1118 — Exploring Blast Exposure Effects Using Bayesian Network (BN) Modeling

Toyinbo PA, HSR&D Center of Innovation on Disability and Rehabilitation Research (CINDRR); Vanderploeg RD, HSR&D Center of Innovation on Disability and Rehabilitation Research (CINDRR); Belanger HG, HSR&D Center of Innovation on Disability and Rehabilitation Research (CINDRR); Spehar AM, HSR&D Center of Innovation on Disability and Rehabilitation Research (CINDRR); Lapcevic WA, HSR&D Center of Innovation on Disability and Rehabilitation Research (CINDRR); Scott SG, HSR&D Center of Innovation on Disability and Rehabilitation Research (CINDRR);

Objectives:
To investigate probabilistic dependencies among the predisposing factors, and persistent physical and mental health effects of blast exposure using an advanced analytic approach.

Methods:
An online survey of members (N = 3098) of the Florida National Guard (FNG) cohort (1443 deployed, 1655 not deployed) provided data on: pre-deployment (demographics, psychological trauma, probable mild TBI (m-TBI) diagnosis), deployment-related (exposure to potentially traumatic deployment experiences, combat exposure amount, blast exposure, m-TBI, and physical injuries), and post-deployment (PTSD, neurobehavioral, anxiety, depressive and physical symptoms) variables. Bayesian Network (BN), an efficient probabilistic graphical model, was used to learn the complex qualitative relationships between these variables, and to estimate their conditional probabilities given blast exposure.

Results:
The marginal (prior) probabilities of all variables served as reference values for inferences. In the absence of blast exposure and probable m-TBI diagnosis, members with less health symptoms were more likely to have been deployed or have had multiple deployments. Blast exposure was the single critical deployment-related factor directly linked with subsequent symptoms. The arousal subset of PTSD symptoms was the bridge between blast exposure and other PTSD and mental health symptoms, while tinnitus was the bridge symptom between blast exposure and other physical symptoms. The conditional probabilities of symptoms (on a dichotomized response scale) given blast exposure without probable m-TBI diagnosis increased relative to the reference values by: 18-25% (moderate to very severe PTSD); 13% (high anxiety); 11% (high depressive); 39% (tinnitus); 27% (dizziness), 4% (pain interference); and 2% (poor general health perception). When given joint presence of probable m-TBI diagnosis and blast exposure, the conditional probabilities were not different from when there was no m-TBI. The Neurobehavioral Symptom Inventory (NSI) factors (symptom groups) were minimally dependent on blast experience or probable m-TBI diagnosis in this sample.

Implications:
Findings suggest that blast exposure may be an important source of persistent PTSD and other mental health symptoms, regardless of comorbid m-TBI. A replication on a larger scale may shed more light on both short-term and long-term effects of deployment-related blast exposure.

Impacts:
Findings will inform future research on blast exposure.