UP TO 20% of the 2 million military personnel deployed to Iraq and Afghanistan in support of Operation Enduring Freedom and Operation Iraqi Freedom have experienced a deployment-related traumatic brain injury (TBI).1–3 A TBI consists of a disruption of brain function due to an external force that results in loss of or alternation of consciousness, posttraumatic amnesia (PTA), neurological deficits, and/or intracranial lesion(s). The Department of Veterans Affairs and the Department of Defense (DOD) classify TBI severity on the basis of the presence or absence and duration of each of these sequelae. A mild TBI (concussion) is a head injury that meets 1 or more of the following criteria consisting of any loss of consciousness (LOC) for less than 30 minutes, alteration of consciousness (AOC) up to 24 hours, and PTA up to 24 hours. A moderate TBI diagnosis requires LOC for more than 30 minutes but less than 24 hours, AOC lasting less than 24 hours, and PTA of more than 1 day but less than 7 days. Severe TBIs consist of LOC and AOC lasting more than 24 hours and PTA more than 7 days. If the patient meets criteria in more than 1 category of severity, the higher severity level is assigned. The majority of deployment-related TBIs are mild.4
The primary method for measuring cognitive impairment following suspected brain injury in a combat zone is the Automated Neuropsychological Assessment Metrics (ANAM).5 The ANAM is the result of 30 years of performance and cognitive test development by the DOD. It is a computerized system designed to assess the cognitive effects of medication used to treat conditions related to environmental stressors such as extreme heat, cold, and chemical weapons.5 The assessment consists of several brief, automated, repeatable measures that lend themselves to longitudinal assessment of cognitive functioning in a cost-effective and time-efficient manner. The ANAM system measures a wide variety of cognitive functions including mood, fatigue, attention and concentration, working memory, mental flexibility, spatial processing, processing efficiency, memory recall, and psychomotor performance using subjective and objective ratings.5,6 It demonstrated adequate construct validity when compared with traditionally used neuropsychological assessments with regard to processing speed, short-term memory, and working memory.6–8
Research regarding the ability of the ANAM to detect cognitive changes has yielded conflicting results. Civilian studies have supported the ability of the ANAM to detect cognitive deficits and monitor recovery among individuals who sustained an mTBI.9,10 Conversely, a study of service members with and without a history of deployment-related mTBI did not find an association between ANAM performance and history of TBI.11 The lack of association found in this study can be attributed to the time delay between the index injury and ANAM administration in which participants completed the ANAM up to 2 years postdeployment. Furthermore, the study used ANAM throughput subtest scores as the primary outcome variable. Throughput scores reflect both accuracy and speed of responding and do not provide information regarding performance on either measure separately, which may prove to be more sensitive measures of impairment or decline. For instance, the ANAM has been shown to detect cognitive deficits in individuals during the days following the index event and may serve as a useful tool for monitoring recovery.9,10 However, current research regarding the relationship between ANAM performance and TBI is limited by the lack of longitudinal studies. The baseline approach to neurocognitive testing has many benefits because it enables clinicians to compare postinjury performance to performance levels when injury-free. Currently, there are no published data demonstrating typical ranges of decline in ANAM performance associated with a TBI evaluation among deployed service members. The primary aim of the current study was to compare the proportion of service members demonstrating declines in ANAM scores relative to baseline performance as part of a TBI evaluation conducted while deployed to Iraq both between groups (TBI vs no TBI) and within groups (duration of LOC among TBI patients).
Participants were referred to a TBI clinic located at a combat support hospital in central Iraq via 1 of 2 mechanisms: (1) direct evacuation from the battlefield immediately following injury or (2) referral from a primary medical provider following self-report of symptoms and history consistent with a possible TBI. Upon arrival at the TBI clinic, participants completed self-report measures of psychological and TBI-related symptoms, a clinical interview with a licensed clinical psychologist or a licensed clinical social worker under the supervision and oversight of the psychologist, a physical examination by a physician, and neurocognitive testing (ie, the ANAM). Only head injuries resulting in a diagnosis of mTBI or no TBI were evaluated and treated at the combat support hospital; all cases meeting criteria for moderate or severe TBI were immediately evacuated from Iraq for more advanced evaluation and treatment. Baseline scores were obtained from the centralized ANAM database located in Washington, District of Columbia, which stores all predeployment baseline ANAM scores. Baseline ANAM scores from testing that occurred immediately preceding the current deployment only were available. Baseline scores were sent to the deployed psychologist via encrypted e-mail, typically within 24 hours of request. These baseline scores were stored locally in the TBI clinic's clinical database along with other clinical variables to assist with decision making and clinical tracking over time. Study approval was obtained from the Brooke Army Medical Center Institutional Review Board, the US Army Medical Research and Materiel Command's Office of Research Protection, and the Multi-National Force–Iraq Institutional Official as exempt research.
Participants included 116 service members referred to a TBI clinic in central Iraq for a TBI evaluation. The sample was 92.2% male with an average age of 27.74 years (SD = 7.07). Racial breakdown was 72.4% white, 13.8% African American, 10.3% Hispanic/Latino, 1.7% Asian, and 1.7% unreported. Participants included members of the Army (79.3%), Air Force (14.7%), and Marines (5.2%), plus 1 civilian contractor (0.9%). Military status was mixed between active duty (44.0%), National Guard (42.2%), Reserve (2.6%), and civilian (0.9%). The average number of years of service was 6.57 (SD = 4.97), with a rank distribution of 56.9% junior enlisted (E1–E4), 34.5% noncommissioned officer (E5–E6), 2.6% senior noncommissioned officer (E7–E9), 5.2% officer, and 0.9% civilian. On average, participants had been deployed a total of 0.73 times previously (SD = 1.04, range: 0–6); had been in Iraq on an average of 4.06 months (SD = 2.55) at the time of injury; were evaluated a median of 2 days postinjury (M = 56.84, SD = 209.23, range: 0–1364 days); and reported an average of 2.53 previous head injuries (SD = 2.79, range: 0–19), although these past head injuries were not assessed to determine whether they met definitional criteria for TBI.
The ANAM is a computerized test widely used within the DOD for measuring neurocognitive performance in 6 domains suspected to be most highly impacted by concussion injuries: simple reaction time, procedural reaction time, learning, working memory, delayed memory, and spatial memory.12 Scaled scores based on previous normative studies with military samples were used in the current study to account for age and gender effects.13 The ANAM battery takes approximately 20 to 30 minutes to complete, and it records performance on each subtest in 3 dimensions: speed, accuracy, and throughput (a score that combines both speed and accuracy by calculating number correct per unit of time). The instructions, stimuli, and data collection are completely computerized, with participants providing responses via a 2-button mouse. A description of the 6 ANAM subtests used at the TBI clinic follows.
In the Simple Reaction Time (SRT) subtest, participants are directed to click the left mouse button each time a stimulus (ie, an asterisk [*]) appears in the center of the screen.
In the Procedural Reaction Time (PRT) subtest, participants are presented with an integer number (either 1, 2, 4, or 5) in the center of the screen and are directed to click the left mouse button when the number is less than the value 3, or click the right mouse button when the number is greater than the value 3.
In the Code Substitution-Learning (CSL) subtest, a “code” consisting of 9 distinct symbols paired with 9 digits is presented at the top of the computer screen. A symbol-digit pairing is then presented at the bottom of the computer screen, and participants are directed to click the left mouse button if the symbol-digit pairing is correct (ie, matches the pairing at the top of the screen), or to click the right mouse button if the symbol-digit pairing is incorrect. Because this is a learning subtest, there is a greater probability of correct pairings (ie, left button clicks).
In the Mathematical Processing (MATH) subtest, participants are presented with an arithmetic equation that includes addition and/or subtraction (eg, 5 + 2 − 3 = ?). Participants are directed to click the left mouse button if the equation results in a value less than 5, or click the right mouse button if it results in a value greater than 5.
In the Code Substitution-Delayed (CSD) subtest, participants are presented with symbol-digit pairings based on the “code” used in the CSL subtest. Participants are directed to click the left mouse button if the symbol-digit pairing correctly matches the original code, or to click the right mouse button if the symbol-digit pairing is incorrect.
In the Spatial Memory (SM) subtest, a checkerboard matrix is presented on the computer screen for 3 seconds, followed by a blank screen for 5 seconds. Two matrices are then presented side by side, only 1 of which matches the original matrix. Participants are directed to click the left mouse button if the matrix on the left matches the original matrix, or to click the right mouse button if the matrix on the right matches.
Mild TBI diagnoses
A diagnosis of mild TBI was made by a licensed clinical psychologist using the TBI Task Force's criteria for mild TBI.14 The Task Force defined TBI as a traumatically induced structural injury and/or physiological disruption of brain function as a result of an external force that is indicated by new onset or worsening of at least 1 of the following clinical signs immediately following the event: (a) any period of loss of or decreased level of consciousness; (b) any loss of memory for events immediately before or after the injury; (c) any alteration in mental state at the time of the injury; (d) neurological deficits that may or may not be transient; and (e) intracranial lesion. Mild TBI is defined as any head injury meeting 1 or more of these criteria which was marked by normal structural imaging, LOC for less than 30 minutes, AOC up to 24 hours, and PTA of less than 24 hours. Diagnoses were based on the combination of neuropsychological testing, clinical interview, computed tomographic imaging, and physical examination.
The ANAM uses a mean standard score of 100, with a standard deviation of 15. Changes in ANAM subtest standard scores were calculated by subtracting the postinjury standard score from the baseline standard score. These raw numbers were then categorized into 4 groups according to the magnitude of score decline, in standard deviation units: less than 0.5 SD (<7.5 points) decline was considered “minimal decline”; 0.5 to 1.0 SDs (7.5–15 points) decline was considered “moderate decline”; 1.0 to 2.0 SDs (15–30 points) decline was considered “large decline”; and greater than 2.0 SDs (>30 points) decline was considered “very large decline.” Proportion of total sample in each category of decline was then computed according to TBI diagnosis and the length of time of LOC.
Analyses were completed in the two steps. First, chi-square analyses were calculated to compare ANAM score declines according to mTBI diagnosis (ie, TBI vs no TBI). Next, chi-square analyses were calculated within the TBI group only to compare ANAM score declines across duration of LOC (ie, no LOC, <1 minute, 1–20 minutes, 20+ minutes). Where cell counts were small, Fisher exact test was used instead of the chi-square statistic. The Bonferroni step-down method was used to correct for multiple comparisons. We first conducted analyses with all participants included regardless of length of time from index event and then repeated the analyses with only those evaluated within the first 72 hours of the index event.
Proportions of decline in ANAM performance: TBI versus no TBI
A significantly larger proportion of patients with TBI demonstrated greater than minimal declines in speed across 5 out of 6 ANAM subtests as compared with patients with no TBI (χ2s > 8.541, Ps < .036, Φs > 0.271; see Table 1): only 10% to 25% of patients with no TBI demonstrated greater than minimal declines in speed across subtests, whereas 50% or more of patients with TBI had greater than minimal declines in speed across subtests. Only the CSD subtest did not reach the threshold for statistical significance, although a strong trend in the expected direction was observed. When considering only those participants seen within 72 hours of the index event, patterns remained by and large the same, although statistically significant differences occurred only on the SRT and PRT subtests (SRT: χ2 = 12.197, P = .007, Φ = 0.409; PRT: χ2 = 7.686, P = .053, Φ = 0.334; see Supplemental Digital Content Table 1 available at http://links.lww.com/JHTR/A54).
TABLE 1-a Proportion...Image Tools
TABLE 1-b Proportion...Image Tools
Although a slightly larger proportion of patients with TBI demonstrated greater than minimal declines in accuracy across subtests relative to patients without TBI, these differences were nonsignificant (χ2s < 2.286, Ps > .131, Φs < 0.175; see Table 1): less than 20% of patients with no TBI demonstrated greater than minimal declines in accuracy, whereas approximately 25% to 33% of patients with TBI demonstrated greater than minimal declines in accuracy. One notable exception to this trend was the SRT subtest, in which minimal declines in accuracy were observed in 95% (for patients with TBI) to 100% (for patients without TBI) of cases, suggesting that SRT accuracy generally does not decline more than 0.5 standard deviations from baseline regardless of injury severity. These trends did not differ when considering only those cases seen within 72 hours of the index event (χ2s < 5.575, Ps > .134, Φs < 0.276; see Supplemental Digital Content Table 1 available at http://links.lww.com/JHTR/A54).
A significantly larger proportion of patients demonstrated greater than minimal declines on throughput scores on 3 of 6 subtests: SRT, PRT, CSL, and SM (χ2s > 11.513, Ps < .009, Φs > 0.316; see Table 1): 5% to 25% of patients with no TBI demonstrated greater than minimal declines in throughput, whereas approximately 50% or more of patients with TBI demonstrated greater than minimal declines, with the exception of MATH. When considering only those cases seen within 72 hours of the index event, only SRT significantly differed between TBI versus no TBI groups (χ2 = 12.850, P = .005, Φ = 0.420) when correcting for multiple comparisons (see Supplemental Digital Content Table 1 available at http://links.lww.com/JHTR/A54). A significantly larger proportion of patients with TBI (51.9%) demonstrated greater than minimal decline on SRT throughput than patients without TBI (5.3%).
Proportions of decline in ANAM performance: LOC duration within TBI
Approximately 50% to 67% of all patients demonstrated greater than minimal declines in speed across all ANAM subtests. Although LOC duration of more than 1 minute seemed to be associated with a slightly larger proportion of greater than minimal declines in speed (approximately 50% for no or <1-minute LOC, approximately 66% for >1 minute), this did not reach statistical significance for any subtest (χ2s < 13.553, Ps > .139, Φs < .378; see Table 2). Among patients with TBI assessed within 72 hours of the index event, approximately 25% to 75% demonstrated greater than minimal declines in speed across the 6 subtests. Longer LOC duration was not associated with greater than minimal declines in speed (χ2s < 11.626, Ps > .235, Φs < .473; see Supplemental Digital Content Table 2 available at http://links.lww.com/JHTR/A55).
TABLE 2-a Proportion...Image Tools
TABLE 2-b Proportion...Image Tools
Greater than minimal declines in accuracy were associated with longer duration LOC on only 2 subtests (PRT and CSD) among all patients with TBI, although these differences were not statistically significant when correcting for multiple comparisons (PRT: χ2 = 20.726, P = .014, Φ = 0.467; CSD: χ2 = 17.048, P = .048, Φ = 0.421; see Table 2). For both subtests, the proportion of greater than minimal declines in accuracy increased as LOC duration lengthened. On the PRT subtest, 35.2% of service members with TBI with no LOC, 41.7% of patients with less than 1-minute LOC, 64.3% of patients with 1- to 20-minute LOC, and 66.7% of patients with more than 20-minute LOC had greater than minimal declines. On the CSD subtest, 27.3% of patients with TBI with no LOC, 50% of patients with less than 1-minute LOC, 57.1% of patients with 1- to 20-minute LOC, and 66.7% of patients with more than 20-minute LOC had greater than minimal declines. Among service members with TBI evaluated within 72 hours of the index event, LOC duration was not associated with an increased proportion of greater than minimal declines in accuracy (χ2s < 15.950, Ps > .068, Φs < .543; see Supplemental Digital Content Table 1 available at http://links.lww.com/JHTR/A54), although a trend in this direction was observed for the PRT subtest.
No differences were seen in proportions of greater than minimal declines in throughput scores as LOC duration increased (χ2s < 12.799, Ps < .172, Φs > 0.367; see Table 2). Approximately 50% to 67% of all patients with TBI demonstrated greater than minimal declines in throughput, regardless of LOC duration. Among those patients with TBI evaluated within 72 hours of the index event, greater than minimal declines in throughput score was associated with longer duration of LOC only on the MATH subtest, although this difference was not significant when correcting for multiple comparisons (χ2 = 19.285, P = .023, Φ = 0.603; see Supplemental Digital Content Table 2 available at http://links.lww.com/JHTR/A55). On the MATH subtest, 33.3% of patients with TBI, 0% of patients with less than 1-minute LOC, 55.6% of patients with 1- to 20-minute LOC, and 50% of patients with more than 20-minute LOC demonstrated greater than minimal declines in throughput score.
The current study investigated the proportion of service members who demonstrated declines in scores on the ANAM during a TBI evaluation conducted in a combat zone as compared with their predeployment baseline ANAM performance. Results indicate that service members with TBI demonstrate greater declines in speed and throughput as compared with those service members without TBI regardless of the timing of assessment, although no differences in accuracy exist. This suggests that the ANAM might be reasonably sensitive to TBI regardless of length of time from injury. It also suggests that for some service members with TBI, cognitive declines observed in the earliest stages postinjury remain impaired over time. Longitudinal studies that repeatedly assess cognitive performance will be necessary in better clarifying trajectories of recovery (or lack of recovery) over time. Some variability within injury categories was observed in performance declines across subtests (eg, greater than minimal declines in speed for patients with TBI but no LOC ranged from 50% to 67% across subtests), which could be due to a range of factors such as previous head injuries, comorbid psychological symptoms, or differential cognitive demands across subtests. Further research is required to shed light on this issue.
One particularly notable finding is that SRT accuracy scores remained largely intact regardless of TBI diagnosis or duration of LOC, with the exception of 3 (2.6%) patients who scored more than 2 standard deviations below their baseline scores. The magnitude of the decline for these 3 patients, all of whom were diagnosed with TBI but did not report any LOC, was much larger than with patients with TBI and longer duration of LOC. Two possible explanations could account for this finding. First, these patients could have misunderstood the instructions or been otherwise distracted during test administration, which could have resulted in inaccurate responding. A second possible explanation is poor effort, in which these 3 participants intentionally responded inaccurately. Unfortunately, effort testing is not routinely assessed as a part of clinical care in combat zones. This is due primarily to the relative infrequency with which poor effort and malingering occurs among deployed service members, as supported by unpublished data demonstrating that less than 5% of deployed service members score below cutoff scores indicative of poor effort (M. Russell, PhD, oral communication, April 2011). The absence of effort testing is a limitation of the current study and is a direction for future research with deployed service members with possible brain injuries. Although we are unable to definitively explain this finding, the overall pattern of results suggests that the SRT accuracy score is quite robust to injury relative to other subtests, both in terms of accuracy and speed. As such, poor performance on this particular measure might be of clinical noteworthiness to medical providers. Researchers and clinicians using computer-based assessments for brain-injured patients might benefit from focusing on reaction times separate from accuracy measures, as the former may serve as a more sensitive measure of cognitive decline (and potentially recovery). The identification of a more sensitive measure of cognitive decline could facilitate early detection and continued monitoring, which are essential for decreasing the severity of post-TBI symptoms and improving outcomes. These findings further support the use of the ANAM as a clinical screening tool among at-risk populations, such as those with a positive history of TBI. The ANAM can be used in addition to corroborating evidence gathered from a clinical interview and subjective and objective measures of functioning to determine the need for further testing, intervention, and return to duty status. Follow-up testing may provide useful clinical data for the provider trying to determine improvement or decline in cognitive functioning over time.
Another limitation of the current study is the ANAM's restriction to visually based stimuli, which could miss potential deficits in language- or auditory-based cognitive performance. Clinicians who also assess performance in these additional areas could therefore provide a more comprehensive evaluation. However, lengthier neuropsychological batteries may not be practical for some clinical settings such as combat zones; the ANAM's sensitivity to TBI, as demonstrated in the current study, suggests it is a reasonable assessment tool for use within combat zones. Another limitation of the current study is the restricted sample, which consisted only of patients who met criteria for mild TBI; patients with more severe TBI were not routinely evaluated in the clinic. Replication of the current study using a longitudinal design with a larger and more heterogeneous sample with regard to TBI severity would provide additional information regarding the clinical use of the ANAM following a TBI. Despite these limitations, this study is among the first to compare pre- and postinjury ANAM performance and identify the proportion of service members who demonstrate declines in scores. Furthermore, given the location and context of the study, results are critical for understanding cognitive performance following brain injury among deployed service members. The setting, context, and findings of the current study lend support for the ecological validity of the ANAM by illustrating the potential for decline in reaction time and accuracy among service members with mTBI in a combat setting. These declines may translate into impaired functioning in high- and low-threat situations, thereby jeopardizing the objectives and safety of the service member and those depending on his/her performance.