The Effects of Blast-Related Neurotrauma on Aurally Aided Visual Search While Standing and Walking

Abstract

Service members (SMs) who have suffered mild traumatic brain injury due to blast exposure (b/TBI) often report post-concussive symptoms consistent with auditory, visual, or vestibular impairments even when they score within the normal range on traditional clinical tests of sensory function. One possible explanation for this discrepancy is that patients who score in the low normal range in more than one sensory modality may be severely impaired in tasks that require multisensory integration. This study evaluated unimodal and multimodal sensory performance in SMs with b/TBI and healthy controls by having them conduct four tasks while walking or standing in an immersive virtual environment: an Auditory Localization task (AL) where they moved a cursor to the perceived location of a sound; a Visual Discrimination task (VD) where they distinguished between two visual targets; an Aurally-Aided Visual Search Task (AAVS) where they used an auditory cue to locate and identify a visual target hidden in a field of visual distractors; and a Visual-Only Visual Search task (VOVS) where they located and identified a visual target in a field of distractors with no auditory cue. The results show the b/TBI and healthy control groups performed equivalently in the AL and VD tasks, but that the b/TBI group responded roughly 15% slower in the AAVS task and 50% slower in the VOVS task. Walking had no effect on performance in the visual-only tasks, but both groups responded faster while walking in the AL and AAVS tasks without any reduction in accuracy.

Introduction

A blast wave may last only milliseconds, but the rapid change in pressure can result in significant organ damage, especially in areas where tissue/fluid and air meet (e.g., brain, auditory system, and vision system). Blast exposure is the most common cause of traumatic brain injury (TBI), including mild TBI (mTBI), among service members (SMs) deployed in support of Operation Iraqi Freedom and Operation Enduring Freedom (OIF/OEF),¹ and has been associated with sensory impairments that impact functioning and quality of life. Studies have shown that the majority of SMs who experience blast-related TBIs (b/TBIs) that are severe enough to warrant treatment in a polytrauma rehabilitation treatment center experience some form of sensory impairment (85%).² These impairments include auditory-only injuries in 19% of patients, as measured by an audiometric threshold greater than 25 dB HL at any frequency; visual-only injuries in 34% of patients, as measured by individuals with impaired visual acuity, hemianopsia, or binocular dysfunction; and both audio and visual impairments in 32% of patients.

Among SMs with b/TBI who experienced chronic symptoms lasting more than 30 days post-blast exposure, the primary manifestations were 1) dizziness (84%) and vertigo (36%), as measured by an evaluation of patient history and a physical exam that included rotary chair testing and computerized dynamic posturography, and 2) hearing loss, which was defined by elevated pure tone thresholds (49%) and/or by the diagnosis of auditory processing disorders (27%).³ Overall, these results show that as many as 60% of SMs with b/TBI who experience chronic symptoms can be expected to have at least some clinically measurable abnormalities in visual, vestibular, and somatosensory integration.

One of the most challenging aspects of treating b/TBI is that many SMs continue to report post-concussive symptoms related to hearing, balance, and vision for many months or years after their injuries even when they no longer exhibit any clinically measurable deficits in objective tests of auditory, visual, or vestibular function. Many of these patients will not seek or receive treatment for their perceived sensory deficits because of the subclinical level of their symptoms and perhaps because of the presence of more pressing medical problems in other domains.⁴ However, these SMs pose a significant problem for the Military Health System (MHS) because it is challenging to assess when brain-injured patients who experience symptoms that are not supported by objective findings should be cleared to return to duty.⁵ The general lack of understanding about the nature of the functional impacts experienced by these SMs also makes it difficult to devise effective rehabilitation and treatment strategies.

One particular area of concern in SMs with chronic b/TBI is that they might be experiencing deficits in multi-sensory integration that do not manifest themselves in traditional clinical tests that are designed to isolate the effects of individual sensory disorders. Most studies of multi-sensory deficits in b/TBI patients focus on a deficit in more than one sensory area, rather than a deficit in integrating information across sensory modalities.⁶ Only a few studies have more directly looked at multi-sensory integration in b/TBI patients.^7,8 However, many researchers and clinicians have observed that deficits in high-functioning patients with b/TBI that are undetectable in standardized clinical tests can be detected when the difficulty of the task increases. Thus, they have designed measures that add more-complex tasks to traditional tests of sensory function (like the head-shake Sensory Organization Test⁹ or the Functional Gait Assessment¹⁰).

Similarly, clinicians within the MHS, who are faced with the task of evaluating the return-to-duty status of SMs with b/TBI, have developed “dual tasks” aimed at capturing the complex activities that SMs must complete as part of their military missions. For example, the Assessment of Military Multitasking Performance (AMMP) task¹¹ includes dual-task objectives like loading ammunition into a rifle magazine while monitoring verbal commands or completing an obstacle course that requires SMs to run, roll, and shoot at targets. Although there are advantages to assessments that incorporate real-world tasks like those included in the AMMP, there may also be benefits to using more abstract tasks that capture the essential elements of real-world military operations, but also allow the relative contributions of auditory, visual, and vestibular function to be evaluated in a more systematic way.

One common component of many real-world military tasks is the requirement for SMs who are walking on patrol to 1) use auditory cues to detect and localize new objects or events in the environment and 2) visually inspect these sound sources to determine whether they represent threats to be avoided or opportunities that require further attention. In this study, we adapted an aurally aided visual search (AAVS) task developed by Perrott and colleagues¹² to assess the auditory, visual, and vestibular capabilities of SMs with a history of b/TBI. The testing paradigm was implemented in a virtual-reality–based system that allowed participants to interact with auditory and visual targets either while standing or while walking on a treadmill. Testing included four tasks: 1) auditory localization (AL)—a unisensory “auditory” condition, where participants were asked to identify the location of a sound source; 2) visual discrimination (VD)—a unisensory “visual” condition, where participants were asked to identify a visual target that appeared at a known location; 3) AAVS—a multi-sensory “audio-visual search” condition where participants were asked to find a visual target in the presence of a larger number of visually similar distractors, where an auditory cue provided information about the location of the target; and 4) visual-only visual search (VOVS)—a visual-only parallel to the AAVS task that was done in quiet, without an audio cue at the location of the visual target.

We hypothesized that the requirement to perform unimodal and multi-modal sensory tasks while walking on a treadmill might provide a greater degree of sensitivity, and a higher level of user acceptability, for the assessment of chronic performance deficits in SMs with b/TBI than traditional clinical measures of sensory function. Further, we hypothesized that the highly controlled virtual environment (VE) used in the tests might produce more repeatable results than the fully realistic tests contained within the AMMP. The following sections describe the four tests in more detail.

Methods

Participants

Participants in the study were active duty U.S. military personnel, 18–55 years of age, recruited at the Walter Reed National Military Medical Center (WRNMMC) in Bethesda, Maryland. All provided written informed consent before their participation and all study procedures were approved by the Institutional Review Board at WRNMMC.

SMs in the healthy control group were screened to ensure they had no history of significant blast exposure. This was determined by administering the 3-Question Defense and Brain Injury Center (DVBIC) TBI Screening Tool¹³ at the time of study enrollment. Members of the healthy control group were also screened to ensure they had normal visual, auditory, and vestibular function. Normal visual function was determined from self-report of normal vision with correction. A Snellen chart was available for participants who were unsure of their status. Normal auditory function was determined by administering a pure-tone audiogram and excluding individuals who did not qualify as an H1 profile under current U.S. Army Regulations (i.e., average threshold level at 500, 1000, and 2000 Hz not greater than 25 dB in the better ear, with no individual level greater than 30 dB at these frequencies, and a threshold not greater than 45 dB at 4000 Hz) or who had asymmetrical hearing with more than a 20-dB difference between ears at 500, 1000, or 2000 Hz. Vestibular function was evaluated by administering the Sensory Organization Test (SOT)¹⁴ and excluding SMs with a composite score <70.

Participants in the b/TBI group were recruited from the National Intrepid Center of Excellence (NICoE), which is a facility on the campus of WRNMMC that provides a 4-week intensive outpatient program for SMs experiencing chronic post-concussive symptoms as a result of blast exposure and/or TBI. All of the participants in the b/TBI group reported at least one deployment to Iraq or Afghanistan and a significant history of blast exposure as indicated by Questions 1 and 2 of the DVBIC TBI Screening Tool. The b/TBI participants had audiometric thresholds that met the same H1 criteria used for the control participants, and again participation was limited to individuals who had an SOT composite score of at least 70.

Table 1 shows mean values and standard deviations for the demographic variables associated with each of the two groups, including United States Special Operations Command (USSOCOM) status and the Audiometric Pure Tone Average, which was defined as the mean audiometric threshold at 500 Hz, 1 kHz, and 2 kHz, averaged across both ears. The table also shows the percentage of participants in the b/TBI group who experienced acute post-concussive symptoms at the time of the blast injury and/or chronic post-concussive symptoms at the time of testing. These responses were taken from Questions 2 and 3 of the DVBIC TBI Screening Tool. Also, for the b/TBI participants, mean scores are given for the military version of the Post-Traumatic Stress Disorder Checklist (PCL-M),¹⁵ which measures the degree to which SMs were bothered by symptoms related to a traumatic event in the past month.¹⁶ Thirty-one percent of the b/TBI participants had PCL-M scores >50, which is often used as the cutoff for diagnosing patients with probable post-traumatic stress disorder (PTSD).

Table 1.

Demographics of Control and b/TBI Subject Groups

Demographics	Control	b/TBI
No. of subjects	29	29
Age, years	30.24 (6.81)	37.00 (5.11)**
Proportion of males	86%	100%*
USSOCOM	0%	76%**
PSTD PCL-M	—	44.21 (18.00)
Pure tone average (dB HL)	6.18 (4.10)	11.01 (5.47)**
SOT composite score	82.76 (4.75)	84.93 (4.37)
Speed (mi/h)	2.42 (0.40)	2.26 (0.53)
Acute symptoms at time of blast
Confusion/seeing stars	0%	93%**
Loss of memory of injury	0%	10%
Loss of consciousness <1 min	0%	38%**
Loss of consciousness 1–20 min	0%	10%
Loss of consciousness >20 min	0%	0%
Concussion symptoms	0%	72%**
TBI	0%	31%**
Chronic symptoms at time of test
Headaches	0%	69%**
Dizziness	0%	28%**
Memory problems	0%	86%**
Balance problems	0%	28%**
Ringing in ears	0%	62%**
Irritability	0%	79%**
Sleep problems	0%	79%**

Symptoms are taken from the DVIBC TBI Screening Tool.

^*Variables significantly different between the control and b/TBI groups at the p < 0.05 level; ^**variables significantly different at the p < 0.01 level.

TBI, blast-related traumatic brain injury; b/TBI, blast-related TBI; USSOCOM, United States Special Operations Command; PSTD PCL-M, military version of the Post-Traumatic Stress Disorder Checklist; SOT, Sensory Organization Test.

Facility

The study was conducted in the Computer Assisted Rehabilitation Environment (CAREN; Fig. 1) Laboratory at the NICoE. The CAREN (Motek Medical BV, Amsterdam, Netherlands) is a multi-modal system that can project immersive VEs in front of SMs as they walk. The CAREN has a six-degree-of-freedom motion platform, 3 m in diameter, mounted on electric actuators that allow it to translate ≈ 1 m and rotate ≈ 20 degrees in all directions. This platform contains an embedded, instrumented dual-belt treadmill that is 2 m long, 1 m wide, and has a maximum velocity of 5 m/s. The system's software allows this platform to be synchronized with virtual environments (VEs) that are projected onto the 180-degree horizontal field of view of a cylindrical screen, 2.5 m high by 4.5 m wide, by four projectors (Projection Design, Fredrikstad, Norway). The CAREN was also equipped with an infrared camera-based motion capture system (Vicon Motion Systems Ltd., Oxford, UK) that allows the position and orientation of each participant's head to be tracked by reflective markers in real time. For the purposes of this experiment, the NICoE CAREN was modified to mount a 64-speaker array (Meyer Sound MX-4 Speakers, Berkeley, CA) in a quasi-random spatial pattern behind the screen. The sound presented by these speakers passed through the fabric screen with minimal distortion or loss of signal intensity, allowing auditory targets to be presented in up to 64 distinct locations within the VE.

FIG. 1.

The CAREN (Computer Assisted Rehabilitation Environment) system at the National Intrepid Center of Excellence (NICoE; left panel) is utilized to implement the experimental tasks. The CAREN system consists of a six-degree-of-freedom motion platform with an embedded treadmill surrounded by a 180-degree field-of-view screen. The NICoE system was modified for this study to include an array of 64-speaker array mounted on a scaffold, behind the fabric screen. The center panel shows the speakers positioned on a truss. The blue dots in the right panel provide a visual representation of each speaker location in the virtual environment, corresponding to the physical location of each loudspeaker. Color image is available online.

In order to reduce the potential confounding effect that treadmill noise might have on auditory localization, all of the trials in the experiment were conducted with simulated treadmill noise emanating from three of the loudspeakers in the array (one to the lower left, one in the center, and one to the lower right). This masking noise was generated by measuring the spectrum of the system's treadmill noise at the location of the listener at three different walking speeds (0.89, 1.12, and 1.32 m/s), inverse filtering these noises to account for the frequency responses of the masking filters, and continuously playing the resulting filtered noises through the three speakers during the course of each trial. This resulted in a background noise level of 62–63 dBA SPL in each trial.

Experimental procedures

After reviewing and signing the informed consent document, each of the healthy control participants completed a 2-h audiometric and vestibular screening session before participating in the CAREN portion of the study. During this session, they completed the six test conditions of the SOT and had their left- and right-ear pure-tone audiometric thresholds measured at eight frequencies by an audiologist using a clinical audiometer in a double-walled, sound-treated booth.

Because b/TBI participants were recruited into the study during the first week of their 4-week tenure at the NICoE, the SOT and the audiometric thresholds were typically completed as part of their routine clinical examination during their second week. Those who consented and also met the inclusion criteria were then scheduled for CAREN testing during either their third or fourth week.

Each testing session in the CAREN lasted approximately 2 h. At the start of the session, participants were familiarized with the system, given a safety brief, and fitted with a safety harness. Once they were secured to the platform's safety stand, they were taken through a series of preliminary VEs, starting with balance tasks called the Balance Balls and Balance Cubes¹⁷ to familiarize them with weight and step shifting on the CAREN platform. Then they completed the Continuous Road VE (flat walking task) to acclimate to walking on the system's treadmill. The speed of the treadmill was synced to a visualization of a road on the screen and was adjusted until it reached what they considered to be a “comfortable walking pace.” This speed, which averaged approximately 2.5 miles per hour (see Table 1), was then recorded and used as the fixed walking pace in all subsequent walking conditions of the experiment. Each participant then completed five training trials of each of the four major tasks in the experiment (AL, VD, AAVS, VOVS) while standing on the platform. Once they were fully acclimated, two blocks of each experimental task were completed, once standing and once walking, in a randomized testing order. The four tasks are described in detail below.

Auditory localization task

The AL task measured how quickly and accurately participants were able to find the location of a pulsed noise presented from a single, random speaker location behind the CAREN screen. In this task, head tracking information from the motion capture system was used to project a white cross-hair on the screen directly in front of the participant. Each trial started with the participant facing a fixation point in the center of the screen, and then an auditory stimulus was presented randomly from one of the 64 speaker locations. The auditory stimulus consisted of pink noise, which was pulsed on and off at a 1-Hz rate with a 50% duty cycle. The noise was adjusted to a level of 70 dBA SPL at the location of the listener's head, and it was band-pass filtered to have a bandwidth from 400 Hz to 12.5 kHz and inverse filtered with 128-point finite impulse response (FIR) filters that were designed to reduce the spectral variability of the stimuli across the different speaker locations.

The participant was asked to move the head-slaved cross-hair to the perceived location of the sound (as shown in Fig. 2) and then press a button on a hand-held controller (Microsoft XBox Wireless) to indicate their response. They were instructed to do this task as quickly and accurately as possible; localization accuracy and response time were measured in each trial.

FIG. 2.

Visual display shown to the participants in the auditory localization task. Color image is available online.

Each participant completed two 50-trial blocks of the AL task, one block while standing on the platform with the treadmill stationary and one while walking at a comfortable pace. In order to eliminate spurious responses, trials where the reaction time was faster than 0.5 sec or where the angular error was more than 3 standard deviations (SDs) greater than the mean value for that participant in that condition were eliminated from the analysis (this eliminated roughly 0.75% of the total trials).

Visual discrimination task

The VD task was a choice reaction time task that measured how quickly and accurately participants were able to identify a visual target presented at a known location on the screen. Participants were instructed that all trials would originate at a fixation point that was directly in the center of the screen, at a distance of approximately 280 cm. Then, at the start of each trial, a visual target with one or three white dots was presented at the fixation point. The screen always had a light blue background, and the orientation of the three-dot cluster was selected randomly for each trial. Each dot was 1.5 cm in diameter, with a 2-cm separation between dots along the edge of the three-dot target stimulus.

Participants were instructed to wait until the one- or three-dot target appeared and then to use one of two buttons on a wireless response controller to identify the number of dots in the target stimulus (Fig. 3). They were instructed to do this task as quickly and accurately as possible; response accuracy and reaction time were measured in each trial. Each participant completed two 50-trial blocks in the VD task, one while standing on the platform with the treadmill stationary and one while walking at a comfortable pace. Trials where the reaction time was more than 3 SDs greater than the mean value for that participant in that condition were eliminated from the analysis (this eliminated roughly 1.2% of the total trials).

FIG. 3.

Visual discrimination task in the experiment. The left panel shows the visual display shown to the participants. The right panel shows examples of target stimuli, containing an odd number of dots. Color image is available online.

The participants generally achieved very high scores on the VD task, with the percent of correct responses ranging from 97.5% to 98.6% across the two participant groups in the two conditions. The incorrect trials were discarded, and only the reaction times of the correct responses were analyzed in the results.

Aurally aided visual search task

The AAVS task combined the AL and VD tasks, requiring the listener to use an auditory cue to locate a visual target and then respond by indicating whether it contained one or three dots. Each trial of the AAVS task started with the participant facing a fixation point in the center of the screen. Then, a visual target with one or three dots (from the VD task) would be randomly presented with a coincident audio cue (the pulsed noise from the AL task) from one of the 64 possible loudspeaker locations, among a field of 263 other visual distractors. These visual distractors were identical to the target, but contained two or four dots (Fig. 4). As in the VD task, the participant's task was to find the visual target as quickly as possible and use the wireless controller to indicate whether the target contained one or three dots.

FIG. 4.

Aurally aided visual search and visual-only visual search tasks in the experiment. The left panel shows the visual display shown to the participants. The right panel shows a zoomed-in view of the target location. Color image is available online.

Each participant completed two 50-trial blocks in the AAVS task, one with the participant standing and one with the participant walking. Trials with reaction times <0.5 sec or those with reaction times >3 SDs worse than the mean score for an individual participant in an individual condition were eliminated from the analysis (these accounted for 1.1% of all the trials in the experiment).

Response accuracy was very high in all conditions, ranging from 98.5% to 99.2% across all the conditions. As in the VD task, trials with incorrect responses were eliminated from further analyses.

Visual-only visual search task

The VOVS task was identical to the AAVS task, except that no coincident audio cue was presented at the location of the visual target. Participants were asked to locate the visual target, among a field of visual distractors, and then respond by indicating whether it contained one or three dots. Each participant completed two 20-trial blocks of the VOVS task, one while standing and one while walking.

The VOVS task was more challenging than the AAVS task, with mean response times that were roughly an order of magnitude slower. These much slower response times necessitated the implementation of a 120-sec timeout period to prevent trials from becoming too long. Trials where the participants spent more than 120 sec searching for the visual target were terminated without a response and were conservatively assumed to have a 120-second response time in the analysis (these timeouts occurred in around 1.3% of trials in the control group and 5.1% of trials in the b/TBI group).

Statistical analysis

In the AL task, accuracy and reaction time were the primary dependent variables. In the VD, AAVS, and VOVS tasks, accuracy was near 100% and reaction time was the primary dependent variable (incorrect responses were eliminated from the response time analysis). In both cases, the within-subject independent variable was treadmill condition (walking vs. standing) and the between-subjects independent variable was participant group (b/TBI vs. healthy control). In all cases, a two-factor mixed-design analysis of variance was used for the statistical analysis.

In order to account for possible order effects, the average scores in the first and second blocks in each task were compared to determine whether presentation order had a significant impact on performance. This analysis revealed that order effects had no significant impact on the AL or VD tasks, but that the participants, on average, responded 15% faster in the second block of trials on the AAVS task and 20% faster in the second block of trials on the VOVS task. Data were corrected for these order effects before conducting the mixed-model ANOVA by increasing the reaction time by half the difference in all of the initial blocks and decreasing the reaction time by half the difference in all the final blocks.

In cases where significant differences were found between the b/TBI and control groups, step-wise regression was used to determine whether these differences could be explained by the demographic differences observed in the two participant groups.

Results

Auditory localization task

The left panel of Figure 5 shows the average localization accuracy, the error from the true sound location to the response location on each trial. The results show that the angular error was approximately 9 degrees in all the conditions tested. A two-factor analysis of variance (ANOVA) confirmed that there was no significant effect of either treadmill condition (F_(1,56) = 0.01; p = 0.94) or participant type (F_(1,56) = 0.72; p = 0.40).

FIG. 5.

The left panel shows mean accuracy (in angular error) for localization responses in the standing and walking conditions of the auditory localization (AL) task, and the right panel shows mean response time in the standing and walking conditions of the AL task. The error bars shows ±1 standard error around the mean subject value. b/TBI, blast-related traumatic brain injury.

Walking on the treadmill did, however, have a significant impact on the response time in the experiment, which is shown in the right panel of Figure 5. Both groups responded substantially faster when they were walking, with no loss of accuracy (F_(1,56) = 15.70; p < 0.001). Overall, 44 of the 59 participants had faster response times in the walking condition, by an average of 240 ms.

Visual discrimination task

Mean response times for each participant group in each condition of the VD task are shown in Figure 6. Mean response times were around 470 ms in all conditions, and the results of a two-factor ANOVA indicate that there was essentially no differences in performance in the reaction time, either across listener group (F_(1,56) = 0.001; p = 0.98) or across the standing and walking treadmill conditions (F_(1,56) = 0.32; p = 0.58).

FIG. 6.

Mean log response times in the standing and walking conditions. The error bars show ±1 standard error around the mean subject value. b/TBI, blast-related traumatic brain injury.

Aurally aided visual search task

Mean reaction times are shown in Figure 7. Results of a two-factor ANOVA indicate that there was a significant main effects of treadmill condition (F(1,56) = 5.93; p = 0.018), with slower results in the standing condition than in the walking condition. There was also a significant main effect of participant group (F(1,56) = 5.55; p = 0.022), with slower response times for the b/TBI group than for the healthy controls. However, the interaction between these factors was not statistically significant (F(1,56) = 0.28; p = 0.60). Thus it appears that a history of b/TBI may contribute to degraded performance in the AAVS task, even in participants who perform no worse than healthy controls in unimodal tasks evaluating localization accuracy and visual reaction time. The data also demonstrate that walking on a treadmill facilitates performance in the AAVS task for participants in both the b/TBI and healthy control groups.

FIG. 7.

Mean log response times in the standing and walking conditions of the aurally aided visual search task. The error bars show ±1 standard error around the mean subject response time. b/TBI, blast-related traumatic brain injury.

One concern in analyzing the data from the experiment is that there were some differences between the b/TBI and healthy control group, highlighted in Table 1, that might have contributed to the difference in performance across the groups. The b/TBI group was slightly older than the control group and had slightly worse hearing (as measured by the pure tone average). The b/TBI group also had a much higher proportion of individuals with USSOCOM training. In order to account for possible intersubject differences that might have influenced localization accuracy in the experiment, a step-wise regression was conducted on the following factors: b/TBI group; age; 0.5–1–2 k pure tone average hearing threshold (dB HL); and whether the participant had received USSOCOM training. The results of this regression, shown in Table 2, indicate that participant group (control or b/TBI) was the only variable to survive this step-wise regression. This suggests that the difference in performance between the two groups cannot be explained by differences in age, hearing thresholds, or specialized military training. An additional analysis was also conducted to see whether the scores of the b/TBI participants were significantly correlated with their PCL-M scores. This analysis did not find a significant correlation (ρ = 0.14; p = 0.48), suggesting that probable PTSD could not account for the degraded performance in the b/TBI group.

Table 2.

Step-Wise Regression of Contributing Variables on Aurally Aided Visual Search Score

Variable	Coeff	Std.Err.	Status	p value
Subject group	0.0630	0.0220	In	0.0050
Age	0.0000	0.0019	Out	0.9997
Pure tone average (dB)	–0.0016	0.0023	Out	0.4982
Special operations	–0.0018	0.0365	Out	0.9610

Visual-only visual search task

Figure 8 shows the results of the VOVS task, in terms of the mean response time in each trial. These results clearly demonstrate that the mean log response times in the b/TBI group (24.5 sec) were substantially slower than those in the control group (16.1 sec).

FIG. 8.

Log response times in the standing and walking conditions for each participant group in the visual-only visual search task. The error bars show ±1 standard error around the mean subject response time. b/TBI, blast-related traumatic brain injury.

This main effect of group was significantly significant at the p < 0.001 level (F_(1,56) = 20.28). The main effect of walking was not significant (F_(1,56) = 1.29; p = 0.26), but there was an interaction between treadmill condition and participant group, with the b/TBI group exhibiting a much larger increase in response times when walking than the control group (F_(1,56) = 4.42; p = 0.040). Thus, it seems that b/TBI status had an even larger impact on performance in the VOVS task than in the AAVS task. Also, it appears that walking, which seemed to improve performance for both groups in the AL and AAVS tasks, tended to interfere with the visual search in the b/TBI group.

In order to determine whether potentially confounding variables might be accounting for the difference between b/TBI and control performance, a step-wise regression was conducted (Table 3). As with the AAVS task, the only variables to survive this step-wise regression was b/TBI status. Also, although there was a trend in that direction, the scores on the PCL-M were not significantly correlated with the VOVS scores of the b/TBI participants (ρ = 0.33; p = 0.08). Thus, it seems that the poor performance observed in the b/TBI participants in the VOVS task cannot be explained by age, hearing status, advanced military training, or PTSD status.

Table 3.

Step-Wise Regression of Contributing Variables on Visual-Only Visual Search Score

Variable	Coeff	Std.Err.	Status	p value
Subject group	0.1117	0.0410	In	0.0075
Age	0.0003	0.0027	Out	0.9252
Pure tone average (dB)	0.0014	0.0034	Out	0.6674
Special operations	0.0098	0.0527	Out	0.8527

Comparison across subject distributions

A distinguishing characteristic of studies of blast exposure, particularly those involving sensory problems associated with blast in those with mTBI, is that there is no gold standard for authoritatively determining who is impaired and who is not. Consequently, many SMs with b/TBI might appear to be normal on any clinical measure tested. A blast exposed population that appears to be only slightly different than a control group may contain a mixture of some individuals with severe symptoms who are masked by other individuals who fall in the normal range.

Because of these possible issues, it is helpful to look at the differences between the control and b/TBI populations not only in terms of the mean differences between the two groups, but also in terms of the distributions across the two groups. The four panels of Figure 9 show the performance of individual participants within each of the two groups for each of the four tasks tested in the experiment. In these panels, performance has been averaged across the walking and standing conditions to get a single average score for each participant.

FIG. 9.

Cumulative distributions of each participant scores in the control and b/TBI groups from each task in the experiment. The three dashed vertical lines show the 25th, 50th, and 75th percentile scores of the normal population in each task. b/TBI, blast-related traumatic brain injury. Color image is available online.

Distributions are shown as cumulative distributions. Each point on the curve represents the proportion of the population who scored at or below the value shown by the abscissa. Also, for reference, three vertical dashed lines have been added to each panel to show the 25th, 50th, and 75th percentile of performance in the control group in each panel. Note that, because the data are plotted in terms of reaction times and angular errors, lower values indicate better performance. Thus, one would expect the curve of the better-performing group to be shifted to the left of each panel.

The VD task, shown in the upper left figure, shows the least difference between the control and b/TBI groups. Distribution curves of the control (solid black line) and b/TBI (dotted line) groups are very similar, and the 25th, 50th, and 75th percentiles of the two groups were almost identical. Thus, it seems that there were no discernible differences between the two groups in the visual baseline task.

The AL task also produced very similar distributions of individual performance across the two groups. There is some indication of a longer tail in the blast-exposed group, indicating that a few b/TBI participants may have had increased difficulty localizing, but in general the control and b/TBI groups performed very similarly.

A larger difference was observed in the AAVS task, in the bottom left of the panel. In this task, we observed similar performance between the top half of the blast group and the top half of the control group, as indicated by comparable values for the 25th and 50th percentile points on the curves. However, the poorer performing b/TBI participants were generally much worse than the poorer performing control participants. In fact, 55% of the b/TBI participants fell in the bottom quartile of the control participants, and almost 50% fell in the bottom 15% of the control participants.

The bottom right panel shows performance in the VOVS condition, which by far showed the biggest differences between the two groups. In this curve, we observed that there appeared to be a limitation on the best performing b/TBI participants: Almost 100% of the b/TBI participants had a slower reaction time than the median control participant, and 70% scored in the bottom quartile of the control group.

One potentially problematic aspect of the data for the AAVS and VOVS tasks is that there was a substantial number of control participants who scored poorly in the task. This makes it difficult to use these tasks in the traditional “clinical” test, where individuals who score in the bottom 5th percentile of the normal range are assumed to be abnormal. Thus, it may be difficult for a clinician to distinguish whether an individual participant who scores poorly on one of these tests is suffering from a b/TBI-related impairment or whether they simply happen to be an unimpaired individual who falls at the low end of the normal range. However, looking at the distributions as a whole, it is apparent that there are some systematic degradations in visual search ability in individuals who have a history of blast and mTBI.

Discussion

The purpose of this study was to compare healthy control SMs to those with a history of b/TBI on a series of tasks related to the detection and identification of audio-visual targets while standing and walking. The intent was to develop a series of tasks with plausible relevance to military tasks that might be capable of providing objective measures of performance deficits in SMs with chronic b/TBI who report persistent symptoms but tend to score in the normal range on traditional clinical tests.

Our b/TBI population, which was drawn from the population of clinical patients at the NICoE, would appear to represent the archetype of patients experiencing chronic negative sequelae of blast or mTBI. All were active duty and most were continuing to perform their military duties at a high level before NICoE admission. Most were current or former members of the USSOCOM who were likely selected from the most physically fit candidates and were subjected to a career of highly rigorous training. However, all had a clear history of blast exposure and mTBI and reported experiencing some chronic post-concussive symptoms that brought them to the intensive NICoE outpatient program.

In the simple VD and AL tasks, our results were consistent with previous studies that have shown very little difference between the control and b/TBI populations. Despite being slightly older and having slightly greater hearing loss, our b/TBI group performed almost identically to the control group in terms of VD and AL tasks. They also performed slightly better than the control group on the dynamic conditions of the SOT and better than the reported mean composite score of 80-81 on the SOT of healthy SMs who are currently active in the USSOCOM.¹⁸ This is consistent with other studies that have generally failed to find reliable differences between SMs with and without a history of b/TBI using objective behavioral measures^19–23 or between participant-reported symptoms and objectively measurable deficits on neuropsychological tests.¹⁹ The failure to find a localization deficit in the b/TBI group is also consistent with a recent study in our laboratory that only found differences between blast-injured and healthy controls in localization tasks that involved more than one simultaneous sound source.²⁴

Our results did, however, find substantial differences between the control and b/TBI populations in the AAVS and the VOVS conditions. The differences were modest in the AAVS task, with a 15% higher mean search time in the b/TBI population. And in the VOVS task, the group differences were profound, with a 50% higher search time in the b/TBI group. There are a number of possible explanations for these differences. Susceptibility to distractors in the VOVS might be related to impaired inhibition, which has been reported for individuals with a history of blast, mTBI, and PTSD.²⁵ Poor performance in the VOVS might also be related to changes in oculomotor function, including changes in the ability to follow a visual target.²⁶ Combs and colleagues²⁷ also reported impaired visual scanning performance in the visual scanning portion of the Delis-Kaplan Executive Function System. There might also be cognitive issues related to maintaining focus on the visual search task and implementing a consistent strategy without rescanning the visual field. Studies of visual search in more severely impaired TBI patients have found that performance in easy “parallel” visual search tasks recovers quickly, but that performance in more difficult search tasks may take much longer.²⁸

Further analysis is needed to fully understand the source of the visual search deficits found in the b/TBI participants. However, whatever its origin, the magnitude of the deficit has concerning implications for SMs who may be suffering from chronic post-concussive symptoms. The visual search tasks used in this experiment are analogous to many of the tasks SMs must accomplish when they are in combat situations, and a mean increase of ≥5 sec in the VOVS task could have a real impact on operational effectiveness in these environments.

The other major focus of this study was on the effect that walking might have on performance in a VOVS task for SMs with mTBI. In a previous study that examined the visual search tasks examined here with healthy control participants, we showed that walking at a constant pace on a treadmill tended to facilitate, rather than impair, both AL performance and AAVS performance.²⁹ This is consistent with other studies that have found that walking can improve performance in cognitive tasks. Tomprowski and Audiffren³⁰ reported that older participants who were required to distinguish between even and odd digits and vowels and consonants while walking at their preferred pace on a treadmill experienced a decrease in response time and in mixed switching costs (but an increase in error rate). Schaefer and colleagues³¹ reported that young adults exhibited reduced variability in gait and improved performance in three- and four-back digit recall tasks when walking on a treadmill at a preferred speed as opposed to when they were stationary.

Further, McMorris and Graydon³² showed exercise tended to improve the speed, but not the accuracy, of complex cognitive tasks. Duncan and colleagues³³ also found that reaction times in a visual discrimination task were faster for older adults (60–77 years of age) when they were walking at a comfortable pace on a treadmill than when standing before or after the walking interval. Thus, there is some reason to believe that the act of walking on the treadmill might have made participants slightly more efficient at the AAVS task.

However, these factors cannot explain why the benefits of walking occurred in the AL and AAVS tasks, but not in the VD or VOVS tasks. At least one previous study³⁴ has reported a dual-cost benefit for healthy individuals asked to perform a speech recognition task while walking, so it is possible that apparent improvement in the AL and AAVS task could be related to the same kind of increased generalized arousal that occurred in the walking conditions of that study. It is also possible that the increased translational and rotational head movements that occur when walking on a treadmill³⁵ might have improved localization performance in the same way that small intentional head movements have been shown to improve localization accuracy in other studies.^36–38 More research is needed to fully understand that advantages that walking might be providing to both the b/TBI and healthy control groups in the AL and AAVS conditions of this study.

The final major finding of the experiment is that the b/TBI group appeared to suffer from a significant degradation in the walking conditions of the VOVS task that was not noted in the healthy control participants. The VOVS task is a challenging task that requires rapid eye movements over a large visual field populated with numerous potential detractors. If there was any degradation the compensatory ocular-motor reflexes that help observers maintain a steady gaze on a fixed object while in motion, it is not surprising that it would emerge in this task.

Conclusions

Overall, the results of this study appear to validate the complaints about vision, hearing, and balance problems often reported by mTBI patients who test in the normal range on conventional tests of vision, hearing, and cognition. These patients may suffer from a modest deficit in AAVS tasks, and they may have more severe deficits in VOVS tasks. For military personnel engaged in combat tasks, these deficits may be particularly alarming, because these individuals are often engaged in challenging tasks that may require them to conduct visual searches while navigating over uneven terrain. More research is needed to determine the underlying mechanisms behind these deficits, evaluate the extent to which they may correlate with other clinical measures of cognitive and sensorimotor function, and develop rehabilitative strategies to address them.

Author Disclosure Statement

No competing financial interests exist.