Economic evaluation of telephone-based concussion management for combat-related mild traumatic brain injury

Abstract

Introduction

Mild traumatic brain injury (mTBI) is an unfortunately common repercussion of military service in a combat zone. The CONTACT study tested an individualized telephone support intervention employing problem solving therapy (PST) for mTBI in soldiers recently returned from deployment. We sought to determine the cost effectiveness of this intervention from a military healthcare system perspective.

Methods

We conducted an intent-to-treat post-hoc analysis by building a decision analytic model that evaluated the choice between using PST or education only (EO). The model included cost-minimization and cost-effectiveness analyses. The incremental cost-effectiveness ratios (ICERs) were calculated as the differences in costs of PST versus EO relative to the differences in the outcomes of participants.

Results

The PST intervention resulted in an annual per-enrolee cost of $1027 (95% CI: $836 to $1248), while EO costs were $32 (95% CI: $25 to $39), resulting in a net incremental cost of $996 per enrolee (95% CI: $806 to $1,217). The ICERs were $68,658/QALY based on EQ-5D (95% CI: -$463,535 to $596,661) and $49,284/QALY based on SF-6D (95% CI: $26,971 to $159,309). Estimates of treatment costs in a real-world setting were accompanied by substantially lower ICERs that are within accepted thresholds for willingness-to-pay.

Discussion

Although the intervention had short-term benefits sufficient to yield acceptable ICERs, there was no long-term effect of PST over EO observed in the study. Consequently, we suggest that future studies examine the use of low-cost approaches, such as booster relapse-prevention calls, that may lead to a sustained treatment benefit for this population.

Keywords

Cost effectiveness cost minimization telehealth

Background

According to the Centers for Disease Control and Prevention (CDC), as many as 2.5 million people experience a traumatic brain injury (TBI) in the United States each year.^1–3 Three quarters of those injuries are categorized as mild TBI (mTBI). Although classified as ‘mild’, the symptoms of mTBI can nonetheless be severe and even life-changing. Physical symptoms can include: headache; nausea or vomiting; fatigue or drowsiness; difficulty sleeping or sleeping more than usual; dizziness; or loss of balance. Sensory symptoms may manifest as blurred vision, ringing in the ears, changes in the ability to smell, and sensitivity to light or sound. Cognitive or mental symptoms may present as memory or concentration problems, mood changes or mood swings, post-traumatic stress, or depression and anxiety.⁴

As well as the human toll on physical and psychological well-being, the economic impact of mTBI is profound. Beyond costs to the health care system, mTBI can cause serious long-term or even permanent impairments in daily activities and ability to return to work. The cost of mTBI to the nation, including direct and indirect medical costs, has been estimated to be at least $76.5 billion annually.⁵

Mild traumatic brain injury (mTBI) is an unfortunately common repercussion of military service in a combat zone. Figures from Operation Enduring Freedom and Operation Iraqi Freedom (OEF/OIF) suggest that up to one quarter of OEF/OIF-deployed American military service members have suffered from mTBI,^6–11 most often emanating from explosive blasts.^11–13 As with civilian victims, the costs of military mTBI are substantial. Consequently, cost-effective interventions to alleviate mTBI symptoms are of considerable interest to senior leaders.

We recently conducted a randomized clinical trial to assess the impact of telephone-delivered problem-solving treatment (PST) on mTBI symptoms in military service members. The CONcussion Treatment After Combat Trauma (CONTACT) Study was a randomized, prospective trial of telephone-delivered PST versus general education only (EO) for service members with mTBI (n = 356).¹⁴ The PST comprised 12 telephone intervention calls over a six-month period and focused on injury-related education, training in problem solving, and focused behavioural strategies for symptoms (e.g., anxiety, post-traumatic stress disorder, depression, sleep, and headache) that commonly co-occur with mTBI. The EO consisted of 12 educational brochures describing common problems following mTBI and deployment. Follow-up for the trial extended six months beyond the intervention for a total of twelve months of follow-up. Results revealed that the PST group showed significant short-term improvement at six months in psychological distress, sleep, depression, post-traumatic stress, and physical functioning.¹⁴ However, post-concussion symptoms improved equally in both groups at six months and, after discontinuation of PST, none of the effects persisted at 12-month follow-up.

We concluded from our clinical results that telephone-delivered PST was a well-accepted treatment that offers promise for reducing psychological distress after combat-related mTBI. However, we could not conclude without further analysis whether or not that promising intervention was also economically feasible. This is especially salient given the lack of maintenance of benefit over a year compared to less resource-intensive alternative. To our knowledge, the economic feasibility of interventions that address mTBI has not been studied in the literature. In the present paper, we estimated the economic costs and outcomes associated with the findings from the CONTACT study, as well as real-world estimates of the costs of these two interventions, in order to evaluate whether PST represents a cost-effective treatment option for the military healthcare system.

Methods

Study design

We conducted an intent-to-treat (ITT) post-hoc analysis of the CONTACT study by building a decision analytic model in Microsoft Excel. Because the 12 month outcomes of CONTACT were not statistically significant, we conducted a cost minimization analysis of the CONTACT study by assuming equal outcomes between the two trial comparators. At six months, however, PST showed significant improvements in outcomes over EO. Therefore, we also examined the cost effectiveness for the temporary improvement in those outcomes relative to the economic costs of achieving those improvements.

This decision analytic model accrued all intervention costs and the choice for cost minimization was the lowest cost treatment arm.

The cost-effectiveness analyses evaluated the choice between using PST or EO, and considered the differences in costs of PST versus EO relative to the differences in the outcomes collected in the trial, or the incremental cost-effectiveness ratios (ICERs).

Subjects

Subjects from the CONTACT study were analysed in the treatment group to which they were randomized. There were no crossovers or protocol violations observed in the trial.¹⁴ Each treatment group included 178 participants. The primary outcomes report of the CONTACT study showed that the characteristics of participants in both the PST and EO groups were evenly balanced.¹⁴ Overall, participants were primarily young (on average 30 years old), male (93%), white (77%), non-Hispanic (81%), and of regular military status (93%). 85% of participants had experienced a blast-related TBI. The majority of participants were from Womack Army Medical Center (69%), and the other 31% were from Madigan Army Medical Center.

Perspective and time horizon

Our analyses took the perspective of the Department of Defense as the health care payer. We examine two potential cost scenarios from this perspective: (1) the costs specifically accumulated in the CONTACT study; (2) the costs if the PST and EO were implemented in a real-world setting at a local or regional level that provided the intervention at a rate of 500 service members over 6 months.

The base case time horizon for the analysis was 12 months as that aligned with the outcomes collected in the CONTACT study.

Cost inputs

Costs were gathered through a series of discussions, after the study was completed, with supervisors, counsellors, and administrative staff that delivered the PST and EO interventions. From these individuals we established consensus around estimates of the amount of time spent on training counsellors, performing supervision, conducting fidelity oversight, providing additional follow-up calls to ensure the safety of a few suicidal participants, filling out paper work with each counselling session, and mailing or emailing participant call summaries and educational brochures.

We were able to use recorded data from the clinical trial to measure the average number of completed sessions per person (7 sessions), the average call time per completed session (45 min), and the average number of attempts not resulting in a counselling session per person over the course of the entire intervention (49 additional attempted calls). We then aggregated these cost estimates and summarize them in Table 1. The hours spent on each of the major activities were multiplied by an average hourly wage of different military mental health providers ($35 per hour for counsellors, $42 per hour for supervisors, and $32 per hour for supporting staff) to calculate the total costs, conservatively estimating that counsellor activities could be performed by a Master’s level trained practitioner and not accounting for fringe benefits. Only 1% of the education materials and call summary reports for those in the PST group were sent by mail, and the rest were sent by email. The cost of postage for this mail was also included in the estimates.

Table 1.

Estimated economic costs of problem solving treatment (PST) and education only (EO) in the CONTACT study (base case).

		Lower	Upper
		credible	credible
Cost components	Mean	limit	limit
Problem solving treatment (PST) n = 178
Counselling sessions (counsellors only)
hours	878	782	974
Attempts (counsellors only) hours	291	145	436
Paperwork (total) hours	353	310	396
Paperwork (counsellors) hours	249	210	288
Paperwork (staff) hours	104	100	108
Mailing costs	$6.48	$5.41	$7.55
Training (total) hours	488	430	546
Training (supervisors) hours	128	94	162
Training (counsellors) hours	360	336	384
Oversight (total) hours	2916	2871	2962
Oversight (supervisors) hours	1342	1321	1363
Oversight (counsellors) hours	1445	1425	1466
Oversight (staff) hours	129	124	133
Total hours	4926	4538	5314
Supervisor total hours	1470	1416	1525
Counsellor total hours	3223	2898	3549
Staff total hours	233	224	241
Hours per enrolee	28	25	30
Education Only (EO) n = 178
Paperwork (staff only) hours	178	171	185
Mailing costs	$11.11	$9.27	$12.94
Total hours	178	171	185
Hours per enrolee	1	1	1

We also discussed with the clinicians who implemented the study and with clinicians in the military how the intervention might be implemented outside of a clinical trial. In particular, the clinicians were asked to consider the scenario of implementing this as an ongoing service offered at a local or regional level for 500 service members with mTBI every 6 months. We estimated the real-world costs of implementing the interventions using two distinct approaches. First, we undertook a micro-costing exercise with clinicians where we estimated a highly efficient clinic setting with 500 patients and estimated the hours required to care for those patients (Table 2(a)). We also took a macro-costing approach from our clinician discussions whereby we estimated the number of full-time employees it would take at each level of clinical staff in order to undertake the intervention with a 500 patient case load (Table 2(b)).

Table 2a.

Estimated economic costs of problem solving treatment (PST) and education only (EO) if implemented in a real-world military setting, enrolling 500 participants every 6 months (micro-costing approach).

		Lower	Upper
		credible	credible
Types of economic costs	Mean	limit	limit
Problem solving treatment (PST) n = 500
Counselling sessions (counsellors only)
hours	2250	2119	2381
Attempts (counsellors only) hours	167	83	250
Paperwork (total) hours	1000	881	1119
Paperwork (counsellors) hours	1000	881	1119
Paperwork (staff) hours	0	0	0
Mailing costs	$0.00	$0.00	$0.00
Training (total) hours	12	0	45
Training (supervisors) hours	2	0	9
Training (counsellors) hours	10	0	36
Oversight (total) hours	270	263	276
Oversight (supervisors) hours	88	86	89
Oversight (counsellors) hours	182	177	187
Oversight (staff) hours	0	0	0
Total hours	3698	3345	4072
Supervisor total hours	90	86	98
Counsellor total hours	3609	3259	3974
Staff total hours	0	0	0
Hours per enrolee	7	7	8
Education only (EO) n = 500
Paperwork (staff only) hours	500	482	518
Mailing costs	$31.20	$26.04	$36.36
Total hours	500	482	518
Hours per enrolee	1	1	1

Notes: For supervisors we used the average hourly wage of an O6 or O5 officer ($41.93; range: $32.24, $49.08); for the three counsellors we used the average hourly wage of O3 to O5 officers ($35.31; range: $24.45, $43.87); for staff we used the average hourly wage of an O3 officer ($31.69; range: $24.45, $37.01). In the real-world setting there would be no additional oversight and fidelity checking by staff members, and all the paper work would be done electronically with electronic medical records by the counsellors rather than having a separate staff member emailing and mailing out paperwork after the PST phone calls. Each enrolee is expected to complete 6 sessions.

Table 2b.

Estimated economic costs of problem solving treatment (PST) if implemented in a real-world military setting, enrolling 500 participants every 6 months (macro-costing approach).

		Lower	Upper
		credible	credible
Type of provider	Mean	limit	limit
Counsellors FTEs	6.67	5.00	10.00
Supervisors FTEs	0.67	0.44	1.11
Total FTEs	7.33	5.44	11.11
FTEs per enrolee	0.01	0.01	0.02
Work hours per enrolee	15.25	11.32	23.11

Notes: FTE: full time employee; for supervisors we used the average hourly wage of an O6 or O5 officer ($41.93; range: $32.24, $49.08); for counsellors we used the average hourly wage of O3 to O5 officers ($35.31; range: $24.45, $43.87). Each enrolee is expected to complete 6 sessions.

Health outcome inputs

The four main health outcomes used in the cost-effectiveness analyses were the Brief Symptom Inventory-18 (BSI-18) global severity index (GSI), the Patient Health Questionnaire-9 (PHQ-9), the Euroquol five dimensions questionnaire (EQ-5D), and the Short Form six dimensions index (SF-6D).

The BSI-18 was a primary outcome measure of the CONTACT study and previous analyses of the study found that there was a significant reduction in the GSI at six months. The GSI is a standardized measure of overall mental and emotional symptom severity with a mean of 50 and standard deviation of 10. The changes at six months were at the magnitude of about one-third of a standard deviation of the measure.

Significant improvements in depression were also observed at six months as measured by the PHQ-9. The average improvement in the PST group’s PHQ-9 score was about one unit better than that of the EO group. A one unit shift in scores is comparable to a shift from a depressive symptom that bothers an individual nearly every day to more than half the days, more than half the days to several days, or several days to not at all. In some studies the PHQ-9 has been used to develop a measure of depression free days,^15,16 but that approach was not used in this study because it would require an interpolation of depression free days over the course of the year using only three data points (baseline, six months, and 12 months).

The EQ-5D and SF-6D are both measures of health related quality of life that can be used to generate preference weighted health utility values. When the incremental change in health utility is multiplied by the duration of the effect, this is the estimated number of quality adjusted life years (QALYs) accrued by individuals receiving the intervention. The EQ-5D responses were preference weighted according to a US sample in order to estimate the health utilities.¹⁷ We generated the SF-6D preference weighted utilities using a proprietary algorithm (OptumInsight Life Sciences) that converts SF-12 responses into SF-6D utilities.

For each of the health outcomes we generated mixed-effects regression models. These models adjusted for the subject’s military status, their baseline GSI, and the study site. From these models we calculated the predicted values at the six month point-estimate. The incremental benefit for the outcomes over the entire 12 months was estimated by taking the difference in predicted values at six months and multiplying them by 0.5, or the estimated duration of the effect in terms of years. The effect was estimated to last half a year because all outcomes for PST and EO were similar by the end of the year. Thus, for the EQ-5D and SF-6D we report health utility point-estimates at 6 months, but the incremental benefit over 12 months is the QALYs gained over the year.

Sensitivity analyses

Model estimate variance and uncertainty was assessed through both one-way and probabilistic sensitivity analyses. Each input parameter’s variance was included in the model and combined with the deterministic parameter estimates. The one-way analysis involved individually varying each input parameter to the 2.5% and 97.5% values of the estimated distribution and storing the resulting cost-minimization finding. The results of the one-way sensitivity analysis are presented in the form of a tornado diagram, with the input parameters organized by descending order of influence on the results. We performed a Monte Carlo simulation, drawing randomly from the appropriate statistical distribution of each parameter simultaneously and recorded the results from 3000 simulations. The results of this probabilistic sensitivity analysis were estimated as the probabilistic mean and 95% confidence range.

The University of California, San Diego Human Research Protections Program approved this study (#101730). The CONTACT study was also approved by the institutional review boards at each study site, the University of Washington, and by the Army Human Rights Protection Office. All analyses on the outcomes of the study were conducted in R statistical package. The model calculations and sensitivity analyses were conducted using Excel.

Results

Utilization of the PST intervention resulted in an annual per-enrolee cost of $1,027 (95% CI: $836 to $1248), while EO costs were $32 (95% CI: $25 to $39), resulting in a net incremental cost of $996 per enrolee (95% CI: $806 to $1217) for delivering the PST intervention. Table 3 presents the model results for incremental differences in costs and outcomes as well as incremental cost-effectiveness ratios for the various outcome measures. Enrolees receiving PST had a 1.56-point improvement on the GSI over 12 months (95% CI: 0.43 to 2.71). The associated ICER for cost per point of improvement in GSI was $639 (95% CI: $342 to $2191) at one year.

Table 3.

Cost effectiveness of problem solving treatment (PST) compared to education only (EO) over 12 months.

Outcome	Cost (perenrolee)	Outcome measures per enrolee
Outcome	Cost (perenrolee)	Time period	GSI (95% CI)	PHQ-9 (95% CI)	EQ-5D (95% CI)	SF-6D (95% CI)
PST	$1,027	6-Month	54.57	7.85	0.780	0.744
PST	$836–$1,248	point-estimate	(52.88–56.23)	(6.93–8.78)	(0.753–0.807)	(0.723–0.764)
EO	$32	6-Month	57.69	9.16	0.751	0.704
EO	$25–$39	point-estimate	(56.20–59.26)	(8.32–9.99)	(0.725–0.775)	(0.684–0.722)
Incremental	$996	Entire 12-months^a	1.56	0.66	0.015	0.020
Incremental	$806–$1,217		(0.43–2.71)	(0.01–1.27)	(−0.003 to 0.034)	(0.006–0.034)
Incremental cost- effectiveness ratio (ICER)	$996	Entire 12-months	$639	$1,519	$68,658	$49,284
Incremental cost- effectiveness ratio (ICER)	$806–$1,217		($342 to $2191)	($591 to $7749)	(−$463,535 to $596,661)	($26,971 to $159,309)
Real-world ICER	$516	Entire 12-months	$331	$787	$35,592	$25,549
Real-world ICER	$303–$789		($147 to $1208)	($287 to $4267)	(−$218,970 to $348,955)	($11,505 to $86,922)
Highly-efficient real-world ICER	$231	Entire 12-months	$148	$352	$15,906	$11,417
Highly-efficient real-world ICER	$163–$308		($74 to $479)	($138 to $1619)	(−$102,138 to $133,678)	($5919 to $38,967)

The incremental benefit for the outcomes over the entire 12 months was estimated by taking the difference in predicted values at 6 months and multiplying them by 0.5, or the estimated duration of the effect in terms of years. The effect was estimated to last half a year because all outcomes for PST and EO were similar by the end of the year. Thus, for the EQ-5D and SF-6D we report health utility point-estimates at 6 months, but the incremental benefit over 12 months is the QALYs gained over the year.

The PST group scored lower in the PHQ-9 over 12 months, by 0.66 points (95% CI: 0.01–1.27), generating an incremental cost per point improvement in the PHQ-9 at $1519 (95% CI: $591 to $7749).

Depending on the measure of health utility, there were on average 0.015 (95% CI: −0.003 to 0.034) to 0.020 (95% CI: 0.006–0.034) QALYs gained per enrolee over the 12 months. The ICERs were $68,658/QALY based on EQ-5D (95% CI: -$463,535 to $596,661, with the negative ICER representing a dominated strategy of higher costs but poorer outcomes) and $49,284/QALY based on SF-6D (95% CI: $26,971 to $159,309). The confidence intervals for the ICER based on the EQ-5D are much wider than those for the SF-6D because in the sensitivity analysis there were more values close to zero incremental benefit as measured by the EQ-5D.

One-way sensitivity analysis revealed that wage rates of care providers, call attempts, counselling sessions, and training were the most influential model parameters (Figure 1). This parameter level analysis also demonstrated that the utility input values and wage rates were the most influential on the incremental cost-effectiveness ratio (Figure 2). The probabilistic sensitivity analysis resulted in similar cost-effectiveness ratios compared to the deterministic findings, with similar interpretation overall (Figure 3).

Figure 1.

One-way sensitivity analysis of the 10 most influential model parameters on incremental costs per person for problem solving treatment vs. education only as implemented in the CONTACT study.

Figure 2.

One-way sensitivity analysis of the 10 most influential model parameters on incremental cost-effectiveness (SF-6D) for problem solving treatment vs. education only as implemented in the CONTACT study.

Figure 3.

Probabilistic sensitivity analysis of the incremental cost-effectiveness for problem solving treatment vs. education only using the EQ-5D and SF-6D.

Real-world costs

Both the macro- and micro-costing based treatment costs for the PST and EO interventions resulted in lower incremental costs for PST compared to the trial-based costs. The more conservative costs estimated an incremental cost of PST at an additional $516 (95% CI: $303 to $789), while the highly efficient clinic costs were estimated at an additional $231 (95% CI: $163 to $308). These lower costs were accompanied by substantially lower cost-effectiveness ratios, well within accepted thresholds for willingness-to-pay in both the EQ-5D and SF-6D models (Table 3).

Discussion

We estimated the economic consequences of choosing to use telephone-delivered PST versus EO to deliver care to individuals with mTBI. Though EO would minimize overall costs compared to PST, we found that the PST intervention can still represent a cost-effective treatment. Acknowledging that the trial-based costs are inflated, particular from a heightened level of oversight that included frequent meetings of supervisors and counsellors, the true incremental cost of adding this new clinical service appears palatable at a likely cost of between $200 and $500 per patient over six months. The oversight in a real-world setting could include less frequent meetings and could also be for a larger group of patients thus reducing the time cost per patient. It should be noted, however, that the lowering of costs in a real-world scenario will have an unknown effect on the treatment efficacy. So, while the costs will likely be lower than those observed for the trial-based application of the intervention, it is possible that the efficacy will also be attenuated. This would increase the incremental cost of PST versus EO.

Limitations

As with any modelling study, there are limitations and assumptions that should be considered when evaluating the results presented herein. First, the effectiveness of both interventions was based on a clinical trial. We attempted to utilize the best available information, generating statistical models from the trial data and adjusting for characteristics that were not balanced by the trial’s randomization. However, real-world effectiveness may differ from what was measured in the trial, so the lower ICERs associated with these analyses may not occur. Second, assumptions were made about the improvements in health outcomes from 0–6 months and 6–12 months. More frequent measures are needed to obtain a more accurate understanding of the intervention over time. Third, we were required to make assumptions regarding costs both in the base case that was developed from the actual costs incurred during the trial and from input received from clinicians regarding real-world scenarios. Those real-world scenarios were developed with clinicians who implement PST interventions in their clinics, lending credibility to the estimates we have generated. Finally, the impact on cost was limited in this study to only the costs of the intervention. We were unable to examine cost impacts on productivity gains from the intervention. We also did not include any changes in medical costs among those receiving PST or EO. If this intervention is implemented more broadly, we suggest that an evaluation be performed that tracks the real-world effectiveness with repeated and more frequent outcome measures, assesses the ongoing time costs associated with the intervention in a real-world setting, and explores the potential costs or gains from changes in medical service utilization and productivity.

Conclusion

The PST intervention provided a measureable benefit on outcomes at six months, but this benefit was indistinguishable from a less resource-intensive alternative by twelve months. The cost-effectiveness analysis found that the cost per increase in several outcomes would likely fall within an acceptable range for the payer. Unfortunately, both the costs and the measured benefit of PST observed in the trial will likely differ in a real-world effectiveness context, thereby affecting the ICER. Given the lack of long-term effect of PST over EO observed in the study, it would be appropriate to further study whether applying resources to low-cost approaches, such as booster relapse-prevention calls, may lead to a sustained treatment benefit for this population.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was partially supported the US Army Medical Research and Material Command (USAMRMC; grant number W81XWH-08-2-0159). Supplementary support was provided by the National Institute on Disability and Rehabilitation Research (grant number H133G070143). The views expressed are those of the authors and do not reflect the official policy or position of the Department of Defense (DoD) or the US Government.

References

Centers for Disease Control and Prevention (CDC). National hospital ambulatory medical care survey (NHAMCS), www.cdc.gov/nchs/ahcd (2010, accessed 9 August 2016).

Centers for Disease Control and Prevention (CDC). National hospital discharge survey (NHDS), www.cdc.gov/nchs/nhds (2010, accessed 9 August 2016).

Centers for Disease Control and Prevention (CDC). National vital statistics system (NVSS), www.cdc.gov/nchs/nvss/ (2010, accessed 9 August 2016).

Mayo Clinic. Traumatic brain injury, www.mayoclinic.org/diseases-conditions/traumatic-brain-injury/basics/symptoms/con-20029302 (2016, accessed 9 August 2016).

Centers for Disease Control and Prevention (CDC). Injury prevention and control: traumatic brain injury and concussion, www.cdc.gov/traumaticbraininjury/severe.html (2016, accessed 9 August 2016).

Burke

Degeneffe

Olney

. A new disability for rehabilitation counselors: Iraq war veterans with traumatic brain injury and post-traumatic stress disorder. J Rehabil 2009; 75(3): 5–14.

Hoge

McGurk

Thomas

et al.

Mild traumatic brain injury in US soldiers returning from Iraq. N Engl J Med 2008; 358: 453–463.

Morissette

Woodward

Kimbrel

et al.

Deployment-related TBI, persistent postconcussive symptoms, PTSD, and depression in OEF/OIF veterans. Rehabil Psychol 2011; 56: 340–350.

Shively

Perl

. Traumatic brain injury, shell shock, and posttraumatic stress disorder in the military: past, present, and future. J Head Trauma Rehabil 2012; 27: 234–239.

10.

Terrio

Brenner

Ivins

et al.

Traumatic brain injury screening: preliminary findings in a US Army Brigade Combat Team. J Head Trauma Rehabil 2009; 24: 14–23.

11.

Warden

. Military TBI during the Iraq and Afghanistan wars. J Head Trauma Rehabil 2006; 21: 398–402.

12.

Taber

Warden

Hurley

. Blast-related traumatic brain injury: what is known? J Neuropsychiatry Clin Neurosci 2006; 18: 141–145.

13.

Wilk

Thomas

McGurk

et al.

Mild traumatic brain injury (concussion) during combat: lack of association of blast mechanism with persistent postconcussive symptoms. J Head Trauma Rehabil 2010; 25: 9–14.

14.

Bell

Fann

Brockway

et al.

Telephone problem solving for service members with mild traumatic brain injury: a randomized, clinical trial. J Neurotrauma 2017; 34: 313–321.

15.

Katon

Russo

Lin

et al.

Cost-effectiveness of a multicondition collaborative care intervention: a randomized controlled trial. Arch Gen Psychiatry 2012; 69: 506–514.

16.

Vannoy

Arean

Unutzer

. Advantages of using estimated depression-free days for evaluating treatment efficacy. Psychiatr Serv 2010; 61: 160–163.

17.

Agency for Healthcare Research and Quality. Calculating the US population-based EQ-5D index score, https://archive.ahrq.gov/professionals/clinicians-providers/resources/rice/EQ5Dscore.html (2005, accessed 25 October 2016).