Posttraumatic Stress Disorder and Justice Involvement Among Military Veterans: A Systematic Review and Meta-Analysis

Abstract

There is evidence of an association between posttraumatic stress disorder (PTSD) and criminal justice involvement among military veterans. For this study, we systematically reviewed the literature to examine the association between PTSD and criminal justice involvement among military veterans, assess the magnitude of this association, and identify strengths and limitations of the underlying evidence. Five databases were searched for a larger scoping review, and observational studies that assessed PTSD and criminal justice involvement were selected from the scoping review database (N = 191). Meta-analyses were conducted, pooling odds ratios (ORs) via restricted maximum likelihood random-effects models. The main outcomes were criminal justice involvement (i.e., documentation of arrest) and PTSD (i.e., PTSD assessment score indicating probable PTSD). Of 143 unique articles identified, 10 studies were eligible for the meta-analysis. Veterans with PTSD had higher odds of criminal justice involvement (OR = 1.61, 95% CI [1.16, 2.23], p = .002) and arrest for violent offenses (OR = 1.59, 95% CI [1.15, 2.19], p = .002) compared to veterans without PTSD. The odds ratio of criminal justice involvement among military veterans with PTSD assessed using the PTSD Checklist was 1.98, 95% CI [1.08, 3.63], p = .014. Considerable heterogeneity was identified, but no evidence of publication bias was found. Criminal justice involvement and PTSD are linked among military veterans, highlighting an important need for clinicians and healthcare systems working with this population to prioritize PTSD treatment to reduce veterans' new and recurring risk of criminal justice involvement.

Figure 1.

PRISMA flow diagram of study selection (Moher et al., 2009).

Figure 2.

Forest plot of odds ratios (ORs) for studies examining posttraumatic stress disorder (PTSD) and any criminal justice involvement (JI) among military veterans. A considerable amount of heterogeneity was observed, Q = 53.77, p = .000, I² = 88.2%, with no indication of publication bias (i.e., no funnel plot asymmetry, p > .10). RE = random effects.