INTRODUCTION
Self-awareness (SA), defined as the ability to recognize problems caused by damaged brain function, is commonly impaired in patients with traumatic brain injury (TBI) (Ben-Yishay et al., Reference Ben-Yishay, Rattok, Lakin, Piasetsky, Ross, Silver, Zide and Ezrachi1985; Prigatano et al., Reference Prigatano, Fordyce, Zeiner, Roueche, Pepping and Wood1986; Sherer et al., Reference Sherer, Hart and Nick2003a). Disturbance of SA may cause reduced motivation for rehabilitation (Malec & Moessner, Reference Malec and Moessner2001) and may interfere with safe and independent functioning (Flashman et al., Reference Flashman, Amador and McAllister1998), leading to poor outcome and difficulty with community integration (Trudel et al., Reference Trudel, Tyron and Purdum1998) and employability (Sherer et al., Reference Sherer, Hart, Nick, Whyte, Thompson and Yablon2003b).
Crosson et al. (Reference Crosson, Barco, Velozo, Bolesta, Cooper, Werts and Brobeck1989) divided awareness into the following areas: intellectual awareness, which represents patients' ability to describe their deficits or impaired functioning; emergent awareness, which represents patients' ability to recognize their difficulties as they are happening; anticipatory awareness, which represents patients' ability to predict when difficulties will arise because of their deficits.
Recent studies have differentiated between metacognitive knowledge (or declarative knowledge) about one's abilities (which incorporates elements of intellectual awareness), and online monitoring of performance during tasks (which relates to emergent awareness and anticipatory awareness) (O'Keeffe et al., Reference O'Keeffe, Dockree, Moloney, Carton and Robertson2007; Toglia & Kirk, Reference Toglia and Kirk2000).
To assess metacognitive SA deficits (O'Keeffe et al., Reference O'Keeffe, Dockree, Moloney, Carton and Robertson2007) the most commonly used scales, that is, the Patient Competency Rating Scale (PCRS) (Prigatano et al., Reference Prigatano, Fordyce, Zeiner, Roueche, Pepping and Wood1986) and the Dysexecutive Questionnaire (DEX) (Wilson et al., Reference Wilson, Alderman, Burgess, Esmlie and Evans1996), compare self-reports of competencies to reports of significant others.
As an alternative to these scales, Sherer et al. (Reference Sherer, Bergloff, Boake, High and Levin1998) developed the Awareness Questionnaire (AQ), consisting of 17 items that evaluate patients' current functional abilities compared with their preinjury abilities. AQ items are rated on a Likert scale ranging from 1 (much worse) to 5 (much better). Scores vary from 17 to 85, with a score of 51 indicating that the patient is functioning “about the same” as his/her preinjury level (Sherer et al., Reference Sherer, Hart and Nick2003a). Like the PCRS and the DEX, the AQ also includes forms for patient self ratings as well as family/significant other and clinician ratings. The degree of the SA deficit is calculated by subtracting family/significant other ratings or clinician ratings from patient self ratings. These discrepancy scores can range from −68 to 68. Higher discrepancy scores are associated with more severe SA deficits, while negative scores are rare and might show a patient's overestimation of his impairment (Cicerone, Reference Cicerone1991; Prigatano & Altman, Reference Prigatano and Altman1990), possibly due to a high level of emotional distress (Fleming et al., Reference Fleming, Strong and Ashton1998; Godfrey et al., Reference Godfrey, Partridge, Knight and Bishara1993) or to the development of self-limiting belief systems in which TBI patients overrate the effects of their injury in everyday life (Moore & Stambrook, Reference Moore and Stambrook1995).
Reliability studies of the AQ revealed internal consistencies (Chronbach's α = 0.88) for both patient and family ratings; however, test–retest reliability has not been reported. Factor analysis of the AQ (Sherer et al., Reference Sherer, Bergloff, Boake, High and Levin1998) revealed three subscales: motor-sensory (four items), cognition (seven items), and behavioral-affective (six items).
However, whichever scales were used many issues about SA impairment in TBI patients are still being debated. Severity of brain injury correlated with measures of impaired awareness in some studies (Leathem et al., Reference Leathem, Murphy and Flett1998; Prigatano & Altman, Reference Prigatano and Altman1990), but not in others (Anson & Ponsford, Reference Anson and Ponsford2006; Bach & David, Reference Bach and David2006; O'Keeffe et al., Reference O'Keeffe, Dockree, Moloney, Carton and Robertson2007; Port et al., Reference Port, Willmott and Charlton2002).
Also, the correlation between SA deficits and neuropsychological disturbances is not clear (Allen & Ruff, Reference Allen and Ruff1990; Boake et al., Reference Boake, Freeland, Ringholz, Nance and Edwards1995) even if executive functions (EF), in particular, are frequently impaired in TBI patients (Mattson & Levin, Reference Mattson and Levin1990; Stablum et al., Reference Stablum, Mogentale and Umiltà1996) and are considered to influence degree of SA following brain damage (Hart et al., Reference Hart, Whyte, Kim and Vaccaro2005; Noè et al., Reference Noè, Ferri, Caballero, Villadre, Sanchez and Chirivella2005).
EF are part of a very complex system that includes behavioral, affective, motivational, and cognitive components (Apollonio et al., Reference Apollonio, Leone, Isella, Piamarta, Consoli, Villa, Forapani, Russo and Nichelli2005). Cognitive disorders suggestive of the dysexecutive syndrome include response initiation and response suppression, focused attention, maintenance and shifting of set, rule deduction, problem solving and planning, and information generation (Apollonio et al., Reference Apollonio, Leone, Isella, Piamarta, Consoli, Villa, Forapani, Russo and Nichelli2005). As a consequence, there is no comprehensive test for the executive system and many tests have been proposed to analyze its single aspects. However, it has been demonstrated that the Wisconsin Card Sorting Test (WCST) is an effective measure of multiple components, such as abstract reasoning, and that it provides data on problem solving, the ability to use response feedback information, the cognitive flexibility, set-shifting and set-persistence capacity, concept identification, hypothesis generation (Hanks et al., Reference Hanks, Rappaport, Millis and Deshpande1999; Mukhopadhyay et al., Reference Mukhopadhyay, Dutt, Das, Basu, Hazra, Dhibar and Roy2008). In particular, in this test perseverative errors have been shown to be an excellent measure of executive dysfunction (Johnstone et al., Reference Johnstone, Hexum and Ashkanazi1995).
In exploring EF, the Verbal Fluency (VF) test also engages several cognitive processes such as working memory, self-monitoring, and cognitive flexibility (Schwartz et al., Reference Schwartz, Baldo, Graves and Brugger2003); another well-established test, the Tower of London (ToL) (Unterrainer & Owen, Reference Unterrainer and Owen2006), has been used extensively to evaluate planning ability in patients with neuropsychological disorders and in control subjects.
Many authors have tried to correlate executive dysfunction with impaired SA in TBI patients, but no conclusive findings have been reported. Bach and David (Reference Bach and David2006), who investigated SA deficits by means of the PCRS, failed to demonstrate that EF disorders (explored by the VF Test, the Trial Making Test and the Gambling Task) are associated with reduced behavioral/social SA. Conversely, using the DEX and the Self Awareness of Deficit Interview (SADI) (Fleming et al., Reference Fleming, Strong and Ashton1996) to evaluate SA deficits, Bogod et al. (Reference Bogod, Mateer and MacDonald2003) found a correlation between the dysexecutive syndrome (diagnosed by means of the Go–no-go Task, the Victoria Stroop Test, and the Self-ordered Pointing Test–SOPT) and unawareness. In a group of 31 patients with TBI, O'Keeffe et al. (Reference O'Keeffe, Dockree, Moloney, Carton and Robertson2007) showed that low SA and high SA groups did not differ on any standard neuropsychological task, but that those with low SA were more likely to exhibit disinhibition, interpersonal problems, and greater difficulty in overall competency.
Finally, Noè et al. (Reference Noè, Ferri, Caballero, Villadre, Sanchez and Chirivella2005) evaluated EF by means of the WCST, the Color Trial Making Test and the VF Test and demonstrated a significant correlation between poor WCST performance and low SA (assessed by the PCRS) in a heterogeneous acquired brain injury population in which both professionals and family members were enrolled as significant others.
The purpose of the present study was to identify the clinical, neuropsychological, neuropsychiatric, and functional variables that correlate with metacognitive self-awareness (SA) in severe traumatic brain injury (TBI) outpatients. Also, the influence of the same indicators on the sensory-motor, cognitive, and behavioral-affective domains of the SA scale was investigated separately.
METHODS
Participants
A total of 40 consecutive severe TBI outpatients were enrolled from May 2006 to June 2007 in the Post-Coma Unit of Santa Lucia Foundation, a neurorehabilitation hospital and research institute in Rome. The study was approved by the Santa Lucia Foundation ethical committee.
The study sample was recruited from an overall sample of 76 outpatients evaluated in that period on the basis of the following inclusion and exclusion criteria: inclusion criteria: (1) age ≥ 15 years; (2) diagnosis of severe TBI (GCS ≤ 8; Teasdale & Jennett, Reference Teasdale and Jennett1974); (3) posttraumatic amnesia (PTA) resolution; (4) capacity to undergo formal psychometric evaluation despite cognitive and sensory-motor deficits; (5) absence of aphasia; (6) availability of informed consent; exclusion criteria: previous history of drug and alcohol addiction, psychiatric diseases, and repeated TBI.
Three patients were excluded after enrollment because they refused to complete the test battery (one patient because of an excessive and unexpected fatigability; the other two because of low tolerance of frustration).
Finally, we evaluated 37 patients, 29 males (78%) and 8 females (22%), with a mean age of 32.3 ± 11.6 years and a mean of 12.6 ± 3.1 years of education. The median interval in years from injury to date of assessment (chronicity) was 0.69 (Interquartile Range, IQR: 0.45/8.52). All patients had severe TBI with a median time to follow commands (TFC − coma duration) of 20 days (IQR: 12/40) and a median PTA length of 60 days (IQR: 40/140). TFC was defined as the interval, in days, from coma onset until the patient was able to follow simple commands. PTA was evaluated prospectively by means of the Galveston Orientation and Amnesia Test (GOAT) (Levin et al., Reference Levin, O'Donnel and Grossman1979) in 23 patients, who were previously hospitalized in our rehabilitation unit as inpatients. It was calculated retrospectively, on the basis of information given by patients and family members, for the last 14 patients who had already recovered from PTA at the time of admission to our rehabilitation hospital. In fact, since it has been reported that retrospective analysis is correlated (r = 0.87) with prospective investigation of PTA (McMillan et al., Reference McMillan, Jongen and Greenwood1996), retrospective and prospective evaluations of PTA were considered equivalent.
Assessment
A neuropsychological battery was administered to all patients and the following assessment was made: memory: Digit Span Test (forward and backward) (Orsini, Reference Orsini2003), Prose Memory Test (Novelli et al., Reference Novelli, Papagno, Capitani, Laiacona, Vallar and Cappa1986); executive functioning: WCST (Heaton et al., Reference Heaton, Chelune, Talley, Kay and Curtiss1993, Reference Heaton, Chelune, Talley, Kay, Curtiss, Hardoy, Carta, Hardoy and e Cabras2000), ToL (Krikorian et al., Reference Krikorian, Bartok and Gay1994), VF Test (Novelli et al., Reference Novelli, Papagno, Capitani, Laiacona, Vallar and Cappa1986); attention: Go–No Go Test of the Test Batterie zur Aufmerksamkeitsprüfung (Zimmerman & Fimm, Reference Zimmerman and Fimm1992).
Neuropsychiatric disturbances were evaluated by means of the Neuropsychiatric Inventory (NPI) (Cummings, Reference Cummings1994). Functional assessment included the following scales: the Disability Rating Scale (DRS) (Rappaport et al., Reference Rappaport, Hall, Hopkins, Belleza and Cope1982), the Levels of Cognitive Functioning Scale (LCF-S) (Hagen et al., Reference Hagen, Malkmus and Durham1972), and the Glasgow Outcome Scale Extended (GOS-E) (Wilson et al., Reference Wilson, Pettigrew and Teasdale1998).
SA level was measured by means of AQ, which was completed by both the patient and a family member in all cases. Only first-degree relatives who were living with patients or at least had daily contact with them were enrolled. Scores obtained by patients and family members on each AQ subscale were calculated as well. Possible SA deficits were evaluated according to the discrepancy between self rating and family rating (Ownsworth et al., Reference Ownsworth, Fleming, Strong, Radel, Chan and Clare2007; Pagulayan et al., Reference Pagulayan, Temkin, Machamer and Dikmen2007; Prigatano, Reference Prigatano1996; Sherer et al., Reference Sherer, Bergloff, Boake, High and Levin1998; Walker et al., Reference Walker, Blankenship, Ditty and Lynch1987). We chose relatives instead of the clinician as significant other raters because the former are in the best position to judge the patient's functional ability in daily life, especially compared with his/her premorbid functioning.
In previous reports (Noè et al., Reference Noè, Ferri, Caballero, Villadre, Sanchez and Chirivella2005; Sherer et al., Reference Sherer, Hart and Nick2003a) a cutoff point for severity of SA deficits, based on different probabilities of employability associated with patient–clinician discrepancy, was established for the AQ. We decided not to use this cutoff point because in our study the “significant other” was always a member of the patient's family, not the clinician.
Statistical Analysis
AQ discrepancy scores for the questionnaire as a whole and for each subscale were computed by subtracting the relative rating from the patient's self rating. Pearson's correlation coefficients were calculated to study the correlation between AQ discrepancy scores and single variables. For these analyses, p values were calculated using Hommel's multiple-comparison procedure (Hommel, Reference Hommel1988). A value of p below 0.05 (two-tail) was considered statistically significant. All analyses were performed using SPSS for Windows, version 10.0.
RESULTS
Table 1 shows the distribution of neuropsychological data, in terms of means, medians, and IQR range.
Distribution of neuropsychological data
Note
WCST, Wisconsin Card Sorting Test.
When all patients' AQ data was considered, a discrepancy (which ranged from −18 to 28) was detected between self ratings and relative ratings. Negative scores were found in 9 patients who were well preserved cognitively. Discrepancy was higher for the behavioral-affective and cognitive subscales than for the sensory–motor subscale. Data regarding the AQ scale and subscales are shown in Table 2.
AQ scale and subscale discrepancy scores
Note
AQ = Awareness Questionnaire.
There were no significant correlations between the AQ patient–relative discrepancy scores and the clinical and functional variables (see Table 3).
Correlations of clinical and neuropsychological data with AQ discrepancy score
Note
PTA = Post-Traumatic Amnesia; DRS = Disability Rating Scale; LCF = Levels of Cognitive Functioning; GOS-E = Glasgow Outcome Scale Extended; WCST = Wisconsin Card Sorting Test.
* p < 0.05.
** p < 0.01.
The p values were calculated using Hommel's multiple-comparison procedure.
Conversely, the AQ discrepancy score was significantly correlated with the WCST number of categories completed (p = .027) and the WCST perseverative responses (p = .005), but not with the WCST nonperserverative errors (p = .26). None of the other neuropsychological variables analyzed correlated with the AQ discrepancy scores.
As shown in Table 4, correlations between the WCST variables and the AQ discrepancy scores were substantially confirmed for the AQ discrepancy subscores. In particular, the correlations seemed stronger for the cognitive subscale than for the behavioral-affective and sensory–motor subscales.
Correlations of WCST with AQ discrepancy subscores
* p < .05.
** p < .01.
The p values were calculated using Hommel's multiple-comparison procedure.
DISCUSSION
The main result of this study was the following: decreased metacognitive self-awareness is significantly correlated with increased problems in some components of executive system, even when the AQ subscales were considered separately.
Consistent with previous studies (Bach & David, Reference Bach and David2006; Borgaro & Prigatano, Reference Borgaro and Prigatano2002; Leathem et al., Reference Leathem, Murphy and Flett1998; Noè et al., Reference Noè, Ferri, Caballero, Villadre, Sanchez and Chirivella2005; O'Keeffe et al., Reference O'Keeffe, Dockree, Moloney, Carton and Robertson2007; Port et al., Reference Port, Willmott and Charlton2002; Prigatano & Altman, Reference Prigatano and Altman1990), we found no correlation between either the SA and chronicity or between the SA and length of PTA.
Conversely, in agreement with Noè et al. (Reference Noè, Ferri, Caballero, Villadre, Sanchez and Chirivella2005) our findings demonstrate a significant correlation between some components of executive functions (flexibility and ability to inhibit the response, ability to benefit from feedback, shifting of set, problem solving), assessed by the WCST, and the metacognitive SA. Unlike that study, we used the AQ (instead of the PCRS), because it compares patients' current functional ability with their preinjury condition. We enrolled only relatives as significant others, rather than professionals or family members, as in the study by Noè et al. Moreover, our sample was more homogeneous because all subjects were severe TBI outpatients who had recovered from PTA and were able to answer the questionnaires reliably; in the study by Noè et al. instead, the patients had sustained different types of acquired brain injury, TBI was not severe in all cases, and some of them were still in PTA.
Our findings expand upon the results of Noè et al. (Reference Noè, Ferri, Caballero, Villadre, Sanchez and Chirivella2005) on the correlation between pathological number of categories and low SA. In fact, we found a correlation between low SA levels and pathological number of categories completed and percentage of perseverative responses, which is a well-known index of poor flexibility and inability to inhibit the response (Johnstone et al., Reference Johnstone, Hexum and Ashkanazi1995).
Considering the significant correlation between the above-mentioned WCST scores and the low metacognitive SA levels, the lack of any relationship between AQ scores and other measures of EF, such as the VF Test and the ToL scores, is worth noting. Perhaps this was due to the fact that the different tests used to assess EF measure different aspects of these functions. While a successful performance on WCST requires some abilities as the cognitive flexibility, set-shifting and set-persistence capacity, concept identification, hypothesis generation, and the ability to use response feedback information (Mukhopadhyay et al., Reference Mukhopadhyay, Dutt, Das, Basu, Hazra, Dhibar and Roy2008), the ToL, particularly in the version used in the present study (Krikorian et al., Reference Krikorian, Bartok and Gay1994), primarily assesses planning and problem solving (Unterrainer & Owen, Reference Unterrainer and Owen2006). Moreover, it has been pointed out that while the WCST and the VF Test discriminate clearly between severe TBI patients and control subjects, the ToL test does not (Cockburn, Reference Cockburn1995). Furthermore, although poor performance on VF tests is generally interpreted as reflecting executive dysfunction (Phillips, Reference Phillips1999) and suggesting frontal lobe damage (Benton, Reference Benton1968), little is known about the cognitive processes involved in fluency tasks (Light, Reference Light, Craik and Salthouse1992; Randolph et al., Reference Randolph, Braun, Goldberg and Chase1993).
The correlations between SA and EF were further analyzed using DEX (Wilson et al., Reference Wilson, Alderman, Burgess, Esmlie and Evans1996), a significant other scale which investigates the awareness of cognitive, affective, and behavioral aspects of the dysexecutive syndrome. When Bogod et al. (Reference Bogod, Mateer and MacDonald2003) compared the DEX questionnaire and the SADI (Fleming et al., Reference Fleming, Strong and Ashton1996) with tests of EF and intelligence quotient (IQ), they found that the SADI correlated better than the DEX with measures of frontal lobe functioning and injury severity. Moreover, Hart et al. (Reference Hart, Whyte, Kim and Vaccaro2005) performed a composite EF assessment (called executive composite, EC) and showed that individuals with TBI who had low EC scores had statistically significant worse SA than controls when assessed by DEX. In our opinion, the present study is not comparable with these previous reports because we investigated different aspects of the executive system and evaluated awareness using different approaches.
Awareness is not a unitary concept and aspects of awareness can be differentiated and linked to different areas of daily functioning. In fact, the correlation between EF deficits and impaired SA was still present even when different AQ subscales were considered, confirming the close relationship between these cognitive functions and SA. Moreover, our data seem to indicate that SA is more impaired with respect to cognitive and social-emotional components and less impaired for physical deficits.
Our study presents some limitations. We included a sample of family members whose ratings may have been unreliable because of their high distress levels (Fleming et al., Reference Fleming, Strong and Ashton1996), also due to the patient's chronicity. In fact, the long-term of the caregiver efforts to cope with the posttrauma condition and the change of quality of their previous/current relationship with the patient worsen with time, because of the lack of further improvement. However, we chose relatives instead of the clinician as significant other raters because the former are in the best position to compare the patient's functional ability in daily life with his/her premorbid functioning, as required by the AQ. Moreover, as only outpatients were included in our study sample, relatives were presumably able to provide more accurate information about performance because they were familiar with the patient's personality.
We preferred to use the AQ instead of the PCRS to evaluate the functional implications of postinjury deficits in determining life changes after TBI because it compares patients' current functional abilities with their preinjury ones; in fact, there are no previous reports of the use of this scale to measure SA compared with patient's premorbid condition. On the other hand, as we did not include an additional “gold standard” measure, such as the PCRS, to provide concurrent validity, only a comparative analysis between the PRCS and the AQ would be able to provide conclusive results.
The present study is also limited by the small sample size, which may not have sufficient power to detect associations between clinical and functional variables and SA.
Studies based on larger samples are needed to investigate more thoroughly the neuropsychological disorders correlated with SA deficits and the possibility that the correlation between SA deficits and executive dysfunction depends on anatomical coincidence, as the frontal lobes are involved in both EF and SA. Moreover, it would be important to focus on a multidimensional assessment of awareness across the three levels of SA, as outlined by different clinical models (Crosson et al., Reference Crosson, Barco, Velozo, Bolesta, Cooper, Werts and Brobeck1989; Toglia & Kirk, Reference Toglia and Kirk2000) and suggested by O'Keeffe et al. (Reference O'Keeffe, Dockree, Moloney, Carton and Robertson2007).
However, the significant correlation between EF and metacognitive SA strongly suggests the importance of integrating an overall assessment of cognitive functions with a specific evaluation of self-awareness and of treating self-awareness contextually in a structured comprehensive rehabilitation program (Port et al., Reference Port, Willmott and Charlton2002). In fact, the inclusion of self-awareness in a multi-disciplinary rehabilitation program (Dirette, Reference Dirette2002) might enhance patients' self-awareness and participation in cognitive and functional tasks.
