Rehabilitation Psychology August 2000 Vol. 45, No. 3, 260-273 © 2000 by the Educational Publishing Foundation For personal use only--not for distribution. Do Neurocognitive Ability and Personality Traits Account for Different Aspects of Psychosocial Outcome After Traumatic Brain Injury? David J. Schretlen Department of Psychiatry and Behavioral Sciences Johns Hopkins University School of Medicine ABSTRACT ABSTRACT. Objective: To examine the contributions of demographic, injury, cognitive, and personality characteristics to psychosocial outcome 8 years after traumatic brain injury (TBI). Design: Multiple regression analyses were used to estimate the variance explained by putative "predictors" of psychosocial outcome. Participants: Thirty-nine TBI survivors and 39 family member informants. On the basis of Glasgow Coma Scale scores and Accident Injury Severity (head) ratings, the patients' brain injuries ranged from mild to critical in severity. Main Outcome Measures: One self-report measure combined putative markers of social role engagement, such as marital status and earned income. Another, based on informant ratings using the Katz Adjustment Scale, was conceptualized as reflecting behavioral adjustment. Results: Whereas cognitive functioning explained significant unique variation in social role engagement, it did not account for variance in behavioral adjustment. Conversely, whereas 3 personality trait ratings explained significant incremental variance in behavioral stability, only 1 did the same with respect to social role engagement. Conclusions: Social role engagement and behavioral adjustment appear to represent 2 related but distinguishable aspects of TBI outcome that are associated with different patient characteristics. Many follow-up studies of traumatic brain injury (TBI) have used the Glasgow Outcome Scale ( Jennet & Bond, 1975 ), Rancho Los Amigos Scale ( Hagen, 1982 ), or Disability Rating Scale ( Rappaport, Hall, Hopkins, Belleza, & Cope, 1982 ) to assess outcome. These measures combine such disparate outcomes as death, return to work, self-care, mobility, psychiatric morbidity, and cognitive impairment into single measures of global outcome. Although valuable, these scales do not differentiate among dimensions of outcome. As a result, they may obscure correlations between specific outcome domains and other patient or injury characteristics. Some studies have examined TBI outcome in terms of return to work, marital status, and need for psychiatric treatment ( Bayless, Varney, & Roberts, 1989 ; Brooks, McKinlay, Symington, Beattie, & Campsie, 1987 ; Fabiano & Crewe, 1995 ; Oddy, Coughlan, Tyerman, & Jenkins, 1985 ; Ruff et al., 1993 ; Thompsen, 1984 ). These can be conceptualized as studies of social role engagement. Others have emphasized selfreported emotional and behavior problems ( Diamond, Barth, & Zillmer, 1988 ; Dikmen & Reitan, 1977 ; Levin & Grossman, 1978 ) or informant ratings of social adjustment ( Fordyce, Roueche, & Prigatano, 1983 ; Hinkeldey & Corrigan, 1990 ; Klonoff, Costa, & Snow, 1986 ). These can be conceptualized as studies of behavioral adjustment. Relatively few studies have examined long-term outcomes after TBI in terms of both social role engagement and behavioral adjustment. In one such study, Bowman (1996) found that demographic and neurocognitive test performance accounted for more variance in long-term occupational outcome than in everyday behavioral impairment. Conversely, self-rated emotional problems accounted for more variance in everyday activity impairment than in occupational functioning. Initial injury severity accounted for relatively little variance in either outcome measure. These results suggest that demographic, injury severity, cognitive, and emotional variables affect different dimensions of psychosocial outcome after TBI to varying degrees. Of course, selfreported emotional distress measured by the Minnesota Multiphasic Personality Inventory-2 ( Hathaway & McKinley, 1989 ) has been conceptualized as an outcome of TBI in many previous studies. Whether personality trait characteristics that are little affected by the development of psychiatric illness also account for some aspect(s) of long-term psychosocial outcome remains unclear. Although measures of injury severity, such as the Glasgow Coma Scale (GCS; Teasdale & Jennett, 1974 ) and duration of posttraumatic amnesia, have been shown to predict psychosocial outcome after TBI, these relationships can be weak or diminish over time ( Bond, 1976 ; Bowman, 1996 ; Brooks et al., 1987 ; Dikmen, Machamer, Temkin, & McLean, 1990 ; Goran, Fabiano, & Crewe, 1997 ). Injury severity and subsequent cognitive performance both predict return to work ( Bowman, 1996 ; Klonoff et al., 1986 ; Ruff et al., 1993 ). But which demographic and injury characteristics are most relevant to specific aspects of psychosocial outcome also remains unclear. This study investigated two aspects of psychosocial outcome (social role engagement and behavioral adjustment) in light of demographic, injury severity, neurocognitive, and personality trait characteristics among patients who had sustained TBIs 8 years earlier. Executive cognitive abilities were emphasized because they are thought to be particularly relevant to everyday functioning among patients with TBI ( Lezak, 1995 ). The first hypothesis was that cognitive test performance and personality trait characteristics would explain significant variance in psychosocial outcome after TBI, even after accounting for the contributions of demographic characteristics and initial injury severity. The second was that the Social Role Engagement (SRE) Interview, developed for this study, and overall ratings on the Katz Adjustment Scale (KAS) would demonstrate related but distinguishable aspects of everyday functioning, as reflected by different patterns of association with demographic, injury, neurocognitive, and personality trait variables. METHOD Participants and Procedure The 78 adults who participated in this study included 39 patients and 39 informants who provided collateral information about each patient. The patients were recruited from a sample of adults who had all sustained TBIs 8 years earlier. Inclusion criteria were as follows: (a) no history of preexisting psychiatric disorder or substance abuse, (b) between 16 and 45 years of age at injury, (c) admission to Johns Hopkins Hospital or the University of Maryland Shock Trauma Unit within 24 hr of injury, (d) injury not due to assault, and (e) an Abbreviated Injury Scale (AIS head; Association for the Advancement of Automotive Medicine, 1980 ) severity rating of 2—5. The patients were mostly young ( M = 25.7 years at injury, SD = 7.5), Caucasian (90%) men (74%), of whom 69% had completed 12 years of school or more. At follow-up, they ranged in age from 23 to 53 years ( M = 33.3, SD = 7.5). Three of these variables (sex, age at injury, and years of education) were used for statistical analyses of functional outcome. The patient informants ranged from 19 to 66 years of age at follow-up ( M = 43.8, SD = 13.0). Most were Caucasian (90%) women (74%), of whom 75% had completed 12 years of school or more. For 46% of the patients, the informant was a spouse or partner; 33% were parents, 13% were friends, and the rest were siblings or "other" (8%). Informants had known their TBI patient counterparts for a mean of 20 years ( SD = 11, range = 2—41). More than 78% knew their patient counterpart before the injury, and only 4 informants reported less than daily contact. Patients and their respective informants were seen on the same day. Written informed consent was obtained from all participants; patients were paid $30, and informants were paid $10, for their participation. Patients were asked, via the SRE Interview, a series of questions about their postinjury psychosocial history, after which they completed a battery of cognitive and personality tests. Informants were asked to rate their patient counterparts using the SRE Interview, as well as the NEO Personality Inventory ( Costa & McCrae, 1985 ) and the KAS ( Katz & Lyerly, 1963 ). Neither outcome measure (SRE Interview or KAS) included any questions about injury severity, and both were scored without reference to injury severity or any predictor measure. Injury severity. Measures of injury severity included admission GCS scores, AIS head ratings, and length of initial hospitalization in days. Admission GCS ( Teasdale & Jennett, 1974 ) scores were recorded for 31 patients ( M = 10.3, SD = 4.6). Twelve of these scores were in the severe range (3—7), 4 were moderate (8—12), and 15 were mild (13— 15). Initial hospitalizations ranged from 3 to 190 days ( M = 44, SD = 47). The AIS yields ratings of injury severity that range from 1 (no or minor) to 6 (unsurvivable) for each of eight body regions. Only AIS head ratings were used for the present study. These ratings were based on information gathered from each patient's medical record at discharge, incorporated evidence of diffuse or focal brain injury, and ranged from 2 to 5, with a mean of 3.7 ( SD = 1.2). On the basis of AIS ratings, the head injuries of 12 patients were rated as critical; those of the remaining 27 were equally distributed (i.e., 9 each) among the three categories of severe, serious, and moderate. Because GCS scores were unavailable for 8 patients, and many other factors can determine length of initial hospitalization, AIS head ratings were used as the primary measure of brain injury severity for the purpose of statistical analyses. Outcome predictors—correlates. One set of "predictor" variables included (a) age at injury, (b) sex, (c) education at injury, and (d) AIS head ratings. Because individual differences on the remaining measures both precede and are affected by TBI, they are better described as correlates than predictors of outcome. The Shipley Institute of Living Scale ( Zachary, 1986 ) was used to estimate Wechsler Adult Intelligence Scale—Revised ( Wechsler, 1981 ) Full Scale IQ scores. The other cognitive measures included the Trail-Making Test ( Reitan, 1958 ), a measure of psychomotor speed, visual search, and mental flexibility for which the number of seconds required to complete each part was recorded; the Wisconsin Card Sorting Test (WCST; Heaton, 1981 ), a measure of concept formation and attentional shifting for which the number of conceptual sorts was used for data analyses; the Tinkertoy Test ( Lezak, 1995 ), a measure of the ability to initiate and execute a potentially complex activity in the absence of a clearly defined external goal, for which total scores based on the sum of nine feature scores ( Bayless et al., 1989 ) were recorded; and the Brief Test of Attention (BTA; Schretlen, 1997 ), a measure of auditory divided attention for which total scores were recorded. Finally, each informant rated his or her patient counterpart using Form R of the NEO Personality Inventory ( Costa & McCrae, 1985 ), which yields scores for the five personality dimensions of neuroticism, extraversion, openness, agreeableness, and conscientiousness. Four of these dimensions (neuroticism, extraversion, agreeableness, and conscientiousness) were used for data analyses. Thus, 14 variables were entered into regression models to "predict" long-term psychosocial outcome after TBI. The psychometric characteristics (i.e., reliability and validity estimates) of these "predictor" measures can be found in the reference for each. All of these measures have been used in studies of persons with TBI ( Bayless et al., 1989 ; Dikmen, Machamer, Winn, & Temkin, 1995 ; Segalowitz, Unsal, & Dywan, 1992 ; Wong, 1999 ). Additional information concerning the psychometric properties of each also can be found in Lezak (1995) and Spreen and Strauss (1998) . Outcome measures. The SRE Interview, developed for this study, consists of 20 factual questions about social role performance (such as marital and employment status, income earned in the past year, being licensed to drive a car, number of criminal arrests since injury, and hospital admissions for psychiatric treatment since injury). A numerical score based on the weighted sum of responses given by each patient was computed such that higher scores reflected greater and more successful social role engagement. 1 Psychometric characteristics of the SRE Interview are described in the Results section. An informant also rated each patient's everyday behavior on the KAS ( Katz & Lyerly, 1963 ). Although the KAS was developed as a measure of psychosocial adjustment for patients with psychiatric disorders, it has been used in studies of TBI ( Fordyce et al., 1983 ; Hinkeldey & Corrigan, 1990 ; Oddy, Humphrey, & Uttley, 1978 ). The KAS contains 127 behavioral observations for which an informant rates the patient from 1 ( almost never ) to 4 ( almost always ). Hogarty and Katz (1971) derived 13 KAS factor scores. For 12 of these (e.g., anxiety, withdrawal, and suspiciousness), higher scores denote poorer functioning. For Factor 13 (i.e., stability), higher scores denote better functioning. In this study, an overall KAS score based on the sum of ratings across Factors 1—12 (aberrant behaviors) minus each patient's rating on Factor 13 (behavioral stability) was taken as the primary index of behavioral adjustment, with higher scores reflecting worse behavioral adjustment. Separate summary ratings for Factors 1—12 (aberrant behaviors) and Factor 13 (behavioral stability) were used in some analyses. Internal consistency estimates of KAS factor scores have been found to range from .41 to .87 ( Katz & Lyerly, 1963 ). Results Pearson correlation analyses revealed that AIS head scores were associated with GCS scores ( r = .83, p < .001) and length of hospitalization ( r = .58, p < .001). Admission GCS scores also correlated ( r = .63, p < .001) with length of hospitalization. Neither AIS nor length of hospitalization was significantly related to age at injury, sex, race, or education (all p s > .05). Basic Characteristics of Patient Outcomes Patient-reported SRE Interview scores ranged from 2 to 86 ( M = 33.3, SD = 23.1). They were distributed fairly normally, with a skewness of .40 and a kurtosis of .32, indicating that the distribution was mildly skewed to the right and slightly leptokurtic (less peaked than expected). The intraclass correlation coefficient ( Shrout & Fleiss, 1979 ) for 26 pairs of informant and patient reports on the TBI Interview (13 informants did not know their corresponding patient's earned income) was excellent (equation [3, 1] ICC = .87). The high level of agreement between patient and informant responses to the SRE Interview probably reflects the factual nature of the questions. Data concerning the test—retest reliability of this instrument are not yet available. Concurrent validity of the SRE Interview scores was explored in terms of both item content and correlation with the KAS. On the basis of discontinuities in the distribution of SRE Interview scores, three distinct outcome groups were defined by SRE scores of or below 20, 21—44, and 45 or above, reflecting poor ( n = 11), fair ( n = 15), and good ( n = 13) outcomes, respectively. As expected, the three outcome groups clearly differed in terms of marital and employment status, earned income, being licensed to drive an automobile, and incidence of psychiatric hospitalization. Thus, the SRE Interview distinguishes among patients in ways that reflect meaningful differences in social role engagement, thereby supporting the content validity of this measure. These findings are shown in Table 1 . Furthermore, although SRE Interview scores did not correlate with overall KAS ratings ( r = .25, p = .12), they did correlate significantly with scores on the KAS behavioral stability factor ( r = .48, p = .002). These findings support the construct validity of the SRE Interview. They also confirm the hypothesis that the SRE Interview and KAS measure partially overlapping but largely distinct aspects of psychosocial outcome. As expected, measures of initial injury severity showed modest but statistically significant correlations with long-term outcome. Specifically, AIS head ratings correlated significantly with overall KAS ratings ( r = .28, p = .04, one-tailed) and marginally with SRE Interview scores ( r = .26, p = .053, one-tailed). Length of initial hospitalization correlated with SRE Interview scores ( r = .33, p = .02, onetailed) but not with KAS ratings ( r = .20, ns ). Multiple Regression Models of Concurrent Psychosocial Outcome Hypotheses regarding the relative contributions of demographic, injury, severity, personality trait, and cognitive variables to behavioral adjustment and social role engagement after TBI were tested via hierarchical, stepwise multiple regression analyses. Predictor variables were grouped into three blocks and made available for model building according to the position they were hypothesized to occupy in the causal sequence of psychosocial outcome. Thus, four demographic and clinical variables (age and education at injury, sex, and AIS head ratings) were grouped together in Block 1, four NEO personality trait scores (neuroticism, extraversion, agreeableness, and conscientiousness) were grouped in Block 2, and six cognitive test scores (IQ, Trail-Making Test Parts A and B, WCST conceptual sorts, BTA, and Tinkertoy Test scores) were grouped in Block 3. Within each block, variables were entered stepwise according to the criterion of an F -to-enter at p < .05. Behavioral adjustment. In the first regression analysis, overall KAS ratings were regressed on the three blocks of predictor variables just described. The results of this analysis are summarized in Table 2 . As shown, not a single variable from Block 1 met entry criteria, whereas three NEO Personality Inventory trait ratings each improved the model fit. Thereafter, none of the neurocognitive test scores explained significant additional variance in behavioral adjustment. Thus, three NEO Personality Inventory trait scores yielded a highly significant, F (3, 35) = 15.34, p < .001, model that accounted for 57% of the variance ( R 2 = .568) in overall KAS ratings. As shown in Table 2 , the beta weights for all three predictor variables were significant, and their signs revealed that better behavioral adjustment was associated with lower trait neuroticism, higher agreeableness, and higher conscientiousness. In a follow-up multiple regression analysis not shown in Table 2 , when all of the demographic and injury severity variables were forced into the equation, they yielded a nonsignificant model that accounted for about 10% of the outcome variance. Thereafter, the same three NEO Personality Inventory variables met entry criteria (according to the stepwise procedure described earlier), and the resulting model accounted for 61% of the outcome variance (i.e., less than 5% more than that explained by the three NEO Personality Inventory scores alone). Social role engagement. In the second analysis, SRE Interview scores were regressed on the same three blocks of predictor variables after identical variable entry procedures. This analysis revealed a very different pattern of findings. First, three variables from Block 1 (AIS ratings, education, and sex) met entry criteria, after which one variable from Block 2 (trait neuroticism) and one variable from Block 3 (BTA scores) also met entry criteria and improved the model fit. Thus, in this analysis, five predictor variables yielded a highly significant, F (3, 35) = 8.52, p < .001, model that accounted for 56% of the variance (adjusted R 2 = .563) in social role engagement. As shown in Table 3 , the final beta weights remained significant for all but one predictor, NEO neuroticism scores. The signs of statistically significant beta weights indicate that better social role engagement was associated with less severe head injuries, more years of education at injury, male gender, and better cognitive test performance. In a final series of exploratory analyses, SRE Interview and overall KAS ratings were regressed on the same 14 variables described earlier, with all variables grouped in a single block, again using a stepwise method of variable entry. For overall KAS ratings, this procedure yielded results that were identical to those described earlier: The same three NEO Personality Inventory trait T scores met entry criteria for a model with an R 2 value of .57. Consequently, these results are not shown in a separate table. For SRE Interview scores, in contrast, when all 14 predictor variables were grouped together for stepwise entry, a very different model emerged. Specifically, just three variables met entry criteria for a highly significant, F (3, 35) = 11.54, p < .001, model that accounted for 50% of the variance ( R 2 = .50) in social role engagement. As shown in Table 4 , these three "predictor" variables included performance on the BTA, total score on the Tinkertoy Test, and age at injury. Thus, absent the a priori hierarchical ordering of potential predictor variables, terms for sex, injury severity, and personality trait characteristics did not enter the equation, whereas the contribution of cognitive test performance became more evident. Discussion Several findings emerged from this study. First, the SRE Interview and KAS behavioral stability factor score yielded moderately correlated measures of psychosocial outcome. The content validity of the SRE Interview also was supported by the finding that three outcome groups defined by it differed in ways that obviously reflect the adequacy of social role engagement, such as earned income, marital status, and being licensed to drive an automobile. They also differed in terms of behavioral stability (but not aberrant behavior) as measured by the KAS. These findings suggest that social role engagement and behavioral adjustment represent partially overlapping aspects of psychosocial outcome after TBI. Second, despite the moderately strong correlation between SRE Interview scores and KAS Factor 13 (behavioral stability) ratings, the fact that only 25% of the variance in each was shared by the other suggests that they tap different aspects of outcome. Other findings also support this inference. For example, SRE Interview scores did not correlate significantly with either overall KAS ratings or the summed ratings for KAS Factors 1—12 (aberrant behaviors). Third, further evidence that social role engagement and behavioral adjustment represent distinguishable aspects of outcome emerged from multiple regression analyses. These analyses showed that SRE Interview scores and overall KAS ratings were best explained by very different variables. Behavioral adjustment, as measured by overall ratings on the KAS, was best explained by informant ratings of personality trait neuroticism, agreeableness, and conscientiousness. Each of these accounted for significant unique variance in behavioral adjustment, whereas age, sex, education, injury severity, and cognitive test performance did not. In contrast, adequacy of selfreported social role engagement, as measured by SRE Interview scores, was best explained by a combination of five variables: years of education at injury, sex, injury severity, trait neuroticism, and cognitive test performance. Not only did personality trait ratings account for relatively little variance in social role engagement, the beta weight for neuroticism was attenuated to a nonsignificant level by entering a term for cognitive test performance. A fourth finding was that preinjury level of education and measures of injury severity made statistically significant but relatively modest contributions to models accounting for psychosocial outcome 8 years later. This is consistent with previous findings that the relationship between injury severity and outcome can be relatively weak or diminish over time ( Bond, 1976 ; Brooks et al., 1987 ; Dikmen et al., 1990 ; Goran et al., 1997 ). In this study, social role engagement was more strongly associated with concurrent cognitive test performance than with injury severity or personality characteristics. Given the inclusion of employment status and earned income in the measure of social role engagement, these findings are strikingly similar to those obtained by Bowman (1996) and Goran et al. (1997) , both of whom found that measures of concurrent intellectual ability were more predictive of return to work than were length of coma or overall injury severity among TBI survivors. For example, using a stepwise multiple regression procedure, Bowman (1996) found that neurocognitive test performance accounted for 21% of the variance, as compared with 2% for injury severity, in occupational attainment 3—4 years after TBI. The finding that heightened distractibility, as measured by the BTA, was especially related to worse social role functioning in the present study also is consistent with previous findings that low scores on the same test were associated with functional disability among psychiatric patients ( Schretlen et al., 2000 ) and an increased 2-year incidence of vehicular accidents among elderly drivers ( Keyl, Rebok, & Gallo, 2000 ). In short, in the present study, patients who were more educated at the time of injury, less distractible at follow-up, and lower in trait neuroticism were more likely to be employed, married, and licensed to drive and more likely not to have required psychiatric hospitalization during the 8 years since their initial injury. Although injury severity and neurocognitive test performance distinguished among the patient groups defined by KAS ratings of behavioral adjustment, NEO Personality Inventory factor scores were more closely associated with this dimension of psychosocial outcome. The most relevant NEO Personality Inventory factor scores were neuroticism (impulsivity and irritability), agreeableness (trustfulness and compliance), and conscientiousness (competence and deliberation). Whether these represent enduring characteristics that preceded each patient's injury or were affected by the brain injury cannot be determined from this cross-sectional study. Costa and McCrae (1985) contended that the NEO Personality Inventory measures normal personality characteristics that are stable over time and relatively insensitive to circumstance or the onset of psychiatric symptoms. Thus, the observed differences in NEO Personality Inventory scores among outcome groups might reflect preinjury personality trait differences that moderate the relationship between injury severity and long-term behavioral adjustment after TBI. Alternatively, the observed differences in NEO Personality Inventory trait scores might be jointly determined by premorbid factors and the effects of TBI. This would be consistent with the observation that brain injuries often seem to exacerbate preinjury personality vulnerabilities or lead to a general coarsening of personality ( Prigatano, 1987 ). Two limitations of this study require comment. First, the variable-to-subject ratio was somewhat excessive. Although a part of the study design, the a priori selection of certain variables (e.g., use of WCST categories sorted) and the use of summary scores (e.g., factor scores from the NEO Personality Inventory) did not eliminate the problem. A somewhat elevated risk of Type I errors was accepted on the basis that this investigation was designed to refine the initial hypotheses for further research rather than to definitively test a specific hypothesis. A second weakness was the relatively coarse manner in which injury severity was assessed. More precise quantification of the extent and location of brain tissue injury with neuroimaging or information about surgical intervention, intracranial pressure, and so forth might have provided more accurate predictions of long-term outcome than GCS scores, AIS ratings, and length of hospitalization. Both of these limitations point to areas in which the study design might be strengthened by future research. Despite limitations, the obtained results suggest that two broad domains of psychosocial outcome can be discerned 8 years after TBI. Although these two domains tap overlapping aspects of everyday functioning, the dimension of social role engagement was more strongly associated with cognitive abilities than personality. In contrast, the dimension of behavioral adjustment was most strongly associated with individual differences in personality. For clinical purposes, these findings suggest that comprehensive neuropsychological evaluations of patients with TBI ought to include measures of both neurocognitive functioning and personality. The former can elucidate rate-limiting targets for rehabilitation interventions that aim to restore competence in social role functioning, and the latter can reveal personality trait characteristics that probably contribute to everyday behavioral adjustment. References Association for the Advancement of Automotive Medicine. (1980). The Abbreviated Injury Scale 1980 revision. (Des Plaines, IL: Author) Bayless, J. D., Varney, N. R. & Roberts, R. J. (1989). Tinker Toy test performance and vocational outcome in patients with closed-head injuries. Journal of Clinical and Experimental Neuropsychology, 11, 913-917. Bond, M. R. (1976). Assessment of the psychosocial outcome of severe head injury. Acta Neurochirurgica, 34, 57-70. Bowman, M. L. (1996). Ecological validity of neuropsychological and other predictors following head injury. Clinical Neuropsychologist, 10, 382-396. Brooks, N., McKinlay, W., Symington, C., Beattie, A. & Campsie, L. (1987). Return to work within the first seven years of severe head injury. Brain Injury, 1, 5-19. Costa, P. T. & McCrae, R. R. (1985). The NEO Personality Inventory. (Odessa, FL: Psychological Assessment Resources) Diamond, R., Barth, J. T. & Zillmer, E. A. (1988). Emotional correlates of mild closed head trauma: The role of the MMPI. International Journal of Clinical Neuropsychology, 10, 35-40. Dikmen, S., Machamer, J., Temkin, N. & McLean, A. (1990). Neuropsychological recovery in patients with moderate to severe head injury: 2 year follow-up. Journal of Clinical and Experimental Neuropsychology, 12, 507-519. Dikmen, S., Machamer, J. E., Winn, H. R. & Temkin, N. R. (1995). Neuropsychological outcome at 1-year post head injury. Neuropsychology, 9, 8090. Dikmen, S. & Reitan, R. M. (1977). Emotional sequela of head injury. Annals of Neurology, 2, 492-494. Fabiano, R. J. & Crewe, N. (1995). Variables associated with employment following severe traumatic brain injury. Rehabilitation Psychology, 40, 223-231. Fordyce, D. J., Roueche, J. R. & Prigatano, G. (1983). Enhanced emotional reactions in chronic head trauma patients. Journal of Neurology, Neurosurgery, and Psychiatry, 46, 620-624. Goran, D. A., Fabiano, R. J. & Crewe, N. (1997). Employment following severe traumatic brain injury: The utility of the individual ability profile system (IAP). Archives of Clinical Neuropsychology, 12, 691-698. Hagen, C. (1982). Language-cognitive disorganization following closed head injury: A conceptualization.(In L. E. Trexler (Ed.), Cognitive rehabilitation (pp. 131—151). New York: Plenum.) Hathaway, S. R. & McKinley, J. C. (1989). The Minnesota Multiphasic Personality Inventory (MMPI-2): Manual for administration and scoring. (Minneapolis: University of Minnesota Press) Heaton, R. K. (1981). A manual for the Wisconsin Card Sorting Test. (Odessa, FL: Psychological Assessment Resources) Hinkeldey, N. S. & Corrigan, J. D. (1990). The structure of head-injured patients' neurobehavioral complaints: A preliminary study. Brain Injury, 4, 115-133. Hogarty, G. E. & Katz, M. M. (1971). Norms of adjustment and social behavior. Archives of General Psychiatry, 25, 470-480. Jennett, B. & Bond, M. (1975). Assessment of outcome after severe brain damage. Lancet, 1, 480-487. Katz, M. M. & Lyerly, S. B. (1963). Methods for measuring adjustment and social behavior in the community: I. Rationale, description, discriminative validity and scale development. Psychological Reports, 13, 503-535. Keyl, P. M., Rebok, G. W. & Gallo, J. J. (2000). Can brief neuropsychological testing in a general medical setting identify unsafe elderly drivers? (Manuscript submitted for publication) Klonoff, P. S., Costa, L. D. & Snow, W. G. (1986). Predictors and indicators of quality of life in patients with closed-head injury. Journal of Clinical and Experimental Neuropsychology, 8, 469-485. Levin, H. S. & Grossman, R. G. (1978). Behavioral sequela of closed head injury. Archives of Neurology, 35, 720-727. Lezak, M. D. (1995). Neuropsychological assessment ((3rd ed.). New York: Oxford University Press) Oddy, M., Coughlan, T., Tyerman, A. & Jenkins, D. (1985). Social adjustment after closed head injury: A further follow-up seven years after injury. Journal of Neurology, Neurosurgery, and Psychiatry, 48, 564-568. Oddy, M., Humphrey, M. & Uttley, D. (1978). Stresses upon the relatives of head injured patients. British Journal of Psychiatry, 133, 507-513. Prigatano, G. (1987). Psychiatric aspects of head injury: Problem areas and suggested guidelines for research.(In H. S. Levin, J. Grafman, & H. M. Eisenberg (Eds.), Neurobehavioral recovery from head injury (pp. 215—231). New York: Oxford University Press.) Rappaport, M., Hall, K. M., Hopkins, K., Belleza, I. & Cope, D. N. (1982). Disability rating scale for severe head trauma: Coma to community. Archives of Physical Medicine and Rehabilitation, 63, 118-123. Reitan, M. (1958). Validity of the Trail-Making Test as an indication of organic brain damage. Perceptual and Motor Skills, 8, 271-276. Ruff, R. M., Marshall, L. F., Crouch, J., Klauber, M. R., Levin, H. S., Barth, J., Kreutzer, J., Blunt, B. A., Foulkes, M. A. & Eisenberg, H. M. (1993). Predictor of outcome following severe head trauma: Follow-up data from the Traumatic Coma Data Bank. Brain Injury, 7, 101-111. Schretlen, D. (1997). Brief Test of Attention professional manual. (Odessa, FL: Psychological Assessment Resources) Schretlen, D., Jayaram, G., Maki, P., Park, K., Abebe, S. & DiCarlo, M. (2000). Demographic, clinical, and neurocognitive correlates of everyday functional impairment in severe mental illness. Journal of Abnormal Psychology, 109, 134138. Segalowitz, S. J., Unsal, A. & Dywan, J. (1992). CNV evidence for the distinctiveness of frontal and posterior neural processes in a traumatic brain-injured population. Journal of Clinical and Experimental Neuropsychology, 14, 545-565. Shrout, P. E. & Fleiss, J. L. (1979). Intraclass correlations: Uses in assessing rater reliability. Psychological Bulletin, 86, 420-428. Spreen, O. & Strauss, E. (1998). A compendium of neuropsychological tests. (New York: Oxford University Press) Teasdale, G. & Jennett, B. (1974). Assessment of coma and impaired consciousness: A practical scale. Lancet, 2, 81-83. Thompsen, I. V. (1984). Late outcome of very severe blunt head trauma: A 10- to 15year second follow-up. Journal of Neurology, Neurosurgery, and Psychiatry, 47, 260268. Wechsler, D. (1981). WAIS-R manual. (New York: Psychological Corporation) Wong, T. M. (1999). Validity and sensitivity of the Brief Test of Attention with acute brain injury and mild head injury patients([Abstract]) Archives of Clinical Neuropsychology, 14, 728-729. Zachary, R. A. (1986). Shipley Institute of Living Scale Revised manual. (Los Angeles: Western Psychological Services) 1 A copy of the SRE Interview is available from David J. Schretlen on request. This study was supported by the U.S. Department of Education, National Institute on Disability and Rehabilitation Research (Grant H133C90054). I gratefully acknowledge John A. Bates for his assistance with data collection, test scoring, and data entry. Correspondence may be addressed to David J. Schretlen, Johns Hopkins Hospital, Meyer 218, Baltimore, Maryland, 21287-7218. Electronic mail may be sent to dschret@jhmi.edu Received: May 24, 1999 Revised: December 5, 1999 Accepted: December 8, 1999 Table 1. Characteristics of Outcome Groups Defined by Responses to the Social Role Engagement (SRE) Interview Table 2. Hierarchical—Stepwise Multiple Regression of Overall Katz Adjustment Scale Ratings (Behavioral Adjustment) on Demographic, Clinical, Personality, and Cognitive Predictor Variables Table 3. Hierarchical—Stepwise Multiple Regression of Social Role Engagement Interview Scores on Demographic, Clinical, Personality, and Cognitive Predictor Variables Table 4. Stepwise Multiple Regression of Social Role Engagement Interview Scores on Demographic, Clinical, Personality, and Cognitive Predictor Variables