Global Disability Trajectories Over the First Decade Following Combat Concussion : The Journal of Head Trauma Rehabilitation

Secondary Logo

Journal Logo

Original Articles

Global Disability Trajectories Over the First Decade Following Combat Concussion

Mac Donald, Christine L. PhD; Barber, Jason MS; Johnson, Ann; Patterson, Jana; Temkin, Nancy PhD

Author Information
Journal of Head Trauma Rehabilitation 37(2):p 63-70, March/April 2022. | DOI: 10.1097/HTR.0000000000000738
  • Free


ANNUAL COSTS for US COMBAT-RELATED traumatic brain injury (TBI) have been previously estimated to be between $591 million and $910 million1; however, this is now thought to be grossly underestimated. Additionally, it has been reported that peak disability payout for veterans of world conflicts is incurred decades after the conflict is over.2 World War I (1917-1918) disability cost reportedly peaked in 1969, World War II (1941-1945) disability cost reportedly peaked in 1980, and Vietnam (1959-1975) disability cost was still on the rise in 2011 when reported.3 With the conflicts in the Middle East (2001 to present) as defined by US policy already exceeding cost projections, the true impact is likely not to be felt for decades.4 Recent efforts have demonstrated that annual healthcare costs for veterans with mild TBI, the majority of TBIs in combat, were 2 to 3 times higher than those without mild TBI, with greatest cost utilization in the behavioral health domain.5 This has, and remains, a major public health burden, as this population ages, motivating efforts to understand these global disability trajectories in our service men and women.

To our knowledge, global disability trajectories over the first decade following TBI have primarily been studied in more moderate to severe civilian cases using the Glasgow Outcome Scale Extended6,7 (GOSE).8,9 In the study by Dr Dams-O'Connor and colleagues, GOSE trajectories were explored in the US-based TBI-Model Systems study to understand the difference between TBI patients who survived and those who died within the first decade post-injury.8 In a Finnish population-based cohort, Dr. Forslund and colleagues reported on GOSE trajectories over the first decade post-TBI finding that key demographic and injury metrics such as duration of posttraumatic amnesia were predictive of decline in later years.9

Additionally, there have been a small number of studies that have reported incremental GOSE disability over this period in largely moderate to severe civilian brain injury. An example is the study by Dr. Ponsford and colleagues, where GOSE disability scores were reported 2, 5, and 10 years post-injury across the TBI severity spectrum.10 While this sheds light on the longer-term impact of these moderate to severe civilian brain injuries, questions remain regarding similar trajectories in milder forms of brain injury, particularly in the service member and veteran populations.

Through the “Evaluation Of Longitudinal outcomes in mild TBI Active-Duty Military and Veterans” (EVOLVE) study, we have been provided the unique opportunity to examine Global War on Terrorism (GWoT) service members with combat-related concussion, at the point of injury in combat, or after medical evacuation to Landstuhl Regional Medical Center in Germany, and follow them out to 1-year,11,12 5-year,13,14 and now 10-year outcome. In parallel, we have followed non-brain-injured combat-deployed services members for comparison. Through this, the GOSE has been collected every 6 months on these patients and participants. The objective of the current study was to use latent class growth analysis to determine global disability outcome trajectories and characterize the profile of the patients in those trajectories. The hope was to understand who is most at risk of a poor long-term outcome to help focus earlier targeted intervention with the ultimate goal of reducing the extremely high public health cost documented following prior conflicts.



This study was approved by the University of Washington Institutional Review Board with additional approval from the US Army Medical Research and Materiel Command Institutional Review Board and carried out in accordance with the approved protocol. Consent and subsequent reconsent for each follow-up evaluation were provided by all participants according to the Declaration of Helsinki; no surrogate consent was allowed.

Design and procedure

Participants were originally enrolled into 1 of 4 previous cohorts from 2008 to 201311–13,15,16 (see Table 1 for demographics). This is a prospective, observational, longitudinal study that has followed these very same patients for 10 years. There are 4 main groups (n = 475), 2 primary and 2 exploratory: (1) combat-deployed controls without a history of blast exposure “non-blast-control” (n = 143), (2) concussive blast TBI “blast-TBI” (n = 236) (primary), (3) combat-deployed controls with a history of blast exposure “blast-control” (n = 54), and (4) patients sustaining a combat concussion not arising from blast “non-blast-TBI” (n = 42) (exploratory). Inclusion criteria have been reported elsewhere.11,12,16 Briefly, inclusion criteria were defined as service members, deployed to the combat theater, in which original enrollment was completed either directly in Afghanistan11 or following medical evacuation to Landstuhl Regional Medical Center in Germany.12,16 For the TBI groups, TBI diagnosis was determined by trained medical personnel working in the TBI clinics in Afghanistan or Germany using the same protocol. First the Military Acute Concussion Evaluation (MACE) was administered by clinic staff followed by examination for diagnosis corroboration by a TBI neurologist. For the concussive blast TBI group, all available clinical histories indicated blast exposure plus another mechanism of head injury such as a fall, motor vehicle crash, or being struck by a blunt object. None suffered an isolated blast injury. All concussive-blast and non-blast TBI patients met the Department of Defense definition for mild, uncomplicated TBI17 defined as a Glasgow Coma Scale (GCS) 13 to 15, loss of consciousness (LOC) 0 to 30 minutes, alteration of consciousness (AOC) less than 24 hours, posttraumatic amnesia (PTA) less than 24 hours, and unremarkable CT or MRI at the time of evaluation. For the control groups, all combat-deployed controls were clinically evaluated to be free of signs and symptoms of head injury for both the “non-blast” and “blast” control groups and additionally no history of blast exposure for the “non-blast-control” group. Prior psychiatric and TBI diagnoses were exclusions for all groups and were ascertained both by clinician evaluation as noted earlier, patient-reported history, as well as medical records review at the time of screening.

TABLE 1 - Patient demographics by GOSE latent class growth trajectorya
Main GOSE disability level of trajectory Overall Trajectory 1 n (%)
Good recovery
Trajectory 2 n (%)
Upper moderate
Trajectory 3 n (%)
Lower moderate
Trajectory 4 n (%) Death P value
Total number of patients 475 113 251 104 7
Patient group
Non-blast-controls 143 80 (71%) 55 (22%) 8 (8%) 0 (0%) <.001
Blast-controls 54 14 (12%) 32 (13%) 7 (7%) 1 (14%)
Blast-TBI 236 15 (13%) 141 (56%) 74 (71%) 6 (86%)
Non-blast-TBI 42 4 (4%) 23 (9%) 15 (14%) 0 (0%)
Mean (SD) 29.5 (7.9) 32.3 (8.5) 28.1 (7.1) 29.7 (8.1) 29.6 (9.5) <.001
Male 439 (92%) 100 (88%) 233 (93%) 99 (95%) 7 (100%) .29
Female 36 (8%) 13 (12%) 18 (7%) 5 (5%) 0 (0%)
Education, y
Mean (SD) 13.7 (2.3) 15.2 (3.1) 13.4 (1.8) 13.1 (1.7) 12.3 (1.0) <.001
Military rank
Enlisted 437 (92%) 91 (81%) 236 (94%) 103 (99%) 7 (100%) <.001
Officer 38 (8%) 22 (19%) 15 (6%) 1 (1%) 0 (0%)
Caucasian 347 (73%) 86 (76%) 186 (74%) 70 (67%) 5 (71%) .36
African American 64 (13%) 15 (13%) 37 (15%) 12 (12%) 0 (0%)
Hispanic/Latinx 53 (11%) 9 (8%) 23 (9%) 19 (18%) 2 (29%)
Asian/Pacific Islander 7 (2%) 2 (2%) 3 (1%) 2 (2%) 0 (0%)
Other 4 (1%) 1 (1%) 2 (1%) 1 (1%) 0 (0%)
Branch of service
Army 368 (77%) 76 (68%) 191 (76%) 94 (90%) 7 (100%) <.001
Marines 50 (11%) 7 (6%) 35 (14%) 8 (8%) 0 (0%)
Navy 30 (6%) 14 (12%) 15 (6%) 1 (1%) 0 (0%)
Air Force 27 (6%) 16 (14%) 10 (4%) 1 (1%) 0 (0%)
Subsequent head injury exposure
0 222 (47%) 70 (62%) 104 (41%) 48 (46%) 0 (0%) <.001
1 73 (15%) 10 (9%) 39 (16%) 24 (23%) 0 (0%)
≥2 47 (10%) 4 (4%) 20 (8%) 22 (21%) 1 (14%)
Not captured 133 (28%) 29 (25%) 88 (35%) 10 (10%) 6 (86%)
Abbreviations: GOSE, Glasgow Outcome Scale Extended; TBI, traumatic brain injury.
aStatistical significance by Kruskal-Wallis and Fisher's exact as appropriate.

Measurement of disability

Through these efforts 475 participants have been prospectively enrolled and assessed over the phone with the GOSE6 at a 6-month frequency. These data were leveraged to understand trajectories of global disability outcome in the first decade following enrollment during deployment. The GOSE is scored from 1 to 8: 1 = dead, 2 = vegetative, 3 to 4 = severe disability, 5 to 6 = moderate disability, and 7 to 8 = good recovery. Moderate disability (GOSE = 5-6) is defined as one or more of the following: (1) inability to work to previous capacity, (2) inability to resume much of regular social and leisure activities outside the home, and (3) psychological problems, which have frequently resulted in ongoing family disruption or disruption of friendships. Severe disability (GOSE = 3-4) is defined as one or more of the following: (1) inability to drive and/or travel locally without assistance, (2) inability to shop or run errands without assistance, and (3) support required for activities of daily living. Standardized, structured interviews were performed per published guidelines.6,7 Participants were instructed to consider deployment and for those with concussion, the brain injury, as the reference point for this interview and to compare current functional level to that of predeployment. As the GOSE can be administered multiple ways, the decision was made to focus on disability from the brain injury in contrast to disability from all bodily injuries of which there were minimal across groups (enrollment Injury Severity Score (ISS) mean ± SD, non-blast control 0.15 ± 1, blast control 0.26 ± 0.96, blast TBI 1.43 ± 2.91, non-blast TBI 1.64 ± 3.87). Also utilized was the consideration of subsequent head injury exposure (SHIE), which was revisited at each study wave (1 year, 5 years, and 10 years) and inquired about with each GOSE evaluation. This included a TBI history intake interview modified from the Brain Injury Screening Questionnaire (BISQ)18 to include more military-specific and combat-specific scenarios, to confirm life history of head injury exposure and identify any subsequent head injuries sustained since last evaluation.

Statistical analysis

Analysis was completed in January to April 2021. GOSE data were analyzed using latent class growth analysis, in which subjects are hypothesized to be clustered into unobserved longitudinal trajectory classes19 based on individual response patterns. We chose this method over mixed effects regression because it does not assume that all members of an injury or control group have a similar outcome. Rather it looks for participants with similar levels and patterns of outcome, called trajectory groups, and then examines these trajectory groups to identify the characteristics of the participants belonging to them. As is suggested, multiple candidate trajectory models were estimated, varying both the number (4-5 based on BIC, Bayesian Information Criteria) and shape of the trajectory curves (linear, quadratic, cubic although cubic was ruled out due to lack of significance) and a single model was selected based on fit indices criteria including BIC, posterior probability, minimum class size, interpretability, and parsimony.20 All of the models reviewed classified the observed deaths into their own trajectory. The decision was made to narrow the search to just the 4-class models, as this was the maximum number that consistently yielded class sizes between 10% and 50% (excluding the deaths) and posterior probabilities all above 80%. Among the 4-class models, consideration was given to various combinations of cubic effects among the individual trajectories, evaluating each on significance, fit indices, and resulting class size. In the end, the model containing only linear and quadratic effects was selected for the analysis, as it minimized BIC among those with sufficient class sizes and had the added feature of parsimony.

Differences in demographic characteristics among the 4 trajectory groups were assessed for statistical significance using Kruskal-Wallis tests for continuous/ordinal variables and Fisher's exact tests for categorical variables. Group membership in each trajectory (excluding the worst due to low membership) was modeled using nominal logistic regression. Univariable significance was used initially to identify potential predictors, and a multivariable model was constructed controlling for sex and other demographic variables found to be significant in the univariable analysis. A sensitivity analysis was also carried out on the subset of patients with known status of SHIE since the time of enrollment to investigate whether additional head injury exposures impacted global outcome and subsequently this modeling. All reported P values are reported prior to adjustment for multiple comparisons. A Benjamini-Hochberg false discovery rate of 5% was then applied across the entire set of P values for each table, with those that did remain statistically significant explicitly noted.21

The trajectory analyses were carried out in SAS22 statistical software version 9.4 using the “proc traj” application available for free download at Additional statistical analyses were carried out in SPSS version 26 (IBM, Armonk, New York). P values of .05 or lower were considered significant.


Figure 1 shows the latent class growth trajectories identified by model fitting. Dotted lines indicate the model trajectory and black vertical lines show the confidence intervals at each time point while solid lines indicate the group means for each trajectory. Four primary trajectories were identified with corresponding mean GOSE values over the first 10 years following deployment displayed for comparison. As the study sought to collect GOSE evaluations from every service member, patient or control, every 6 months, the general frequency of the GOSE scores as shown is biannual. The primary GOSE disability range corresponding to each trajectory included: good recovery (trajectory 1), upper moderate disability (trajectory 2), lower moderate disability (trajectory 3), and death (trajectory 4). There were no appreciable differences in follow-up rates at each time point among the trajectories and so for this outcome analysis, the missingness was assumed to be random. As we previously reported, all of the known deaths to date were in blast-exposed service members and were primarily death by suicide.14 It is worth noting that even trajectory 1, the good recovery trajectory, was found to have a downward trend beginning around year 8.

Figure 1.:
Latent growth class trajectories of global disability in the first decade following combat concussion.

As we enrolled both combat concussion and combat-deployed controls, this provided the opportunity to examine whether concussion exposure may impact the service member's long-term outcome separate from deployment exposure. By group, 143 non-blast-controls, 54 blast-controls, 236 blast-TBI, and 42 non-blast-TBI were followed (see Table 1). While there was no significant difference in sex or race across the trajectories, there were significant differences by trajectory group in the proportion of each study group, age, education, military rank, branch of service, and where captured, SHIE. Evaluation of the missingness of SHIE by patient group did not reveal any significant differences across trajectories (P = .47, not significant). As military rank is a surrogate for education and there were a few missing entries for education but complete reporting on military rank, all subsequent analyses were interrogated and adjusted for military rank along with patient group, age, and branch of service. Given the interest in sex as a biological variable possibly impacting outcomes, we included sex in all further analyses as well even though there were no significant differences across trajectories. As SHIE since enrollment in combat was captured in a proportion of the sample, further analysis focusing just on this subsample was also examined.

Univariable analyses of patient group, age, sex, military rank, and branch of service from the entire cohort were compared among trajectories (see Table 2). Overall, each parameter other than sex was found to be significantly related to the GOSE trajectories. As the death trajectory (trajectory 4) had very few members, comparative analysis focused on the top 3 trajectories using multinomial logistic regression modeling. Comparing the lower moderate disability trajectory (trajectory 3) to the good recovery disability trajectory (trajectory 1), we found that participants were much more likely to have sustained a concussion in combat (odds ratio [OR] = 49.33 blast-TBI, OR = 37.50 non-blast-TBI, P < .001 for both compared with non-blast control) and more likely to have been enlisted (OR = 24.90, P = .002). Blast-controls still had 5 times the odds of being in the lower moderate disability category than non-blast-controls (OR = 5.00 blast control, P = .007). They also had 4 times the odds of having served in the Army (OR = 4.58, P < .001) and were more likely to be younger (OR = 1.45, P = .03, per 10-year decrease), though the latter did not remain significant after adjustment for multiple comparisons. Comparing the upper moderate disability trajectory (trajectory 2) to the good recovery disability trajectory (trajectory 1) again revealed a similar profile with odds ratios of smaller magnitude although with similar significance. Those who fell into trajectory 2, were more likely to have sustained a concussion in combat (OR = 13.67 blast-TBI, OR = 8.36 non-blast-TBI, P < .001 for both compared with non-blast-control) or have sustained blast exposure (OR = 3.32 blast-control, P = .001), and more likely to have been enlisted (OR = 3.80, P < .001). They were also more likely to be younger (OR = 1.93, P < .001, per 10-year decrease). Comparing the 2 middle trajectories of lower moderate disability group to the upper moderate disability group, we found the lower moderate disability group was significantly more likely to have sustained a concussion in combat (OR = 3.61 blast-TBI, P = .002, OR = 4.48 non-blast-TBI, P = .003) and have been in the Army (OR = 2.95, P = .003) compared with the upper moderate disability group.

TABLE 2 - Univariable analysis of GOSE trajectoriesa
Overall P value Lower moderate vs good recovery Upper moderate vs good recovery Lower moderate vs upper moderate
OR (95% CI) P value OR (95% CI) P value OR (95% CI) P value
Patient group <.001
Blast-control vs non-blast-control 5.00 (1.56-15.99) .007 3.32 (1.63-6.80) .001 1.50 (0.50-4.54) .47
Blast-TBI vs non-blast-control 49.33 (19.77-123.11) <.001 13.67 (7.26-25.76) <.001 3.61 (1.63-7.98) .002
Non-blast-TBI vs non-blast-control 37.50 (10.01-140.50) <.001 8.36 (2.74-25.53) <.001 4.48 (1.67-12.02) .003
Blast-TBI vs non-blast-TBI 1.32 (0.38-4.52) .66 1.63 (0.50-5.36) .42 0.80 (0.40-1.63) .55
Age (per 10-y decrease) <.001 1.45 (1.05-2.01) .03 1.93 (1.46-2.56) <.001 0.75 (0.55-1.02) .07
Sex (male vs female) .17 2.57 (0.88-7.49) .08 1.68 (0.79-3.57) .17 1.53 (0.55-4.23) .41
Military rank (enlisted vs officer) <.001 24.90 (3.29-188.42) .002 3.80 (1.89-7.66) <.001 6.55 (0.85-50.22) .07
Branch of service (Army vs other) <.001 4.58 (2.14-9.80) <.001 1.55 (0.95-2.53) .08 2.95 (1.45-6.03) .003
Abbreviations: CI, confidence interval; GOSE, Glasgow Outcome Scale Extended; OR, odds ratio; TBI, traumatic brain injury.
aEstimates based on multinomial logistic regression modeling, with the death trajectory excluded due to low cell counts. All significant P values (P < .05) remained so after applying a Benjamini-Hochberg 5% false discovery rate (m = 29).

Univariable analysis was followed by multivariable analysis of the entire sample adjusting for patient group, age, sex, military rank, and branch of service (see Table 3). Comparing the lower moderate disability trajectory (trajectory 3) to the good recovery disability trajectory (trajectory 1) by multivariable regression further confirmed the higher odds of combat concussion to the worse disability trajectories. Patients in trajectory 3 had over 40 times the odds of having sustained a blast-related concussion and over 30 times the odds of having sustained a non-blast concussion compared with non-blast-controls (OR = 43.31 blast-TBI, OR = 31.06 non-blast-TBI, P < .001 for both) and still more likely to have been enlisted (OR = 14.93, P = .01). Among those not sustaining a concussion in combat, they had 4 times the odds of having experienced blast exposure (OR = 4.09 blast-controls, P = .02). Comparing the upper moderate disability trajectory (trajectory 2) to the good recovery disability trajectory (trajectory 1) by multivariable regression again revealed a similar profile with odds ratios of smaller magnitude although with similar significance. Interestingly, comparing the 2 middle trajectories of lower moderate disability group to the upper moderate disability group by multivariable regression still found a greater odds of those in the worse disability trajectory for combat concussion (OR = 3.70 blast-TBI, P = .002; OR = 4.08 non-blast-TBI, P = .01).

TABLE 3 - Multivariable analysis of GOSE trajectoriesa
Overall P value Lower moderate vs good recovery Upper moderate vs good recovery Lower moderate vs upper moderate
OR (95% CI) P value OR (95% CI) P value OR (95% CI) P value
Patient group
Blast-control vs non-blast-control <.001 4.09 (1.25-13.41) .02 3.51 (1.67-7.39) .001 1.16 (0.38-3.58) .79
Blast-TBI vs non-blast-control 43.31 (16.64-112.72) <.001 11.70 (5.99-22.87) <.001 3.70 (1.62-8.47) .002
Non-blast TBI vs non-blast-control 31.06 (8.10-119.10) <.001 7.62 (2.46-23.58) <.001 4.08 (1.48-11.25) .01
Blast-TBI vs non-blast-TBI 1.39 (0.39-4.94) .61 1.54 (0.46-5.09) .48 0.91 (0.43-1.92) .80
Age (per 10-y decrease) .02 0.85 (0.56-1.30) .46 1.34 (0.95-1.90) .10 0.64 (0.46-0.89) .01
Sex (male vs female) .65 1.01 (0.27-3.84) .99 0.70 (0.28-1.76) .45 1.44 (0.48-4.32) .52
Military rank (enlisted vs officer) .005 14.93 (1.74-128.02) .01 2.42 (1.01-5.76) .05b 6.18 (0.77-49.55) .09
Branch of service (Army vs other) .07 2.16 (0.90-5.21) .09 0.96 (0.53-1.73) .89 2.26 (1.08-4.72) .03b
Abbreviations: CI, confidence interval; GOSE, Glasgow Outcome Scale Extended; OR, odds ratio; TBI, traumatic brain injury.
aEstimates based on multinomial logistic regression modeling, with the death trajectory excluded due to low cell counts.
bUnless noted, all significant P values (P < .05) remained so after applying a Benjamini-Hochberg 5% false discovery rate (m = 29).

To account for the possible relationship of SHIE to these outcome trajectories, we performed a sensitivity analysis using multivariable regression on the subset where SHIE was captured (see Table 4). In this subset analysis, SHIE was not found to add predictive power to the model compared with the other measures examined (overall P = .17). Given that the general significance stayed roughly the same for the other factors, we interpret this nonsignificant contribution to also mean that it is not likely confounding the effect of other measures in our models that include the entire cohort (see Tables 2 and 3).

TABLE 4 - Multivariable analysis of GOSE trajectories in the subset with known subsequent head injury exposurea
Overall P value Lower moderate vs good recovery Upper moderate vs good recovery Lower moderate vs upper moderate
OR (95% CI) P value OR (95% CI) P value OR (95% CI) P value
Patient group
Blast-control vs non-blast-control <.001 3.26 (0.82-12.99) .09 2.75 (1.13-6.89) .03b 1.19 (0.33-4.33) .79
Blast-TBI vs non-blast-control 80.84 (23.88-273.67) <.001 17.28 (6.90-43.29) <.001 4.68 (1.76-12.41) .002
Non-blast-TBI vs non-blast-control 95.85 (10.26-895.25) <.001 17.10 (2.10-139.33) .01 5.61 (1.72-18.26) .004
Blast-TBI vs non-blast-TBI 0.84 (0.09-7.58) .88 1.01 (0.11-8.94) .99 0.83 (0.35-1.98) .68
Age (per 10-y decrease) .07 0.58 (0.33-1.00) .05b 0.87 (0.55-1.37) .55 0.67 (0.45-0.98) .04b
Sex (male vs female) .13 0.36 (0.07-1.81) .22 0.30 (0.09-0.99) .05b 1.20 (0.37-3.93) .76
Military rank (enlisted vs officer) .22 33.83 (3.40-336.38) .003 4.51 (1.60-12.74) .004 7.50 (0.85-65.85) .07
Branch of service (Army vs other) <.001 1.89 (0.68-5.26) .23 0.93 (0.45-1.91) .84 2.04 (0.89-4.66) .09
Subsequent head injury exposure
1 vs 0 .17 2.19 (0.80-6.04) .13 1.68 (0.69-4.10) .26 1.31 (0.68-2.51) .42
≥2 vs 0 3.36 (0.89-12.71) .07 1.62 (0.47-5.65) .45 2.07 (0.99-4.32) .05b
Abbreviations: CI, confidence interval; GOSE, Glasgow Outcome Scale Extended; OR, odds ratio; TBI, traumatic brain injury.
aEstimates based on multinomial logistic regression modeling, with the death trajectory excluded due to low cell counts.
bUnless noted, all significant P values (P < .05) remained so after applying a Benjamini-Hochberg 5% false discovery rate (m = 36).


In summary we found very high odds of being in a trajectory of worse long-term outcome for those who sustained a concussion in combat and were younger at the time of exposure well above the risk of deployment alone. Furthermore, the risk profile included those with lower education and those who had enlisted in the Army. Also worth noting was the downward trend even in the highest functioning group, which included the majority of combat-deployed controls starting around the 8-year mark post-deployment. Taken together, we believe these findings help inform targeting of more aggressive treatment strategies in service members meeting this profile of greatest risk following deployment to aide in reducing the extremely high public health burden identified with prior conflicts. Additionally, this trajectory analysis brings to light the long-term effects of these seemingly more mild brain injuries, which we have also seen substantiated by continued evolution of both clinical outcome measures14 and neuroimaging13 changes in these very same patients. This study adds to the literature on global disability trajectories previously focused on moderate to severe civilian TBI,8–10 by extending the findings to the service member population with milder brain injuries.

Strengths of the study include the prospective, observational, longitudinal study design with initial evaluation at the point of injury reducing the likelihood of recall bias, which often plagues chronic injury studies, the repeated collection of the primary outcome measure (GOSE) every 6 months over the 10-years of follow-up evaluation to date providing granularity to the trajectory data, the relatively robust sample size in our 2 primary groups of non-blast-controls and blast-TBI, utilization of 2 different control groups and TBI groups to be able to directly examine impact of combat exposure plus head injury via blast or non-blast mechanism relative to combat exposure alone, as well as impact of subconcussive blast injuries in our blast-control patients, and consideration of additional head injury exposures that may have ensued since original enrollment in the study.

Limitations of this study include the inability to control for the heterogeneity of treatment centers in the United States in which our patients and participants sought care and the impact this may have on global disability outcome, lack of predeployment information that could have yielded insight into baseline global disability, the relative paucity of female service members at the time of enrollment to more adequately examine sex as a biological variable, and unmeasured covariates that may have influenced the outcome trajectories.

Overall, the United States is facing a rapidly expanding public health burden from these conflicts, as mortality rates have notably decreased but morbidity rates have substantially risen. Survival does not come without financial and psychological costs to the service members, their families, and the community. There are over 23 million US veterans of all previous conflicts alive today, with TBI diagnosis from prior conflicts24 and mild TBI in particular from recent conflicts25,26 impacting 20%25,27 to 40%24 of this population; even a small increase in life quality could have significant impact on reducing the public health burden. We believe by being informed from longitudinal studies such as this one, the medical community can be proactive in mitigating the potentially negative and extremely costly impact of these combat-related injuries.


1. Taniellian T, Jaycox L. Invisible Wounds of War: Psychological and Cognitive Injuries, Their Consequences, and Services to Assist Recovery. RAND Corporation; 2008.
2. Edwards RD. U.S. war costs: two parts temporary, one part permanent. J Public Econ. 2014;113:54–66. doi:10.1016/j.jpubeco.2014.03.008
3. Bilmes L, Stiglitz J. The long-term costs of conflict: the case of the Iraq War. In: Braddon D, Hartley K, eds. Elgar Handbook on the Economics of Conflict: Edward Elgar Publishers; 2011.
4. Bilmes L. Soldiers Returning From Iraq and Afghanistan: The Long-Term Costs of Providing Veterans Medical Care and Disability Benefits. Harvard University, John F. Kennedy School of Government; 2007.
5. Taylor BC, Hagel Campbell E, Nugent S, et al. Three year trends in veterans health administration utilization and costs after traumatic brain injury screening among veterans with mild traumatic brain injury. J Neurotrauma. 2017;34(17):2567–2574. doi:10.1089/neu.2016.4910
6. Wilson JT, Pettigrew LE, Teasdale GM. Structured interviews for the Glasgow Outcome Scale and the extended Glasgow Outcome Scale: guidelines for their use. J Neurotrauma. 1998;15(8):573–585. doi:10.1089/neu.1998.15.573
7. Wilson L, Boase K, Nelson LD, et al. A Manual for the Glasgow Outcome Scale-Extended (GOSE) Interview. J Neurotrauma. Published online April 6, 2021. doi:10.1089/neu.2020.7527
8. Dams-O'Connor K, Pretz C, Billah T, Hammond FM, Harrison-Felix C. Global outcome trajectories after TBI among survivors and nonsurvivors: a National Institute on Disability and Rehabilitation Research Traumatic Brain Injury Model Systems Study. J Head Trauma Rehabil. 2015;30(4):E1–E10. doi:10.1097/HTR.0000000000000073
9. Forslund MV, Perrin PB, Roe C, et al. Global Outcome trajectories up to 10 years after moderate to severe traumatic brain injury. Front Neurol. 2019;10:219. doi:10.3389/fneur.2019.00219
10. Ponsford JL, Downing MG, Olver J, et al. Longitudinal follow-up of patients with traumatic brain injury: outcome at two, five, and ten years post-injury. J Neurotrauma. 2014;31(1):64–77. doi:10.1089/neu.2013.2997
11. Mac Donald CL, Adam OR, Johnson AM, et al. Acute post-traumatic stress symptoms and age predict outcome in military blast concussion. Brain. 2015;138(pt 5):1314–1326. doi:10.1093/brain/awv038
12. Mac Donald CL, Johnson AM, Cooper D, et al. Detection of blast-related traumatic brain injury in U.S. military personnel. N Engl J Med. 2011;364(22):2091–2100. doi:10.1056/NEJMoa1008069
13. Mac Donald CL, Barber J, Andre J, Panks C, Zalewski K, Temkin N. Longitudinal neuroimaging following combat concussion: sub-acute, 1 year and 5 years postinjury. Brain Commun. 2019;1(1):fcz031. doi:10.1093/braincomms/fcz031
14. Mac Donald CL, Barber J, Patterson J, et al. Comparison of clinical outcomes 1 and 5 years postinjury following combat concussion. Neurology. 2021;96(3):e387–e398. doi:10.1212/WNL.0000000000011089
15. Mac Donald CL, Johnson AM, Wierzechowski L, et al. Prospectively assessed clinical outcomes in concussive blast vs nonblast traumatic brain injury among evacuated US Military personnel. JAMA Neurol. 2014;71(8):994–1002. doi:10.1001/jamaneurol.2014.1114
16. Mac Donald CL, Johnson AM, Wierzechowski L, et al. Outcome trends after US Military concussive traumatic brain injury. J Neurotrauma. 2017;34(14):2206–2219. doi:10.1089/neu.2016.4434
17. VA DoD. Clinical Practice Guideline: Management of Concussion/mild Traumatic Brain Injury. Department of Veterans Affairs (VA) and the Department of Defense (DoD); 2009.
18. Dams-O'Connor K, Cantor JB, Brown M, Dijkers MP, Spielman LA, Gordon WA. Screening for traumatic brain injury: findings and public health implications. J Head Trauma Rehabil. 2014;29(6):479–489. doi:10.1097/HTR.0000000000000099
19. Nagin DS, Land KC. Age, criminal careers, and population heterogeneity specification and estimation of a nonparametric, mixed Poisson model. Criminology. 1993;31(3):327–362. doi:10.1111/j.1745-9125.1993.tb01133.x
20. Jung T, Wickrama K. An introduction to latent class growth analysis and growth mixture modeling. Soc Pers Psychol Compass. 2008;10(1):302–317. doi:10.1111/j.1751-9004.2007.00054.x
21. Benjamini Y, Hochberg Y. Controlling the False discovery rate: a practical and powerful approach to multiple testing. J Roy Statist Soc Ser B. 1995;57(1):289–300.
22. Jones BL, Nagin DS. Advances in group-based trajectory modeling and an SAS procedure for estimating them. Sociol Methods Res. 2007;35:542–571. doi:10.1177/0049124106292364
23. Jones BL. PROC TRAJ. Published 2005. Accessed February 23, 2020.
24. Raymont V, Salazar AM, Krueger F, Grafman J. “Studying injured minds”—the Vietnam head injury study and 40 years of brain injury research. Front Neurol. 2011;2:15. doi:10.3389/fneur.2011.00015
25. Warden D. Military TBI during the Iraq and Afghanistan wars. J Head Trauma Rehabil. 2006;21(5):398–402. doi:10.1097/00001199-200609000-00004
26. Defense and Veterans Brain Injury Center. DoD Worldwide TBI Numbers (2000-2018, Q1-Q2). Defense and Veterans Brain Injury Center; 2018.
27. Taylor BC, Hagel EM, Carlson KF, et al. Prevalence and costs of co-occurring traumatic brain injury with and without psychiatric disturbance and pain among Afghanistan and Iraq War Veteran V.A. users. Med Care. 2012;50(4):342–346. doi:10.1097/MLR.0b013e318245a558

concussion; global disability; long-term outcome; military; trajectory analysis; veteran

© 2022 Wolters Kluwer Health, Inc. All rights reserved.