Abstract

One hundred persons with mild traumatic brain injury (TBI) and their informants completed the Behavior Rating Inventory of Executive Function—Adult Version (BRIEF–A) within 1–12 months after injury. Exploratory maximum-likelihood factor analysis with oblique rotation revealed that although a traditional 2-factor model fit the informant-report data well, a 3-factor solution fit the self-report data relatively best. These factors were labeled Metacognition, Behavioral Regulation, and Emotional Regulation. The presence of a premorbid history of outpatient psychiatric treatment was strongly predictive of higher scores (reflecting more perceived problems) on each of these 3 factors. Lower educational attainment was associated with higher scores on the Behavioral Regulation factor, whereas absence of intracranial findings on neuroimaging was associated with higher scores on the Emotional Regulation factor. It is concluded that, after mild TBI, self-report data on the BRIEF–A can be interpreted along a 3-factorial model and that high elevations on this instrument are strongly affected by premorbid complications.

Introduction

The evaluation of executive functions is an integral part of neuropsychological assessment because of their presumed relevance to real-life behavior in various clinical populations (Laloyaux et al., 2014; Scott et al., 2011). Executive functions can be broadly defined as a group of interrelated cognitive processes and skills that have a coordinating and organizing role for goal-directed and future-oriented behavior (Anderson, 2008). Although the specific functions subsumed under this rubric vary in the literature, most often they include both cognitive (e.g., direction of attention, planning, and implementation of problem-solving activity) and emotional/behavioral (e.g., adjustment to change and inhibitory control) components. These processes are dependent on integrated brain networks that involve reciprocal connections between the prefrontal cortex and subcortical structures such as the basal ganglia, cerebellum, and thalamus (Bonelli & Cummings, 2007). There are, however, concerns about the ecological validity of traditional laboratory tests of executive functioning. Even in cognitively intact individuals, the degree to which such tests account for variance in daily functioning is typically fairly modest (Mitchell & Miller, 2008; Van der Elst, Van Boxtel, Van Breukelen, & Jolles, 2008). This may be because the assessment environment of most neuropsychological evaluations has limited resemblance to the real world, possibly providing so much structure that it might obscure subtle executive function deficits (Chaytor & Schmitter-Edgecombe, 2003). For these reasons, standardized rating scales of actual behavior have become increasingly used as a supplement to traditional laboratory tests. The Behavior Rating Inventory of Executive Function―Adult Version (BRIEF–A) (Roth, Isquith, & Gioia, 2005) is one of such rating scales. The purpose of this evaluation was to evaluate the latent structure of the BRIEF–A in a clinical sample of persons with mild traumatic brain injury (TBI).

Several studies have demonstrated sensitivity of the BRIEF–A to executive dysfunction in conditions ranging from mild cognitive impairment (Rabin et al., 2006) to focal lesions of the orbitofrontal cortex (Løvstad et al., 2012). In addition, there is evidence for significant correlations between BRIEF–A findings and ratings of competence in daily life after moderate to severe TBI (García-Molina, Tormos, Bernabeu, Junqué, & Roig-Rovira, 2012), although BRIEF–A ratings were not meaningfully associated with passing or failing a driving assessment in another investigation of persons with various forms of acquired brain injury (Rike, Ulleberg, Schultheis, Lundqvist, & Schanke, 2014).

In a recent study in our laboratory in persons with mild TBI (Donders, Oh, & Gable, 2015), higher levels of education and the presence of intracranial findings on neuroimaging were associated with lower (i.e., better) BRIEF–A ratings, whereas longer time since injury and premorbid outpatient psychiatric treatment were associated with higher (i.e., worse) BRIEF–A ratings. It was also notable that the self-ratings of persons with mild TBI were, on average, about half a standard deviation higher (i.e., reflecting worse perceived functioning) than those of their informants. These findings suggested that subjective perceptions of poor executive functioning in individuals with mild TBI were related primarily to premorbid factors instead of brain impairment. However, that study utilized the traditional two summary scales of the BRIEF–A, that is, Behavioral Regulation and Metacognition. There is an emerging evidence to suggest that the BRIEF–A may actually be better conceptualized along a 3-factor model. Roth, Lance, Isquith, Fischer, and Giancola (2013) used confirmatory factor analysis of the BRIEF–A in a large sample of healthy adults and found some support for splitting the Behavioral Regulation factor into two separate Behavioral Regulation (composed of the Inhibit and Self-Monitor scales) and Emotional Regulation (composed of the Emotional Control and Shift scales) factors. The remaining scales (Initiate, Working Memory, Plan/Organize, Organization of Materials, and Task Monitor) defined the third factor, Metacognition.

The three-factor model for the BRIEF–A has not yet been adequately validated in specific clinical populations. Power, Dragović, and Rock (2012) did perform a principal component analysis of BRIEF–A ratings by a mental health practitioner in a schizophrenia sample, but used only a limited pool of items from the instrument, which compromised external validity. The purpose of the present investigation was to determine whether a one-, two-, or three-factor model might best represent the constructs measured by the BRIEF–A in persons with mild TBI. This was considered important because it might aid clinicians in the assessment of patients' functioning. Specifically, a better understanding of the degree to which this instrument measures different latent constructs, as well as the nature of such constructs, could help clinicians make a more nuanced analysis of the individual's daily functioning. Mild TBI is also one of the most common acquired conditions that neuropsychologists may encounter in their clinical practice, as it accounts for about 800,000 emergency room visits in the United States per year (Langlois, Rutland-Brown, & Wald, 2006). In addition, test instruments need to be validated for the specific purposes and with the various clinical populations with which they are used (American Educational Research Association, American Psychological Association, & National Council on Measurement in Education, 2014). Furthermore, the study by Roth and colleagues (2013) was based only on the self-report form of the BRIEF–A. No prior study has evaluated the factor structure of this instrument for both self-report and informant report.

Because of the paucity of the literature on latent constructs measured by the BRIEF–A, particularly with regard to the informant version, no a priori hypotheses were formulated with regard to which factor structure would best fit the data.

Method

Participants

Participant selection procedures have been described in detail elsewhere (Donders et al., 2015) and this investigation used the same sample of participants. By means of summary, all participants had a diagnosis of mild TBI, were ≥18 years of age, did not have significant premorbid neurological or inpatient psychiatric histories, and had self- and informant BRIEF–A completed within 30–360 days after injury. The previous study did not address the factor structure of the BRIEF–A.

The final sample consisted of 100 patients with mild TBI who completed valid BRIEF–A self-report forms and their respective informants who completed valid BRIEF–A informant reports. Cases with critical elevations on any of the three BRIEF–A validity scales (Negativity, n = 4; Infrequency, n = 5; or Inconsistency, n = 3) had been eliminated. Mild TBI was defined as having duration to follow verbal commands <30 min, duration of post-traumatic amnesia <24 hr, and Glasgow Coma Scale scores (when available) >12. Persons with preinjury (e.g., outpatient psychiatric treatment; n = 49) or postinjury (e.g., financial compensation-seeking; n = 18) complicating factors were not excluded because we intended to investigate the relationship of such variables to any new factor scores.

BRIEF–A data were obtained from 52 male and 48 female patients with mild TBI and their informants (48 spouses, 34 parents, and 18 others) within 1–12 months after injury (M = 162.25 days, SD = 93.95). The patient sample was largely Caucasian (n = 92) and moderately educated (M = 13.56 years, SD = 2.67). The most common cause of injury was motor vehicle collision (n = 46). Almost a quarter of the patient sample (n = 23) had intracranial findings on neuroimaging, reflecting complicated mild injuries.

Procedures

This study was completed with approval from the Institutional Review Board at Mary Free Bed Rehabilitation Hospital. The BRIEF–A self- and informant-report versions both include 75 items of behavioral descriptors along a 3-point Likert scale (“never,” “sometimes,” and “often”), pertaining to the last month. These items divide into nine nonoverlapping clinical scales that are combined to form two broad-band indices, the Behavioral Regulation Index and the Metacognition Index, which in turn contribute to an overall Global Executive Composite. Traditionally, the Inhibit, Shift, Emotional Control, and Self-Monitor scales comprise the Behavioral Regulation Index, whereas the Initiate, Working Memory, Plan/Organize, Task Monitor, and Organization of Materials scales form the Metacognitive Index. All BRIEF–A variables are expressed as T scores (M = 50, SD = 10), with higher scores reflecting more reported problems.

Statistical Analyses

All statistical analyses were based on BRIEF–A clinical scale T scores. Exploratory factor analysis was used instead of confirmatory factor analysis because the latter would have required a much larger sample size. However, with nine BRIEF–A clinical scales, a sample size of 100 was sufficient to have an acceptable participant-to-variable ratio of >10:1 (Hatcher, 1994). Because we were interested in latent structure rather than data reduction, we used maximum-likelihood factor analysis instead of principal component analysis. Furthermore, in light of the theoretical likelihood that the factors in any multifactor model would be correlated with each other to some degree, we used an oblique instead of an orthogonal rotation. One-, two-, and three-factor models were considered for both the self-report data and the informant-report data. Smaller values of χ2/df (preferably ≤2) and higher values of the Tucker–Lewis index (preferably ≥0.95) were considered to represent better fit (Hatcher, 1994). In addition, we gave consideration to Akaike's information criterion, for which lower values are desirable. Differences between two competing models in the value of this criterion >4 suggest some support for the model with the relatively lowest value, whereas differences >10 offer unequivocally strong support for such a preference (Burnham & Anderson, 2004).

We saved factor scores for subsequent analyses to determine which variables predicted statistically significant variance in any new factors. For this purpose, we used linear regression analysis. We retained both statistically significant and nonsignificant variables in the regression model, as recommended in the literature (Millis, 2003) to allow reflection of their relative contributions. Variance inflation factors were all <1.17, ruling out collinearity (Fox, 1991). We considered the R2 statistic to determine the amount of variance accounted for by the regression models. Consistent with conventional standards (Murphy & Myors, 2004), we interpreted values <0.10 as small, values 0.10–0.25 as medium, and values >0.25 as large.

Results

Table 1 presents the average results on the BRIEF–A. Patients typically rated themselves as worse than their informants did (see Donders et al., 2015 for further details regarding informant vs. self-discrepancies). Table 2 presents the fit indices for the one-, two-, and three-factor models. Kaiser's measure of sampling adequacy was 0.90 for the informant-report data and 0.95 for the self-report data, with no values for any of the individual scales below 0.85, which is excellent by conventional standards for the purpose of factor analysis (Hatcher, 1994).

Table 1.

Informant and self-BRIEF–A ratings after mild traumatic brain injury (n = 100)

Clinical scale Informant
 
Self
 
M SD M SD 
Inhibit 55.24 11.58 59.61 12.34 
Shift 57.29 10.80 61.18 13.68 
Emotional Control 56.71 11.03 60.34 14.80 
Self-Monitor 52.43 11.54 55.97 13.17 
Initiate 59.37 11.41 62.44 14.22 
Working Memory 64.75 13.70 71.26 16.73 
Plan/Organize 57.29 11.18 62.23 13.61 
Task Monitor 57.77 11.19 61.84 13.46 
Organization of Materials 55.33 11.03 57.53 12.52 
Clinical scale Informant
 
Self
 
M SD M SD 
Inhibit 55.24 11.58 59.61 12.34 
Shift 57.29 10.80 61.18 13.68 
Emotional Control 56.71 11.03 60.34 14.80 
Self-Monitor 52.43 11.54 55.97 13.17 
Initiate 59.37 11.41 62.44 14.22 
Working Memory 64.75 13.70 71.26 16.73 
Plan/Organize 57.29 11.18 62.23 13.61 
Task Monitor 57.77 11.19 61.84 13.46 
Organization of Materials 55.33 11.03 57.53 12.52 

Note: BRIEF–A = Behavior Rating Inventory of Executive Function—Adult Version.

Table 2.

Fit indices for various factor models of the BRIEF–A

Model χ2 df χ2/df TLI AIC 
Informant report 
 One factor 111.85 27 4.14 0.81 63.18 
 Two factors 33.42 19 1.76 0.96 2.74 
Self-report 
 One factor 87.15 27 3.23 0.90 37.31 
 Two factors 37.78 19 1.99 0.95 1.86 
 Three factors 19.01 12 1.58 0.97 −3.80 
Model χ2 df χ2/df TLI AIC 
Informant report 
 One factor 111.85 27 4.14 0.81 63.18 
 Two factors 33.42 19 1.76 0.96 2.74 
Self-report 
 One factor 87.15 27 3.23 0.90 37.31 
 Two factors 37.78 19 1.99 0.95 1.86 
 Three factors 19.01 12 1.58 0.97 −3.80 

Notes: BRIEF–A = Behavior Rating Inventory of Executive Function—Adult Version; TLI = Tucker–Lewis index; AIC = Akaike's information criterion.

Inspection of Table 2 reflects that for the informant-report data, a one-factor model did not meet the a priori specified criteria for model fit. However, a two-factor model had clearly acceptable values of χ2/df and the Tucker–Lewis index. When an attempt was made to explore a three-factor model, the solution would not converge due to Heywood cases (i.e., communalities exceeding 1), which indicated instability. For these reasons, we concluded that there was no support for a three-factor model and that a two-factor model fit the BRIEF–A informant-report data sufficiently. Table 3 presents this model. Inspection of this table indicates that the factor structure pretty much mirrored the one that was originally presented by the BRIEF–A authors for informant reports in the manual for the instrument. These two factors were strongly correlated (r = .63). The composite reliability of this factor model, based on the methods and criteria described by Hatcher (1994), was excellent (0.83).

Table 3.

Final factor models for the BRIEF–A

Clinical scale Informant
 
Self
 
Factor 1 Factor 2 Factor 1 Factor 2 Factor 3 
Inhibit 0.18 0.87 0.36 0.55 0.06 
Shift 0.39 0.44 0.11 0.20 0.71 
Emotional Control 0.18 0.59 0.09 0.29 0.48 
Self-Monitor 0.03 0.89 0.01 0.85 0.13 
Initiate 0.56 0.31 0.26 0.44 0.31 
Working Memory 0.74 0.12 0.70 0.15 0.41 
Plan/Organize 0.87 0.07 0.74 0.21 0.08 
Task Monitor 0.91 0.02 0.58 0.12 0.28 
Organization of Materials 0.68 0.01 0.67 0.26 0.07 
Clinical scale Informant
 
Self
 
Factor 1 Factor 2 Factor 1 Factor 2 Factor 3 
Inhibit 0.18 0.87 0.36 0.55 0.06 
Shift 0.39 0.44 0.11 0.20 0.71 
Emotional Control 0.18 0.59 0.09 0.29 0.48 
Self-Monitor 0.03 0.89 0.01 0.85 0.13 
Initiate 0.56 0.31 0.26 0.44 0.31 
Working Memory 0.74 0.12 0.70 0.15 0.41 
Plan/Organize 0.87 0.07 0.74 0.21 0.08 
Task Monitor 0.91 0.02 0.58 0.12 0.28 
Organization of Materials 0.68 0.01 0.67 0.26 0.07 

Notes: BRIEF–A = behavior Rating Inventory of Executive Function—Adult Version. Factor loadings >0.40 are italicized.

The findings were different for the self-report data. Again, a one-factor model did not meet the a priori specified criteria for model fit, whereas a two-factor model was a minimally acceptable fit in terms of the χ2/df and Tucker–Lewis indices. However, in light of the fact that a three-factor model had a value of Akaike's information criterion that was more than 5 points lower than that of the two-factor model, while also having slightly better χ2/df and Tucker–Lewis indices, we concluded that there was psychometric support for a three-factor model for the BRIEF–A self-report data. In addition, the three-factor model appeared to be interpretable from a theoretical point of view, which is another important consideration in factor analysis. The composite reliability of the three-factor model (0.69) was somewhat lower than that of the two-factor model (0.78) but was still well above the minimum standard of 0.60 (Hatcher, 1994). Furthermore, none of the individual factors in the three-factor model had a value of Cronbach's α below 0.81, again supporting the reliability of this solution.

The findings in Table 3 suggest that, with regard to the self-report data, Factor 1 was a Metacognition factor, composed of Working Memory, Plan/Organize, Task Monitor, and Organization of Materials. One clinical scale that is included in the Metacognition index in the traditional scoring of BRIEF–A protocols (Initiate) did not have a strong loading on Factor 1 in this sample. Factor 2 resembled the Behavioral Regulation construct identified by Roth and colleagues (2013) but with the distinction that Initiate joined Inhibit and Self-Monitor in defining this factor. Finally, Factor 3 represented the Emotional Control construct described by Roth and colleagues (2013), with primary loadings by Shift and Emotional Control, although there was an additional secondary loading by Working Memory. Factor 1 was moderately correlated to Factor 2 (r = −.36) and to a larger extent to Factor 3 (r = −.50), whereas there was a small correlation between Factors 2 and 3 (r = −.29).

Because of the fact that the self-report data suggested a different factor structure than the one along which BRIEF–A data are typically reported, we explored correlates of the three identified constructs in that dataset. For this purpose, we used regression analysis and selected the following independent variables: level of education, interval between injury and assessment with the BRIEF–A, presence versus absence of a prior psychiatric history, presence versus absence of any other premorbid complicating factors (i.e., attention-deficit/hyperactivity disorder [ADHD], learning disability, personal abuse, or substance abuse), presence versus absence of financial compensation-seeking, and presence versus absence of an intracranial lesion on neuroimaging. Table 4 presents the resulting regression models.

Table 4.

Regression models for three BRIEF–A self-report factor scores

Variable Factor 1
 
Factor 2
 
Factor 3
 
SRC t p SRC t p SRC t p 
Years of completed education −0.09 −1.06 .29 −0.29 −3.18 .002 −0.11 −1.21 .23 
Days since injury 0.25 2.66 .009 0.12 1.26 .21 0.20 2.15 .04 
Prior psychiatric history 0.29 3.20 .002 0.34 3.63 .001 0.34 3.74 .001 
Prior other complicating factors 0.07 0.72 .48 0.17 1.80 .08 0.06 0.69 .49 
Financial compensation-seeking 0.16 1.71 .09 0.06 0.58 .57 0.11 1.14 .26 
Intracranial neuroimaging findings −0.29 −3.23 .002 −0.17 −1.86 .07 −0.29 −3.19 .002 
Variable Factor 1
 
Factor 2
 
Factor 3
 
SRC t p SRC t p SRC t p 
Years of completed education −0.09 −1.06 .29 −0.29 −3.18 .002 −0.11 −1.21 .23 
Days since injury 0.25 2.66 .009 0.12 1.26 .21 0.20 2.15 .04 
Prior psychiatric history 0.29 3.20 .002 0.34 3.63 .001 0.34 3.74 .001 
Prior other complicating factors 0.07 0.72 .48 0.17 1.80 .08 0.06 0.69 .49 
Financial compensation-seeking 0.16 1.71 .09 0.06 0.58 .57 0.11 1.14 .26 
Intracranial neuroimaging findings −0.29 −3.23 .002 −0.17 −1.86 .07 −0.29 −3.19 .002 

Notes: BRIEF–A = Behavior Rating Inventory of Executive Function—Adult Version; SRC = standardized regression coefficient.

The model for Factor 1 (Metacognition) was statistically significant, F(6, 93) = 6.85, p < .0001, and explained a large amount of the variance (adjusted R2 = .26). The presence of a prior psychiatric history as well as longer time since injury were both associated with worse self-ratings (i.e., higher factor scores), whereas the presence of intracranial findings on neuroimaging was associated with better self-ratings (i.e., lower scores on this factor).

The model for Factor 2 (Behavioral Regulation) was also statistically significant, F(6, 93) = 5.49, p< .0001, and explained a moderate amount of the variance (adjusted R2 = .21). The presence of a prior psychiatric history was associated with worse self-ratings (i.e., higher scores on this factor), whereas a higher level of education was associated with better self-ratings (i.e., lower factor scores).

Finally, the model for Factor 3 (Emotional Regulation) was also statistically significant, F(6, 93) = 6.50, p< .0001, and explained a moderate amount of the variance (adjusted R2 = .25). The presence of a prior psychiatric history was associated with worse self-ratings (i.e., higher factor scores), whereas the presence of intracranial findings on neuroimaging was associated with better self-ratings (i.e., lower factor scores). There was also a less influential but still statistically significant influence of duration of time since injury, with longer intervals being associated with worse self-ratings.

Discussion

The purpose of this investigation was to clarify the latent structure of the BRIEF–A in persons with mild TBI. For the informant reports, the factor solution was very comparable with that reported in the test manual for the standardization sample, including a Metacognition factor and a Behavioral Regulation factor. However, for the self-reports, a three-factor solution fit the data relatively better, with the second factor split into a Behavioral Regulation factor and an Emotional Regulation factor. These findings were compatible with those of Roth and colleagues (2013) in a large sample of healthy young adults. There was, however, one notable difference in that the Initiate scale loaded on the Behavioral Regulation factor instead of the Metacognition factor.

The Behavioral Regulation factor was defined by the Inhibit, Self-Monitor, and Initiate scales. The Inhibit scale pertains to the ability to resist or stop specific behaviors, whereas the Self-Monitor scale measures the degree to which a person can keep track of his/her behavior and its effects on others. Given that many of the items on the Initiate scale pertain to difficulties getting started with specific behaviors, the current factor loading pattern suggests that, at least in this sample of persons with mild TBI, the Initiate scale relates more to the behavioral and motivational aspects of daily functioning than to the metacognitive components of problem-solving.

The Emotional Regulation factor was defined by the Shift and Emotional Control scales. The former pertains to the level of comfort that an individual has with internal flexibility and adjusting to changes in the psychosocial environment, whereas the latter addresses one's ability to modulate emotional responses to events in that environment. An association between poor set shifting ability and emotional adjustment issues has been seen in other populations with psychological disorders. For example, Roberts, Tchanturia, and Treasure (2010) found poor set shifting to be associated with higher levels of depression and anxiety in those with eating disorders.

The fact that the Emotional Regulation factor was distinct from the Behavioral Regulation one suggests that, at least in persons with mild TBI, control over one's overt behavior and control over one's internal emotional reactions are two distinct constructs. This finding is compatible with functional neuroimaging studies that suggest differential engagement of frontal neural networks when controlling cognitive/behavioral versus emotional content (Kompus, Hugdahl, Ohman, Marklund, & Nyberg, 2009; Mohanty et al., 2007). In their study, Roth and colleagues (2013) also found differentiation of behavioral and emotional regulation in a sample of young adults with ADHD. Specifically, respondents with ADHD reported higher levels of behavioral versus emotional regulation problems on the BRIEF–A; furthermore, higher levels of depression and anxiety were correlated with the emotional regulation, but not with the behavioral regulation, factor. The distinction between behavioral and emotional regulation is particularly relevant in patients with mild TBI, given that emotional maladjustment postinjury is most predictive of the long-term outcome, above and beyond factors such as injury severity and neuroimaging findings (Waljas et al., 2015).

The finding that Working Memory also had a secondary loading on the Emotional Regulation factor was not anticipated. There is some literature to suggest that individuals who are more distressed have difficulties updating information in working memory, being slower to disengage from sad stimuli and faster to disengage from happy stimuli (Levens & Gotlib, 2010). It is also possible that high levels of perceived stress may deplete working memory capacity, which in turn may lead to cognitive inefficiency and reactive emotional distress (Jha, Stanley, Kiyonaga, Wong, & Gelfand, 2010). In the mild TBI population, in which no objective persistent cognitive impairment is expected, another likely explanation is that complaints of working memory dysfunction are more a function of mood problems than of objective working memory impairments. That is, depression is commonly associated with the poor outcome after a mild TBI (Waljas et al., 2015), and those with high levels of depression have been found to report significantly more problems with memory on self-report measures, despite objective test results suggesting intact cognitive performance (Rohling, Green, Allen, & Iverson, 2002). These possibilities will need further exploration in future research that also includes more specific measures of coping and adjustment as well as objective cognitive data.

The correlates of the factor scores in the self-report data may clarify further the distinction between the Behavioral Regulation and Emotional Regulation factors in persons with mild TBI. Both factors were affected strongly by a history of premorbid outpatient psychiatric treatment. Those with such prior needs rated themselves as doing much more poorly in both areas than did those without such a history. This is consistent with the fact that poor premorbid mental health is a significant risk factor for poor outcome after mild TBI (Carroll et al., 2004; Cassidy et al., 2014). Reasons for this may range from the possibility that persons with preexisting psychiatric dysfunction may be more vulnerable to diagnosis threat, affecting perceived self-efficacy (Trontel, Hall, Ashendorf, & O'Connor, 2013), and/or more likely to reattribute premorbid difficulties to the socially more acceptable impact of a “brain injury” (Iverson, Lange, Brooks, & Rennison, 2010).

For the Behavioral Regulation factor, an additional predictor was level of education. Advanced education was associated with fewer reported problems in this domain. This finding likely reflects greater resiliency of those with higher educational attainment. Similar findings regarding the protective effects of educational attainment have been found in other populations (e.g., Llewellyn et al., 2013; Pietrzak & Cook, 2013). In addition, a lower level of education may simply be a correlate of limited vocational achievement, consistent with the literature that limited independence or decision-making capacity in jobs is associated with the relatively poorer outcome after mild TBI (Cancelliere et al., 2014). It is possible that persons who have limited experience with responsibilities such as showing initiative and autonomy may lack self-confidence after mild TBI.

With regard to the Emotional Regulation factor, there were two variables (in addition to prior psychiatric history) that influenced the self-ratings. Positive neuroimaging findings were a strong predictor of fewer self-reported problems in this domain. It may be tempting to speculate that this may reflect lack of deficit awareness, but this is unlikely, because in our previous study (Donders et al., 2015), neuroimaging findings had no impact on BRIEF–A ratings by informants in this sample. It is relatively more likely that those with no objective findings on neuroimaging may feel even more inclined to reattribute preexisting psychiatric dysfunction to an injury that they presume to be more significant or permanent than it really is. The fact that there was an additional small but statistically significant influence of time since injury on this factor, with increasing intervals associated with worse self-ratings, suggests that this inclination may become more ingrained over time, leading to a magnified perception of deficits as related to mild TBI (Tsanadis et al., 2008). It is unlikely that this effect of interval since injury reflects worse recovery overall in those seeking care further from the time of injury because in our previous investigation, we found that these people still performed well within normal limits on various laboratory tests of executive functioning such as the Wisconsin Card Sorting Test (Heaton, Chelune, Talley, Kay, & Curtiss, 1993) and the Tower of London (Culbertson & Zillmer, 2005). On the other hand, it is possible that, to some extent, these findings may have been affected by selection bias as those with good immediate recovery would not have presented for specialty care in the first place.

It is important to appreciate that financial compensation-seeking was not a statistically significant predictor in any of the regression models. This was somewhat surprising because financial compensation-seeking has been reported to be associated with a risk for poor outcome after mild TBI (Carroll et al., 2004; Paniak et al., 2002). The current findings may be related in part to the fact that we eliminated from this investigation any BRIEF–A data that had elevations on the Negativity (raw score ≥6) or Infrequency (raw score ≥3) indices, which would have been suggestive of over-endorsement or otherwise atypical reporting. Another, and not mutually exclusive, possibility is that after accounting for the influence of premorbid psychiatric history, financial compensation-seeking may not always have incremental value in the prediction of neuropsychological outcomes after mild TBI. This would be consistent with a previous investigation in our laboratory with a completely independent sample (Donders & Boonstra, 2007). Thus, highly elevated BRIEF–A ratings after mild TBI may not necessarily suggest deliberate symptom magnification but rather reattribution of longstanding adjustment difficulties.

Limitations of this investigation must also be considered. We used a convenience sample of referred patients, and replication in an independent sample of consecutive emergency room admissions would be desirable. It is also important to appreciate that this investigation was limited to mild TBI, so the findings cannot simply be generalized to those with more severe injuries. Our sample was also largely Caucasian, so replication in an ethnically more diverse sample is still needed. Finally, it needs to be realized that the support for the three-factor model was modest and not truly unequivocal by conventional statistical standards. Replication in a larger sample with application of confirmatory factor analysis is a distinct goal for future research.

With these considerations in mind, we conclude that the BRIEF–A most likely measures three at least partially separable latent constructs in persons with mild TBI: Metacognition, Behavioral Regulation, and Emotional Regulation. This suggests a different model of interpretation of the results than suggested in the test manual. A goal for future research is to evaluate the degree to which these various constructs are meaningfully related to variables such as participation in society as well as satisfaction with life beyond the first year after mild TBI.

Conflict of Interest

None declared.

References

American Educational Research Association, American Psychological Association, & National Council on Measurement in
Education
. (
2014
).
Standards for educational and psychological testing
 .
Washington, DC
:
Author
.
Anderson
P. J.
(
2008
).
Towards a developmental model of executive function
. In
Anderson
V.
,
Jacobs
R.
,
Anderson
P. J.
(Eds.),
Executive functions and the frontal lobes: A lifespan perspective
  (pp.
3
21
).
New York
:
Taylor & Francis
.
Bonelli
R. M.
,
Cummings
J. L.
(
2007
).
Frontal-subcortical circuitry and behavior
.
Dialogues in Clinical Neuroscience
 ,
9
,
141
151
.
Burnham
K. P.
,
Anderson
D. R.
(
2004
).
Multimodel inference: Understanding AIC and BIC in model selection
.
Sociological Methods and Research
 ,
33
,
261
304
.
Cancelliere
C.
,
Kristman
V. L.
,
Cassidy
J. D.
,
Hincapié
C. A.
,
Côté
P.
,
Boyle
E.
et al
. (
2014
).
Systematic review of return to work after mild traumatic brain injury: Results of the International Collaboration on Mild Traumatic Brain Injury Prognosis
.
Archives of Physical Medicine and Rehabilitation
 ,
95
,
S201
S209
.
Carroll
L. J.
,
Cassidy
J. D.
,
Peloso
P. M.
,
Borg
J.
,
von Holst
H.
,
Holm
L.
et al
. (
2004
).
Prognosis for mild traumatic brain injury: Results of the WHO Collaborating Centre Task Force on Traumatic Brain Injury
.
Journal of Rehabilitation Medicine
 ,
43
,
84
105
.
Cassidy
J. D.
,
Cancelliere
C.
,
Carroll
L. J.
,
Côté
P.
,
Hincapié
C. A.
,
Holm
L. W.
et al
. (
2014
).
Systematic review of self-reported prognosis in adults after mild traumatic brain injury: Results of the International Collaboration on Mild Traumatic Brain Injury Prognosis
.
Archives of Physical Medicine and Rehabilitation
 ,
95
,
S132
S151
.
Chaytor
N.
,
Schmitter-Edgecombe
M.
(
2003
).
The ecological validity of neuropsychological tests: A review of the literature on everyday cognitive skills
.
Neuropsychology Review
 ,
13
,
181
197
.
Culbertson
W.
,
Zillmer
E.
(
2005
).
Tower of LondonDrexel University (
 
2nd ed
.).
Chicago
:
Multi-Health Systems
.
Donders
J.
,
Boonstra
T.
(
2007
).
Correlates of invalid neuropsychological test performance after traumatic brain injury
.
Brain Injury
 ,
21
,
319
326
.
Donders
J.
,
Oh
Y. I.
,
Gable
J.
(
2015
).
Self and informant ratings of executive functioning after mild traumatic brain injury
.
Journal of Head Trauma Rehabilitation
 ,
30
,
E30
39
.
Fox
J.
(
1991
).
Regression diagnostics
 .
Newbury Park, CA
:
Sage
.
García-Molina
A.
,
Tormos
J. M.
,
Bernabeu
M.
,
Junqué
C.
,
Roig-Rovira
T.
(
2012
).
Do traditional executive measures tell us anything about daily-life functioning after traumatic brain injury in Spanish-speaking individuals?
Brain Injury
 ,
26
,
864
874
.
Hatcher
L.
(
1994
).
A step-by-step approach to using the SAS system for factor analysis and structural equation modeling
 .
Cary, NC
:
SAS Institute
.
Heaton
R. K.
,
Chelune
C. J.
,
Talley
J. L.
,
Kay
G. G.
,
Curtiss
G.
(
1993
).
Wisconsin Card Sorting Test manual: Revised and expanded
 .
Odessa, FL
:
Psychological Assessment Resources
.
Iverson
G. L.
,
Lange
R. T.
,
Brooks
B. L.
,
Rennison
V. L. A.
(
2010
).
“Good old days” bias following mild traumatic brain injury
.
The Clinical Neuropsychologist
 ,
24
,
17
37
.
Jha
A. P.
,
Stanley
E. A.
,
Kiyonaga
A.
,
Wong
L.
,
Gelfand
L.
(
2010
).
Examining the protective effects of mindfulness training on working memory capacity and affective experience
.
Emotion
 ,
10
,
54
64
.
Kompus
K.
,
Hugdahl
K.
,
Ohman
A.
,
Marklund
P.
,
Nyberg
L.
(
2009
).
Distinct control networks for cognition and emotion in the prefrontal cortex
.
Neuroscience Letters
 ,
467
,
76
80
.
Laloyaux
J.
,
Van der Linden
M.
,
Levaux
M.-N.
,
Mourad
H.
,
Pirri
A.
,
Bertrand
H.
et al
. (
2014
).
Multitasking capacities in persons diagnosed with schizophrenia: A preliminary examination of their neurocognitive underpinnings and ability to predict real world functioning
.
Psychiatry Research
 ,
217
,
163
170
.
Langlois
J. A.
,
Rutland-Brown
W.
,
Wald
M. W.
(
2006
).
The epidemiology and impact of traumatic brain injury: A brief overview
.
Journal of Head Trauma Rehabilitation
 ,
21
,
375
378
.
Levens
S. M.
,
Gotlib
I. H.
(
2010
).
Updating positive and negative stimuli in working memory in depression
.
Journal of Experimental Psychology: General
 ,
139
,
654
664
.
Llewellyn
C. D.
,
Horney
D. J.
,
McGurk
M.
,
Weinman
J.
,
Herold
J.
,
Altman
K.
et al
. (
2013
).
Assessing the psychological predictors of benefit finding in patients with head and neck cancer
.
Psychooncology
 ,
22
,
97
105
.
Løvstad
M.
,
Funderud
I.
,
Endestad
T.
,
Due-Tønnessen
P.
,
Meling
T. R.
,
Lindgren
M.
et al
. (
2012
).
Executive functions after orbital or lateral prefrontal lesions: Neuropsychological profiles and self-reported executive functions in everyday living
.
Brain Injury
 ,
26
,
1586
1598
.
Millis
S. R.
(
2003
).
Statistical practices: The seven deadly sins
.
Child Neuropsychology
 ,
9
,
221
233
.
Mitchell
M.
,
Miller
L. S.
(
2008
).
Prediction of functional status in older adults: The ecological validity of four Delis-Kaplan Executive Function System tests
.
Journal of Clinical and Experimental Neuropsychology
 ,
30
,
683
690
.
Mohanty
A.
,
Engels
A. S.
,
Herrington
J. D.
,
Heller
W.
,
Ringo-Ho
M-H.
,
Banich
M. T.
et al
. (
2007
).
Differential engagement of anterior cingulate cortex subdivisions for cognitive and emotional function
.
Psychophysiology
 ,
44
,
343
351
.
Murphy
K. R.
,
Myors
B.
(
2004
).
Statistical power analysis
  (
2nd ed
.).
Mahwah, NJ
:
Lawrence Erlbaum
.
Paniak
C.
,
Reynolds
S.
,
Toller-Lobe
G.
,
Melnyk
A.
,
Nagy
J.
,
Schmidt
D.
(
2002
).
A longitudinal study of the relationship between financial compensation and symptoms after treated mild traumatic brain injury
.
Journal of Clinical and Experimental Neuropsychology
 ,
24
,
187
193
.
Pietrzak
R. H.
,
Cook
J. M.
(
2013
).
Psychological resilience in older U.S. veterans: Results from the national health and resilience in veterans study
.
Depression and Anxiety
 ,
30
,
432
443
.
Power
B. D.
,
Dragović
M.
,
Rock
D.
(
2012
).
Brief screening for executive dysfunction in schizophrenia in a rehabilitation hospital
.
Journal of Neuropsychiatry and Clinical Neurosciences
 ,
24
,
215
222
.
Rabin
L. A.
,
Roth
R. M.
,
Isquith
P. K.
,
Wishart
H. A.
,
Nutter-Upham
K. E.
,
Pare
N.
et al
. (
2006
).
Self and informant reports of executive function on the BRIEF–A in MCI and older adults with cognitive complaints
.
Archives of Clinical Neuropsychology
 ,
21
,
721
732
.
Rike
P.-O.
,
Ulleberg
P.
,
Schultheis
M. T.
,
Lundqvist
A.
,
Schanke
A.-K.
(
2014
).
Behavioral ratings of self-regulatory mechanisms and driving behavior after an acquired brain injury
.
Brain Injury
 ,
28
,
1687
1699
.
Roberts
M. E.
,
Tchanturia
K.
,
Treasure
J. L.
(
2010
).
Exploring the neurocognitive signature of poor set-shifting in anorexia nervosa and bulimia nervosa
.
Journal of Psychiatric Research
 ,
44
,
964
970
.
Rohling
M. L.
,
Green
P.
,
Allen
L. M.
,
Iverson
G. L.
(
2002
).
Depressive symptoms and neurocognitive test performance in patients passing symptom validity tests
.
Archives of Clinical Neuropsychology
 ,
17
,
205
222
.
Roth
R. M.
,
Isquith
P. K.
,
Gioia
G. A.
(
2005
).
Behavior Rating Inventory of Executive Function—Adult Version
 .
Lutz, FL
:
Psychological Assessment Resources
.
Roth
R. M.
,
Lance
C. E.
,
Isquith
P. K.
,
Fischer
A. S.
,
Giancola
P. R.
(
2013
).
Confirmatory factor analysis of the Behavior Rating Inventory of Executive Function—Adult Version in healthy adults and application to attention-deficit/hyperactivity disorder
.
Archives of Clinical Neuropsychology
 ,
28
,
425
434
.
Scott
J. C.
,
Woods
S. P.
,
Vigil
O.
,
Heaton
R. K.
,
Schweinsburg
B. C.
,
Ellis
R. J.
et al
. (
2011
).
A neuropsychological investigation of multitasking in HIV infection: Implications for everyday functioning
.
Neuropsychology
 ,
25
,
511
519
.
Trontel
H. G.
,
Hall
S.
,
Ashendorf
L.
,
O'Connor
M. K.
(
2013
).
Impact of diagnosis threat on academic self-efficacy in mild traumatic brain injury
.
Journal of Clinical and Experimental Neuropsychology
 ,
35
,
960
970
.
Tsanadis
J.
,
Montoya
E.
,
Hanks
R. A.
,
Millis
S. R.
,
Fichtenberg
N. L.
,
Axelrod
B. N.
(
2008
).
Brain injury severity, litigation status, and self-report of postconcussive symptoms
.
The Clinical Neuropsychologist
 ,
22
,
1080
1092
.
Van der Elst
W.
,
Van Boxtel
M. P. J.
,
Van Breukelen
G. J. P.
,
Jolles
J.
(
2008
).
A large-scale cross-sectional and longitudinal study into the ecological validity of neuropsychological test measures in neurologically intact people
.
Archives of Clinical Neuropsychology
 ,
23
,
787
800
.
Waljas
M.
,
Iverson
G. L.
,
Lange
R. T.
,
Hakulinen
U.
,
Dastidar
P.
,
Huhtala
H.
et al
. (
2015
).
A prospective biopsychosocial study of the persistent post-concussion symptoms following mild traumatic brain injury
.
Journal of Neurotrauma
 ,
32
,
534
547
.