- Email: [email protected]
Concordance of patients' and family members' ratings of neurobehavioral functioning after traumatic brain injury
1254
Concordance of Patients’ and Family Members’ Ratings of Neurobehavioral Functioning After Traumatic Brain Injury Ronald T. Seel, MS, Jeffrey S. Kreutzer, PhD, Angelle M. Sander, PhD ABSTRACT. See1 RT, Kreutzer JS, Sander AM. Concordance of patients’ and family members’ ratings of neurobehavioral functioning after traumatic brain injury. Arch Phys Med Rehabil 1997; 78: 1254-9. Objective: To examine differences in family and patient evaluation of neurobehavioral functioning in adults with traumatic brain injury (TBI). Design: Differences were examined by conducting 70 paired sample t tests on scale items and 6 paired sample t tests on scale scores from a neurobehavioral inventory. Setting: Medical center outpatient clinic. Participants: Three hundred one consecutive adult patients with TBI and 301 informants, primarily family members, completed the neurobehavioral inventory. Main Outcome Measure: Neurobehavioral Functioning Inventory (NFI) comprised of six scales with items describing symptoms and daily living problems. Results: Paired t test analysesof the six scalesindicated that patients reported a significantly greater level of communication problems than did their matched family members. No differences were found for the other five scales.Paired t test analyses of the 70 scale items revealed significant differences in patient and family ratings for only 13 items. In all 13 instances,patients reported greater levels of dysfunction than were reported by their family members. Analysis of variance (ANOVA) indicated a main effect of injury severity for only the Communication and Memory/Attention scales. Conclusions: Findings indicate general agreement between family members and patients regarding patients’ everyday problems. Results do not support contentions that patients tend to underestimate difficulties. Agreement levels appear related to injury severity, item specificity, and item content. More research is needed to identify other variables relating to agreement levels, including age, injury severity, and amount of contact between patients and family members. 0 1997 by the American Congress of Rehabilitation Medicine and the American tation
Academy of Physical Medicine
and Rehabili-
OR MORE THAN TWO decades, investigators have atF tempted to describe psychosocial outcome after traumatic brain injury (TBI).1-6 In these studies, a preference has been expressed for relying on the reports of family members rather than patients.2U4,6 Greater emphasis on family members’ reports From the Department of Physical Medicine and Rehabilitation, Medical College of Viwinia. Virginia Commonwealth Universitv. Richmond. VA. Submitted fo;publication March 3, 1997. Accepted April 7, 1997. Supported in part by grant H133B80029 from the National Institute on Disabil-. ity and Rehabilitation Research, United States Department of Education. No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit upon the authors or mm am cmanization with which the authors are associated. Reprint requests to Jeffrey S. Kreutzer, PhD, Department of Physical Medicine and Rehabilitation, Medical College of Virginia, MCV Box 980542, Richmond, VA 23298.0542. 0 1997 by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation 0003-9993/97/7811-4412$3.00/O
Arch
Phys Mad
Rehabil
Vol78,
November
1997
has been justified, based on assertions that cognitive impairments, including poor self-awareness, compromise the validity of self-report information. ‘,’ However, empirical examination of actual differences between patients’ and family members’ perceptions of neurobehavioral functioning has yielded varying results. Some investigators have collected psychosocial outcome data from both patients and family members, comparing their reports. For example, using a structured interview, Thomsen’ gathered information about injury-related changes from 50 patients with severe TBI and their family members. Interviews were conducted between 1 and 5 years after injury. While the majority of relatives reported negative behavioral and personality changes, patients rarely made such complaints. Oddy and colleagues9 administered a neurobehavioral symptom checklist to 54 patients with severe TBI and their relatives at 6 months after injury. The type and frequency of symptoms reported by patients and relatives was highly similar. In spite of similarities, the authors reported that “there were marked discrepancies in individual cases, which underlines the hazards of accepting the patient’s report as definitive.” 9 No statistical tests, however, were performed to support their conclusion. In a subsequent study conducted by Oddy et al5 at 7 years after injury, a greater discrepancy was noted in the type and frequency of symptoms reported by patients and relatives. Differences in proportions of patients and relatives reporting the same problem ranged from 4% to 26%. Forty percent of relatives reported that their loved ones refused to admit difficulties. Once again, despite the discrepancies noted, no statistical tests were conducted to compare patients’ and relatives’ ratings. Furthermore, no information was provided on the validity of the neurobehavioral checklist used. Investigators have also provided evidence that patient and family agreement levels vary depending on areas of functioning assessed.Lower agreement rates have been found for items pertaining to patients’ emotional and behavioral changes. For example, using a neurobehavioral checklist, M&inlay and Brooks7 investigated agreement rates between 52 severely injured patients and their relatives 6 months after injury. They found agreement rates ranging from 52% to 85% between patients and relatives. The highest rate of agreement was for sensory and motor difficulties, and the greatest disagreement was for items pertaining to emotional and behavioral changes. However, no information regarding patients’ injury severity or the validity of the checklist was provided, and an unspecified method of analysis was used. Cavallo, Kay, and Ezrachilosllstudied patients’ and relatives’ ratings on the 34-item Problem Checklist of the New York Head Injury Family Interview. Subjects were 34 patients with primarily moderate to severeinjuries. Patients and their relatives rated neurobehavioral symptoms at an average of 2 years after injury. In 35% of the cases,patients and family members were categorized as being in “high agreement,” as defined by agreement on at least 75% of the items. Twenty-three percent of the patients endorsed difficulties at least twice as often as their relatives, and 32% of the patients showed the opposite pattern, underestimating difficulties compared with family members’
NEUROBEHAVIORAL
reports. On items pertaining to interpersonal and emotional difficulties, patients tended to rate problems as less frequent. However, conclusions were drawn primarily on the basis of descriptive statistics. Hendryx” found a similar pattern of results in a sample of 20 patients with severe TBI. Patients and family members rated perceived change from before to after injury in the areas of physical, emotional, and cognitive functioning. Although no significant difference was found between patients’ and family members’ ratings of physical and cognitive changes, family members perceived patients as exhibiting greater emotional changes than reported by patients themselves. Unfortunately, no validity information was provided for the rating scale. Prigatano and colleagues’3 developed the Patient Competency Rating Scale (PCRS) to assess a wide range of difficulties commonly encountered after brain injury. The PCRS is a 30-item rating scale with separate forms available for completion by patients and family members. Patients’ abilities are rated on a 5-point scale, with 1 conveying an inability to perform the activity and 5 conveying performance of the activity with ease. Prigatano et all4 compared patients’ and family members’ PCRS ratings for 64 patients representing the full range of injury severity. Comparisons were made for 18 of the 30 items. Agreement rates ranged from 20% to 77%. Consistent with earlier studies, the highest rates of agreement were for items pertaining to self-care activities (eg, “dressing self,” “personal hygiene”), whereas the lowest rates were for emotional and behavioral characteristics (eg, “handling arguments,” “keep from being depressed”). Similar results were found in an investigation of PCRS ratings in a sample of 41 English-speaking New Zealand patients and their relativesi Although the PCRS has now been used in several studies of self-awareness, no information has been provided regarding psychometric properties. Research has also suggested that differences in perceptions of neurobehavioral functioning relate to the specificity of questions or items. For example, Gasquoinei6 compared neurobehavioral ratings made by patients and staff. He found greater levels of agreement for more clearly definable items. Sherer and colleagues17 developed the Awareness Questionnaire to allow comparisons between patients’, family members’ and clinicians’ ratings of patients’ difficulties. The consistency of patients’ and family members’ perceptions is used to determine self-awareness. Working mostly with severe brain injury patients, Sherer et al’* found that impaired awareness was more commonly observed on items addressing general rather than specific behaviors. In summary, studies comparing patients’ and relatives’ perceptions of difficulties have found widely varying rates of agreement. Agreement levels have been related to item specificity and content. More specific items tend to elicit higher levels of agreement between patients’ and family members’ perceptions. Furthermore, higher levels of agreement have been observed for items relating to self-care and physical functioning. Disagreement in perceptions has been observed for items relating to emotional and behavioral changes. Variable research findings likely result from several methodologic limitations and differences. First, the relation between injury severity and agreement levels has not been investigated. With a few exceptions, studies have included primarily persons with severe injuries, leaving questions about patients with mild and moderate injuries. Second, sample sizes have generally been small, ranging from 20 to 64. Third, the majority of studies have used descriptive statistics or visual inspection to compare patients’ and family members’ perceptions. Finally, researchers have used instruments varying in content, response format, and
RATINGS,
1255
See1
item specificity. A comparison of instruments used to compare patients’ and family members’ perceptions of problems is provided in table 1. Some researchers have examined responses to individual items, whereas others have organized items into scales based on face validity considerations. The use of empirically constructed scales with demonstrated validity, however, has been lacking. The current study attempted to clarify the relation between patients’ and family members’ perceptions of injury-related changes by improving on the methodology of previous studies (table 1). Investigation focused on a large sample of patients representing the full range of injury severity, and the relation between injury severity and agreement rates was examined. The study also used a factor-analytically derived, validated assessment tool, the Neurobehavioral Functioning Inventory (NFI).19 The NFI items are highly specific and can be combined into six scales representing distinct categories of neurobehavioral functioning. The present investigation was unique in comparing responses to 70 individual items and six scale scores using between-groups statistical tests. Based on previous research, patients were expected to report less frequent emotional and behavioral difficulties than family members, as indicated by comparisons between NFI Depression and Aggression scores. Perceptions regarding physical functioning were not expected to differ, as indicated by comparisons involving the Motor and Somatic Scales. In addition, greater discrepancies were anticipated for patients with more severe injuries.
METHOD Subjects The original sample consisted of 310 pairs of patients and informants referred for comprehensive outpatient assessment at a regional level I trauma center. Informants were selected by the patients, who were asked to identify a single individual who was either a primary caretaker or who knew the patient well. To control for unreliable responses, nine pairs of informants and patients were eliminated because of high levels of missing data. For these nine cases, more than 25% of items were not completed by either one or both members of the pair. The final sample for analysis consisted of 301 pairs of patients and informants. The following demographic and medical information for patients was identified: 55% were men; with regard to ethnicity, 80% were Caucasian, 18% were African American, and 2% were of other ethnic backgrounds; 79% had been involved in vehicular accidents; mean age at the time of injury was 35.9 years (SD = 12.2). Patients were classified into one of four injury severity groups based on number of days unconscious: (1) “mild” = 20 minutes or less (n = 161); (2) “moderate” = greater than 20 minutes and less than 24 hours (n = 21); (3) “severe” = greater than or equal to 24 hours and less than 7 days (n = 20); and, (4) “very severe’ ’ = 7 days or greater (n = 47). The mean length of time in coma was 9.6 days (SD = 26.8), and the median length of time in coma was 0 days. Average time postinjury for evaluation was 2.2 years (SD = 3.8), with a median value of 0.8 years. Informants largely consisted of immediate family members: spouses (42%); parents (30%); children (7%); and siblings (5%). A small proportion of informants were nonfamily members that included: significant others (10%); friends (4%); and others (2%). Seventy-three percent of the informants were women. Informants’ mean age at the time of evaluation was 41.7 years (SD = 13.4).
Arch
Phys Med
Rehabil
Vol78,
November
1997
1256
NEUROBEHAVIORAL
Table instrument
Characteristics
Thomsen
(1974)’
Number of items Areas of functioning
Not reported Emotional, behavioral, cognitive symptoms
Response
Structured
format
Item specificity Validation Construction of scales Statistically derived scales
interview
Uncertain None reported None reported None reported
1: Characteristics Oddy et al (1978)’
of Assessment Prigatano
Methods et al (1990)”
Cavallo et al (1992)‘O
Emotional, somatic, cognitive, communication Checklist, yes or no
34 Emotional, behavioral, physical, cognitive, communication
Low to high None reported None reported None reported
Low to high None reported Yes None reported
Dichotomous; converted from Likert scale High None reported Yes None reported
Present Investigation
70 Aggression, depression, memory/attention, communication, somatic, motor Likert scale
High Yes Yes Yes
RESULTS
The NFI is contained within a larger questionnaire, the General Health and History Questionnaire (GHHQ). The NFI is comprised of 70 scored items grouped into six categories based on principal components and confirmatory factor analytic methodology: Depression; Somatic Complaints; Memory/Attention Difficulties; Communication Deficits; Aggressive Behaviors; and Motor Impairment.” Frequency of problem occurrence is rated by patients or informants on a 4-point scale that allows the following choices: (1) never; (2) sometimes; (3) often; or (4) always. For a subset of items, raters have an opportunity to rate problems as not applicable. Research has provided support for criterion-related validity and high internal consistency within NFI scales.” For all scales, Chronbach’s alpha has ranged from .86 to .95. In addition, scores for each of the NFI scales have been found to correlate logically with other measures of neurobehavioral dysfunction.
Procedure The neurobehavioral inventory was mailed to patients and informants in advance of the patient’s clinic appointment. On arrival for evaluation, responses to the inventory were reviewed for compliance with instructions. Forms were corrected or missing information was obtained by interviewing the informant in person or by phone. In cases where patients were re-evaluated, more than one NFI was often available. In such cases, the most recent questionnaires completed by both the patient and informant were selected for analysis. For each NFI scale, a total score was calculated based on the sum of ratings for all scale items. Descriptive statistics were calculated for all items and for the six NFI scales. Seventy-six matched-pair t tests were conducted to examine differences in family and patient ratings on the 70 scored items and the six scale scores of the NFI. The t tests were calculated using the pairwise deletion method. Nonmissing values for each pair of variables were correlated, regardless of whether other variables for the patient/family set were missing. As a result, the number of participants for each t test differed based on the number of valid variable pairs. Six one-way analysis of variance (ANOVA) tests were conducted to determine whether patient injury severity had an impact on differences in patents’ and family members’ ratings of scale scores. When a main effect for injury severity was found, a Tukey post hoc test was conducted to identify significant differences between the four injury severity categories.“’ Given the number of planned comparisons, a decision was made to set a conservative alpha level for t tests and ANOVA tests, minimizing the chance of a Type I error. Family-wise error rates and the potential for Type II errors were also considered. Consequently, an alpha level ofp < .Ol was deemed acceptable.
Phys Med
See1
30 Emotional, behavioral, somatic, motor, cognitive, self-care, communication Likert scale
37 behavioral, sensory,
Measures
Arch
RATINGS,
Rehabil
Vol78,
November
1997
Differences for Individual Items The first series of analyses attempted to determine the similarities and differences between responses given by family members and patients for the same item. Each pair of responses was categorized in one of three ways: (1) the patient rated the problem as occurring more frequently than the family member (P > F); (2) the patient and family member were in agreement, rating the problem as occurring with equal frequency (P = F); and (3) the family member rated the problem as occurring more frequently than the patient (F > P). For the 70 NFI items, the proportion of paired ratings falling into each category was calculated (table 2). For all 70 NFI items, the percentage of agreement ranged from 48% to 84%. For each item, fewer than 35% of patients (range, 9% to 35%) rated their problems as occurring more frequently than rated by family members. For each item, approximately one fourth of family members (range, 7% to 27%) rated the patient’ problems as occurring more frequently than indicated by the patient’s self-report. The data suggest that family and patients tend to rate problems as occurring with equal frequency. The second set of analyses measured the magnitude of difference between family and patient responses for NFI items. For each pair of responses, the patient’s item rating based on the four point Likert scale was subtracted from the family member’s rating. Difference values ranged from -3 to +3. Negative difference scores indicated that the family member rated the problem as occurring less frequently than the patient; positive scores indicated that the patient rated the problem as occurring less frequently than the family member. A difference score of “0” indicated that the family member and patient rated the problem as occurring with the same frequency. The greater the score deviated from zero regardless of direction, the greater the magnitude of disagreement between the family member’s and patient’s rating. For each NFI scale, the proportion of difference values for all scale items was calculated (table 3). The magnitude of similarities and differences between family and patient responses was then analyzed. The percentage of paired responses with identical item scores (0 difference) ranged from 55% to 69%. For more than 92% of the paired responses, there was either exact agreement or a difference of one point. The difference between patient and family ratings exceeded a value of two in fewer than 1% percent of the responses. The results strongly suggest that family members and patients rate problems as occurring with similar frequency. The third set of analyses identified NFI items for which there were statistically different responses between patients and family members. Matched-pair t tests were conducted individually for all 70 items. These tests found no statistical differences
NEUROBEHAVIORAL
Table
2: Family
Versus
Patient
Depression Items Uncomfortable around others Feels worthless Lonely Restless Sad, blue Sits with nothing to do Scared or frightened Bored Difficulty enjoying activities Feels hopeless Cannot get mind off certain thoughts No confidence Frustrated Aggression Items Threatens to hurt others Hits or pushes others Breaks or throws things Curses at others Rude to others Curses at self Screams or yells Argues Inappropriate comments or behavior Somatic Items Food does not taste right Nightmares Nauseous Stomach hurts Ringing in ears Poor appetite Dizzy Stomach bloated Trouble hearing Trouble falling asleep Headaches Motor Items Difficulty lifting heavy objects Drops things Loses balance Weak Trips over things Muscles tingle or twitch Moves slowly Difficulty performing chores Memory/Attention Items Forgets to take medication Late for appointments Forgets to turn off appliances Forgets or misses appointments Loses way, gets lost Misplaces things Forgets yesterday’s events Forgets people’s names Forgets phone numbers Confused Loses train of thought Disorganized Easily distracted Concentration poor Loses track of time, day, or date Forgets to do chores or work Forgets if he or she has done things Forgets what he or she reads Trouble following instructions Communication Items Speech does not make sense Talks too fast or slow Writes slowly Makes spelling mistakes Trouble understanding conversation Difficulty thinking of the right word Writing is hard to read Difficulty pronouncing words Reads slowly Difficulty making conversation
Ratings
of NFI item
Frequency
RATINGS,
Table
P>F
P=F
F>P
26 20 26 z: 24 23 24 22 26 21 22 26
60 60 58 58 57 56 55 55 54 54 53 52 48
14 20 16 22 23 20 22 21 24 20 26 26 26
9 10 15 17 17 26 19 21 17
a4 80 71 68 66 61 59 59 56
7 10 14 15 17 13 22 20 27
14 13 14 15 22 18 15 ia 17 21 20
75 74 72 71 71 70 69 68 66 62 61
11 13 14 14 7 12 16 14 17 17 19
15 16 20 17 ia 22 17 19
65 63 61 61 61 59 59 57
20 21 19 22 21 19 24 24
22 19 20 21 20 23 23 28 27 20 25 24 23 26 27 26 30 30 23
66 63 63 62 60 58 57 57 57 56 56 55 55 53 52 51 51 50 50
12 18 17 17 20 19 20 15 16 24 19 21 22 21 21 23 19 20 27
22 24 25 29 28 29 27 35 28 35
70 63 61 59 58 58 57 53 52 51
8 13 14 12 14 13 16 12 20 14
1257
See1
3: Distribution
of Patient for Each Item Difference
NFI Scale Depression Somatic Memory/Attention Communication Aggression Motor
-3 -
* Negative values denote lems on an item compared reported as “-“.
and Family Difference by NFI Scale
Scores
Between Family and Patient Ratings*
-2
-1
4% 2% 4% 5% 3% 2%
19% 15% 20% 23% 15% 16%
0 55% 69% 56% 58% 69% 61%
+1
+2
+3
18% 12% 17% 11% 12% 18%
3% 2% 3% 2% 2% 3%
-
patients reported a higher frequency of probto family members; values less than 1% are
between family and patient ratings for 57 of the 70 items (81%). Conversely, significant differences were found for only 13 of the 70 items (table 4). For all 13 statistically different items, patients rated their neurobehavioral problems as occurring more frequently than did corresponding family members. Not a single NFI item was rated as occurring more frequently by family members. Seven of the 13 different items, including 5 of the 6 largest discrepancies, were on the Communication scale. The Memory/Attention scale had the second largest number of differences (3). Only the Motor scale had no statistically different items. Differences for Scale Scores The fourth set of analyses examined the consistency between family and patient perceptions as indicated by differences in mean NFI scale scores. Family members’ and patients’ mean scores were calculated for each scale. Six matched-pair t tests were then conducted (table 5). These tests revealed no statistical differences between family and patient mean scores for 5 of the 6 NFI scales. Only the Communication scale showed a difference with family mean scores (M = 1.827) being significantly less than patient mean scores (M = 1.992) (t = -6.03, p < .Ol). The results strongly suggest that communication is the only area where patients perceive more frequent problems than are perceived by family members. Differences by Injury Severity For each patient, a difference score was calculated for each item by subtracting the patient’s rating from the family member’s rating. For each NFI scale, a mean difference score was
Table
4: NFI Items With Significantly and Patient Responses Item/Scale
Difficulty pronouncing words (C) Difficulty making conversation (C) Ringing in ears (S) Makes spelling mistakes (C) Speech does not make sense (C) Difficulty thinking of right word (C) Curses at self (A) Trouble understanding conversation Forgets people’s names (M/A) Uncomfortable around others (D) Forgets phone numbers (M/A) Forgets if he or she has done things (M/A) Writing is hard to read (C) Abbreviations: Memory/Attention; * p < .Ol.
C, Communications; D, Depression.
Arch
Phys
(C)
Different
Family
df
Family Mean
Patient Mean
tva1ue
296 293 290 277 294 297 293 290 292 294 293
1.815 1.684 1.784 2.050 1.400 2.181 1.612 1.674 2.133 1.827 2.133
2.104 1.939 1.955 2.259 1.549 2.369 1.789 1.835 2.300 1.983 2.282
-6.31* -5.07” -4.77” -4.47* -4.13* -4.06* -3.70” -3.69* -3.62* -3.45% -3.35*
294 292
2.197 1.751
2.336 i .a87
-2.87’ -2.66,
S, Somatic;
Med Rehabil
A, Aggression,
Vol78,
November
M/A,
1997
1258
NEUROBEHAVIORAL
Table
5: NFI Scale
Differences
Scale
Communication Motor Attention/Memory Somatic Depression Aggression
Between
Family
and Patient
Ratings
df
Family Mt3W
Patient M6W-l
t Value
291 291 290 291 296 297
1.827 2.140 2.054 1.853 2.043 1.586
1.992 2.092 2.102 1.880 2.063 1.593
-6.03* 1.98 -1.65 -1.57 -.68 -.30
* p c .Ol.
obtained and utilized as a dependent measure. One-way ANOVA tests with injury severity as the independent variable were used to compare mean difference scores on each of the six NFI scales. Injury severity main effects were observed for only two scales: Memory/Attention (F = 9.41, p < .Ol) and Communication (F = 5.96, p < .Ol). Tukey post hoc tests for the Memory/Attention scale indicated significant differences between the very severe (M difference = .26) and mild (M difference = -.14) injury groups. For the Communication Scale, the very severely injured group (M difference = .09) differed from both the mild (M difference = -.21) and moderate (M difference = -.21) injury groups.
RATINGS,
See1
items (3), and did not evaluate the potential influence of injury severity. Only a few researchershave directly compared patients’ and relatives’ perceptions of memory and attention problems. Investigators have provided minimal evidence of differing perceptions and have not considered injury severity.10~‘2 In our investigation, problems related to memory/attention were generally perceived similarly by patients and family members. When injury severity was taken into account, however, a different pattern of findings emerged. Persons with very severe injuries tended to underrate their problems. In contrast, their mild injury counterparts typically rated problems as occurring slightly more frequently than did family members. In regard to aspectsof physical functioning, the present investigation provided little evidence of differences in perception. Regardless of injury severity, comparisons between family members’ and patients’ mean scores on both the Somatic Scale and the Motor Scales showed no statistical differences. These results are consistent with previous research, which has shown high agreement rates for physical functioning.‘0,‘2,14
Limitations and Future Research Drawing valid conclusions from these findings requires consideration of study limitations. Several limitations relate to sample composition factors, including ethnicity and referral source. DISCUSSION The sample was composed of persons from a single level I Researchersin the United States and Europe have contended trauma center. A study with patients from multiple regional that patients typically report less severe emotional and behavcenters might provide important benefits for sample representaioral problems than do their family members.‘~5~7~g~10~14~15 Our tiveness, particularly in regard to minority groups other than investigation did not support this contention. Comparisons beAfrican American. Additionally, the sample consisted of pertween patients’ and family members’ responses on the NFI sons referred for outpatient services, a group more likely to Aggression and Depression scalesrevealed that identical ratings have a negative outcome than those not requesting outpatient were assigned to the same items in a vast majority of cases. services. Furthermore, comparisons between mean scoresfor both scales Our investigation provides evidence that patients and family revealed no significant differences. In fact, mean patient and members have similar perceptions about many areas of funcfamily ratings for each scale differed by less than .02. Lastly, tioning. There was limited evidence that injury severity is an analysis by injury severity level indicated no differences beimportant consideration. Injury severity in our investigation, tween mean ratings for either scale. however, was based on the number of days patients were unconDifferential findings may be attributable to methodologic difscious. The use of the Glascow Coma Scale (GCS) or posttrauferences relating to questionnaire content and item specificity. matic amnesia data might give a more comprehensive view of In the present investigation, the NFI Aggression and Depression injury severity. Although mean agreement levels for the sample scaleswere comprised of 22 highly specific items. Other investiwere relatively high, differences between individual patient’s gators have measured emotional functioning using only a few and family member’s responsesvaried. More research is needed items with varying specifity. 1s,7,g~10~14~15 A few recent studies to help elaborate factors related to agreement levels. Future have suggested that agreement levels between patients and faminvestigations might focus more carefully on variable and interily members may be higher when item content is specific and action effects for injury severity, age, nature of family relationa large number of items are used.‘6~1s For example, Kinsella et ship, family members’ mental health, and amount of contact alzl compared family members’ and patients’ ratings of the between patients and family members. patient’s depression. No relation between ratings was observed Implications for the single item Visual Analogue Scale, but investigators found a relation (Y = .66) between ratings on the 15item Leeds Researchers have distinguished three hierarchical levels of awareness: intellectual, emergent, and anticipatory.22X23 IntellecDepression Scale. The Communication Scale was the only NFI scale for which tual awareness is the most basic knowledge of abilities and significant between-group differences were found. Statistical actions. Emergent awarenessrefers to recognition of ability that comparisons of Communication Scale items found differences develops during the course of completing a task. Anticipatory between patients’ and family members’ ratings for 7 of the 10 awareness refers to recognition before task initiation. The curitems. Contrary to expectation, patients rated problems as more rent study contributes to our understanding of the most basic frequent than did family members for all seven items. When level of self knowledge, intellectual awareness.Other methodolinjury severity was analyzed, patients with mild and moderate ogies are needed to help quantify perceptions about emergent injuries tended to rate problems as occurring more frequently and anticipatory awareness. than did the family members. Family members’ and patients’ Rehabilitation professionals have been encouraged to careratings were in closer agreement for persons with very severe fully consider patients’ perspectives regarding the sequelae of injuries. In contrast, Cavallo et al” found no evidence of subinjury,24 yet researchershave traditionally given more consideration to the reports of relatives than patients. Our findings sugstantial differences in their comparison of patients’ and family members’ ratings of communication problems. However, they gest that outpatients and family members have similar perceptions regarding the patients’ difficulties except in the area of used a relatively small sample (n = 34), a small number of Arch
Phys Med
Rehabil
Vol78,
November
1997
NEUROBEHAVIORAL
communication. Professionals are likely to find themselves in situations in which no one is available to corroborate the patient’s report or the patient is either unwilling or unable to provide a report. In such cases, a single report may serve as a reasonable estimate of the patient’s difficulties. In casesof severe cognitive or behavioral impairment, clinicians will find patients who are clearly unable to provide accurate or meaningful information about difficulties. In such cases, clinicians’ and family members’ judgments will seem far more important. Perhaps the most difficult situation is encountered when patients are capable of living independently and many areas of cognitive functioning are within normal limits. Discounting the perceptions and desires of these patients is a questionable practice even though they may seem disparate from those of family and staff members. To further clarify the validity of patients’ and family members’ perceptions of disability, research is needed comparing these perceptions with standardized measures of daily living skills and cognitive ability. References
1. Thomsen IV. The patient with severe head injury and his family. Stand J Rehabil Med 1974;6:180-3. 2. Thomsen IV. Late outcome of very severe blunt head trauma: a lo15 year second follow-up. J Neurol Neurosurg Psychiatry 1984;47: 260-8. 3. Weddell R, Oddy M, Jenkins D. Social adjustment after rehabilitation: a two-year follow-up of patients with severe head injury. Psychol Med 1980; 10:257-63. 4. M&inlay WW, Brooks DN, Bond MR. The short term outcome of severe blunt head injury as reported by the relatives of the injured person. J Nemo1 Neurosurg Psychiatry 1981;44:527-33. 5. Oddy M, Coughlan T, Tyerman A, Jenkins D. Social adjustment after closed head injury: a further follow-up seven years after injury. J Neurol Neurosurg Psychiatry 1985;48:564-8. 6. Brooks N, Campsie L, Symington C, Beattie A, McKinlay W. The five year outcome of severe blunt head injury: a relative’s view. J Neurol Neurosurg Psychiatry 1986;49:764-70. 7. M&inlay WW, Brooks DN. Methodological problems in assessing psychosocial recovery following severe head injury. J Clin Neuropsycho1 1984;6:87-99. 8. Prigatano GP. Disturbances of self-awareness of deficit after brain injury. In: Prigatano GP, Schacter DL, editors. Awareness of deficit after brain injury. New York: Oxford University Press, 1991. p. 111-26.
RATINGS,
1259
Seel
9. Oddy M, Humphrey M, Uttley D. Subjective impairment and social recovery after closed head injury. J Neurol Neurosurg Psychiatry 1978;41:611-16. 10. Cavallo MM, Kay T, Ezrachi 0. Problems and changes after traumatic brain injury: differing perceptions within and between families. Brain Inj 1992;6:327-35. 11. Kay T, Cavallo MM, Ezrachi 0. Administration Manual, N.Y.U. Head Injury Family Interview, Version 1.2. New York: N.Y.U. Research and Training Center on Head Trauma and Stroke; 1988. 12. Hendryx PM. Psychosocial changes perceived by closed-head-injured adults and their families. Arch Phys Med Rehabil 1989;70: 526-30. 13. Prigatano GP. Neuropsychological rehabilitation after brain injury. Baltimore (MD): The Johns Hopkins University Press; 1986. 14. Prigatano GP, Altman IM, O’Brien KP. Behavioral limitations that traumatic-brain-injured patients tend to underestimate. Clin Neuropsycho1 1990;4:163-76. 15. Prigatano GP, Leathem JM. Awareness of behavioral limitations after traumatic brain injury: a cross-cultural study of New Zealand Maoris and Non-Maoris. Clin Neuropsychol 1993;7:123-35. 16. Gasquoine PG. Affective state and awareness of sensory and cognitive effects after closed head injury. Neuropsychol 1992;4: 187-96. 17. Sherer M. Boake C. Silver B. Levin E. Rinzholz G. Wilde MC. et al. Assessing awareness of deficits following acquired brain injury: the Awareness Questionnaire [abstract]. J Int Neuropsychol Sot 1995; 1:163. 18. Sherer M, Boake C, Clement V, Levin E, High W. Awareness of deficits after traumatic brain injury: comparison of patient, family, and clinician ratings [abstract].-J Int Neuropsychol Sot 1996;2: 17. 19. Kreutzer JS. Marwitz JH. See1 R. Serio CD. Validation of a Neurobehavioral Functioning Inventory for adults with traumatic brain injury. Arch Phys Med Rehabil 1996177: 116-24. 20. Keppel G. Design and analysis: a researcher’s handbook. Englewood Cliffs (NJ): Prentice Hall; 1991. 21. Kinsella G, Moran C, Ford B, Ponsford J. Emotional disorder and its assessment within the severe head injury population. Psycho1 Med 1988;18:57-63. 22. Crosson B, Barco PP, Velozo CA, Bolesta MM, Cooper PV, Werts D, et al. Awareness and compensation in postacute head injury rehabilitation. J Head Trauma Rehabil 1989;4:46-54. 23. Barco PP, Crosson B, Bolesta MM, Werts D, Stout R. Training awareness and compensation in postacute head injury rehabilitation. In: Kreutzer JSI Wehman P, editors. Cognitive rehabilitation for persons with traumatic brain injury. Baltimore (MD): Paul H. Brookes; 1991. p. 129-46. 24. Commission on Accreditation of Rehabilitation Facilities. 1996 Standards for Medical Rehabilitation. Tucson (AZ): CARF; 1996. v
Arch
Phys Med
Rehabil
Vol78,
November
1997
Concordance of Patients’ and Family Members’ Ratings of Neurobehavioral Functioning After Traumatic Brain Injury Ronald T. Seel, MS, Jeffrey S. Kreutzer, PhD, Angelle M. Sander, PhD ABSTRACT. See1 RT, Kreutzer JS, Sander AM. Concordance of patients’ and family members’ ratings of neurobehavioral functioning after traumatic brain injury. Arch Phys Med Rehabil 1997; 78: 1254-9. Objective: To examine differences in family and patient evaluation of neurobehavioral functioning in adults with traumatic brain injury (TBI). Design: Differences were examined by conducting 70 paired sample t tests on scale items and 6 paired sample t tests on scale scores from a neurobehavioral inventory. Setting: Medical center outpatient clinic. Participants: Three hundred one consecutive adult patients with TBI and 301 informants, primarily family members, completed the neurobehavioral inventory. Main Outcome Measure: Neurobehavioral Functioning Inventory (NFI) comprised of six scales with items describing symptoms and daily living problems. Results: Paired t test analysesof the six scalesindicated that patients reported a significantly greater level of communication problems than did their matched family members. No differences were found for the other five scales.Paired t test analyses of the 70 scale items revealed significant differences in patient and family ratings for only 13 items. In all 13 instances,patients reported greater levels of dysfunction than were reported by their family members. Analysis of variance (ANOVA) indicated a main effect of injury severity for only the Communication and Memory/Attention scales. Conclusions: Findings indicate general agreement between family members and patients regarding patients’ everyday problems. Results do not support contentions that patients tend to underestimate difficulties. Agreement levels appear related to injury severity, item specificity, and item content. More research is needed to identify other variables relating to agreement levels, including age, injury severity, and amount of contact between patients and family members. 0 1997 by the American Congress of Rehabilitation Medicine and the American tation
Academy of Physical Medicine
and Rehabili-
OR MORE THAN TWO decades, investigators have atF tempted to describe psychosocial outcome after traumatic brain injury (TBI).1-6 In these studies, a preference has been expressed for relying on the reports of family members rather than patients.2U4,6 Greater emphasis on family members’ reports From the Department of Physical Medicine and Rehabilitation, Medical College of Viwinia. Virginia Commonwealth Universitv. Richmond. VA. Submitted fo;publication March 3, 1997. Accepted April 7, 1997. Supported in part by grant H133B80029 from the National Institute on Disabil-. ity and Rehabilitation Research, United States Department of Education. No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit upon the authors or mm am cmanization with which the authors are associated. Reprint requests to Jeffrey S. Kreutzer, PhD, Department of Physical Medicine and Rehabilitation, Medical College of Virginia, MCV Box 980542, Richmond, VA 23298.0542. 0 1997 by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation 0003-9993/97/7811-4412$3.00/O
Arch
Phys Mad
Rehabil
Vol78,
November
1997
has been justified, based on assertions that cognitive impairments, including poor self-awareness, compromise the validity of self-report information. ‘,’ However, empirical examination of actual differences between patients’ and family members’ perceptions of neurobehavioral functioning has yielded varying results. Some investigators have collected psychosocial outcome data from both patients and family members, comparing their reports. For example, using a structured interview, Thomsen’ gathered information about injury-related changes from 50 patients with severe TBI and their family members. Interviews were conducted between 1 and 5 years after injury. While the majority of relatives reported negative behavioral and personality changes, patients rarely made such complaints. Oddy and colleagues9 administered a neurobehavioral symptom checklist to 54 patients with severe TBI and their relatives at 6 months after injury. The type and frequency of symptoms reported by patients and relatives was highly similar. In spite of similarities, the authors reported that “there were marked discrepancies in individual cases, which underlines the hazards of accepting the patient’s report as definitive.” 9 No statistical tests, however, were performed to support their conclusion. In a subsequent study conducted by Oddy et al5 at 7 years after injury, a greater discrepancy was noted in the type and frequency of symptoms reported by patients and relatives. Differences in proportions of patients and relatives reporting the same problem ranged from 4% to 26%. Forty percent of relatives reported that their loved ones refused to admit difficulties. Once again, despite the discrepancies noted, no statistical tests were conducted to compare patients’ and relatives’ ratings. Furthermore, no information was provided on the validity of the neurobehavioral checklist used. Investigators have also provided evidence that patient and family agreement levels vary depending on areas of functioning assessed.Lower agreement rates have been found for items pertaining to patients’ emotional and behavioral changes. For example, using a neurobehavioral checklist, M&inlay and Brooks7 investigated agreement rates between 52 severely injured patients and their relatives 6 months after injury. They found agreement rates ranging from 52% to 85% between patients and relatives. The highest rate of agreement was for sensory and motor difficulties, and the greatest disagreement was for items pertaining to emotional and behavioral changes. However, no information regarding patients’ injury severity or the validity of the checklist was provided, and an unspecified method of analysis was used. Cavallo, Kay, and Ezrachilosllstudied patients’ and relatives’ ratings on the 34-item Problem Checklist of the New York Head Injury Family Interview. Subjects were 34 patients with primarily moderate to severeinjuries. Patients and their relatives rated neurobehavioral symptoms at an average of 2 years after injury. In 35% of the cases,patients and family members were categorized as being in “high agreement,” as defined by agreement on at least 75% of the items. Twenty-three percent of the patients endorsed difficulties at least twice as often as their relatives, and 32% of the patients showed the opposite pattern, underestimating difficulties compared with family members’
NEUROBEHAVIORAL
reports. On items pertaining to interpersonal and emotional difficulties, patients tended to rate problems as less frequent. However, conclusions were drawn primarily on the basis of descriptive statistics. Hendryx” found a similar pattern of results in a sample of 20 patients with severe TBI. Patients and family members rated perceived change from before to after injury in the areas of physical, emotional, and cognitive functioning. Although no significant difference was found between patients’ and family members’ ratings of physical and cognitive changes, family members perceived patients as exhibiting greater emotional changes than reported by patients themselves. Unfortunately, no validity information was provided for the rating scale. Prigatano and colleagues’3 developed the Patient Competency Rating Scale (PCRS) to assess a wide range of difficulties commonly encountered after brain injury. The PCRS is a 30-item rating scale with separate forms available for completion by patients and family members. Patients’ abilities are rated on a 5-point scale, with 1 conveying an inability to perform the activity and 5 conveying performance of the activity with ease. Prigatano et all4 compared patients’ and family members’ PCRS ratings for 64 patients representing the full range of injury severity. Comparisons were made for 18 of the 30 items. Agreement rates ranged from 20% to 77%. Consistent with earlier studies, the highest rates of agreement were for items pertaining to self-care activities (eg, “dressing self,” “personal hygiene”), whereas the lowest rates were for emotional and behavioral characteristics (eg, “handling arguments,” “keep from being depressed”). Similar results were found in an investigation of PCRS ratings in a sample of 41 English-speaking New Zealand patients and their relativesi Although the PCRS has now been used in several studies of self-awareness, no information has been provided regarding psychometric properties. Research has also suggested that differences in perceptions of neurobehavioral functioning relate to the specificity of questions or items. For example, Gasquoinei6 compared neurobehavioral ratings made by patients and staff. He found greater levels of agreement for more clearly definable items. Sherer and colleagues17 developed the Awareness Questionnaire to allow comparisons between patients’, family members’ and clinicians’ ratings of patients’ difficulties. The consistency of patients’ and family members’ perceptions is used to determine self-awareness. Working mostly with severe brain injury patients, Sherer et al’* found that impaired awareness was more commonly observed on items addressing general rather than specific behaviors. In summary, studies comparing patients’ and relatives’ perceptions of difficulties have found widely varying rates of agreement. Agreement levels have been related to item specificity and content. More specific items tend to elicit higher levels of agreement between patients’ and family members’ perceptions. Furthermore, higher levels of agreement have been observed for items relating to self-care and physical functioning. Disagreement in perceptions has been observed for items relating to emotional and behavioral changes. Variable research findings likely result from several methodologic limitations and differences. First, the relation between injury severity and agreement levels has not been investigated. With a few exceptions, studies have included primarily persons with severe injuries, leaving questions about patients with mild and moderate injuries. Second, sample sizes have generally been small, ranging from 20 to 64. Third, the majority of studies have used descriptive statistics or visual inspection to compare patients’ and family members’ perceptions. Finally, researchers have used instruments varying in content, response format, and
RATINGS,
1255
See1
item specificity. A comparison of instruments used to compare patients’ and family members’ perceptions of problems is provided in table 1. Some researchers have examined responses to individual items, whereas others have organized items into scales based on face validity considerations. The use of empirically constructed scales with demonstrated validity, however, has been lacking. The current study attempted to clarify the relation between patients’ and family members’ perceptions of injury-related changes by improving on the methodology of previous studies (table 1). Investigation focused on a large sample of patients representing the full range of injury severity, and the relation between injury severity and agreement rates was examined. The study also used a factor-analytically derived, validated assessment tool, the Neurobehavioral Functioning Inventory (NFI).19 The NFI items are highly specific and can be combined into six scales representing distinct categories of neurobehavioral functioning. The present investigation was unique in comparing responses to 70 individual items and six scale scores using between-groups statistical tests. Based on previous research, patients were expected to report less frequent emotional and behavioral difficulties than family members, as indicated by comparisons between NFI Depression and Aggression scores. Perceptions regarding physical functioning were not expected to differ, as indicated by comparisons involving the Motor and Somatic Scales. In addition, greater discrepancies were anticipated for patients with more severe injuries.
METHOD Subjects The original sample consisted of 310 pairs of patients and informants referred for comprehensive outpatient assessment at a regional level I trauma center. Informants were selected by the patients, who were asked to identify a single individual who was either a primary caretaker or who knew the patient well. To control for unreliable responses, nine pairs of informants and patients were eliminated because of high levels of missing data. For these nine cases, more than 25% of items were not completed by either one or both members of the pair. The final sample for analysis consisted of 301 pairs of patients and informants. The following demographic and medical information for patients was identified: 55% were men; with regard to ethnicity, 80% were Caucasian, 18% were African American, and 2% were of other ethnic backgrounds; 79% had been involved in vehicular accidents; mean age at the time of injury was 35.9 years (SD = 12.2). Patients were classified into one of four injury severity groups based on number of days unconscious: (1) “mild” = 20 minutes or less (n = 161); (2) “moderate” = greater than 20 minutes and less than 24 hours (n = 21); (3) “severe” = greater than or equal to 24 hours and less than 7 days (n = 20); and, (4) “very severe’ ’ = 7 days or greater (n = 47). The mean length of time in coma was 9.6 days (SD = 26.8), and the median length of time in coma was 0 days. Average time postinjury for evaluation was 2.2 years (SD = 3.8), with a median value of 0.8 years. Informants largely consisted of immediate family members: spouses (42%); parents (30%); children (7%); and siblings (5%). A small proportion of informants were nonfamily members that included: significant others (10%); friends (4%); and others (2%). Seventy-three percent of the informants were women. Informants’ mean age at the time of evaluation was 41.7 years (SD = 13.4).
Arch
Phys Med
Rehabil
Vol78,
November
1997
1256
NEUROBEHAVIORAL
Table instrument
Characteristics
Thomsen
(1974)’
Number of items Areas of functioning
Not reported Emotional, behavioral, cognitive symptoms
Response
Structured
format
Item specificity Validation Construction of scales Statistically derived scales
interview
Uncertain None reported None reported None reported
1: Characteristics Oddy et al (1978)’
of Assessment Prigatano
Methods et al (1990)”
Cavallo et al (1992)‘O
Emotional, somatic, cognitive, communication Checklist, yes or no
34 Emotional, behavioral, physical, cognitive, communication
Low to high None reported None reported None reported
Low to high None reported Yes None reported
Dichotomous; converted from Likert scale High None reported Yes None reported
Present Investigation
70 Aggression, depression, memory/attention, communication, somatic, motor Likert scale
High Yes Yes Yes
RESULTS
The NFI is contained within a larger questionnaire, the General Health and History Questionnaire (GHHQ). The NFI is comprised of 70 scored items grouped into six categories based on principal components and confirmatory factor analytic methodology: Depression; Somatic Complaints; Memory/Attention Difficulties; Communication Deficits; Aggressive Behaviors; and Motor Impairment.” Frequency of problem occurrence is rated by patients or informants on a 4-point scale that allows the following choices: (1) never; (2) sometimes; (3) often; or (4) always. For a subset of items, raters have an opportunity to rate problems as not applicable. Research has provided support for criterion-related validity and high internal consistency within NFI scales.” For all scales, Chronbach’s alpha has ranged from .86 to .95. In addition, scores for each of the NFI scales have been found to correlate logically with other measures of neurobehavioral dysfunction.
Procedure The neurobehavioral inventory was mailed to patients and informants in advance of the patient’s clinic appointment. On arrival for evaluation, responses to the inventory were reviewed for compliance with instructions. Forms were corrected or missing information was obtained by interviewing the informant in person or by phone. In cases where patients were re-evaluated, more than one NFI was often available. In such cases, the most recent questionnaires completed by both the patient and informant were selected for analysis. For each NFI scale, a total score was calculated based on the sum of ratings for all scale items. Descriptive statistics were calculated for all items and for the six NFI scales. Seventy-six matched-pair t tests were conducted to examine differences in family and patient ratings on the 70 scored items and the six scale scores of the NFI. The t tests were calculated using the pairwise deletion method. Nonmissing values for each pair of variables were correlated, regardless of whether other variables for the patient/family set were missing. As a result, the number of participants for each t test differed based on the number of valid variable pairs. Six one-way analysis of variance (ANOVA) tests were conducted to determine whether patient injury severity had an impact on differences in patents’ and family members’ ratings of scale scores. When a main effect for injury severity was found, a Tukey post hoc test was conducted to identify significant differences between the four injury severity categories.“’ Given the number of planned comparisons, a decision was made to set a conservative alpha level for t tests and ANOVA tests, minimizing the chance of a Type I error. Family-wise error rates and the potential for Type II errors were also considered. Consequently, an alpha level ofp < .Ol was deemed acceptable.
Phys Med
See1
30 Emotional, behavioral, somatic, motor, cognitive, self-care, communication Likert scale
37 behavioral, sensory,
Measures
Arch
RATINGS,
Rehabil
Vol78,
November
1997
Differences for Individual Items The first series of analyses attempted to determine the similarities and differences between responses given by family members and patients for the same item. Each pair of responses was categorized in one of three ways: (1) the patient rated the problem as occurring more frequently than the family member (P > F); (2) the patient and family member were in agreement, rating the problem as occurring with equal frequency (P = F); and (3) the family member rated the problem as occurring more frequently than the patient (F > P). For the 70 NFI items, the proportion of paired ratings falling into each category was calculated (table 2). For all 70 NFI items, the percentage of agreement ranged from 48% to 84%. For each item, fewer than 35% of patients (range, 9% to 35%) rated their problems as occurring more frequently than rated by family members. For each item, approximately one fourth of family members (range, 7% to 27%) rated the patient’ problems as occurring more frequently than indicated by the patient’s self-report. The data suggest that family and patients tend to rate problems as occurring with equal frequency. The second set of analyses measured the magnitude of difference between family and patient responses for NFI items. For each pair of responses, the patient’s item rating based on the four point Likert scale was subtracted from the family member’s rating. Difference values ranged from -3 to +3. Negative difference scores indicated that the family member rated the problem as occurring less frequently than the patient; positive scores indicated that the patient rated the problem as occurring less frequently than the family member. A difference score of “0” indicated that the family member and patient rated the problem as occurring with the same frequency. The greater the score deviated from zero regardless of direction, the greater the magnitude of disagreement between the family member’s and patient’s rating. For each NFI scale, the proportion of difference values for all scale items was calculated (table 3). The magnitude of similarities and differences between family and patient responses was then analyzed. The percentage of paired responses with identical item scores (0 difference) ranged from 55% to 69%. For more than 92% of the paired responses, there was either exact agreement or a difference of one point. The difference between patient and family ratings exceeded a value of two in fewer than 1% percent of the responses. The results strongly suggest that family members and patients rate problems as occurring with similar frequency. The third set of analyses identified NFI items for which there were statistically different responses between patients and family members. Matched-pair t tests were conducted individually for all 70 items. These tests found no statistical differences
NEUROBEHAVIORAL
Table
2: Family
Versus
Patient
Depression Items Uncomfortable around others Feels worthless Lonely Restless Sad, blue Sits with nothing to do Scared or frightened Bored Difficulty enjoying activities Feels hopeless Cannot get mind off certain thoughts No confidence Frustrated Aggression Items Threatens to hurt others Hits or pushes others Breaks or throws things Curses at others Rude to others Curses at self Screams or yells Argues Inappropriate comments or behavior Somatic Items Food does not taste right Nightmares Nauseous Stomach hurts Ringing in ears Poor appetite Dizzy Stomach bloated Trouble hearing Trouble falling asleep Headaches Motor Items Difficulty lifting heavy objects Drops things Loses balance Weak Trips over things Muscles tingle or twitch Moves slowly Difficulty performing chores Memory/Attention Items Forgets to take medication Late for appointments Forgets to turn off appliances Forgets or misses appointments Loses way, gets lost Misplaces things Forgets yesterday’s events Forgets people’s names Forgets phone numbers Confused Loses train of thought Disorganized Easily distracted Concentration poor Loses track of time, day, or date Forgets to do chores or work Forgets if he or she has done things Forgets what he or she reads Trouble following instructions Communication Items Speech does not make sense Talks too fast or slow Writes slowly Makes spelling mistakes Trouble understanding conversation Difficulty thinking of the right word Writing is hard to read Difficulty pronouncing words Reads slowly Difficulty making conversation
Ratings
of NFI item
Frequency
RATINGS,
Table
P>F
P=F
F>P
26 20 26 z: 24 23 24 22 26 21 22 26
60 60 58 58 57 56 55 55 54 54 53 52 48
14 20 16 22 23 20 22 21 24 20 26 26 26
9 10 15 17 17 26 19 21 17
a4 80 71 68 66 61 59 59 56
7 10 14 15 17 13 22 20 27
14 13 14 15 22 18 15 ia 17 21 20
75 74 72 71 71 70 69 68 66 62 61
11 13 14 14 7 12 16 14 17 17 19
15 16 20 17 ia 22 17 19
65 63 61 61 61 59 59 57
20 21 19 22 21 19 24 24
22 19 20 21 20 23 23 28 27 20 25 24 23 26 27 26 30 30 23
66 63 63 62 60 58 57 57 57 56 56 55 55 53 52 51 51 50 50
12 18 17 17 20 19 20 15 16 24 19 21 22 21 21 23 19 20 27
22 24 25 29 28 29 27 35 28 35
70 63 61 59 58 58 57 53 52 51
8 13 14 12 14 13 16 12 20 14
1257
See1
3: Distribution
of Patient for Each Item Difference
NFI Scale Depression Somatic Memory/Attention Communication Aggression Motor
-3 -
* Negative values denote lems on an item compared reported as “-“.
and Family Difference by NFI Scale
Scores
Between Family and Patient Ratings*
-2
-1
4% 2% 4% 5% 3% 2%
19% 15% 20% 23% 15% 16%
0 55% 69% 56% 58% 69% 61%
+1
+2
+3
18% 12% 17% 11% 12% 18%
3% 2% 3% 2% 2% 3%
-
patients reported a higher frequency of probto family members; values less than 1% are
between family and patient ratings for 57 of the 70 items (81%). Conversely, significant differences were found for only 13 of the 70 items (table 4). For all 13 statistically different items, patients rated their neurobehavioral problems as occurring more frequently than did corresponding family members. Not a single NFI item was rated as occurring more frequently by family members. Seven of the 13 different items, including 5 of the 6 largest discrepancies, were on the Communication scale. The Memory/Attention scale had the second largest number of differences (3). Only the Motor scale had no statistically different items. Differences for Scale Scores The fourth set of analyses examined the consistency between family and patient perceptions as indicated by differences in mean NFI scale scores. Family members’ and patients’ mean scores were calculated for each scale. Six matched-pair t tests were then conducted (table 5). These tests revealed no statistical differences between family and patient mean scores for 5 of the 6 NFI scales. Only the Communication scale showed a difference with family mean scores (M = 1.827) being significantly less than patient mean scores (M = 1.992) (t = -6.03, p < .Ol). The results strongly suggest that communication is the only area where patients perceive more frequent problems than are perceived by family members. Differences by Injury Severity For each patient, a difference score was calculated for each item by subtracting the patient’s rating from the family member’s rating. For each NFI scale, a mean difference score was
Table
4: NFI Items With Significantly and Patient Responses Item/Scale
Difficulty pronouncing words (C) Difficulty making conversation (C) Ringing in ears (S) Makes spelling mistakes (C) Speech does not make sense (C) Difficulty thinking of right word (C) Curses at self (A) Trouble understanding conversation Forgets people’s names (M/A) Uncomfortable around others (D) Forgets phone numbers (M/A) Forgets if he or she has done things (M/A) Writing is hard to read (C) Abbreviations: Memory/Attention; * p < .Ol.
C, Communications; D, Depression.
Arch
Phys
(C)
Different
Family
df
Family Mean
Patient Mean
tva1ue
296 293 290 277 294 297 293 290 292 294 293
1.815 1.684 1.784 2.050 1.400 2.181 1.612 1.674 2.133 1.827 2.133
2.104 1.939 1.955 2.259 1.549 2.369 1.789 1.835 2.300 1.983 2.282
-6.31* -5.07” -4.77” -4.47* -4.13* -4.06* -3.70” -3.69* -3.62* -3.45% -3.35*
294 292
2.197 1.751
2.336 i .a87
-2.87’ -2.66,
S, Somatic;
Med Rehabil
A, Aggression,
Vol78,
November
M/A,
1997
1258
NEUROBEHAVIORAL
Table
5: NFI Scale
Differences
Scale
Communication Motor Attention/Memory Somatic Depression Aggression
Between
Family
and Patient
Ratings
df
Family Mt3W
Patient M6W-l
t Value
291 291 290 291 296 297
1.827 2.140 2.054 1.853 2.043 1.586
1.992 2.092 2.102 1.880 2.063 1.593
-6.03* 1.98 -1.65 -1.57 -.68 -.30
* p c .Ol.
obtained and utilized as a dependent measure. One-way ANOVA tests with injury severity as the independent variable were used to compare mean difference scores on each of the six NFI scales. Injury severity main effects were observed for only two scales: Memory/Attention (F = 9.41, p < .Ol) and Communication (F = 5.96, p < .Ol). Tukey post hoc tests for the Memory/Attention scale indicated significant differences between the very severe (M difference = .26) and mild (M difference = -.14) injury groups. For the Communication Scale, the very severely injured group (M difference = .09) differed from both the mild (M difference = -.21) and moderate (M difference = -.21) injury groups.
RATINGS,
See1
items (3), and did not evaluate the potential influence of injury severity. Only a few researchershave directly compared patients’ and relatives’ perceptions of memory and attention problems. Investigators have provided minimal evidence of differing perceptions and have not considered injury severity.10~‘2 In our investigation, problems related to memory/attention were generally perceived similarly by patients and family members. When injury severity was taken into account, however, a different pattern of findings emerged. Persons with very severe injuries tended to underrate their problems. In contrast, their mild injury counterparts typically rated problems as occurring slightly more frequently than did family members. In regard to aspectsof physical functioning, the present investigation provided little evidence of differences in perception. Regardless of injury severity, comparisons between family members’ and patients’ mean scores on both the Somatic Scale and the Motor Scales showed no statistical differences. These results are consistent with previous research, which has shown high agreement rates for physical functioning.‘0,‘2,14
Limitations and Future Research Drawing valid conclusions from these findings requires consideration of study limitations. Several limitations relate to sample composition factors, including ethnicity and referral source. DISCUSSION The sample was composed of persons from a single level I Researchersin the United States and Europe have contended trauma center. A study with patients from multiple regional that patients typically report less severe emotional and behavcenters might provide important benefits for sample representaioral problems than do their family members.‘~5~7~g~10~14~15 Our tiveness, particularly in regard to minority groups other than investigation did not support this contention. Comparisons beAfrican American. Additionally, the sample consisted of pertween patients’ and family members’ responses on the NFI sons referred for outpatient services, a group more likely to Aggression and Depression scalesrevealed that identical ratings have a negative outcome than those not requesting outpatient were assigned to the same items in a vast majority of cases. services. Furthermore, comparisons between mean scoresfor both scales Our investigation provides evidence that patients and family revealed no significant differences. In fact, mean patient and members have similar perceptions about many areas of funcfamily ratings for each scale differed by less than .02. Lastly, tioning. There was limited evidence that injury severity is an analysis by injury severity level indicated no differences beimportant consideration. Injury severity in our investigation, tween mean ratings for either scale. however, was based on the number of days patients were unconDifferential findings may be attributable to methodologic difscious. The use of the Glascow Coma Scale (GCS) or posttrauferences relating to questionnaire content and item specificity. matic amnesia data might give a more comprehensive view of In the present investigation, the NFI Aggression and Depression injury severity. Although mean agreement levels for the sample scaleswere comprised of 22 highly specific items. Other investiwere relatively high, differences between individual patient’s gators have measured emotional functioning using only a few and family member’s responsesvaried. More research is needed items with varying specifity. 1s,7,g~10~14~15 A few recent studies to help elaborate factors related to agreement levels. Future have suggested that agreement levels between patients and faminvestigations might focus more carefully on variable and interily members may be higher when item content is specific and action effects for injury severity, age, nature of family relationa large number of items are used.‘6~1s For example, Kinsella et ship, family members’ mental health, and amount of contact alzl compared family members’ and patients’ ratings of the between patients and family members. patient’s depression. No relation between ratings was observed Implications for the single item Visual Analogue Scale, but investigators found a relation (Y = .66) between ratings on the 15item Leeds Researchers have distinguished three hierarchical levels of awareness: intellectual, emergent, and anticipatory.22X23 IntellecDepression Scale. The Communication Scale was the only NFI scale for which tual awareness is the most basic knowledge of abilities and significant between-group differences were found. Statistical actions. Emergent awarenessrefers to recognition of ability that comparisons of Communication Scale items found differences develops during the course of completing a task. Anticipatory between patients’ and family members’ ratings for 7 of the 10 awareness refers to recognition before task initiation. The curitems. Contrary to expectation, patients rated problems as more rent study contributes to our understanding of the most basic frequent than did family members for all seven items. When level of self knowledge, intellectual awareness.Other methodolinjury severity was analyzed, patients with mild and moderate ogies are needed to help quantify perceptions about emergent injuries tended to rate problems as occurring more frequently and anticipatory awareness. than did the family members. Family members’ and patients’ Rehabilitation professionals have been encouraged to careratings were in closer agreement for persons with very severe fully consider patients’ perspectives regarding the sequelae of injuries. In contrast, Cavallo et al” found no evidence of subinjury,24 yet researchershave traditionally given more consideration to the reports of relatives than patients. Our findings sugstantial differences in their comparison of patients’ and family members’ ratings of communication problems. However, they gest that outpatients and family members have similar perceptions regarding the patients’ difficulties except in the area of used a relatively small sample (n = 34), a small number of Arch
Phys Med
Rehabil
Vol78,
November
1997
NEUROBEHAVIORAL
communication. Professionals are likely to find themselves in situations in which no one is available to corroborate the patient’s report or the patient is either unwilling or unable to provide a report. In such cases, a single report may serve as a reasonable estimate of the patient’s difficulties. In casesof severe cognitive or behavioral impairment, clinicians will find patients who are clearly unable to provide accurate or meaningful information about difficulties. In such cases, clinicians’ and family members’ judgments will seem far more important. Perhaps the most difficult situation is encountered when patients are capable of living independently and many areas of cognitive functioning are within normal limits. Discounting the perceptions and desires of these patients is a questionable practice even though they may seem disparate from those of family and staff members. To further clarify the validity of patients’ and family members’ perceptions of disability, research is needed comparing these perceptions with standardized measures of daily living skills and cognitive ability. References
1. Thomsen IV. The patient with severe head injury and his family. Stand J Rehabil Med 1974;6:180-3. 2. Thomsen IV. Late outcome of very severe blunt head trauma: a lo15 year second follow-up. J Neurol Neurosurg Psychiatry 1984;47: 260-8. 3. Weddell R, Oddy M, Jenkins D. Social adjustment after rehabilitation: a two-year follow-up of patients with severe head injury. Psychol Med 1980; 10:257-63. 4. M&inlay WW, Brooks DN, Bond MR. The short term outcome of severe blunt head injury as reported by the relatives of the injured person. J Nemo1 Neurosurg Psychiatry 1981;44:527-33. 5. Oddy M, Coughlan T, Tyerman A, Jenkins D. Social adjustment after closed head injury: a further follow-up seven years after injury. J Neurol Neurosurg Psychiatry 1985;48:564-8. 6. Brooks N, Campsie L, Symington C, Beattie A, McKinlay W. The five year outcome of severe blunt head injury: a relative’s view. J Neurol Neurosurg Psychiatry 1986;49:764-70. 7. M&inlay WW, Brooks DN. Methodological problems in assessing psychosocial recovery following severe head injury. J Clin Neuropsycho1 1984;6:87-99. 8. Prigatano GP. Disturbances of self-awareness of deficit after brain injury. In: Prigatano GP, Schacter DL, editors. Awareness of deficit after brain injury. New York: Oxford University Press, 1991. p. 111-26.
RATINGS,
1259
Seel
9. Oddy M, Humphrey M, Uttley D. Subjective impairment and social recovery after closed head injury. J Neurol Neurosurg Psychiatry 1978;41:611-16. 10. Cavallo MM, Kay T, Ezrachi 0. Problems and changes after traumatic brain injury: differing perceptions within and between families. Brain Inj 1992;6:327-35. 11. Kay T, Cavallo MM, Ezrachi 0. Administration Manual, N.Y.U. Head Injury Family Interview, Version 1.2. New York: N.Y.U. Research and Training Center on Head Trauma and Stroke; 1988. 12. Hendryx PM. Psychosocial changes perceived by closed-head-injured adults and their families. Arch Phys Med Rehabil 1989;70: 526-30. 13. Prigatano GP. Neuropsychological rehabilitation after brain injury. Baltimore (MD): The Johns Hopkins University Press; 1986. 14. Prigatano GP, Altman IM, O’Brien KP. Behavioral limitations that traumatic-brain-injured patients tend to underestimate. Clin Neuropsycho1 1990;4:163-76. 15. Prigatano GP, Leathem JM. Awareness of behavioral limitations after traumatic brain injury: a cross-cultural study of New Zealand Maoris and Non-Maoris. Clin Neuropsychol 1993;7:123-35. 16. Gasquoine PG. Affective state and awareness of sensory and cognitive effects after closed head injury. Neuropsychol 1992;4: 187-96. 17. Sherer M. Boake C. Silver B. Levin E. Rinzholz G. Wilde MC. et al. Assessing awareness of deficits following acquired brain injury: the Awareness Questionnaire [abstract]. J Int Neuropsychol Sot 1995; 1:163. 18. Sherer M, Boake C, Clement V, Levin E, High W. Awareness of deficits after traumatic brain injury: comparison of patient, family, and clinician ratings [abstract].-J Int Neuropsychol Sot 1996;2: 17. 19. Kreutzer JS. Marwitz JH. See1 R. Serio CD. Validation of a Neurobehavioral Functioning Inventory for adults with traumatic brain injury. Arch Phys Med Rehabil 1996177: 116-24. 20. Keppel G. Design and analysis: a researcher’s handbook. Englewood Cliffs (NJ): Prentice Hall; 1991. 21. Kinsella G, Moran C, Ford B, Ponsford J. Emotional disorder and its assessment within the severe head injury population. Psycho1 Med 1988;18:57-63. 22. Crosson B, Barco PP, Velozo CA, Bolesta MM, Cooper PV, Werts D, et al. Awareness and compensation in postacute head injury rehabilitation. J Head Trauma Rehabil 1989;4:46-54. 23. Barco PP, Crosson B, Bolesta MM, Werts D, Stout R. Training awareness and compensation in postacute head injury rehabilitation. In: Kreutzer JSI Wehman P, editors. Cognitive rehabilitation for persons with traumatic brain injury. Baltimore (MD): Paul H. Brookes; 1991. p. 129-46. 24. Commission on Accreditation of Rehabilitation Facilities. 1996 Standards for Medical Rehabilitation. Tucson (AZ): CARF; 1996. v
Arch
Phys Med
Rehabil
Vol78,
November
1997