PTSD and Military Health Research

Randomized trials, biomarker analysis, and mediation methods for PTSD, blast injury, and veteran health

2026-06-07 18:16 PDT

Overview

Post-traumatic stress disorder (PTSD) and blast-related injury represent major unmet clinical needs in veteran populations, with substantial overlap between PTSD symptomatology and the neurological sequelae of blast exposure. This program supports statistical analysis of randomized trials testing prazosin – an alpha-1 adrenergic antagonist with evidence for PTSD nightmares and post-traumatic headache – as well as causal mediation analysis of PTSD comorbidities and biomarker trajectory analysis in blast-exposed cohorts.

The program operates in close collaboration with investigators at the VA Puget Sound Health Care System and draws on data from the VA Cooperative Studies Program (CSP-563) and investigator-initiated trials.

Current work

  • Prazosin for PTSD nightmares: CSP-563 reanalysis. Full reanalysis of the VA Cooperative Studies Program prazosin RCT (Raskind et al., NEJM 2018) in NEJM manuscript format, using contemporary mixed-model methods applied to the original SAS-format trial data (39 CRF forms). Analysis includes Table 1 demographics, Table 2 primary outcomes, and subgroup analyses.

  • Prazosin for post-traumatic headache: Murray/Yeilim RCT. Primary analysis of an investigator-initiated prazosin RCT for post-traumatic headache in veterans, using longitudinal mixed-effects models with emmeans-based marginal contrasts and mediation analysis of secondary outcomes. Manuscript in preparation.

  • GFAP trajectories in blast-exposed veterans. Comprehensive analysis of plasma GFAP (glial fibrillary acidic protein) biomarker trajectories in a prazosin blast-exposure RCT, comparing parametric LMM, robust regression, and non-parametric permutation tests to assess treatment effects under potential distributional violations.

  • PTSD, depression, and glycemic control: mediation analysis. Causal mediation analysis estimating the pathway from PTSD and depression to HbA1c in veterans, using LASSO variable selection and multiple imputation to handle high-dimensional covariate sparsity. Manuscript near completion.

Methods

Linear mixed-effects models (lme4, lmerTest, nlme); robust regression (robustbase, MASS::rlm); permutation tests (coin); emmeans marginal contrasts; causal mediation decomposition (causalweight, mediation); LASSO variable selection via glmnet; multiple imputation via mice.

Publications

The full publications list filtered by military-health, blast-injury, or va-cooperative-studies provides the relevant record, which includes papers from several completed VA Cooperative Studies Program trials and investigator-initiated studies in veteran populations.