J Head Trauma Rehabil Vol. 30, No. 1, pp. E57–E64 c 2015 Wolters Kluwer Health, Inc. All rights reserved. Copyright
Population Estimates of Undocumented Incident Traumatic Brain Injuries Among Combat-Deployed US Military Personnel Rachel P. Chase, BSc; Remington L. Nevin, MD, MPH Objective: To estimate the number of undocumented incident traumatic brain injuries (TBIs) among active component US military personnel serving in Iraq and Afghanistan prior to policy changes implemented in late 2006 and 2010 that improved TBI documentation. Methods: Negative binomial regression was used to model monthly incident TBI counts between December 2010 and June 2012 (N = 19) and then estimate expected monthly counts of incident TBIs during 2 periods: January 2003-October 2006 and November 2006-November 2010. Monthly amputation counts from Department of Defense surveillance data were used as a proxy for changing injury rates. Monthly active component deployment estimates derived from the Congressional Research Service, Brookings Institution, and Defense Manpower Data Center were used to estimate the size of the at-risk population each month. The difference between expected monthly incident TBI counts and reported counts is presented as the estimated number of undocumented incident TBIs. Results: The full model estimates that 21 257 active component military personnel experienced undocumented incident TBIs while deployed in Iraq or Afghanistan between January 2003 and October 2006, more than 4 times the 5272 incident TBIs documented during that period. Conclusions: A sizeable majority of Iraq and Afghanistan combat veterans who experienced incident TBI while deployed prior to November 2006 are likely to have had their injuries undocumented, creating challenges for clinical care, disability evaluation, and future research. Key words: Afghan war, Iraq war, late diagnosis, military, traumatic brain injury
T
RAUMATIC BRAIN INJURIES (TBIs), particularly those classified as “mild,” have been regarded as “signature injuries” and “invisible wounds” of the conflicts in Iraq and Afghanistan due to both their pernicious effects and their systematic underdocumentation, particularly at the outset of each war.
Author Affiliations: Departments of International Health (Ms Chase) and Mental Health (Dr Nevin), Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland. The authors thank Shannon McMahon and Roger Peng for their helpful comments during manuscript preparation and review and the 3 anonymous reviewers for their valuable recommendations. The authors are also indebted to the staff of the Armed Forces Health Surveillance Center for providing the monthly traumatic brain injury and amputation count data on which these analyses are based. Dr Nevin reports serving as consultant and expert witness in cases involving claims of antimalarial toxicity, a relevant conflict given the potential for confounding of this condition with traumatic brain injury in military populations. Ms Chase reports no conflicts of interest. This work was conducted without external funding. Corresponding Author: Rachel P. Chase, BSc, Department of International Health, Johns Hopkins Bloomberg School of Public Health, 615 N Wolfe St, Baltimore, MD 21205 ([email protected]). DOI: 10.1097/HTR.0000000000000061
Prior to military-wide policy changes following legislation passed in October 2006,1 US military personnel were not regularly evaluated for TBIs unless they were evacuated to a medical center such as the Walter Reed Army Medical Center after severe injuries.2 Systematic predeployment and in-theater TBI assessments were initiated only in November 2006 after investigations revealed high rates of previously undocumented TBIs, especially among patients with blast-induced injuries.2 Efforts to better document TBIs led to dramatically increased rates of TBI incidence beginning in late 2006,3 but it is not clear how many TBIs had gone undocumented prior to these changes or how many went undocumented as the improved policies went into effect. As of 2010, estimates of undocumented TBIs were reported in the popular media as numbering into the “tens of thousands”4 of cases. However, as a result of historical underdocumentation, significant uncertainties regarding the true historical burden of deployment-related TBI persist. Estimating the total number of incident TBIs over this period is difficult not only due to changing documentation policies and practices but also due to temporal fluctuations in rates of underlying injury causing events, which are typically subject to military E57
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
E58
JOURNAL OF HEAD TRAUMA REHABILITATION/JANUARY–FEBRUARY 2015
secrecy and not publicly available to researchers. In the absence of such information, such fluctuations may be accounted for through the use of an available proxy measure of combat injury. Recent trends support the use of amputations as one such measure. While steady improvements in the use of armored vehicles and body armor as well as forward medical assistance have lessened mortality in recent conflicts as compared with previous wars,5 rates of certain injuries including amputations have increased.6 The majority of combat-related injuries within the US military now result from the blast effects of improvised explosive devices and other explosive devices.7 Blast effects are often categorized as primary (caused by the blast wave itself), secondary (caused by items propelled by the blast, such as shrapnel), or tertiary (caused by the body being thrown against other objects).8 Advances in the use of protective equipment including body armor and ballistic goggles have been especially effective at limiting the incidence of penetrating injuries from many secondary and tertiary blast effects but have left the unprotected extremities susceptible to penetrating injury and the head susceptible to nonpenetrative injuries to the brain.2 While mild blast injuries to the brain are often hard to discern and diagnose,7 injuries to other parts of the head, such as the scalp, are easy to discern and are not the focus of this article. Blast injuries to the extremities are highly visible injuries that in their most extreme form result in amputation. Amputations have been documented consistently over the course of the conflicts in Iraq and Afghanistan and, unlike counts of injury causing events, are publicly available. On this basis, amputations should prove an accessible and reasonably valid and reliable proxy for changing combat injury rates when studying TBI incidence. Brain injuries are important to diagnose and manage because they may cause a wide range of short- and longterm problems. These include neuropsychiatric and somatic symptoms such as memory loss, cognitive deficits, mood volatility, substance use disorders, personality changes, sleep difficulties, and sensory deficits.9 Blastrelated and multiple TBIs are associated with visual and auditory deficits, posttraumatic epilepsy, headaches, major depression, and suicide risk.10,11 In deployed settings, brain injuries are associated with an increased risk of posttraumatic stress disorder,12 and there are growing concerns that recurrent and blast-related TBIs are associated with progressive neurodegenerative processes such as chronic traumatic encephalopathy13 and disorders resembling Alzheimer disease.14 The burden of disability from combat-related TBIs may be more debilitating than other TBIs due to both the extreme nature of the injuryproducing event15 and the limited ability to diagnose and treat the injury in a timely manner, which may lead to repeat exposure to injury.16 In some cases, symptoms
of TBI may complicate processes of postdeployment recovery and reintegration.17 Among veterans, a formally diagnosed and documented TBI is critical in adjudicating individual medical disability claims, receiving health benefits, and facilitating military distinction.18 Similarly, among military personnel remaining on active duty, documentation of prior TBI may preclude subsequent deployment, thus minimizing the risk of reinjury.19 Military personnel with a documented TBI and experiencing persistent symptoms may be eligible for medical retirement, whereas those lacking documentation of TBI risk misattribution of such symptoms to other conditions such as malingering4,14 or personality disorder17 that may preclude eligibility for medical retirement. Accurate population estimates of the burden of deployment-related TBIs are also critical to informing an appropriate public health response. Since guidance issued in late 2006—first in the form of legislation, followed shortly by Department of Defense (DoD) policy implementation1,20 —the US military has attempted to provide more accurate counts through the use of regular screenings and automated medical surveillance.21 Between 2006 and 2010, the DoD made several improvements to policies and practices associated with TBI documentation. These included publishing a standard definition for TBI (in October 2007),20 requiring that all military personnel be evaluated using a standardized neuropsychological assessment (in March 2008),1 more consistently tracking and treating TBIs in the field (in June and December 2010),22–24 and progressively deploying electronic medical record systems in field settings.25 However, despite these improvements, considerable disagreement and controversy remain regarding the accuracy of historical deployment-related TBI count data.4 METHODS In this analysis, negative binomial regression was used to model monthly incident TBI counts between December 2010 and June 2012 (N = 19) and then estimate expected monthly counts of incident TBIs during 2 periods: January 2003-October 2006 and November 2006-November 2010. Two models are presented in which monthly active component deployment estimates served as the size of the at-risk population each month. In the full model, monthly counts of amputations were included as a proxy for changing injury rates, as opposed to the alternative model in which these counts were excluded. The model was then applied to earlier observations to calculate predicted incident TBI counts. The difference between the number of documented incident TBIs and the estimated number each month was interpreted as the number of undocumented incident
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Estimates of Undocumented Incident Combat-Related TBIs TBIs that month. In the case of negative estimates, the estimate was interpreted as the excess number of TBIs documented that month. The full negative binomial regression model took the form of: log(y) = b 0 + b 1 x1 + log(x2 ) + ε where y is the estimated number of incident TBIs in a given month, x1 is the number of amputations, and x2 is the estimated number of deployed active component military personnel (in thousands). For illustration, a sample calculation from the full model for January 2005, when 13 amputations were recorded, an estimated 108 674 personnel were at risk, and 81 TBIs were documented, is provided in the following equation using the estimates of b0 and b1 presented in Table 1: log(y ) = b 0 + b 1 x1 + log(x2 ) = 1.569 + 0.013(13) +log(108.674) = 6.426 By exponentiation, the estimate of the total number of incident TBIs in January 2005 is found to be y = 617. Since 81 of these TBIs were documented, it is estimated that approximately 617 − 81 = 536 incident TBIs were not documented in January 2005 in this population. The incident TBI and amputation data were obtained from the DoD. The DoD maintains a passive medical surveillance system, known as the Defense Medical Surveillance System, which integrates deployment and medical diagnosis data among all active component military personnel.26 The DoD has used the Defense Medical Surveillance System to publish monthly
E59
tabulations of deployment-related conditions of special surveillance interest, including amputations since May 200527 and incident cases of TBI of any severity since September 2007.28 Methods used in these tabulations, including relevant International Classification of Diseases, Ninth Revision, Clinical Modification (ICD9-CM), medical diagnosis codes used to define an incident case of TBI, including ICD-9-CM codes 310.2, 800-801, 803-804, 850-854, 907.0, 950.1-950.3, 959.01, and supplementary DoD ICD-9-CM codes V15.5_19, V15.5_A-F, V15.52_0-9, V15.52_A-F, V15.59_1-9, V15.59_A-F, have been previously described.29 Medical diagnosis data are derived primarily from electronic medical records from medical encounters and patient charts at fixed facilities upon return from deployment, supplemented by a small but rising percentage of diagnosis data derived from electronic medical records at temporary medical facilities at deployed locations.29 Estimates of monthly combat-deployed personnel in Iraq and Afghanistan published by the Congressional Research Service30 and the Brookings Institution,31,32 as well as deployment figures from the Defense Manpower Data Center,33 were used to create the exposure variable estimating the size of the at-risk population. Total combat-deployed personnel estimates (as reported by the Congressional Research Service and the Brookings Institution) were multiplied by the proportion of all deployed personnel who were active component each month (as reported by the Defense Manpower Data Center). The resulting monthly estimates of active component combat-deployed personnel in Iraq and Afghanistan were used for modeling.
Estimated number of undocumented incident TBIs, full and alternative negative binomial models, with model estimates and fit statisticsa TABLE 1
Estimated number of undocumented incident TBIs January 2003-October 2006 November 2006-November 2010 Model estimates Constant (standard error) Amputation coefficient (standard error) Dispersion parameter (standard error) Model fit statistics Log likelihood AIC BIC Deviance residuals Degrees of freedom Deviance to degrees of freedom ratio
Full model
Alternative model
21 257 11 565
24 033 15 123
1.569 (0.114) 0.013 (0.005) 0.040 (0.013)
1.826 (0.053)
−117.3b 12.6 −30.9 19.2 17 1.1
−119.9 12.7 −33.8 19.2 18 1.1
0.052 (0.017)
Abbreviations: AIC, Akaike information criteria; BIC, Bayesian information criteria; TBI, traumatic brain injury. a Alternative model excludes monthly amputation counts as a parameter. b Significantly better than the alternative model (P < .05).
www.headtraumarehab.com Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
E60
JOURNAL OF HEAD TRAUMA REHABILITATION/JANUARY–FEBRUARY 2015
Including monthly deployment figures as an exposure variable or offset is appropriate, as troop surges and drawdowns resulted in substantial changes in the size of the at-risk population between 2003 and 2012, and the negative binomial model would otherwise have to assume a constant population size. In the full model, counts of monthly deployment-related amputations were included as a predictor of each month’s incident TBI counts. As described earlier, including amputation counts in the model as a proxy for injury-causing events is appropriate, given strong clinical and ecological evidence of common mechanisms of injury,7,34 which suggest that trends in the burden of deployment-related TBI should be correlated with those of amputations. The alternative model that includes only deployment figures has been included for comparison with the full model (see Table 1). Monthly incident TBI counts between December 2010 and June 2012 (inclusive) were modeled using negative binomial regression rather than Poisson regression to account for overdispersion in the data (mean = 591; variance = 27 829). The full model with amputations as a parameter and deployment estimates as an offset was compared with the alternative model using log likelihood, Akaike and Bayesian information criteria, and the ratio of deviance residuals to degrees of freedom. Standard errors in model estimates are reported (see Table 1). However, total population data were used in place of randomly selected sample data, so these errors are not used to infer confidence intervals or similar statistics that rely on random sampling assumptions. Only publicly available and de-identified count data were used in this analysis. This study was reviewed by the institutional review board of the Johns Hopkins Bloomberg School of Public Health and deemed exempt. All analyses were conducted using Stata (Version 13.1; StataCorp, College Station, Texas).
incident TBIs would have been identified between November 2006 and November 2010 beyond the 26 904 that were documented during that time. To demonstrate the statistical influence of including amputation counts in the final model, the results of the alternative model that omits these are presented. This alternative, which is nested within the full model, eliminates amputation counts from the analysis and therefore assumes constant rates of combat injury. This alternative model provides somewhat similar estimates of undocumented incident TBIs. Between January 2003 and October 2006, the alternative model predicts 24 033 additional incident TBIs. Between November 2006 and November 2010, it predicts 15 123 additional incident TBIs. The alternative model might be considered an improvement over the full model if changes in the number of amputations each month were uncorrelated with changes in the number of TBIs each month, or if changes in the true relationship between amputations and TBIs over time varied sufficiently that a constant rate of amputations would serve as a better approximation of true trends. Neither of these alternative assumptions is strongly supported, as during the 2 periods of relatively stable TBI documentation practices (before policy changes in November 2006 and after the last major changes in December 2010), a strong association between counts of amputations and counts of incident TBIs is observed. The association is visible by inspection (see Figure 1) and analysis with a simple Pearson statistic (r = 0.55 before the policy changes and r = 0.71 after). Although the full model is most justified on both empirical and theoretical grounds, certain results are qualitatively robust across models. Both models estimate that between January 2003 and October 2006, approximately 4 to 5 TBIs were undocumented for each TBI
RESULTS The 2 statistical models exhibited comparable fit statistics, and neither model distinguished itself by dramatically improved Akaike and Bayesian information criteria or deviance to degrees of freedom ratio, although the full model fit was significantly better by the log likelihood test (P < .05). Drawing on the full model that includes monthly amputation counts as a proxy for combat injury, an estimated 32 822 incident TBIs were undocumented among combat-deployed personnel from January 2003 to November 2010 (Table 1). This model estimates that if current policies were applied to past cases, 21 257 additional incident TBIs would have been documented between January 2003 and October 2006 in addition to the 5272 TBIs that were documented. The full model also suggests that an additional 11 565
Figure 1. Predicted active component monthly incident deployment-related TBIs, according to the full model, with documented monthly incident TBI and amputation counts, and estimated deployment figures. TBI indicates traumatic brain injury.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Estimates of Undocumented Incident Combat-Related TBIs documented among deployed personnel. Both models also suggest major improvements in documentation of incident TBIs following the November 2006 implementation of policy changes. DISCUSSION This analysis provides the first estimate of undocumented incident TBIs among US military personnel serving in Iraq and Afghanistan prior to recent militarywide policy changes that served to improve documentation, particularly of mild forms of TBI. The full model estimates that between January 2003 and November 2010, there were 32 822 undocumented incident TBIs among combat-deployed active component personnel, suggesting that more than half of all incident deployment-related TBIs during that period went undocumented. The full model estimates that for each incident TBI documented prior to November 2006, approximately 4 were undocumented. In comparison, after November 2006, roughly 3 in 10 incident TBIs were undocumented. It is anticipated that these remain conservative estimates for a number of reasons. Although the full model was fit to data derived from a period during which TBIs were increasingly diagnosed in a timely manner commensurate with the timely documentation of amputations, TBIs may have still been underdiagnosed or inadequately documented during this period. The effect of this would have been to underestimate incident TBIs. However, delayed documentation after 2010 of prior incident deployment-related TBIs may have had the opposite effect on the estimate. In addition, US military training, armor, and tactics may have improved over the period of interest to prevent TBIs. The model estimates do not reflect the effects of any successful efforts to reduce rates of TBI over the course of the conflicts. Although a major limitation of this analysis is that the models assume a proportional rate of injury over the analysis period, if proportionally greater strides were made in protecting military personnel from TBIs than from amputations over the period of study, the model would offer an even more conservative estimate. However, if changes in enemy weaponry and tactics caused an increase in TBI incidence relative to both deployment figures and amputations between December 2010 and June 2012 as compared to earlier periods, that change would once again have had the opposite effect on the estimate. At the monthly level, the model predictions occasionally contrast markedly with patterns of deploymentrelated TBI documentation observed between 2006 and 2010. As seen in Figure 1, following the November 2006 implementation of TBI policy changes, a large number of incident TBIs were identified en masse when
E61
soldiers returned from major deployments, resulting in spikes such as that in May and June of 2008 when the surge of troops to Iraq began to return home.30 Each model underestimates observed TBI counts during those months when postdeployment screening was accounting for TBIs that had occurred months prior. However, in aggregate, the models are consistent with what is suggested by inspection of Figure 1: TBIs were significantly underdocumented prior to regular screening for brain injuries, and efforts to improve TBI identification became more effective over time. It is also important to note that these estimates do not reflect incident TBIs in the reserve or National Guard components of the military (which, by summation of annual data,35 comprise approximately 20% of all TBIs documented in the military between 2003 and 2012) or in active component units deployed to theaters outside of Iraq and Afghanistan. Finally, the frequent redeployment of those with prior documented TBI may have limited the number of deployed soldiers who were at risk for an incident TBI in the modeling data during the model fit period. Owing to the surveillance and reporting methods used by the DoD, these estimates may not reflect incident cases of deployment-related TBI among those who had experienced a TBI prior to deployment, nor do they account for the increasing number of multiple deploymentrelated TBIs.36 The DoD has made significant advances in improving the identification and documentation of TBIs, and this in turn has facilitated guideline-concordant care to enhance recovery and to help improve the likelihood of return to duty. Despite these advances, the large number of estimated undocumented TBIs remains troubling, particularly as a lack of documentation may affect the risk of reinjury. Although the incidence of deploymentrelated TBI has been shown to be independent of earlier events,37 reinjury tends to result in more and longerterm morbidity as well as an increased suicide risk.12 Therefore, identifying service members with prior TBI, reducing their exposure to the risk of reinjury, and engaging them in early interventions could be a highly useful strategy to prevent further morbidity.36 Indeed, according to deployment guidance in effect since at least March 2010,19 any history of TBI was considered disqualifying for deployment to Iraq or Afghanistan, unless a medical waiver was granted. This guidance states that military personnel “should not deploy (unless a waiver is approved)” if they have any “history of clinically diagnosed traumatic brain injury (mTBI/TBI) of any severity, including mild.”19(p5) Unfortunately, given both previous uneven enforcement of deployment medical standards38 and manpower pressures that led to the frequent granting of medical waivers for otherwise disqualifying conditions,39 www.headtraumarehab.com
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
E62
JOURNAL OF HEAD TRAUMA REHABILITATION/JANUARY–FEBRUARY 2015
personnel with even a clinically diagnosed and documented history of TBI appear to have been at continued risk of reexposure to injury.1 Although published data suggest that only a few thousands service members with prior TBI have been deployed to Iraq or Afghanistan,36 these few thousand may have in fact sustained a second documented TBI associated with deployment. Careful analysis of monthly figures of deploymentrelated TBIs published by the DoD reveals this practice. In its TBI report published January 2010,40 summation of average monthly counts by year reveal that 27 683 cases of deployment-related TBI were reported from 2003 to 2009. Yet, the following month, in February 2010, after a methodological change, total reported TBIs during the same period dropped by 1710, that is, to 25 973. A footnote accompanying the figure noted the updated totals now excluded individuals with a prior TBI encounter,41 presumably those who had previously been counted as incident cases. A similar footnote for data through June 2012 noted the updated totals excluded 3754 such individuals,42 or fully 9% of the 41 500 documented. These observations imply that a sizeable and growing percentage of combat-related TBIs occurred among those with prior documented injury. As even service members with documented prior TBI do not appear to be protected against reinjury, those without documentation are likely to be at similar, if not greater, risk. Just as lack of documentation may affect risk during service, veterans living with persistent TBI symptoms after their period of service may be affected by the lack of documentation of their injuries. Among those with undocumented TBI, some may have experienced symptom-related disciplinary action and dismissal from the military, nullifying their access to anticipated benefits.43 Even veterans separated from the military with an honorable or general discharge may not have been able to take full advantage of Veterans Health Administration benefits reserved for veterans with service connected disabilities if they were unaware of their TBI or unable to prove the service connectedness of their symptoms.44 Proving symptoms are service connected is a necessary step toward obtaining a disability rating, but it is a challenge compounded by the fact that many necessary records have been lost or are otherwise unattainable by veterans.4 Because veterans tend to seek more care outside the Veterans Health Administration system than from within,45 civilian providers may play an especially important role in veterans’ efforts to obtain a disability rating and appropriate care. Retrospectively addressing the barriers faced by veterans who experienced undocumented brain injuries is a challenge but is not unprecedented. Since 1921, federal legislation and subsequent regulations have specified that certain medical conditions can be presumed to be
service connected.46 These policies change how veterans apply for benefits because, under ordinary conditions, veterans need to have a disability rating to receive benefits such as extended and priority care, training, and disability compensation. Obtaining a disability rating usually requires a diagnosis and proof that the injury was caused by service. By presuming service connectedness, veterans need only prove their diagnosis, not how it was caused by their military service. More recent legislation has been passed (such as the Agent Orange Act of 1991) that establishes both presumption of service connectedness and a process by which to determine whether the presumption is warranted.46 For those with service connected TBIs, new rules will soon go into effect that presume the secondary service connectedness of certain conditions such as depression47 —that is, it will be assumed that these conditions have been caused by the service connected TBI, but only if documentation of the TBI and its service connectedness is available. Given the inability of a significant majority of veterans who experienced a TBI prior to the November 2006 policy changes to access any documentation that would prove their service-related injury, further consideration is clearly warranted to determine how best to implement this policy of presumptive service connectedness among veterans who served during that period. While far fewer TBIs appear to have been undocumented after October 2006, considerations for those who served after this date may prove similarly beneficial. CONCLUSION From January 2003 to October 2006, there were an estimated 21 257 to 24 033 undocumented incident TBIs among deployed active component US military personnel serving in Iraq and Afghanistan. To the authors’ knowledge, this is the first published estimate of such undocumented incident deployment-related TBIs. Future estimates may improve upon this analysis by incorporating more granular data. Estimating the total number of subsequent TBIs per incident TBI would be a valuable way to better understand military TBI estimates, as multiple TBIs present unique risks.12 However, this estimate is a useful starting point. It suggests that veterans who experienced an incident TBI in Iraq or Afghanistan prior to November 2006 are between 4 and 5 times as likely to have had their TBI undocumented as to have it documented. Without documentation of an injury, these veterans may have been more likely to sustain a subsequent TBI while in the service and have been less likely to seek appropriate healthcare, receive a disability rating, or take advantage of the healthcare benefits associated with such a disability rating upon separation from service. This analysis also has important
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Estimates of Undocumented Incident Combat-Related TBIs implications for evaluating deployment-related TBI data derived from automated surveillance systems, including the Defense Medical Surveillance System, when conducting research over the periods identified.
E63
Healthcare providers, researchers, and policy makers should be aware of the undocumented nature of most TBIs in this population to appropriately address their needs.
REFERENCES 1. United States Government Accountability Office. GAO-08-615: mental health and traumatic brain injury screening efforts implemented, but consistent predeployment medical record review policies needed. http://www.gao.gov/assets/280/276029.pdf . Published 2008. Accessed January 7, 2014. 2. Warden D. Military TBI during the Iraq and Afghanistan wars. J Head Trauma Rehabil. 2006;21:398–402. 3. Cameron KL, Marshall SW, Sturdivant RX, Lincoln AE. Trends in the incidence of physician-diagnosed mild traumatic brain injury among active duty U.S. military personnel between 1997 and 2007. J Neurotrauma. 2012;29:1313–1321. 4. Miller TC, Zwerdling D. Military still failing to diagnose, treat brain injuries. http://www.npr.org/2010/06/08/127402993/ military-still-failing-to-diagnose-treat-brain-injuries. Published 2010. Accessed January 7, 2014. 5. Wallace D. Improvised explosive devices and traumatic brain injury: the military experience in Iraq and Afghanistan. Aust Psychiatry. 2009;17:218–224. 6. Lechner R, Achatz G, Hauer T, Palm H-G, Lieber A, Willy C. [Patterns and causes of injuries in a contemporary combat environment]. Unfallchirurg. 2010;113:106–113. 7. MacGregor AJ, Dougherty AL, Galarneau MR. Injury-specific correlates of combat-related traumatic brain injury in Operation Iraqi Freedom. J Head Trauma Rehabil. 2011;26:312–318. 8. Cernak I, Noble-Haeusslein LJ. Traumatic brain injury: an overview of pathobiology with emphasis on military populations. J Cereb Blood Flow Metab. 2010;30:255–266. 9. Dolan S, Martindale S, Robinson J, et al. Neuropsychological sequelae of PTSD and TBI Following war deployment among OEF/OIF Veterans. Neuropsychol Rev. 2012;22:21–34. 10. Risdall JE, Menon DK. Traumatic brain injury. Philos Trans R Soc B Biol Sci. 2011;366(1562):241–250. 11. Bryan CJ, Clemans TA. Repetitive traumatic brain injury, psychological symptoms, and suicide risk in a clinical sample of deployed military personnel. JAMA Psychiatry. 2013;70:1–6. 12. Yurgil KA, Barkauskas DA, Vasterling JJ, et al. Association between traumatic brain injury and risk of posttraumatic stress disorder in active-duty marines. JAMA Psychiatry. 2014;71:149–157. 13. Baugh CM, Stamm JM, Riley DO, et al. Chronic traumatic encephalopathy: neurodegeneration following repetitive concussive and subconcussive brain trauma. Brain Imaging Behav. 2012;6:244– 254. 14. Shively S, Scher AI, Perl DP, Diaz-Arrastia R. Dementia resulting from traumatic brain injury: what is the pathology? Arch Neurol. 2012;69:1–7. 15. French LM. Military traumatic brain injury: an examination of important differences. Ann N Y Acad Sci. 2010;1208:38–45. 16. United States Government Accountability Office. GAO-08-276: VA Health Care: mild traumatic brain injury screening and evaluation implemented for OEF/OIF Veterans, but challenges remain. http://www.gao.gov/assets/280/271988.pdf . Published 2008. Accessed January 7, 2014. 17. Kors J. How specialist Town lost his benefits. The Nation. http://www.thenation.com/article/how-specialist-town-losthis-benefits. Published 2007. Accessed January 7, 2014.
18. Miller TC, Zwerdling D. Soldiers with brain trauma denied Purple Hearts, adding insult to injury. http://www.propublica.org/article/ soldiers-with-brain-trauma-denied-purple-hearts-adding-insult-toinjury. Published 2010. Accessed January 7, 2014. 19. United States Central Command (USCENTCOM). Amplification of the minimal standards of fitness for deployment to the CENTCOM AOR; to accompany MOD Ten to USCENTCOM individual protection and individual/unit deployment policy. http://cpol.army.mil/library/mobil/docs/CEW-Guide-AttachIX-C.pdf. Published 2010. Accessed January 7, 2014. 20. Casscells W. Assistant Secretary of Defense (Health Affairs) Memorandum 07-30, subject: traumatic brain injury: definition and reporting. http://health.mil/libraries/HA Policies and Guidelines/07-030.pdf . Published 2007. Accessed January 7, 2014. 21. Wojcik BE, Stein CR, Orosco J, Bagg KA, Humphrey RJ. Creation of an expanded Barell matrix to identify traumatic brain injuries of U.S. Military members. J Def Model Simul Appl Methodol Technol. 2010;7:157–166. 22. Lynn WJ. Deputy Secretary of Defense Directive-type memorandum 09-33, subject: policy guidance for management of concussion/mild traumatic brain injury in the deployed setting. http://dod-executiveagent.osd.mil/attachment.aspx?DOC= DTM-09-033.pdf . Published 2010. Accessed January 7, 2014. 23. Rooney JA. Office of the Under Secretary of Defense (Personnel and Readiness) report to Congress: National Defense Authorization Act for fiscal year 2012. Section 724: report on memorandum regarding traumatic brain injuries. http://tricare.mil/tma/ congressionalinformation/downloads/Report%20on%20 Memorandum%20Regarding%20TBI.pdf. Published 2012. Accessed January 7, 2014. 24. Edgar E. A look inside CENTCOM medical policies. http:// www.mc4.army.mil/story/A_Look_Inside_CENTCOM_ Medical_Policies. Published 2012. Accessed January 7, 2014. 25. Snethen B, Steen R. Army hospital saves time with medical record system upgrade in Afghanistan. http://www.army.mil/ article/45167. Published 2010. Accessed January 7, 2014. 26. Rubertone MV, Brundage JF. The Defense Medical Surveillance System and the Department of Defense serum repository: glimpses of the future of public health surveillance. Am J Public Health. 2002;92:1900–1904. 27. Army Medical Surveillance Activity. Deployment-related conditions of special surveillance interest, U.S. Armed Forces, by month and service, January 2003-June 2005. Med Surveill Mon Rep. 2005;11(3):22–24. 28. Armed Forces Health Surveillance Center. Deployment-related conditions of special surveillance interest, U.S. Armed Forces, by month and service, January 2003-September 2007. Med Surveill Mon Rep. 2007;14(6):30–33. 29. Armed Forces Health Surveillance Center. Deriving case counts from medical encounter data: considerations when interpreting health surveillance reports. Med Surveill Mon Rep. 2009;16(12): 2–8.
www.headtraumarehab.com Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
E64
JOURNAL OF HEAD TRAUMA REHABILITATION/JANUARY–FEBRUARY 2015
30. Belasco A. The cost of Iraq, Afghanistan, and other global war on terror operations since 9/11. http://www.fas.org/sgp/ crs/natsec/RL33110.pdf. Published 2011. Accessed January 7, 2014. 31. O’Hanlon M, Livingston I. Iraq Index: tracking variables of reconstruction & security in post-Saddam Iraq. http://www.brookings.edu/∼/media/Centers/saban/iraq index/index20111130.pdf. Published 2011. Accessed January 7, 2014. 32. Livingston I, O’Hanlon M. Afghanistan Index: also including selected data on Pakistan. http://www.brookings.edu/∼ /media/programs/foreign%20policy/afghanistan%20index/ index20120731.pdf . Published 2012. Accessed January 7, 2014. 33. O’Neil C. Changing of the guard: a look back at 10 years of war, military health care. http://hiddensurge.nationalsecurityzone.org/ nsjihs special pages/changing-of-the-guard. Published 2012. Accessed January 7, 2014. 34. Rauh MJ, Aralis HJ, Melcer T, et al. Effect of traumatic brain injury among U.S. service members with amputation. J Rehabil Res Dev. 2013;50:161–172. 35. Defense and Veterans Brain Injury Center. DoD worldwide numbers for traumatic brain injury. http://www.dvbic.org/ dod-worldwide-numbers-tbi. Published 2013. Accessed January 7, 2014. 36. Hampton T. Traumatic brain injury a growing problem among troops serving in today’s wars. JAMA. 2011;306:477–479. 37. Vanderploeg RD, Belanger HG, Horner RD, et al. Health outcomes associated with military deployment: mild traumatic brain injury, blast, trauma, and combat associations in the Florida National Guard. Arch Phys Med Rehabil. 2012;93: 1887–1895. 38. Hicks MHR. Mental health screening and coordination of care for soldiers deployed to Iraq and Afghanistan. Am J Psychiatry. 2011;168:341–343. 39. Nevin RL. Mental health standards for combat deployment. Psychiatr Serv. 2011;62:805.
40. Armed Forces Health Surveillance Center. Deployment-related conditions of special surveillance interest, U.S. Armed Forces, by month and service, January 2003-December 2009. Med Surveill Mon Rep. 2010;17(1):18–19. 41. Armed Forces Health Surveillance Center. Deployment-related conditions of special surveillance interest, U.S. Armed Forces, by month and service, January 2003-January 2010. Med Surveill Mon Rep. 2010;17(2):21–23. 42. Armed Forces Health Surveillance Center. Deployment-related conditions of special surveillance interest, U.S. Armed Forces, by month and service, January 2003-June 2012. Med Surveill Mon Rep. 2012;19(7):24–27. 43. Ader M, Cuthbert R, Hoechst K, Simon EH, Strassburger Z, Wishnie M. Casting troops aside: the United States Military’s illegal personality disorder discharge problem. http://www.vva .org/PPD-Documents/WhitePaper.pdf. Published 2012. Accessed January 7, 2014. 44. US Department of Veterans Affairs. Veterans benefits timetable: information for veterans recently separated from active military service. http://www.vba.va.gov/pubs/forms/VBA-21-0501-ARE .pdf. Published 2009. Accessed January 7, 2014. 45. Auerbach DI, Weeks WB, Brantley I. Health care spending and efficiency in the U.S. Department of Veterans Affairs. http://www.rand.org/content/dam/rand/pubs/research reports/RR200/RR285/RAND RR285.pdf . Published 2013. Accessed January 7, 2014. 46. Panangala SV, Scott C, Weimer DR. Veterans Affairs: presumptive service connection and disability compensation. http://www.dtic .mil/cgi-bin/GetTRDoc?AD=ADA543665. Published 2011. Accessed January 7, 2014. 47. US Department of Veterans Affairs. Secondary service connection for diagnosable illnesses associated with traumatic brain injury. Fed Regist. 2013;78(242):76196–76209. https://www.federalregister.gov/articles/2013/12/17/2013-29911/ secondary-service-connection-for-diagnosable-illnesses-associatedwith-traumatic-brain-injury. Published 2013. Accessed January 7, 2014.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Population Estimates of Undocumented Incident Traumatic Brain Injuries Among Combat-Deployed US Military Personnel Rachel P. Chase, BSc; Remington L. Nevin, MD, MPH Objective: To estimate the number of undocumented incident traumatic brain injuries (TBIs) among active component US military personnel serving in Iraq and Afghanistan prior to policy changes implemented in late 2006 and 2010 that improved TBI documentation. Methods: Negative binomial regression was used to model monthly incident TBI counts between December 2010 and June 2012 (N = 19) and then estimate expected monthly counts of incident TBIs during 2 periods: January 2003-October 2006 and November 2006-November 2010. Monthly amputation counts from Department of Defense surveillance data were used as a proxy for changing injury rates. Monthly active component deployment estimates derived from the Congressional Research Service, Brookings Institution, and Defense Manpower Data Center were used to estimate the size of the at-risk population each month. The difference between expected monthly incident TBI counts and reported counts is presented as the estimated number of undocumented incident TBIs. Results: The full model estimates that 21 257 active component military personnel experienced undocumented incident TBIs while deployed in Iraq or Afghanistan between January 2003 and October 2006, more than 4 times the 5272 incident TBIs documented during that period. Conclusions: A sizeable majority of Iraq and Afghanistan combat veterans who experienced incident TBI while deployed prior to November 2006 are likely to have had their injuries undocumented, creating challenges for clinical care, disability evaluation, and future research. Key words: Afghan war, Iraq war, late diagnosis, military, traumatic brain injury
T
RAUMATIC BRAIN INJURIES (TBIs), particularly those classified as “mild,” have been regarded as “signature injuries” and “invisible wounds” of the conflicts in Iraq and Afghanistan due to both their pernicious effects and their systematic underdocumentation, particularly at the outset of each war.
Author Affiliations: Departments of International Health (Ms Chase) and Mental Health (Dr Nevin), Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland. The authors thank Shannon McMahon and Roger Peng for their helpful comments during manuscript preparation and review and the 3 anonymous reviewers for their valuable recommendations. The authors are also indebted to the staff of the Armed Forces Health Surveillance Center for providing the monthly traumatic brain injury and amputation count data on which these analyses are based. Dr Nevin reports serving as consultant and expert witness in cases involving claims of antimalarial toxicity, a relevant conflict given the potential for confounding of this condition with traumatic brain injury in military populations. Ms Chase reports no conflicts of interest. This work was conducted without external funding. Corresponding Author: Rachel P. Chase, BSc, Department of International Health, Johns Hopkins Bloomberg School of Public Health, 615 N Wolfe St, Baltimore, MD 21205 ([email protected]). DOI: 10.1097/HTR.0000000000000061
Prior to military-wide policy changes following legislation passed in October 2006,1 US military personnel were not regularly evaluated for TBIs unless they were evacuated to a medical center such as the Walter Reed Army Medical Center after severe injuries.2 Systematic predeployment and in-theater TBI assessments were initiated only in November 2006 after investigations revealed high rates of previously undocumented TBIs, especially among patients with blast-induced injuries.2 Efforts to better document TBIs led to dramatically increased rates of TBI incidence beginning in late 2006,3 but it is not clear how many TBIs had gone undocumented prior to these changes or how many went undocumented as the improved policies went into effect. As of 2010, estimates of undocumented TBIs were reported in the popular media as numbering into the “tens of thousands”4 of cases. However, as a result of historical underdocumentation, significant uncertainties regarding the true historical burden of deployment-related TBI persist. Estimating the total number of incident TBIs over this period is difficult not only due to changing documentation policies and practices but also due to temporal fluctuations in rates of underlying injury causing events, which are typically subject to military E57
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
E58
JOURNAL OF HEAD TRAUMA REHABILITATION/JANUARY–FEBRUARY 2015
secrecy and not publicly available to researchers. In the absence of such information, such fluctuations may be accounted for through the use of an available proxy measure of combat injury. Recent trends support the use of amputations as one such measure. While steady improvements in the use of armored vehicles and body armor as well as forward medical assistance have lessened mortality in recent conflicts as compared with previous wars,5 rates of certain injuries including amputations have increased.6 The majority of combat-related injuries within the US military now result from the blast effects of improvised explosive devices and other explosive devices.7 Blast effects are often categorized as primary (caused by the blast wave itself), secondary (caused by items propelled by the blast, such as shrapnel), or tertiary (caused by the body being thrown against other objects).8 Advances in the use of protective equipment including body armor and ballistic goggles have been especially effective at limiting the incidence of penetrating injuries from many secondary and tertiary blast effects but have left the unprotected extremities susceptible to penetrating injury and the head susceptible to nonpenetrative injuries to the brain.2 While mild blast injuries to the brain are often hard to discern and diagnose,7 injuries to other parts of the head, such as the scalp, are easy to discern and are not the focus of this article. Blast injuries to the extremities are highly visible injuries that in their most extreme form result in amputation. Amputations have been documented consistently over the course of the conflicts in Iraq and Afghanistan and, unlike counts of injury causing events, are publicly available. On this basis, amputations should prove an accessible and reasonably valid and reliable proxy for changing combat injury rates when studying TBI incidence. Brain injuries are important to diagnose and manage because they may cause a wide range of short- and longterm problems. These include neuropsychiatric and somatic symptoms such as memory loss, cognitive deficits, mood volatility, substance use disorders, personality changes, sleep difficulties, and sensory deficits.9 Blastrelated and multiple TBIs are associated with visual and auditory deficits, posttraumatic epilepsy, headaches, major depression, and suicide risk.10,11 In deployed settings, brain injuries are associated with an increased risk of posttraumatic stress disorder,12 and there are growing concerns that recurrent and blast-related TBIs are associated with progressive neurodegenerative processes such as chronic traumatic encephalopathy13 and disorders resembling Alzheimer disease.14 The burden of disability from combat-related TBIs may be more debilitating than other TBIs due to both the extreme nature of the injuryproducing event15 and the limited ability to diagnose and treat the injury in a timely manner, which may lead to repeat exposure to injury.16 In some cases, symptoms
of TBI may complicate processes of postdeployment recovery and reintegration.17 Among veterans, a formally diagnosed and documented TBI is critical in adjudicating individual medical disability claims, receiving health benefits, and facilitating military distinction.18 Similarly, among military personnel remaining on active duty, documentation of prior TBI may preclude subsequent deployment, thus minimizing the risk of reinjury.19 Military personnel with a documented TBI and experiencing persistent symptoms may be eligible for medical retirement, whereas those lacking documentation of TBI risk misattribution of such symptoms to other conditions such as malingering4,14 or personality disorder17 that may preclude eligibility for medical retirement. Accurate population estimates of the burden of deployment-related TBIs are also critical to informing an appropriate public health response. Since guidance issued in late 2006—first in the form of legislation, followed shortly by Department of Defense (DoD) policy implementation1,20 —the US military has attempted to provide more accurate counts through the use of regular screenings and automated medical surveillance.21 Between 2006 and 2010, the DoD made several improvements to policies and practices associated with TBI documentation. These included publishing a standard definition for TBI (in October 2007),20 requiring that all military personnel be evaluated using a standardized neuropsychological assessment (in March 2008),1 more consistently tracking and treating TBIs in the field (in June and December 2010),22–24 and progressively deploying electronic medical record systems in field settings.25 However, despite these improvements, considerable disagreement and controversy remain regarding the accuracy of historical deployment-related TBI count data.4 METHODS In this analysis, negative binomial regression was used to model monthly incident TBI counts between December 2010 and June 2012 (N = 19) and then estimate expected monthly counts of incident TBIs during 2 periods: January 2003-October 2006 and November 2006-November 2010. Two models are presented in which monthly active component deployment estimates served as the size of the at-risk population each month. In the full model, monthly counts of amputations were included as a proxy for changing injury rates, as opposed to the alternative model in which these counts were excluded. The model was then applied to earlier observations to calculate predicted incident TBI counts. The difference between the number of documented incident TBIs and the estimated number each month was interpreted as the number of undocumented incident
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Estimates of Undocumented Incident Combat-Related TBIs TBIs that month. In the case of negative estimates, the estimate was interpreted as the excess number of TBIs documented that month. The full negative binomial regression model took the form of: log(y) = b 0 + b 1 x1 + log(x2 ) + ε where y is the estimated number of incident TBIs in a given month, x1 is the number of amputations, and x2 is the estimated number of deployed active component military personnel (in thousands). For illustration, a sample calculation from the full model for January 2005, when 13 amputations were recorded, an estimated 108 674 personnel were at risk, and 81 TBIs were documented, is provided in the following equation using the estimates of b0 and b1 presented in Table 1: log(y ) = b 0 + b 1 x1 + log(x2 ) = 1.569 + 0.013(13) +log(108.674) = 6.426 By exponentiation, the estimate of the total number of incident TBIs in January 2005 is found to be y = 617. Since 81 of these TBIs were documented, it is estimated that approximately 617 − 81 = 536 incident TBIs were not documented in January 2005 in this population. The incident TBI and amputation data were obtained from the DoD. The DoD maintains a passive medical surveillance system, known as the Defense Medical Surveillance System, which integrates deployment and medical diagnosis data among all active component military personnel.26 The DoD has used the Defense Medical Surveillance System to publish monthly
E59
tabulations of deployment-related conditions of special surveillance interest, including amputations since May 200527 and incident cases of TBI of any severity since September 2007.28 Methods used in these tabulations, including relevant International Classification of Diseases, Ninth Revision, Clinical Modification (ICD9-CM), medical diagnosis codes used to define an incident case of TBI, including ICD-9-CM codes 310.2, 800-801, 803-804, 850-854, 907.0, 950.1-950.3, 959.01, and supplementary DoD ICD-9-CM codes V15.5_19, V15.5_A-F, V15.52_0-9, V15.52_A-F, V15.59_1-9, V15.59_A-F, have been previously described.29 Medical diagnosis data are derived primarily from electronic medical records from medical encounters and patient charts at fixed facilities upon return from deployment, supplemented by a small but rising percentage of diagnosis data derived from electronic medical records at temporary medical facilities at deployed locations.29 Estimates of monthly combat-deployed personnel in Iraq and Afghanistan published by the Congressional Research Service30 and the Brookings Institution,31,32 as well as deployment figures from the Defense Manpower Data Center,33 were used to create the exposure variable estimating the size of the at-risk population. Total combat-deployed personnel estimates (as reported by the Congressional Research Service and the Brookings Institution) were multiplied by the proportion of all deployed personnel who were active component each month (as reported by the Defense Manpower Data Center). The resulting monthly estimates of active component combat-deployed personnel in Iraq and Afghanistan were used for modeling.
Estimated number of undocumented incident TBIs, full and alternative negative binomial models, with model estimates and fit statisticsa TABLE 1
Estimated number of undocumented incident TBIs January 2003-October 2006 November 2006-November 2010 Model estimates Constant (standard error) Amputation coefficient (standard error) Dispersion parameter (standard error) Model fit statistics Log likelihood AIC BIC Deviance residuals Degrees of freedom Deviance to degrees of freedom ratio
Full model
Alternative model
21 257 11 565
24 033 15 123
1.569 (0.114) 0.013 (0.005) 0.040 (0.013)
1.826 (0.053)
−117.3b 12.6 −30.9 19.2 17 1.1
−119.9 12.7 −33.8 19.2 18 1.1
0.052 (0.017)
Abbreviations: AIC, Akaike information criteria; BIC, Bayesian information criteria; TBI, traumatic brain injury. a Alternative model excludes monthly amputation counts as a parameter. b Significantly better than the alternative model (P < .05).
www.headtraumarehab.com Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
E60
JOURNAL OF HEAD TRAUMA REHABILITATION/JANUARY–FEBRUARY 2015
Including monthly deployment figures as an exposure variable or offset is appropriate, as troop surges and drawdowns resulted in substantial changes in the size of the at-risk population between 2003 and 2012, and the negative binomial model would otherwise have to assume a constant population size. In the full model, counts of monthly deployment-related amputations were included as a predictor of each month’s incident TBI counts. As described earlier, including amputation counts in the model as a proxy for injury-causing events is appropriate, given strong clinical and ecological evidence of common mechanisms of injury,7,34 which suggest that trends in the burden of deployment-related TBI should be correlated with those of amputations. The alternative model that includes only deployment figures has been included for comparison with the full model (see Table 1). Monthly incident TBI counts between December 2010 and June 2012 (inclusive) were modeled using negative binomial regression rather than Poisson regression to account for overdispersion in the data (mean = 591; variance = 27 829). The full model with amputations as a parameter and deployment estimates as an offset was compared with the alternative model using log likelihood, Akaike and Bayesian information criteria, and the ratio of deviance residuals to degrees of freedom. Standard errors in model estimates are reported (see Table 1). However, total population data were used in place of randomly selected sample data, so these errors are not used to infer confidence intervals or similar statistics that rely on random sampling assumptions. Only publicly available and de-identified count data were used in this analysis. This study was reviewed by the institutional review board of the Johns Hopkins Bloomberg School of Public Health and deemed exempt. All analyses were conducted using Stata (Version 13.1; StataCorp, College Station, Texas).
incident TBIs would have been identified between November 2006 and November 2010 beyond the 26 904 that were documented during that time. To demonstrate the statistical influence of including amputation counts in the final model, the results of the alternative model that omits these are presented. This alternative, which is nested within the full model, eliminates amputation counts from the analysis and therefore assumes constant rates of combat injury. This alternative model provides somewhat similar estimates of undocumented incident TBIs. Between January 2003 and October 2006, the alternative model predicts 24 033 additional incident TBIs. Between November 2006 and November 2010, it predicts 15 123 additional incident TBIs. The alternative model might be considered an improvement over the full model if changes in the number of amputations each month were uncorrelated with changes in the number of TBIs each month, or if changes in the true relationship between amputations and TBIs over time varied sufficiently that a constant rate of amputations would serve as a better approximation of true trends. Neither of these alternative assumptions is strongly supported, as during the 2 periods of relatively stable TBI documentation practices (before policy changes in November 2006 and after the last major changes in December 2010), a strong association between counts of amputations and counts of incident TBIs is observed. The association is visible by inspection (see Figure 1) and analysis with a simple Pearson statistic (r = 0.55 before the policy changes and r = 0.71 after). Although the full model is most justified on both empirical and theoretical grounds, certain results are qualitatively robust across models. Both models estimate that between January 2003 and October 2006, approximately 4 to 5 TBIs were undocumented for each TBI
RESULTS The 2 statistical models exhibited comparable fit statistics, and neither model distinguished itself by dramatically improved Akaike and Bayesian information criteria or deviance to degrees of freedom ratio, although the full model fit was significantly better by the log likelihood test (P < .05). Drawing on the full model that includes monthly amputation counts as a proxy for combat injury, an estimated 32 822 incident TBIs were undocumented among combat-deployed personnel from January 2003 to November 2010 (Table 1). This model estimates that if current policies were applied to past cases, 21 257 additional incident TBIs would have been documented between January 2003 and October 2006 in addition to the 5272 TBIs that were documented. The full model also suggests that an additional 11 565
Figure 1. Predicted active component monthly incident deployment-related TBIs, according to the full model, with documented monthly incident TBI and amputation counts, and estimated deployment figures. TBI indicates traumatic brain injury.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Estimates of Undocumented Incident Combat-Related TBIs documented among deployed personnel. Both models also suggest major improvements in documentation of incident TBIs following the November 2006 implementation of policy changes. DISCUSSION This analysis provides the first estimate of undocumented incident TBIs among US military personnel serving in Iraq and Afghanistan prior to recent militarywide policy changes that served to improve documentation, particularly of mild forms of TBI. The full model estimates that between January 2003 and November 2010, there were 32 822 undocumented incident TBIs among combat-deployed active component personnel, suggesting that more than half of all incident deployment-related TBIs during that period went undocumented. The full model estimates that for each incident TBI documented prior to November 2006, approximately 4 were undocumented. In comparison, after November 2006, roughly 3 in 10 incident TBIs were undocumented. It is anticipated that these remain conservative estimates for a number of reasons. Although the full model was fit to data derived from a period during which TBIs were increasingly diagnosed in a timely manner commensurate with the timely documentation of amputations, TBIs may have still been underdiagnosed or inadequately documented during this period. The effect of this would have been to underestimate incident TBIs. However, delayed documentation after 2010 of prior incident deployment-related TBIs may have had the opposite effect on the estimate. In addition, US military training, armor, and tactics may have improved over the period of interest to prevent TBIs. The model estimates do not reflect the effects of any successful efforts to reduce rates of TBI over the course of the conflicts. Although a major limitation of this analysis is that the models assume a proportional rate of injury over the analysis period, if proportionally greater strides were made in protecting military personnel from TBIs than from amputations over the period of study, the model would offer an even more conservative estimate. However, if changes in enemy weaponry and tactics caused an increase in TBI incidence relative to both deployment figures and amputations between December 2010 and June 2012 as compared to earlier periods, that change would once again have had the opposite effect on the estimate. At the monthly level, the model predictions occasionally contrast markedly with patterns of deploymentrelated TBI documentation observed between 2006 and 2010. As seen in Figure 1, following the November 2006 implementation of TBI policy changes, a large number of incident TBIs were identified en masse when
E61
soldiers returned from major deployments, resulting in spikes such as that in May and June of 2008 when the surge of troops to Iraq began to return home.30 Each model underestimates observed TBI counts during those months when postdeployment screening was accounting for TBIs that had occurred months prior. However, in aggregate, the models are consistent with what is suggested by inspection of Figure 1: TBIs were significantly underdocumented prior to regular screening for brain injuries, and efforts to improve TBI identification became more effective over time. It is also important to note that these estimates do not reflect incident TBIs in the reserve or National Guard components of the military (which, by summation of annual data,35 comprise approximately 20% of all TBIs documented in the military between 2003 and 2012) or in active component units deployed to theaters outside of Iraq and Afghanistan. Finally, the frequent redeployment of those with prior documented TBI may have limited the number of deployed soldiers who were at risk for an incident TBI in the modeling data during the model fit period. Owing to the surveillance and reporting methods used by the DoD, these estimates may not reflect incident cases of deployment-related TBI among those who had experienced a TBI prior to deployment, nor do they account for the increasing number of multiple deploymentrelated TBIs.36 The DoD has made significant advances in improving the identification and documentation of TBIs, and this in turn has facilitated guideline-concordant care to enhance recovery and to help improve the likelihood of return to duty. Despite these advances, the large number of estimated undocumented TBIs remains troubling, particularly as a lack of documentation may affect the risk of reinjury. Although the incidence of deploymentrelated TBI has been shown to be independent of earlier events,37 reinjury tends to result in more and longerterm morbidity as well as an increased suicide risk.12 Therefore, identifying service members with prior TBI, reducing their exposure to the risk of reinjury, and engaging them in early interventions could be a highly useful strategy to prevent further morbidity.36 Indeed, according to deployment guidance in effect since at least March 2010,19 any history of TBI was considered disqualifying for deployment to Iraq or Afghanistan, unless a medical waiver was granted. This guidance states that military personnel “should not deploy (unless a waiver is approved)” if they have any “history of clinically diagnosed traumatic brain injury (mTBI/TBI) of any severity, including mild.”19(p5) Unfortunately, given both previous uneven enforcement of deployment medical standards38 and manpower pressures that led to the frequent granting of medical waivers for otherwise disqualifying conditions,39 www.headtraumarehab.com
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
E62
JOURNAL OF HEAD TRAUMA REHABILITATION/JANUARY–FEBRUARY 2015
personnel with even a clinically diagnosed and documented history of TBI appear to have been at continued risk of reexposure to injury.1 Although published data suggest that only a few thousands service members with prior TBI have been deployed to Iraq or Afghanistan,36 these few thousand may have in fact sustained a second documented TBI associated with deployment. Careful analysis of monthly figures of deploymentrelated TBIs published by the DoD reveals this practice. In its TBI report published January 2010,40 summation of average monthly counts by year reveal that 27 683 cases of deployment-related TBI were reported from 2003 to 2009. Yet, the following month, in February 2010, after a methodological change, total reported TBIs during the same period dropped by 1710, that is, to 25 973. A footnote accompanying the figure noted the updated totals now excluded individuals with a prior TBI encounter,41 presumably those who had previously been counted as incident cases. A similar footnote for data through June 2012 noted the updated totals excluded 3754 such individuals,42 or fully 9% of the 41 500 documented. These observations imply that a sizeable and growing percentage of combat-related TBIs occurred among those with prior documented injury. As even service members with documented prior TBI do not appear to be protected against reinjury, those without documentation are likely to be at similar, if not greater, risk. Just as lack of documentation may affect risk during service, veterans living with persistent TBI symptoms after their period of service may be affected by the lack of documentation of their injuries. Among those with undocumented TBI, some may have experienced symptom-related disciplinary action and dismissal from the military, nullifying their access to anticipated benefits.43 Even veterans separated from the military with an honorable or general discharge may not have been able to take full advantage of Veterans Health Administration benefits reserved for veterans with service connected disabilities if they were unaware of their TBI or unable to prove the service connectedness of their symptoms.44 Proving symptoms are service connected is a necessary step toward obtaining a disability rating, but it is a challenge compounded by the fact that many necessary records have been lost or are otherwise unattainable by veterans.4 Because veterans tend to seek more care outside the Veterans Health Administration system than from within,45 civilian providers may play an especially important role in veterans’ efforts to obtain a disability rating and appropriate care. Retrospectively addressing the barriers faced by veterans who experienced undocumented brain injuries is a challenge but is not unprecedented. Since 1921, federal legislation and subsequent regulations have specified that certain medical conditions can be presumed to be
service connected.46 These policies change how veterans apply for benefits because, under ordinary conditions, veterans need to have a disability rating to receive benefits such as extended and priority care, training, and disability compensation. Obtaining a disability rating usually requires a diagnosis and proof that the injury was caused by service. By presuming service connectedness, veterans need only prove their diagnosis, not how it was caused by their military service. More recent legislation has been passed (such as the Agent Orange Act of 1991) that establishes both presumption of service connectedness and a process by which to determine whether the presumption is warranted.46 For those with service connected TBIs, new rules will soon go into effect that presume the secondary service connectedness of certain conditions such as depression47 —that is, it will be assumed that these conditions have been caused by the service connected TBI, but only if documentation of the TBI and its service connectedness is available. Given the inability of a significant majority of veterans who experienced a TBI prior to the November 2006 policy changes to access any documentation that would prove their service-related injury, further consideration is clearly warranted to determine how best to implement this policy of presumptive service connectedness among veterans who served during that period. While far fewer TBIs appear to have been undocumented after October 2006, considerations for those who served after this date may prove similarly beneficial. CONCLUSION From January 2003 to October 2006, there were an estimated 21 257 to 24 033 undocumented incident TBIs among deployed active component US military personnel serving in Iraq and Afghanistan. To the authors’ knowledge, this is the first published estimate of such undocumented incident deployment-related TBIs. Future estimates may improve upon this analysis by incorporating more granular data. Estimating the total number of subsequent TBIs per incident TBI would be a valuable way to better understand military TBI estimates, as multiple TBIs present unique risks.12 However, this estimate is a useful starting point. It suggests that veterans who experienced an incident TBI in Iraq or Afghanistan prior to November 2006 are between 4 and 5 times as likely to have had their TBI undocumented as to have it documented. Without documentation of an injury, these veterans may have been more likely to sustain a subsequent TBI while in the service and have been less likely to seek appropriate healthcare, receive a disability rating, or take advantage of the healthcare benefits associated with such a disability rating upon separation from service. This analysis also has important
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Estimates of Undocumented Incident Combat-Related TBIs implications for evaluating deployment-related TBI data derived from automated surveillance systems, including the Defense Medical Surveillance System, when conducting research over the periods identified.
E63
Healthcare providers, researchers, and policy makers should be aware of the undocumented nature of most TBIs in this population to appropriately address their needs.
REFERENCES 1. United States Government Accountability Office. GAO-08-615: mental health and traumatic brain injury screening efforts implemented, but consistent predeployment medical record review policies needed. http://www.gao.gov/assets/280/276029.pdf . Published 2008. Accessed January 7, 2014. 2. Warden D. Military TBI during the Iraq and Afghanistan wars. J Head Trauma Rehabil. 2006;21:398–402. 3. Cameron KL, Marshall SW, Sturdivant RX, Lincoln AE. Trends in the incidence of physician-diagnosed mild traumatic brain injury among active duty U.S. military personnel between 1997 and 2007. J Neurotrauma. 2012;29:1313–1321. 4. Miller TC, Zwerdling D. Military still failing to diagnose, treat brain injuries. http://www.npr.org/2010/06/08/127402993/ military-still-failing-to-diagnose-treat-brain-injuries. Published 2010. Accessed January 7, 2014. 5. Wallace D. Improvised explosive devices and traumatic brain injury: the military experience in Iraq and Afghanistan. Aust Psychiatry. 2009;17:218–224. 6. Lechner R, Achatz G, Hauer T, Palm H-G, Lieber A, Willy C. [Patterns and causes of injuries in a contemporary combat environment]. Unfallchirurg. 2010;113:106–113. 7. MacGregor AJ, Dougherty AL, Galarneau MR. Injury-specific correlates of combat-related traumatic brain injury in Operation Iraqi Freedom. J Head Trauma Rehabil. 2011;26:312–318. 8. Cernak I, Noble-Haeusslein LJ. Traumatic brain injury: an overview of pathobiology with emphasis on military populations. J Cereb Blood Flow Metab. 2010;30:255–266. 9. Dolan S, Martindale S, Robinson J, et al. Neuropsychological sequelae of PTSD and TBI Following war deployment among OEF/OIF Veterans. Neuropsychol Rev. 2012;22:21–34. 10. Risdall JE, Menon DK. Traumatic brain injury. Philos Trans R Soc B Biol Sci. 2011;366(1562):241–250. 11. Bryan CJ, Clemans TA. Repetitive traumatic brain injury, psychological symptoms, and suicide risk in a clinical sample of deployed military personnel. JAMA Psychiatry. 2013;70:1–6. 12. Yurgil KA, Barkauskas DA, Vasterling JJ, et al. Association between traumatic brain injury and risk of posttraumatic stress disorder in active-duty marines. JAMA Psychiatry. 2014;71:149–157. 13. Baugh CM, Stamm JM, Riley DO, et al. Chronic traumatic encephalopathy: neurodegeneration following repetitive concussive and subconcussive brain trauma. Brain Imaging Behav. 2012;6:244– 254. 14. Shively S, Scher AI, Perl DP, Diaz-Arrastia R. Dementia resulting from traumatic brain injury: what is the pathology? Arch Neurol. 2012;69:1–7. 15. French LM. Military traumatic brain injury: an examination of important differences. Ann N Y Acad Sci. 2010;1208:38–45. 16. United States Government Accountability Office. GAO-08-276: VA Health Care: mild traumatic brain injury screening and evaluation implemented for OEF/OIF Veterans, but challenges remain. http://www.gao.gov/assets/280/271988.pdf . Published 2008. Accessed January 7, 2014. 17. Kors J. How specialist Town lost his benefits. The Nation. http://www.thenation.com/article/how-specialist-town-losthis-benefits. Published 2007. Accessed January 7, 2014.
18. Miller TC, Zwerdling D. Soldiers with brain trauma denied Purple Hearts, adding insult to injury. http://www.propublica.org/article/ soldiers-with-brain-trauma-denied-purple-hearts-adding-insult-toinjury. Published 2010. Accessed January 7, 2014. 19. United States Central Command (USCENTCOM). Amplification of the minimal standards of fitness for deployment to the CENTCOM AOR; to accompany MOD Ten to USCENTCOM individual protection and individual/unit deployment policy. http://cpol.army.mil/library/mobil/docs/CEW-Guide-AttachIX-C.pdf. Published 2010. Accessed January 7, 2014. 20. Casscells W. Assistant Secretary of Defense (Health Affairs) Memorandum 07-30, subject: traumatic brain injury: definition and reporting. http://health.mil/libraries/HA Policies and Guidelines/07-030.pdf . Published 2007. Accessed January 7, 2014. 21. Wojcik BE, Stein CR, Orosco J, Bagg KA, Humphrey RJ. Creation of an expanded Barell matrix to identify traumatic brain injuries of U.S. Military members. J Def Model Simul Appl Methodol Technol. 2010;7:157–166. 22. Lynn WJ. Deputy Secretary of Defense Directive-type memorandum 09-33, subject: policy guidance for management of concussion/mild traumatic brain injury in the deployed setting. http://dod-executiveagent.osd.mil/attachment.aspx?DOC= DTM-09-033.pdf . Published 2010. Accessed January 7, 2014. 23. Rooney JA. Office of the Under Secretary of Defense (Personnel and Readiness) report to Congress: National Defense Authorization Act for fiscal year 2012. Section 724: report on memorandum regarding traumatic brain injuries. http://tricare.mil/tma/ congressionalinformation/downloads/Report%20on%20 Memorandum%20Regarding%20TBI.pdf. Published 2012. Accessed January 7, 2014. 24. Edgar E. A look inside CENTCOM medical policies. http:// www.mc4.army.mil/story/A_Look_Inside_CENTCOM_ Medical_Policies. Published 2012. Accessed January 7, 2014. 25. Snethen B, Steen R. Army hospital saves time with medical record system upgrade in Afghanistan. http://www.army.mil/ article/45167. Published 2010. Accessed January 7, 2014. 26. Rubertone MV, Brundage JF. The Defense Medical Surveillance System and the Department of Defense serum repository: glimpses of the future of public health surveillance. Am J Public Health. 2002;92:1900–1904. 27. Army Medical Surveillance Activity. Deployment-related conditions of special surveillance interest, U.S. Armed Forces, by month and service, January 2003-June 2005. Med Surveill Mon Rep. 2005;11(3):22–24. 28. Armed Forces Health Surveillance Center. Deployment-related conditions of special surveillance interest, U.S. Armed Forces, by month and service, January 2003-September 2007. Med Surveill Mon Rep. 2007;14(6):30–33. 29. Armed Forces Health Surveillance Center. Deriving case counts from medical encounter data: considerations when interpreting health surveillance reports. Med Surveill Mon Rep. 2009;16(12): 2–8.
www.headtraumarehab.com Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
E64
JOURNAL OF HEAD TRAUMA REHABILITATION/JANUARY–FEBRUARY 2015
30. Belasco A. The cost of Iraq, Afghanistan, and other global war on terror operations since 9/11. http://www.fas.org/sgp/ crs/natsec/RL33110.pdf. Published 2011. Accessed January 7, 2014. 31. O’Hanlon M, Livingston I. Iraq Index: tracking variables of reconstruction & security in post-Saddam Iraq. http://www.brookings.edu/∼/media/Centers/saban/iraq index/index20111130.pdf. Published 2011. Accessed January 7, 2014. 32. Livingston I, O’Hanlon M. Afghanistan Index: also including selected data on Pakistan. http://www.brookings.edu/∼ /media/programs/foreign%20policy/afghanistan%20index/ index20120731.pdf . Published 2012. Accessed January 7, 2014. 33. O’Neil C. Changing of the guard: a look back at 10 years of war, military health care. http://hiddensurge.nationalsecurityzone.org/ nsjihs special pages/changing-of-the-guard. Published 2012. Accessed January 7, 2014. 34. Rauh MJ, Aralis HJ, Melcer T, et al. Effect of traumatic brain injury among U.S. service members with amputation. J Rehabil Res Dev. 2013;50:161–172. 35. Defense and Veterans Brain Injury Center. DoD worldwide numbers for traumatic brain injury. http://www.dvbic.org/ dod-worldwide-numbers-tbi. Published 2013. Accessed January 7, 2014. 36. Hampton T. Traumatic brain injury a growing problem among troops serving in today’s wars. JAMA. 2011;306:477–479. 37. Vanderploeg RD, Belanger HG, Horner RD, et al. Health outcomes associated with military deployment: mild traumatic brain injury, blast, trauma, and combat associations in the Florida National Guard. Arch Phys Med Rehabil. 2012;93: 1887–1895. 38. Hicks MHR. Mental health screening and coordination of care for soldiers deployed to Iraq and Afghanistan. Am J Psychiatry. 2011;168:341–343. 39. Nevin RL. Mental health standards for combat deployment. Psychiatr Serv. 2011;62:805.
40. Armed Forces Health Surveillance Center. Deployment-related conditions of special surveillance interest, U.S. Armed Forces, by month and service, January 2003-December 2009. Med Surveill Mon Rep. 2010;17(1):18–19. 41. Armed Forces Health Surveillance Center. Deployment-related conditions of special surveillance interest, U.S. Armed Forces, by month and service, January 2003-January 2010. Med Surveill Mon Rep. 2010;17(2):21–23. 42. Armed Forces Health Surveillance Center. Deployment-related conditions of special surveillance interest, U.S. Armed Forces, by month and service, January 2003-June 2012. Med Surveill Mon Rep. 2012;19(7):24–27. 43. Ader M, Cuthbert R, Hoechst K, Simon EH, Strassburger Z, Wishnie M. Casting troops aside: the United States Military’s illegal personality disorder discharge problem. http://www.vva .org/PPD-Documents/WhitePaper.pdf. Published 2012. Accessed January 7, 2014. 44. US Department of Veterans Affairs. Veterans benefits timetable: information for veterans recently separated from active military service. http://www.vba.va.gov/pubs/forms/VBA-21-0501-ARE .pdf. Published 2009. Accessed January 7, 2014. 45. Auerbach DI, Weeks WB, Brantley I. Health care spending and efficiency in the U.S. Department of Veterans Affairs. http://www.rand.org/content/dam/rand/pubs/research reports/RR200/RR285/RAND RR285.pdf . Published 2013. Accessed January 7, 2014. 46. Panangala SV, Scott C, Weimer DR. Veterans Affairs: presumptive service connection and disability compensation. http://www.dtic .mil/cgi-bin/GetTRDoc?AD=ADA543665. Published 2011. Accessed January 7, 2014. 47. US Department of Veterans Affairs. Secondary service connection for diagnosable illnesses associated with traumatic brain injury. Fed Regist. 2013;78(242):76196–76209. https://www.federalregister.gov/articles/2013/12/17/2013-29911/ secondary-service-connection-for-diagnosable-illnesses-associatedwith-traumatic-brain-injury. Published 2013. Accessed January 7, 2014.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
