Propensity Score Analysis Examining Behavioral and Physical Health Conditions Associated with Army Warrior Transition Unit Referrals

BACKGROUND/PURPOSE: Warrior Transition Units (WTUs) are military units co-located with military treatment facilities providing six months or more of comprehensive health care to service members with physical and/or psychological conditions limiting performance of military occupations. This study identified behavioral and physical health conditions associated with the probability of WTU referrals, FY2008-FY2015.

METHODS: This observational study analyzed Military Health System data from Army service members (N=883,091) returning from Operations Enduring Freedom and Iraqi Freedom deployments, FY2008-FY2015. The WTU cohort (n=31,094) included service members who were referred to WTUs during post-deployment and the non-WTU cohort included service members who were not referred to WTUs (n=851,997), but received MHS treatment as usual. Baseline was FY2008; index deployment was identified by the first deployment end date, FY2008-FY2015. Baseline demographic (age, sex, race, marital status) and military (service component, rank, TRICARE eligibility) characteristics were identified in the Defense Eligibility Enrollment Record System; deployment characteristics (deployment frequency, deployment length) in the Contingency Tracking System. Primary behavioral health diagnoses were defined by Major Diagnostic Codes (MDC) 19 (mental disorders/diseases) and 20 (alcohol/other drugs). Physical injuries [amputations, musculoskeletal, pain, shrapnel, spinal cord, traumatic brain injury (TBI)] and posttraumatic stress disorder (PTSD) were identified in the Ill, Injured, and Wounded cohort file. SAS %GMATCH macro estimated propensity scores and developed a matched WTU (n=28,148) and non-WTU (n=28,148) analytic sample (N=56,296). Descriptive statistics described baseline characteristics. Chi-square and t-tests estimated standardized differences of original and matched samples to assess confounding bias. A standardized difference <0.20 indicated reduced confounding bias. Propensity score model of the matched sample estimated the probability of a WTU referral using SAS PROC LOGISTIC.

RESULTS: In the original sample, WTU and non-WTU cohorts differed on demographic, military, physical and behavioral health characteristics. In original and matched samples, mean age was 28.1 (sd=8.0) years and the majority were Active Duty, male, White, married, enlisted, and deployed on average <12 months during the index deployment. The matched sample had a large proportion (41.5%) of service members ages 26-30. Original sample standardized differences ranged from 0.015-1.291 and 0.001-0.067 for matched sample, indicating reduced confounding bias in the matched sample. National Guard (AOR 1.16, 95% CI: 1.11-1.22, p<.0001) and Reservists (AOR 1.11, 95% CI: 1.05-1.18, p<.0005) had increased odds of a WTU referral compared to Active Duty. Post-deployment primary diagnoses of anxiety disorders (excluding PTSD; AOR 1.10, 95% CI: 1.02-1.12, p<.005), amputation (AOR 1.74, 95% CI: 1.30-2.33, p<.0005), TBI (AOR 1.15, 95% CI: 1.06-1.25, p<.005), and pain (AOR 1.20, 95% CI: 1.13-1.28, p<.0001) was also associated with increased odds of a WTU referral.

Conclusions/Implications: From FY2008-FY2015, anxiety disorders, amputations, TBI, pain, and being National Guard or Reservist increased the probability of deployed Army service members being referred to a WTU during post-deployment. Assessing and monitoring functional impairment associated with anxiety and pain management in Active, National Guard, and Reserve Soldiers facilitates patient-centered care coordination, self-management, and potentially return to duty. Future research should examine differences in factors associated with WTU referrals in Active and Reserve component Soldiers and treatment effects of WTUs.