This study was a single cohort, pre-/post-treatment design. Veterans reporting cognitive deficits to Speech Services were recruited and treated in two sites (North Florida/South Georgia VA Medical Center and Durham, N. Carolina VA Medical Center. Treatment fidelity across the two sites was ensured with methods including the following: manualized GMT was used; intervention was conducted by the site PI from each facility; and weekly meetings were held of the two PI's, covering review of GMT weekly session content. Research was conducted with the oversight of the Internal Review Boards from the University of Florida (#201601606) and Durham VA Medical Center (#01887).

Subject inclusion criteria were as follows: diagnosis of a blast-related mTBI documented in the medical record by a physician or neuropsychologist; frontal lobe-based cognitive impairment as determined by more than one standard deviation below the mean for the matching age group from the technical manual (29) on trial 3 or 4 of the Delis-Kaplan Executive Function System, Color-Word Interference Test (30); 18 to 55 years old; at least 6-months post injury; no history of pre-morbid learning disability; no psychiatric diagnosis sufficiently severe to have resulted in inpatient hospitalization or a neurological disease unrelated to TBI; score >90 on National Adult Reading Test, Estimated IQ (31); passing score of the validity testing on the Test of Memory Malingering (32); no alcohol or substance abuse within the past year; not involved in litigation; fluent in English.

For the NIH EXAMINER Unstructured Task, cTOL and CRIS variables, we conducted a pre-/post-treatment paired t-test to obtain p-values for assessing statistical significance, calculated effect sizes, and computed 95% confidence interval using the bias-corrected and accelerated (BCa) bootstrap method (47). Effect size was calculated by dividing the change score by the pooled estimate for standard deviation of the outcome (48). The statistical software R was used to conduct analyses and create figures (49). If there was a missing data point, the participant data was removed from analysis. Because the sample size was small, we conducted exploratory analyses. We used IBM SPS Statistics v28.0.0.0 to analyse Pearson correlation models investigating any potential relationship between gain in the Unstructured Task outcome measure and characteristics of depression, IQ, and PTSD score. We conducted the same correlation analysis for the cTOL outcome measure and subject characteristics. Descriptive analysis was conducted including the range of scores and the percent of participants showing change; and scatterplots were generated for visual inspection.

Forty-eight participants were screened of which 19 did not meet inclusion criteria and 3 declined to participate after testing. Subsequently, there were 12 dropouts [starting a new job (n = 2)]; demands at school increased (n = 2); moving (n = 1); new baby (n = 1); coming to clinic increased anxiety (n = 3); and no reason given (n = 3). Thus, 14 participants completed GMT (8 at the NF/SG site and 6 at the Durham site). Subject characteristics are given in Table 2. All participants, except one (Subject 13), had a PTSD diagnosis. All participants exhibited some signs of depression, with scores ranging from 12 (Subject 13) to 54 (Subject 4).

CRIS. There was statistically significant improvement in Life Role Participation, as follows: Extent of Participation in life roles (p < .01; effect size = .75; Table 3) and Perceived Limitations in life role participation (p < .05; effect size = .59; Table 3). The Satisfaction in life roles participation subdomain did not reach significance (p = .18; Table 3). We encountered difficulties with the available software for the CRIS, and subsequently five data points for the CRIS were lost, for which those participants were removed from analysis, with the remaining analysed sample size of 9 participants.

There was no significant relationship between the cTOL and either depression, PTSD symptom severity, or IQ (r = .04, p = .45; r = .116, p = .35; r = .036. p = .45, respectively). There was no significant relationship between the Unstructured Task and either depression or IQ (r = .12, p = .31; r = .15, p = .35, respectively). There was a relatively higher, but non-significant correlation between the Unstructured Task and PTSD severity (r = .439, p = .07). These results should be interpreted with some caution, due to the small sample size of the study.

Data inspection for individuals showed that 93% of participants (13/14) had improvement in one or more outcome measures, in terms of score change. The NIH Examiner Unstructured Task showed that 67% of the participants showed a gain in score (Figure 1); in general, those with a lower baseline score showed greater improvement in change score in comparison with those with higher baseline scores (negative slope of the regression line (red line; Figure 1).

The cTOL showed that 77% had a gain in score (Figure 2). The improvement in score (i.e., higher negative number for the change score) was, in general, greater for those with a worse baseline performance (longer time to complete at pre-treatment). An example of this is subject 14 (lower, right corner of Figure 2), who had the greatest change score (improvement) at post-treatment, but began at pre-treatment with the longest time (worst performance) to completion at pre-treatment.

For the CRIS, 89% of participants improved according to the Extent of Participation in life roles (Figure 3, Panel A). In general, those with the lowest baseline score (horizontal axis), showed greater improvement (vertical axis; Figure 3, Panel A). For the CRIS Perceived Limitations, 78% of participants improved (Figure 3, Panel B). On the CRIS subdomain of Satisfaction, 67% improved (Figure 3, Panel C). In terms of the MDC for the CRIS Satisfaction in Life Roles Participation, the group mean gain score exceeded the MDC threshold (Table 3), indicating a mathematically detectable, and therefore, meaningful difference. This occurred even though three veterans, who had high baseline scores, did not show improvement. For the CRIS Extent of Participation in Life Roles, the group mean gain score closely approached the minimal detectable change (MDC), just shy of the threshold by 0.03 points (Table 3).

In terms of individuals, subject 13 showed the least depression, the least severe PTSD symptomatology, and was in the top quartile of treatment response according to the cTOL (for example, Figure 2); and subject 4 showed the worst depression, the greatest PTSD symptomatology, and was close to the bottom quartile of treatment response according to the cTOL. There was only one subject in the group who did not show some improvement across measures (Subject 9; for example, almost no change, Figure 1); he had the second least depression score (14 points), the second least PTSD symptoms (44 points), and the highest IQ (122).

The dearth of executive function training studies on behalf of mTBI/PTSD is likely due to the inherent difficult challenges. Those challenges include persistent executive dysfunction, poor attention-concentration, and memory difficulties preventing re-entry into societal roles (50, 51). These challenges are compounded in the 13%–33% of Veterans who have both mTBI and PTSD (52–54). The neural disruptions shared by PTSD and TBI include asymmetrical white matter tract abnormalities and gray matter changes in the basolateral amygdala, hippocampus, and prefrontal cortex (55, 56). For both mTBI and PTSD, disruptions occur in the fronto-congulo-parietal cognitive control network circuit, various executive function domains such as the anterior cingulate cortex involving cognitive control, the dorsolateral prefrontal cortex mediating working memory, the inferior frontal gyrus and (pre-) supplementary motor area regulating response inhibition, and the parietal lobes involving attention and its control (55, 57).

There are few studies reporting consistently convincing evidence on the effectiveness of executive function training (58), and studies on the mTBI and PTSD comorbid condition are scarce. A few recent reports documented modest improvement from cognitive training in some areas of cognitive functioning in military members and Veterans with mTBI (17–22). Although promising, there were limitations as follows: intervention that lacked training of cognitive processes essential to executive function (17), lacked sufficient training time (18, 21, 22), or training time for cognitive components was not reported (18, 20). The current study addresses some of these limitations by having enrolled and treated those with mTBI/PTSD. Therefore, this study provides preliminary results on which to further develop evidence-based practice for clinical care, given that “…symptoms from each [condition (mTBI or PTSD)] may be often indistinguishable suggesting that assessment and treatment of mTBI and PTSD benefits from better clinical integration” (59).

In a prior meta-analysis of GMT studies (60), no studies reported measures of community participation. Life role participation is the ultimate goal of rehabilitation (61). It is critical to develop interventions that can effect change and improvement in life role participation for veterans with mTBI/PTSD. In the current study, GMT resulted in sufficiently robust improvement in executive function so as to produce significant improvement in life role and community participation.

A prior meta-analysis of other GMT studies (60) combined multiple diagnoses of brain injury (ABI, MS, ADHD, CVD, Spina Bifida, SUD) but included only two studies (n = 2/33 studies) of those with mild TBI). In their analysis, they accepted studies with multiple outcome measures of executive function including the following: Multiple Errands Test, Hotel Task, and BADS Zoo task. From this information, they reported that their meta-analysis showed a small effect size (Hedges' g = 0.297) immediately following training. Their meta-analysis is an important study as a beginning investigation of GMT across patient diagnostic categories. In fairness, our relatively stronger results cannot be directly compared to the meta-analysis described above (59) because our sample was composed of one diagnostic category compared to their multiple diagnoses; further, our measures were selected according to their ability to assess the cognitive processing and task performance that was practiced during GMT. These study design differences could explain, in part, the higher effect sizes in the current study [Unstructured Task and Extent of Participation in Life Roles (Cohen's d = 0.67, and.75, respectively)].

Additionally, the promising results of the current study could be the function of a number of treatment protocol features. First, the GMT protocol itself encourages the learner to identify three complex tasks that they care about performing and with which they are having difficulty. This feature is an example of precision (customized) neurorehabilitation. A second advantageous feature of GMT is the instruction regarding division of a complex activity into main goal, subgoals, and action steps; this process is guided for success by the therapist, if needed. A third treatment advantage was the use of the VTM app, used for home practice as a memory aid, for keeping in mind the main goal, subgoals, and action steps, during a given activity. Fourth, the VTM provided an alert signal to assist with focus of attention on a given action step. Fifth, the VTM provided a feature that may have improved ability to plan the time for a given task; that is, the VTM timer was used for assessing task completion compared to the prior-predicted time given by the user.

Given these strongly supportive features of the GMT itself and the use of the VTM, it is reasonable to consider that they had a strong impact on the gains in response to treatment.

Our results showed that the NIH Examiner Unstructured Task was sensitive to change in those with mTBI/PTSD in response to GMT (p = .02). The EXAMINER Unstructured Task presents a problem to the user, which is a simulation of a functional task requiring cognitive processes and strategies similar to those cognitive processes and strategies instructed and practiced during GMT task training. These include mindfulness, stating the main goal, generating a plan and evaluating results. In severe TBI, GMT has been offered, with reports of significant improvement in measures, such as the Hotel Task, the Modified Multiple Errands Task and the Zoo Map Test (62, 63). These studies were for severe TBI and the measures are specific to a particular environment(s). In the current study, we contributed to the literature by enrolling mTBI/PTSD, and we used the NIH Examiner Unstructured Task, which assesses cognitive processing for complex tasks using a standardized test, with relevance to any environment.

In the current study, GMT significantly improved cTOL (time to completion), indicating improved efficiency. These prior studies reported varying results, according to TOL variables. One study reported no improvement (64). The other two studies enrolled severe TBI and reported significant improvement on two TOL variables, the “TOL achievement score” and the ‘TOL rule violation score (65, 66). The achievement score is the sum of nine puzzles that are correctly built within the time frame with the least number of moves possible; the rule violation score is the number of times the participant picks up more than one disk at a time or places a larger disc onto a smaller disc. Both studies had significant improvement on the achievement score and rule violations (65, 66). We found significant gain in cTOL in a prior study (25) and in the current study (p < .01).

Our descriptive data showed varied results that potentially support the individual nature of treatment response in those with mTBI and PTSD and thus, the need for precision neurorehabilitation. First, for example, for the group, the negative slope of the regression line for the Unstructured Task (Figure 1) shows the general trend that those with lower baseline scores showed greater change in response to treatment vs. those with higher baseline scores. Second, and in contrast, there was little to no relationship between executive function gains and depression level, PTSD severity, and IQ, according to group correlation analyses. Third, however, two subjects at the extremes of either high symptom severity or low symptom severity showed least or most improvement, respectively, as might be expected. There may be a complex interaction of characteristics and treatment response, given the expected qualitative individual results for these two subjects, along with possible baseline influence (for example, Figure 1), as well as the absence of group correlation with symptomatology and IQ. These results do appear to call for a larger future study to explore individual characteristics and their contribution to treatment response. We designed the intervention according to precision medicine intervention criteria specifically because of the understanding of the individually unique constellation of symptoms, the requirements of GMT itself, the current dearth of literature regarding effective executive function intervention, and improvement for those with mTBI and PTSD. Therefore, we were able to administer treatment that was customizable; taking into account the constellation of symptoms for a given individual. Our results showed that the GMT protocol along with the VTM practice “buddy” could address the unique complexities of mTBI/PTSD, while treating the individual, even in the presence of the obstacles inherent in the complexities of pathologies and cognitive impairments in mTBI/PTSD. Because the GMT is manualized and the VTM is available as a Smartphone app, this combined intervention could, as warranted, be easily deployed to clinical practice.

In this single-cohort study, we would like to note a number of limitations. First, the sample size was small, indicating the standard limitations of generalization of the results. For example, a larger sample size should be studied regarding the influence of individual characteristics on treatment response, including the correlation analyses provided here. Second, given the heterogeneous nature of TBI, there are limitations to generalization of these findings. Third, in this study, we acquired and report two secondary measures, without multiple p-value corrections; therefore, these results should be considered in that light. Fourth, we did not acquire follow-up data at a time point following the end of treatment. At the same time, given the need and the promising results, a larger randomized, controlled trial can be justified for GMT with Veterans Task Manager for those with mTBI and PTSD.