁵³ and less likely to misuse substances.⁵⁴ Thus, it is conceivable these individuals may have been better equipped to navigate the stressors and uncertainties that accompanied the COVID-19 pandemic, which ultimately decreased the likelihood of developing suicidal ideation. Conversely, individuals who entered the pandemic with low purpose in life, individuals who tended to view their lives as less meaningful and lacking direction,⁴⁹ may have been more vulnerable to thoughts of suicide. Collectively, these findings underscore the potential importance of interventions to bolster purpose in life⁵⁵ as part of suicide prevention efforts. Indeed, there is evidence that interventions designed to promote purpose in life and engagement with personally valued activities can help attenuate suicide risk.^56,57

Although our sample was representative of the US veteran population, it primarily comprised older, male, White, non-Hispanic veterans. Future research in more diverse samples (eg, veterans from racial and ethnic minority groups) is needed to elucidate the risk and protective factors of STBs in these subgroups. Screening instruments were used to assess study measures; whether results would differ if clinician-administered interviews were used is unclear. Assessments of suicidal ideation and planning were obtained retrospectively over long periods (eg, the past year); consequently, assessments may reflect current or more chronic STBs.

Results of this study demonstrate that, in spite of predictions to the contrary,^4,6,58 the prevalence of STBs did not markedly increase for most US military veterans during the first 3 years of the COVID-19 pandemic. Nevertheless, a significant minority of veterans did develop new-onset suicidal ideation and planning during the pandemic, which was largely driven by preexisting psychiatric problems and lower purpose in life and social connectedness. Collectively, these results suggest that efforts to promote purpose in life and social connectedness, as well as treatments that target ongoing psychiatric distress and substance use problems, should be prioritized to help mitigate suicide risk in US military veterans. Further research is needed to replicate these results in other population-based samples, identify within-person factors that contribute to fluctuations in suicide risk, and evaluate the effectiveness of interventions targeting key risk and protective factors that mitigate suicide risk as veterans and the world at large begin to emerge from the COVID-19 pandemic.

Accepted for Publication: January 26, 2023.

Published Online: April 5, 2023. doi:10.1001/jamapsychiatry.2023.0393

Corresponding Author: Ian C. Fischer, PhD, US Department of Veterans Affairs National Center for PTSD, VA Connecticut Healthcare System, 950 Campbell Ave, West Haven, CT 06516 (ian.fischer@yale.edu).

Author Contributions: Drs Fischer and Pietrzak had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Fischer, Nichter, Norman, Pietrzak.

Acquisition, analysis, or interpretation of data: Fischer, Na, Krystal, Pietrzak.

Drafting of the manuscript: Fischer, Nichter, Pietrzak.

Critical revision of the manuscript for important intellectual content: Na, Norman, Krystal, Pietrzak.

Statistical analysis: Fischer, Nichter, Pietrzak.

Administrative, technical, or material support: Nichter, Krystal, Pietrzak.

Supervision: Pietrzak.

Conflict of Interest Disclosures: Dr Na reported royalties from Wolters Kluwer. Dr Norman reported royalties from Elsevier Press; payment from UptoDate and Psychiatry Times; and grants funded by the Department of Veterans Affairs, Department of Defense, Patient-Centered Outcomes Research Institute, and National Institutes of Health. Dr Krystal reported personal fees, income, and/or stock and/or options from Freedom Biosciences, Biohaven Pharmaceuticals, Neumora Pharmaceuticals, Spring Health, Temporo Biosciences, Terran Biosciences, EpiVario, Aptinyx, Biogen, Bionomics, Boehringer Ingelheim, Epiodyne, EpiVario, Janssen, Jazz Pharmaceuticals, Otsuka, Sunovion, BioXcel Therapeutics, Cerevel, Eisei, Neurocrine, Novartis, Psychogenics, AstraZeneca, Idec, Biomedisyn, Concert Pharmaceuticals, Heptares Therapeutics, L.E.K. Consulting, Perception Neuroscience Holdings, Sage Pharmaceuticals, BlackThorn Therapeutics, Spring Care, Sunovion Pharmaceuticals, Takeda Industries, and Taisho Pharmaceutical outside the submitted work; nonfinancial support from AstraZeneca, Delix, and Novartis; serving as a scientific advisor to Biohaven Pharmaceuticals, BioXcel Therapeutics, Cadent Therapeutics, PsychoGenics, Stanley Center for Psychiatric Research at the Broad Institute of MIT and Harvard, and Lohocla Research Corporation; serving as Editor of Biological Psychiatry (Society of Biological Psychiatry); holding patents for ketamine (royalties paid from Janssen Pharmaceuticals), riluzole (royalties paid from Biohaven Pharmaceuticals), extending ketamine efficacy (licensed to Freedom Biosciences), and reducing ketamine abuse liability (licensed to Freedom Biosciences); and having US patent applications related to the treatment of mental disorders, depression, suicidal ideation, and addiction. No other disclosures were reported.

Data Sharing Statement: See Supplement 2.

Additional Contributions: We thank the veterans who participated in the National Health and Resilience in Veterans Study.