Introduction We explore disparities in awarding post-traumatic stress disorder (PTSD) service-connected disability benefits (SCDB) to veterans based on gender, race/ethnicity, and misconduct separation. Methods Department of Defense data on service members who separated from October 1, 2001 to May 2017 were linked to Veterans Administration (VA) administrative data. Using adjusted logistic regression models, we determined the odds of receiving a PTSD SCDB conditional on a VA diagnosis of PTSD. Results A total of 1,558,449 (79% of separating service members) had at least one encounter in VA during the study period (12% female, 4.5% misconduct separations). Females (OR 0.72) and Blacks (OR 0.93) were less likely to receive a PTSD award and were nearly equally likely to receive a PTSD diagnosis (OR 0.97, 1.01). Other racial/ethnic minorities were more likely to receive an award and diagnosis, as were those with misconduct separations (award OR 1.3, diagnosis 2.17). Conclusions Despite being diagnosed with PTSD at similar rates to their referent categories, females and Black veterans are less likely to receive PTSD disability awards. Other racial/ethnic minorities and those with misconduct separations were more likely to receive PTSD diagnoses and awards. Further study is merited to explore variation in awarding SCDB.
Keywords: afghanistan; awards and prizes; iraq; military personnel; post-traumatic stress disorder; veterans; diagnosis; disability; health disparity
Recent conflicts in Afghanistan and Iraq (Operations Enduring Freedom, Iraqi Freedom, New Dawn, Freedom's Sentinel, and Inherent Resolve) have become known for both physical and mental injuries, with post-traumatic stress disorder (PTSD) being a major comorbidity for which veterans receive care in the U.S. Department of Veterans Affairs (VA).[[
An important aspect of caring for veterans is a national policy that provides them with disability benefit awards for loss of function/injury or mental health conditions that are a direct result of or are aggravated by their military service.[
Studies using small samples of veterans have shown that women[[
The study population comprised the entire cohort of active duty SM listed in the VA Defense Manpower Data Center file (VA/DoD Identity Repository [VADIR]) who were separated from the military from October 1, 2001 to May 2017. VA administrative data were accessed using the Veterans Informatics and Computing Infrastructure (VINCI),[
The primary outcome for our study was the award of SC disability benefit for PTSD. Disability ratings are in 10% increments from 0% to 100%, and a veteran's benefit level is tied to his or her rating which is determined after a medical and mental health evaluation in VA (compensation and pension examination, C&P).[
The key independent predictor variable used to test our hypothesis was type of separation. The DoD assigns each active duty SM an interservices separation code that indicates the circumstances related to discharge from service. Routine separations are those that occurred after completing a certain number of years on contract, whereas nonroutine separations could be due to misconduct, disability, disqualification, or early separations.[
To account for the possibility of a relationship of a clinical diagnosis of PTSD on the receipt of a PTSD SCDB, using logistic regression models adjusted for demographic variables listed above, we first determined the odds ratios (OR) for receiving any PTSD SC disability award and VA diagnosis of PTSD separately (marginal estimates). Then, we determined the odds of receiving a PTSD SC disability award conditional on either having a VA diagnosis of PTSD or not, using logistic regression models adjusted for covariates listed above and key predictor variables (type of separation). The coefficient for the OR was calculated for a doubling of time in service for years spent in service for both active duty and National Guard/Reservists to enable easier interpretation of the effect of the time in service. Significance was set at the P = 0.05 level or lower. Analyses were performed using R Version 3.5.1.[
During a nearly 16-year study period, a total of 1,969,724 SM experienced 3,472,031 separations from the military (from October 2001 to May 2017). A third of the separations were routine (33%), 17% were nonroutine (misconduct, disability, disqualified, or early), and the remainder (50%) were either null, unknown, or applicable (likely representing National Guard/Reserve SM, as only active duty SM receive an inter-services separation code). Of the total SM, 1,558,449 (79%) had at least one encounter in VA administrative data during the study period (Table I). Females comprised 12% of the total number of SM seen in the VA. The racial/ethnic distribution of the study cohort generally followed the distribution of veterans seen in the VA with non-Hispanic Whites forming the majority.
TABLE I Demographic Characteristics of Study Cohort of 1,558,449 Active Duty Service Members From Recent Conflicts in Iraq and Afghanistan
Variable Demographic Characteristic All Service-Connected Disability Award for PTSD Yes, No, Has VA PTSD diagnosis Yes 398,987 (25.60) 267,993 (67.17) 130,994 (32.83) Sex Female 187,976 (12.06) 36,993 (9.71) 150,983 (12.82) Male 1,370,458 (87.94) 344,054 (90.29) 1,026,404 (87.18) Year of birth <1970 356,072 (22.85) 68,932 (18.09) 287,140 (24.39) 1970–1979 400,904 (25.72) 101,191 (26.56) 299,713 (25.46) 1980–1989 743,770 (47.73) 199,376 (52.32) 544,394 (46.24) 1990+ 57,703 (3.70) 11,552 (3.03) 46,151 (3.92) Race/ethnicity American Indian/Alaskan Native 21,179 (1.36) 5,225 (1.37) 15,954 (1.36) Asian/Native Hawaiian or other Pacific Islander 102,171 (6.56) 31,173 (8.18) 70,998 (6.03) Black or African American 245,345 (15.74) 69,261 (18.18) 176,084 (14.96) Caucasian/White 1,011,768 (64.92) 227,669 (59.75) 784,099 (66.60) Hispanic 146,265 (9.39) 39,580 (10.39) 106,685 (9.06) Other 16,626 (1.07) 3,848 (1.01) 12,778 (1.09) Unknown 15,095 (0.97) 4,295 (1.13) 10,800 (0.92) Education 00-no HS diploma 29,377 (1.89) 6,703 (1.76) 22,674 (1.93) 10-high school equivalent 108,492 (6.96) 38,837 (10.19) 69,655 (5.92) 20-high school 939,044 (60.26) 249,000 (65.35) 690,044 (58.61) 30-some college 115,839 (7.43) 28,026 (7.35) 87,813 (7.46) 40-associates degree 107,410 (6.89) 20,065 (5.27) 87,345 (7.42) 50-baccalaureate degree 170,311 (10.93) 27,518 (7.22) 142,793 (12.13) 60-graduate/professional degree 85,632 (5.49) 10,675 (2.80) 74,957 (6.37) 99-unknown 2,344 (0.15) 227 (0.06) 2,117 (0.18) Combat flag No 67,207 (4.31) 13,181 (3.46) 54,026 (4.59) Yes 1,491,220 (95.69) 367,867 (96.54) 1,123,353 (95.41) Marital status Annulled 329 (0.02) 102 (0.03) 227 (0.02) Divorced 45,862 (2.94) 11,012 (2.89) 34,850 (2.96) Legally separated 843 (0.05) 250 (0.07) 593 (0.05) Married 471,157 (30.23) 109,516 (28.74) 361,641 (30.72) Never married 1,014,728 (65.11) 254,222 (66.72) 760,506 (64.59) Unknown 24,984 (1.60) 5,824 (1.53) 19,160 (1.63) Widowed 546 (0.04) 125 (0.03) 421 (0.04) Separation category Disability 121,314 (7.78) 58,023 (15.23) 63,291 (5.38) Disqualified 65,214 (4.18) 16,293 (4.28) 48,921 (4.16) Early 125,931 (8.08) 27,805 (7.30) 98,126 (8.33) Misconduct 70,496 (4.52) 21,776 (5.71) 48,720 (4.14) Normal 1,175,494 (75.43) 257,154 (67.49) 918,340 (78.00)
1 aExcludes 15 service members with unknown sex.
2 bExcludes 22 service members with missing combat flag status.
All examinations of group differences between those who received (Yes) and those who did not receive (No) service-connected disability award for PTSD were statistically significant by chi-square analyses.
Overall, 381,051 (24.5%) of the cohort had evidence of PTSD SCDB in VA administrative data. Similarly, one-fourth of the SM seen in VA during the study period had a diagnosis of PTSD in the VA (
After adjusting for covariates, female veterans were less likely to receive a PTSD award (OR 0.72) and were nearly equally likely to receive a PTSD diagnosis in the VA (OR 0.97). The disparity of less likelihood of receiving an award persisted on examining the odds of receiving an award conditional on diagnosis for females (Table II).
TABLE II Results of Models for Service Members Receiving PTSD Disability Award, Any PTSD Diagnosis Code, and Award Conditional on Diagnosis
Variable Level Marginal Models Award Conditional on Award Diagnosis Has Diagnosis No Diagnosis OR Sig. OR Sig. OR Sig. OR Sig. Education 10-high school equivalent 1.11 *** 1.25 *** 0.95 *** 1.11 *** » 20-high school (referent) — — — — — — — — 30-some college 0.81 *** 0.76 *** 0.96 * 0.85 *** 40-associates degree 0.72 *** 0.70 *** 0.94 ** 0.73 *** 50-baccalaureate degree 0.59 *** 0.50 *** 0.97 * 0.64 *** 60-graduate/professional degree 0.41 *** 0.33 *** 1.01 0.47 *** Time in service (log2) Air force/active 0.83 *** 0.84 *** 0.93 *** 0.87 *** Air force/guard 0.95 *** 0.97 *** 0.97 *** 0.94 *** Air force/reserve 1.05 *** 1.05 *** 1.01 * 1.04 *** Army/active 1.01 *** 0.98 *** 1.02 *** 1.04 *** Army/guard 1.05 *** 1.08 *** 1.01 *** 1.01 *** Army/reserve 1.05 *** 1.09 *** 1.01 *** 1.01 ** Coast guard/all 0.86 *** 0.88 *** 0.95 ** 0.87 *** Marine corps/active 1.00 0.98 *** 1.02 *** 1.03 *** Marine corps/reserve 1.12 *** 1.12 *** 1.05 *** 1.08 *** Navy/active 0.78 *** 0.83 *** 0.87 *** 0.80 *** Navy/reserve 1.08 *** 1.09 *** 1.03 *** 1.04 *** Racial/ethnic American Indian/Alaskan Native 1.19 *** 1.25 *** 1.01 1.15 ** Asian/Native Hawaiian or other Pacific Islander 1.18 *** 1.51 *** 1.08 *** 0.86 *** Black or African American 0.93 *** 1.01 * 0.82 *** 0.99 » Caucasian/White (referent) — — — — — — — — Hispanic 1.14 *** 1.60 *** 1.03 0.81 *** Separation category Disability 0.80 *** 1.50 *** 0.65 *** 0.56 *** Disqualification 0.94 *** 1.51 *** 0.69 *** 0.86 *** Early 1.06 *** 1.01 0.97 1.11 *** Misconduct 1.31 *** 2.17 *** 0.73 *** 1.26 *** » Normal (referent) — — — — — — — — Sex Female 0.72 *** 0.97 *** 0.62 *** 0.76 *** » Male (referent) — — — — — — — —
The first two columns indicate the estimated odds ratios (OR) for awarding of a service-connected disability (SCD) award (regardless of having a PTSD diagnosis) and the OR for the presence of any ICD diagnosis code that indicates PTSD (regardless of if a SCD was awarded), respectively. The remaining two columns indicate the OR for awarding of a SCD award conditional on having received a PTSD diagnosis in the VHA, or not, respectively. The OR listed for time in service is for a doubling of time in the indicated branch/component. Asterisks indicate the significance level and are as follows:
Racial/ethnic minorities other than Blacks (Hispanic, Asian/Native Hawaiian, and American Indian/Alaskan Natives) were more likely to receive an award (OR 1.14, 1.19, and 1.18, respectively) and diagnosis (OR 1.60, 1.51, and 1.25, respectively) as compared to non-Hispanic Whites. The increased odds of receiving an award persisted only for American Indian/Alaskan Natives conditioned on diagnosis (OR 1.15 to receive an award with no diagnosis) and to a certain extent for Asian/Native Hawaiians (OR 1.08 to receive an award with diagnosis). For Blacks, the lower likelihood of receiving an award persisted after conditioning on receiving a PTSD (OR 0.82) diagnosis.
Those with levels of education of high school equivalent were more likely than those with higher education to receive an award (OR 1.11) and diagnosis (OR 1.25); this trend persisted for an award after conditioning for no diagnosis (OR 1.11).
Though the "time in service" variable demonstrated consistency in direction and effect size across the SC disability award and the VA diagnosis models, there was variation noted among branches of service and active duty/reserve components. The Air Force and Navy behaved similarly showing reduced odds of receiving either an award (OR 0.83 and 0.78, respectively) or diagnosis (OR 0.84 and 0.83, respectively) the longer a SM served in active duty. This was in contrast to those in Air Force and Navy Reserves who had an increased odds for receiving an award with increased time in the Reserves (OR 1.05 and 1.08, respectively). The Army and Marine Corps showed either a small increase or no effect in likelihood for award (OR 1.01 and 1.00, respectively), or a slight decrease for diagnosis (OR 0.98 for both), with increased time in active duty, while also showing the same increased odds for reserve forces (OR > 1).
Those with misconduct separations had the highest odds of receiving SC disability award (OR = 1.31) and VA diagnosis (OR = 2.17). Those with disability (OR = 0.80) and disqualification (OR = 0.94) separations were less likely to have PTSD SC disability awards compared to those with routine separations while still at significantly higher odds of having PTSD diagnoses in the VA. Those with early separation were at slightly elevated odds for receiving an award (OR = 1.06) but there was no observable effect after conditioning on diagnosis.
Acknowledging disparities in the award of service-connected disability for PTSD is an important first step in understanding the complex process involved in such awards with the goal of mitigating them. To the best of our knowledge, our study is the first to demonstrate disparities in awards for PTSD based on a large sample of SM (1.5 million) who are now seeking care in VA as veterans. Studies have demonstrated that veterans file claims for tangible need as well as a belief that PTSD disability award justifies and legitimizes their condition.[
With regard to disparities based on demographic characteristics, we confirm the gender-based disparity reported by smaller studies[
We confirmed the disparity for Blacks noted by Murdoch et al.,[
The seemingly "protective" effect of higher education on a diagnosis of PTSD and lower odds of receiving a PTSD award is an important finding in light of a third of the study cohort having higher than high school education. This may have implications for recruiting, military occupation assignments, military-sponsored training/education, and re-enlistments at a higher grade.
The modest increase in odds of receiving awards for National Guard/Reservists as compared to active duty SM based on the time in service is also an important finding that may have implications for policy and planning, as well as recruitment/retention in the military. The apparent relationship may be explained by variation in combat theater roles and duration of deployment among SM from different branches and component (roles more proximal to combat vs. remote support).
The second set of results with regard to separation category supported our hypothesis that those with misconduct separation were at the highest risk for a diagnosis of PTSD and appropriately were receiving awards. The seemingly discordant result of these SM receiving awards at a higher rate in the absence of a diagnosis (as compared to the presence of a diagnosis) raises important questions regarding whether we are able to retain these SM in care in the VA. Regardless, these results merit mention and further study in assuring that those who are disabled by PTSD are appropriately cared for in the VA. Those with disability and disqualification separations were more likely to have a diagnosis of PTSD and less likely to receive a PTSD SCDB. One possible explanation is that these veterans may have been receiving benefits for other service-connected diagnoses. Those with "early" separations pose special challenges; these SM were separated for reasons generally related to family and social stress such as sole-surviving family member, marriage, parenthood/pregnancy, or lack of dependent support.[
We acknowledge several limitations. Our study cohort is biased in favor of those who seek care in the VA, and such visits may be associated with either an increased or decreased probability of applying for SCDB that is difficult to measure. Thus, our results on disparities should be interpreted in the context of those SM who seek care in the VA, have applied for disability benefits, and have been successfully awarded SCDB for PTSD. Because of the nature of the administrative data used in our study, we are not able to determine who filed claims for disability for PTSD and those who were denied. Studies have demonstrated that those denied disability benefits are more vulnerable, likely lack social support, and generally have poorer health.[
We used ICD codes to detect a diagnosis of PTSD; as with any administrative data, there may be variation in coding for diagnoses at the provider and facility level. However, for a diagnosis such as PTSD in the VA, administrative data are likely reliable for estimation of prevalence of the condition. Our study dataset that combined DoD (VADIR) and VA administrative data excludes those that do not seek care in the VA. The proportion of veterans who do not seek care in the VA is variable and not capturing those veterans places a limitation on our ability to determine who had a diagnosis of PTSD. It has been shown that not all veterans with PTSD seek and continue care in the VA for reasons ranging from access to care to treatment beliefs;[
Although the VA has recognized disparities in the provision of health care to minorities,[
Funding for this project was provided by grant F3Z4537031GW01 (Travis Air Force Base, PIs: IJS and AVG), and VA Center of Innovation Award #I50HX001240 from the Health Services Research and Development of the Office of Research and Development of the US Department of Veterans Affairs.
Presented as an oral presentation at the 2018 Military Health System Research Symposium, August 2018, Kissimmee, FL; abstract #MHSRS-18-2069
The views expressed in this article are those of the authors and do not necessarily represent the position or policy of the US Department of Veterans Affairs, US Army, US Air Force, Department of Defense, or the US government
By Andrew M Redd; Adi V Gundlapalli; Ying Suo; Warren B P Pettey; Emily Brignone; David L Chin; Lauren E Walker; Eduard A Poltavskiy; Jud C Janak; Jeffrey T Howard; Jonathan A Sosnov and Ian J Stewart
Reported by Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author