Latent Profiles of Psychosocial Risk and Associations with Suicidality in Military Veterans

Background and Purpose: Military veterans are at increased risk for suicide relative to their civilian counterparts, but half of veterans do not seek services through the Veterans Health Administration (VHA) and are under-represented in the veterans’ literature. Suicide risk is contingent on a constellation of protective factors, including a sense of purpose and social connection, and risk factors like depression, PTSD, and the related but distinct construct of moral injury. Military veterans are also a heterogenous population, and variable-centered analyses may elide important subgroup differences. This study applied a person centered, latent profile analytic (LPA) approach to investigate heterogeneity in patterns of risk and protective factor variables in a community sample of VHA enrolled and non- enrolled veterans (n=3,180). The study aimed to 1) model latent profiles across risk and protective factor variables; 2) test associations between latent profiles and a suicide risk outcome variable; 3) examine associations between demographic and VHA service use characteristics and latent profile membership.

Analyses: To address aim 1, an iterative LPA model fitting process assuming independence of continuous indicator variables tested additional profiles at each iteration across parsimony and separation indices. To address aim 2, latent profiles from the best fitting solution were specified as independent variables in a logistic regression model with suicide risk as outcome. To address aim 3, a multinomial logistic regression model tested associations between demographic characteristics, VHA service use, and latent class membership.

Results: LPA fit indices supported a 5-profile solution to the data, including a normative profile reproducing full sample averages, three high-risk classes, and one low-risk class. The highest risk class had the highest depression scores, lowest purpose and social support scores, and 6 times the suicide risk odds compared to the normative class (OR=5.91, 95% CI: 4.25, 8.22). Membership in this profile was associated with age (OR 0.98, 95% CI: 0.97, 0.99), high school level education (OR 1.60, 95% CI: 1.17, 2.20), and VHA utilization of both mental and physical health services (OR=2.77, 95% CI: 1.94, 3.97). The lowest risk class had the lowest odds of using VHA mental or physical health services (OR=0.21, 95% CI: 0.10, 0.45).

Conclusions and Implications: Results showed considerable nuance across latent profile and suicide risk outcomes. Risk and protective factors varied significantly between profiles, but profiles with higher depression scores had lower protective factor scores and greater odds of suicide risk compared to the normative profile and profiles with higher PTSD and moral injury scores. Consistent with previous literature, older, married, White, nondeployed veterans were more likely to be in the lowest risk profile. This profile was less likely to use VHA physical or mental health services compared to the normative or high-risk profiles, but it is unclear whether this dynamic reflects a healthier subgroup overall or a preference for alternative service providers. Future studies must investigate this understudied subgroup in much greater depth. Finally, the profile comprising younger veterans with low educational attainment and VHA mental and physical service use requires close monitoring and support to prevent suicide deaths.