duration of posttraumatic amnesia and the glasgow coma scale as

1

Upload: others

Post on 12-Feb-2022

2 views

Category:

Documents


0 download

TRANSCRIPT

DURATION OF POSTTRAUMATIC AMNESIA AND THE GLASGOW

COMA SCALE AS MEASURES OF SEVERITY AND

THEIR RELATIONSHIP TO COGNITIVE OUTCOME

FOLLOWING CLOSED HEAD INJURY

by

MICHAEL T. ROPACKI, B.A., M.A.

A DISSERTATION

IN

PSYCHOLOGY

Submitted to the Graduate Faculty of Texas Tech University in

Partial Fulfillment of the Requirements for

the Degree of

DOCTOR OF PHILOSOPHY

Approved

Acc/^pted

August, 2000

ACKNOWLEDGEMENTS

This dissertation would not have been possible without the never ending

assistance, commitment, and dedication of Jeff Elias, Ph.D.. Jeff went above and beyond

in all aspects of my training-I owe a great deal of who and where I am today to this

extraordinary man. His dedication to teaching and loyalty to his students is unyielding.

His knowledge base is amazing, and devotion to research and great science unparalleled.

Jeff has been a wonderful advisor, instructor, mentor and fiiend. He was my first official

introduction to neuropsychological research and theory, and a major reason I have chosen

this area of specialization for my career path. Jeff is unselfish, giving of his time, and

always willing to let his students, including me, pursue their own research endeavors. I

am Jeffs last doctoral student at Texas Tech University. After his official departure fi-om

the Texas Tech University Department of Psychology, Jeff remained steadfast in his

commitment to seeing my training to completion. For all these reasons, I wish to thank

Jeff, and dedicate this dissertation to him.

I would also like to express my eternal gratitude and thanks to Julie Treland,

Ph.D. for her role in my training and professional development. Julie was my first

clinical supervisor of and introduction to neuropsychological assessment. She helped

plant the seed that has blossomed into my career. Julie (Jeffs wife) never complained

when Jeff and I put in long hours working on my research, and was always willing to

help out in any way possible. She has been extremely supportive throughout my

professional development. Julie always provided excellent advice regarding my clinical

n

training in particular, and life in general. Her guidance and inspiration is extremely

appreciated. Jeff and Julie have consistently been there for me throughout my training,

even attending national meetings when I have presented our research. They have gone

the extra mile not only in their dedication to training, loyalty, and support, but also in

their hospitality and fiiendship. I look forward to continuing personal and professional

contacts with Jeff and JuUe throughout my lifetime. Jeff and Julie, I truly appreciate all

you have both done for me personally and professionally.

I would also like to express my appreciation to Stephanie Harter, Ph.D.. Dr.

Harter always put in extra time and effort to help assist with my training, even though I

was not a student in her laboratory. She was the second reader for my master's paper,

and put extra care and time in editing both my master's paper and Doctoral dissertation.

Dr. Harter's attention to detail and commitment to helping students put out top quality

work is unrivaled. She was an excellent instructor, supervisor, and resource throughout

my training at Texas Tech University. Dr. Harter has been an extraordinary advisor

regarding psychology in general, but also clinical neuropsychology in particular. She

advised me on many aspects of the internship application, interview, and selection

process. Dr. Harter was the primary reason I applied to and completed my internship

training at the University of Oklahoma Health Sciences Center-the best career move I

could have ever made in clinical neuropsychology. I would also like to express my

gratitude to Drs. Epkins and Fireman for all their support, guidance, and contributions to

my training. Their willingness to participate on my dissertation committee will always be

appreciated, thank you.

HI

This dissertation would not have been possible without the assistance of

Elizabeth Linder, Ed.D., and Roger Wolcott, M.D.. Through an external practicum

placement with Dr. Linder, I was able to collect the clinical data required for this

dissertation. Dr. Wolcott's support and assistance through this process was appreciated.

I would also like to thank Drs. Linder and Wolcott, as well as the other individuals in our

office who assisted with my dissertation.

I also wish to express my continued appreciation and many thanks to my family

for their support through this long process, but especially my wonderful, supportive, and

loving parents. They always helped make my dreams possible and have been unwavering

in their dedication to seeing me through my doctoral training. Finally, I would like to

thank Ann Ewing, Ph.D. who was my first advisor, instructor, and mentor in psychology.

Ann is the reason I became a psychology major, someone I have attempted to emulate

throughout my training and development as a professional, and a great fiiend.

IV

TABLE OF CONTENTS

ACKNOWLEDGEMENTS ii

ABSTRACT vii

LIST OF TABLES ix

LIST OF FIGURES xi

CHAPTER

L INTRODUCTION 1

Statement of the Problem 8

n. METHODOLOGY 11

Participants 11

Procedure 13

Measures 13

Wechsler Adult Intelligence Scale-Revised 13

Wechsler Memory Scale/Wechsler Memory Scale-Revised 14

Controlled Oral Word Association Test 15

Stroop 15

Clock Face Drawing Test 16

Trail Making Test 16

IIL RESULTS 18

Demographics 18

Glasgow Coma Scale and Posttraumatic Amnesia Analyses 23

Patterns of correlations between GCS/PTA and

cognitive outcome variables 24

Correlational patterns across domains 38

Comparison by group on cognitive outcome variables 39

Comparison by group: Pre/Post Intellectual Functioning 39

IV. DISCUSSION AND CONCLUSIONS 49

Summary and suggestions for future research 55

REFERENCES 59

APPENDICES

A. EXTENDED LITERATURE REVIEW 67

B. SEVERITY CLASSIFICATION SYSTEMS 107

VI

ABSTRACT

The Glasgow Coma Scale (GCS) and duration of posttraumatic amnesia (PTA) are

the most commonly used clinical instruments for determining severity and predicting

outcome in closed head injury (CHI). Many investigations have attempted to determine

the relationship between head injury severity and cognitive outcome, as measured by

neuropsychological assessment performance, but found mixed results with respect to

utility of prediction. The present study proposed that a major reason for these mixed

results was other investigations' inclusion of individuals with conditions known to

detrimentally affect neuropsychological test performance (i.e., history of alcoholism,

drug abuse, psychiatric history, or previous neurologic insult) into their closed head

injury sample. This investigation directly examined this hypothesis by adding an

additional comparison group, a closed head injury-clean group (CHI-C), and comparing

this groups' performance against a closed head injury not-clean group (CHI-NC). The

effects of combining the CHI-C and CHI-NC groups into one closed head injury

combined (CHI-Comb) group, as previous studies have done, was also examined.

Fifty-one consecutive patients (N = 51) underwent neuropsychological assessment

following brain injury. Participants were included if they had exited PTA, demonstrated

uncompromised upper extremity use, displayed adequate verbal communication, and

were judged capable of completing full Wechsler Adult Intelligence Scale-Revised

(WAIS-R), Wechsler Memory Scale, Controlled Oral Word Association Test, Stroop, and

Trail Making Tests.

vn

Separating the CHI individuals into the CHI-C and CHI-NC groups, and use of a

logarithmic transformation, resulted in better prediction fi-om severity measures to post-

injury cognitive performance than observed in previous investigations. Moreover,

despite the increased power of adding the CHI-C and CHI-NC groups into one CHI-

Comb group, this combination obscured findings as predicted. Duration of PTA was a

more reliable injury severity predictor than the GCS. Combining both measures in a

prediction equation did not improve prediction. This investigation did not find the unique

effects of closed head injury above that of other brain insults. Although the CHI-C and

CHI-NC groups did not significantly differ on PTA duration and GCS severity level, the

use of a pre-morbid PIQ estimate showed the CHI-NC group to have a greater post-injury

PIQ loss than the CHI-C group. Furthermore, the mean VIQ/PIQ discrepancy was

significantly larger for the CHI-NC group as compared to the CHI-C group. In

conclusion, these findings as well as others fi-om this investigation were discussed in

terms of cognitive reserve theory.

vni

LIST OF TABLES

3.1 Demographic Characteristics of Sample by Percentage 19

3.2 Correlations between GCS/PTA and demographic variables by group ... 20

3.3 Correlations between GCS/PTA and pre-morbid history variables 22

3.4 Significant correlations by group between GCS/PTA and cognitive outcome variables (CHI-C) 27

3.5 Significant correlations by group between transformed GCS/PTA and cognitive outcome variables (CHI-C) 29

3.6 Significant correlations by group between GCS/PTA and cognitive outcome variables (CHI-NC) 30

3.7 Significant correlations by group between transformed GCS/PTA and cognitive outcome variables (CHI-NC) 31

3.8 Significant correlations by group between GCS/PTA and cognitive outcome variables (CHI-Combined) 32

3.9 Significant correlations by group between transformed GCS/PTA and cognitive outcome variables (CHI-Combined) 33

3.10 Significant correlations by group between GCS/PTA and cognitive outcome variables (MN) 35

3.11 Significant correlations by group between transformed GCS/PTA and cognitive outcome variables (MN) 36

3.12 Means and Standard Deviations by group across WAIS-R subtests and indices 40

3.13 Means and Standard Deviations by group across WMS/WMS-R subtests and indices, as well as the Stroop and Trail Making Test 41

3.14 Means and Standard Deviations by group for the Clock Face Drawing Test and COWA 42

IX

3.15 ANOVA comparison of OPIE formulas by group 44

3.16 Mean Difference Between Obtained and Predicted IQ's Using the OPIE Best Formula Presented by Diagnostic Category and the Total Sample ... 45

3.17 Means and Standard Deviations by group for Predicted versus Obtained FSIQ, VIQ, and PIQ 48

LIST OF FIGURES

3.1 The relationship between PTA duration and FSIQ scores

for the CHI-C group 25

3.2 The relationship between the GCS and FSIQ for the CHI-C group 26

3.3 Mean WAIS-R subtest scores for the CHI-C and CHI-NC groups 46

XI

CHAPTER I

INTRODUCTION

The incidence of head injury in the United States has been estimated to be

between 500,000 to more than 1.9 million per year with 400,000 to 500,000 individuals

requiring hospitalization or dying fi-om their injuries (Lezak, 1995). Initial severity

classification of a closed head injury (CHI) became measurable and quantifiable by the

emergence of the Glasgow Coma Scale (GCS) (Diller, 1994). According to Teasdale and

Jennette (1974), the GCS is a short assessment that was designed to assess severity,

describe the posttraumatic states of altered consciousness, and predict outcome following

head injuries (see Appendix B). A patient's GCS total score is the sum of their scores for

eye-opening, verbal performance, and motor responsiveness. Although the GCS is

widely used and accepted, there is considerable disagreement on how well the GCS can

determine severity and predict outcome (Coppens, 1995; Garcia, Garrette, Stetz,

Emanuel, & Brandt 1990; Gupta & Ghai, 1996; Haut & Shutty, 1992; Katz, 1992;

McMillan, Jongen, and Greenwood 1996; Nelson & Adams, 1997; Reid & Kelly, 1993;

Richardson, 1990; Stambrook, Moore, Lubusko, Peters, & Blumenschein 1993; Vilkki et

al., 1994).

McMillan et al. (1996) reported that the duration of posttraumatic amnesia (PTA)

is the "best yardstick" for predicting outcome that exists. Duration of PTA has been

defined as the period from the time of injury until the individual can form new memories

and continuously recall day-to-day events (Oder et al., 1992). Posttraumatic amnesia

duration varies considerably by case, but once determined it is compared to severity

classification criteria (see Appendix B). Many investigators concur that duration of PTA

has greater utility for determination of head injury severity and prediction of outcome

(Katz, 1992; Gupta & Ghai, 1996; McClelland, 1988) but other investigators have

reported its limitations (Alfano, Neilson, & Fink, 1993; Bohnen & Jolles, 1992; Coppens,

1995; Gass &Apple, 1997; Levin et al., 1987; Lezak, 1995; Vilkki et al., 1994).

Haslam, Batchelor, Feamside, Haslam, Hawkins, and Kenway (1994) stated that

there is considerable variation in the literature regarding the ability of severity indices'

such as the GCS and PTA duration to predict cognitive outcome. Levin et al. (1990)

listed many impediments that have hindered progress in CHI research. Specifically,

some of these problems included inadequate analysis of cognitive performance in relation

to neurologic severity indices (e.g., the GCS and duration of PTA), non-uniform methods

of characterizing type of head injury and severity, failure to exclude pre-existing

neurological or psychiatric disorders, and wide variation in the measures employed to

assess outcome.

Bennett-Levy (1984), Brooks, Aughton, Bond, Jones, and Rizvi (1980), Haslam et

al. (1994), Haslam, Batchelor, Feamside, Haslam, and Hawkins (1995), Levin et al.

(1990), Stuss et al. (1999), and Vilkki, Poropudas, and Servo (1988) all commented on

the failure of the CHI research literature to consistently demonstrate the relationship

between measures of severity (e.g., the GCS and PTA duration) and cognitive outcome.

Early investigations using the Wechsler AduU Intelligence Scale (WAIS) (Mandleberg,

1976; Mandleberg & Brooks, 1975) found that the duration of PTA was not related to

VIQ performance after three months and PIQ performance after six months. Levin,

Grossman, Rose, and Teasdale (1979) used the WAIS and other measures and found that

the relationship between severity level, as measured by the GCS, and cognitive outcome

was "not invariable." Furthermore, they found that cognitive deficits on the WAIS were

evident for up to three years following CHI, contrary to the results of Mandleberg (1976),

and Mandleberg and Brooks (1975).

Brooks et al. (1980) employed the Raven's Progressive Matrices, Mill Hill

Vocabulary Scale, Logical Memory subtest. Story 1 of Form 1 from the Wechsler

Memory Scale (WMS), Inglis Paired Associate Learning test, Rey Picture, "three

measures of word fluency," Part 5 of De-Renzi and Vignolo Token Test, a copy condition

of the Rey Picture, and the WAIS Block Design subtest and found that severity level, as

measured by the GCS, was not related to performance on these measures. However,

Brooks et al. (1980) did find that severity level, determined by PTA duration, was related

to Associate learning (delayed condition) and Logical Memory Story 1 Form 1 of the

WMS, Rey Picture (immediate and delay conditions), and one word fluency condition out

of three.

Alexandre, Colombo, Nertempi, and Benedetti (1983) also employed a large

number of cognitive measures including the Wechsler-Bellevue Form 1, Benton's test of

visual memory, Rey Auditory Verbal Learning Test (RAVLT), Corsi's test for spatial

memory, Kimura's test for recognition memory, and the Token test and did not find any

significance between severity of injury, determined by length of PTA, and cognitive

outcome. Moreover, Alexandre et al. (1983) found that severity level, as measured by the

GCS, was not consistently reliable for predicting outcome, except when the outcome

measure was mortality.

Vilkki et al. (1988) looked at the relationship between Digit Span, Similarities,

and Block Design of the WAIS and coma duration, but found no relationship. Paniak,

Silver, Finlayson, and Tuff (1992) investigated the relationship of cognitive outcome, as

measured by the Wechsler Aduh Intelligence Scale-Revised (WAIS-R), to injury

severity as determined by PTA duration. They found that only one subtest. Digit Span,

was related to severity of injury. Moreover, Haut and Shutty (1992) investigated the

relationship between cognitive outcome, as measured by the WAIS-R, to severity level,

as measured by the GCS, but failed to find any significant relationships.

The relationship between cognitive outcome, as measured by memory and

attention/concentration measures, and head injury severity measures has also produced

varying results. Brooks (1974) looked at the relationship between recognition memory

and severity of injury, as determined by PTA duration. He found that duration of PTA

was related to total amount correct and number of false positives produced, but not the

number of false negatives produced. To extend his earlier work. Brooks (1976)

investigated the relationship between memory performance on the WMS and injury

severity, as determined by PTA duration. Of all the subtests and conditions. Brooks

(1976) found that PTA duration was only related to Logical Memory and Associate

Learning performance.

Gronwall and Wrightson (1981) completed a two-study investigation to determine

the relationship between attention/concentration, memory, and injury severity as

determined by the duration of PTA. In the first study, attention/concentration abilities

were measured by the Paced Auditory Serial Addition Task (PASAT), while the WMS

was used to assess memory performance. Overall, these investigators found moderate

correlations between severity level and performance on the PASAT, and the Mental

Control and Associate Learning subtests of the WMS. In the second study, they again

used the PASAT, but to assess memory they used the Selective Reminding Task and the

Visual Sequential Memory subtest of the Illinois Test of Psycholinguistic Abilities.

Severity level was related to PASAT performance and total amount correct on the

Selective Reminding Task, but not to the Visual Sequential Memory test.

In Bennett-Levy's (1984) investigation, a large number of memory assessment

devices were compared to severity level, as determined by PTA duration. Overall, the

only significant finding for assessment performance by severity level was on the Logical

Memory subtest of the WMS (delayed recall & percent forgetting conditions), the

Keeping Track Test, and the Rey-Osterrieth (recall condition).

Vilkki et al. (1988) looked at the relationship between memory performance on

free and cued recall conditions of the WAIS Similarities subtest and the Benton Visual

Recognition Test (BVRT) with homogeneous interference, and coma duration as

determined by the GCS. They only found significant difference by severity level on the

free and cued recall conditions of the BVRT and percentage forgetting of the BVRT.

Crosson, Novak, Trenerry, and Craig (1988) investigated performance on the

California Verbal Learning Test (CVLT) in relation to severity level, as determined by

coma duration and length of PTA, but found no relationship between any of the CVLT

trials and severity indices. Haut and Shutty (1992) also examined CVLT performance to

severity of injury, as determined by the GCS, but also failed to find any significant

relationships. Geffen, Butterworth, Forrester, and Geffen (1994) examined memory

functioning, as measured by the RAVLT, and severity of injury determined by PTA

duration. Overall, the RAVLT includes seven trials and severity of injury was related

only to performance on the seventh trial, remote verbal memory.

Reid and Kelly (1993) investigated the relationship between severity of injury, as

measured by both PTA duration and the GCS, to memory performance on the Wechsler

Memory Scale-Revised (WMS-R). They failed to find a relationship between severity of

injury and memory performance across any of the subtests. Gupta and Ghai (1996)

examined memory performance with three WMS subtests (Digit Span, Associate

Learning, and Visual Reproduction) and attention/concentration with the PASAT.

Specifically, these investigators compared the relationship between performance on these

measures to injury severity, as measured by PTA duration. Overall, these authors failed

to find significant differences by group except on the Visual Reproduction subtest.

Haslam et al. (1994) and Haslam et al. (1995) investigated memory performance

with the RAVLT and attention/concentration with the PASAT and Symbol Digit

Modalities Test (SDMT), and examined their relationship to injury severity. Their

measures of severity were numerous, but the ones important here were the GCS and PTA

duration. These authors performed a principal component factor analysis on the

cognitive measures and a square root transformation on PTA duration in both

investigations. Haslam et al. (1994) found that PTA duration, and transformed PTA

duration were significant predictors of recent memory performance, while transformed

PTA duration was a significant predictor of information processing speed up to twelve

months post-injury. Haslam et al. (1995) determined that PTA duration, transformed

PTA, and coma duration were significantly related to recent memory performance, while

speed of information processing was significantly related to transformed PTA duration,

PTA duration, and coma duration up to two-years post-injury.

Loken, Thorton, Otto, and Long (1995) assessed severe CHI patients' abihty to

sustain attention on a visual continuous performance task "of sufficient ease and duration

to assess stability of performance over time," (p. 593) and compared these performances

to severity of injury as measured by duration of PTA. According to Loken et al. (1995),

correlational analyses failed to reveal a significant relationship between performance on

this task and injury severity.

The most striking finding of this body of literature was the lack of consistency

between investigations (Bennett-Levy, 1984; Brooks et al., 1980; Haslam et al, 1994;

Haslam et al., 1995; Levin et al, 1990; Loken et al., 1995; Vilkki et al., 1988). Duration

of PTA and the GCS have both been related to measures of cognitive outcome in some

investigations, but not in others. Most of the positive findings in these reviewed

investigations were minute considering they often gave numerous measures (e.g., only

WMS Logical Memory Story 1, not Story 2), and the positive findings varied by

investigation.

The second most striking finding from this body of literature was the lack of

refinement in participant selection. Only one study (Haslam et al, 1995) excluded

participants with a history of previous head trauma, alcoholism, drug abuse, neurological

condition, and psychiatric problems. Interestingly these investigators found a significant

relationship between severity level and cognitive outcome on measures of memory and

attention/concentration up to two-years after injury.

Third, Levin et al (1990) make an excellent point that CHI research on severity

and outcome is also hindered by the wide variation of measures employed to assess

outcome from one study to the next. However, with the exception of Haslam et al. (1994,

1995), many of the more recent investigations using newer, well-known, commonly used,

and accepted assessment measures (e.g., Crosson et al, 1988; Geffen et al, 1994; Haut &

Shutty, 1992; Paniak et al, 1992; Reid & Kelly, 1993) have failed to demonstrate

consistent and significant findings.

Statement of the Problem

Considering the issues that were raised in the review of the CHI literature, it

would seem that some of the inconsistency in predicting outcome from severity indices

could be due to the inclusion of individuals in study groups without regard to their

previous history of head trauma, neurologic problem, alcohol or drug abuse, or

psychiatric involvement. Therefore, the primary purpose of the present study was to use

CHI groups that were both "clean" and "not clean" with respect to prior history. Both

PTA duration and the GCS were used to determine initial injury severity level so that a

comparison could be made as to which is the better index of severity and predictor of

cognitive outcome. Furthermore, they were examined in combination to determine if

8

both could contribute significant individual variance to the prediction of cognitive

outcome. As a general hypothesis it was expected that the refinement of closed head

injury groups into "clean" and "not clean" groups would provide better prediction of

cognitive outcome from PTA duration and the GCS severity indices than would

combining them into one group, which is the typical procedure in most closed head injury

research. Data from neurologically impaired populations frequently presents

distributional problems for analyses involving linear relationships. Therefore in addition

to refining the study groups, the distributional aspects of data were closely observed for

their effect on linear analyses.

A "mixed neurologic" comparison group was also included in this investigation,

based upon the advice of Satz et al. (1999), who proposed that simply contrasting groups

of healthy normals to closed head injury groups adds nothing to our understanding of the

unique effects of the head injury, to help determine what persistent cognitive deficits

were unique to the closed head injury and not other factors such as pre-morbid history

(e.g., alcoholism) or general brain insuh (e.g., cerebral vascular accident, aneurysm,

neoplasm, and so on). Specifically, a range of tests (these measures are discussed in

detail in the procedures) were used to examine the degree to which refining the study

groups would illustrate differences across measures. The WAIS-R subtests, indices (i.e..

Freedom from Distractibility, Perceptual Organizational, and Verbal Comprehension),

and summary scores (i.e.. Verbal IQ, Performance IQ, and Full Scale IQ) were compared

to assess general intellectual/cognitive functioning by group. The Wechsler Memory

Scale/Wechsler Memory Scale-Revised subtests and indices (i.e.. Verbal Memory,

Visual Memory, General Memory, Attention-Concentration, and Delayed Memory) were

examined for group differences on attention and memory functioning. The Stroop test.

Trail Making Test, and Clock Face Drawing Test were included for their role in

measuring attention/concentration/executive functioning differences across groups.

Finally, the Controlled Oral Word Association Test was used to examine how verbal

fluency may differ by group.

10

CHAPTER n

METHODOLOGY

Participants

Fifty-one consecutive patients (N = 51), twenty-six males and twenty-five

females, were referred for a neuropsychological assessment from various hospitals (i.e.,

Methodist Hospital, St. Mary's Hospital, and University Medical Center) in the Lubbock,

Texas area following brain insult. Participants were included if they had uncompromised

use of their upper extremities as determined during clinical interview (i.e., a brief motor

screen was completed), were able to verbally communicate, and judged capable of

completing a fiill length WAIS-R, WMS-R, Controlled Oral Word Association Test

(COWA), Stroop, and Trail Making Tests (TMT-A; TMT-B). All participants were

tested after their transition to inpatient rehabilitation or as soon as possible after their

discharge from the hospital.

Participants in the CHI-clean group (CHI-C; n = 17) were excluded for a history

of alcoholism, drug abuse, head trauma, psychiatric problems, or neurologic disorder.

Moreover, participants' charts were examined (e.g., history and physical, lab results,

physician notes) to ensure that there was no confounding of assessment findings due to

complications resuhing from alcohol or drug use at the time of the participants' head

injury. The CHI-C group had a mean age of 42.94 years (SD = 23.57) range 18-86 with

an average of 12.06 years of education (SD = 3.01) range 5-16. Participants in the CHI-

not clean group (CHI-NC; n = 9) were not excluded for these criteria. The CHI-NC

11

group's mean age was 62.11 (SD = 25.03) range 19-90 with 12.56 years of education

(SD = 3.47) range 7-16. The Mixed Neurologic group (MN; n = 25) included individuals

referred following cerebral vascular accident, subdural hematoma, aneurysm, anoxia, and

neoplasm. Participants in the MN group were not excluded for a history of alcoholism,

drug abuse, neurologic condition, or psychiatric history. The mean age of the MN group

was 63.48 years (SD = 15.89) range 25-84, and education level 12.14 years (SD = 3.43)

range 6-18.

All participants were judged to have cleared PTA before assessment.

Furthermore, a check to ensure that the patients PTA had remitted and that the

participants were suitable for and able to withstand a neuropsychological evaluation was

completed by the author. Specifically, participants' GCS score was obtained from

hospital records when available and compared to their PTA duration severity

classification as determined by the author. Duration of PTA was determined through the

course of clinical interview questioning (i.e., last memory prior to incident, first memory

following the incident, and assessment for retum of continuous day-to-day memory

return), an accepted, reliable, and valid method for assessing PTA (McMillan et al,

1996). Participants were judged to have exited PTA from the results of their mental

status examination, memory surrounding the incident, combined with assessment to

ensure continuous memory for day-to-day events had returned.

12

Procedure

Participants' severity level, according to GCS was retrieved from hospital records

for all but three MN patients who, because of the nature of their injuries, did not have a

GCS score recorded in their hospital records. Severity level was also determined

according to duration of PTA using the guidelines proposed by McMillan and colleagues

(1996). As with the GCS scores, two MN patients' duration of PTA was not

determinable because of medical conditions that confounded an accurate estimate of PTA

duration. All participants' assessment data was retrieved from archived records.

Following data extraction, premorbid intelligence was estimated using the standardized

procedures for obtaining the Best Full Scale IQ estimate with the Oklahoma Premorbid

Intelligence Estimation (OPIE), a standardized, validated, reliable linear prediction

algorithm developed from the WAIS-R standardization sample (Spreen & Strauss, 1998;

Tremont, Hoffinan, Scott, Adams, & Naldone, 1997). The OPIE takes into account

demographic variables of age, education, occupation, and race, in addition to

performance on the WAIS-R Picture Completion and Vocabulary subtests (Krull, Scott,

& Sherer, 1995; Spreen & Strauss, 1998).

Measures

Wechsler Adult Intelligence Scale-Revised. The WAIS-R is considered the "gold

standard" in intelligence testing and one of the most frequently used measures in

neuropsychological batteries (Lees-Haley, Smith, Williams, & Dunn 1996; Scott, Sherer,

& Adams, 1995). The WAIS-R is comprised of eleven subtests that yield a Full Scale IQ

13

(FSIQ) score. Of these eleven subtests, six are verbal subtests while five are performance

subtests. The verbal subtests combine to form a Verbal IQ (VIQ) score and the

performance subtests combine to form a Performance IQ (PIQ) score. With the

exceptions of the WAIS-R Object Assembly (r = .68) and Picture Arrangement (r = .74),

it has extremely high overall split-half reliability (ranging from r = .81 to .97). Finally,

test-retest reliability ranged from r = .69 to .95 for 25-34 year-olds and from r = .67 to .97

for 45-54 year olds depending upon the subtest (Wechsler, 1981).

Wechsler Memory Scale/Wechsler Memorv Scale-Revised. The WMS-R is a

widely used and accepted individually administered memory assessment device. The

WMS-R contains nine subtests and was designed as a diagnostic instrument for use

during general neuropsychological examinations or other clinical examinations where

memory performance is at question (Wechsler, 1987). Assessment of immediate and

delayed verbal and visual memory, as well as assessment of auditory memory for

numerical information, cognitive speed and flexibility, and orientation information is

accomplished with the WMS-R. Average intemal consistency ranges from r = .41

(Verbal Paired Associates II) to r = .88 (Digit Span) across the subtests. Interrater

reliability for Logical Memory and Visual Reproduction subtests were r = .99 and r = .97

respectively (Wechsler, 1987). Some subjects in this investigation received the WMS

(Wechsler, 1945) instead of the WMS-R. The WMS has also been a widely used an

accepted memory measurement device with a large body of literature supporting its use

across a variety of clinical populations (Wechsler, 1987). Differences between the two

versions include the addition of the Figural Memory, Visual Paired Associates, and

14

Visual Memory Span subtests in the newer WMS-R. Moreover, in comparison, minor

changes (e.g., easier items added to Visual Reproduction, Information and Orientation

subtests lengthened, et cetera) were made across some subtests from the WMS to the

WMS-R. It should be noted that Memory Quotients were not used in this investigation.

Finally, this study standardized the two versions by converting all scores to standard z-

scores so that the different versions could be compared on the same metric.

Controlled Oral Word Association Test. The COWA (a.k.a. FAS) is a measure of

phonemic fluency that measures an individual's ability to spontaneously produce word

beginning with a specific letter (i.e., F, A, and S) under time constraints. Specifically,

individuals are given one-minute to produce as many words as possible that begin with

the letters F, A, and S. Testing rules include that individuals may not provide words

which are proper names (e.g., Phyllis, Phoenix, et cetera) or words that have the same

beginning but different endings (e.g., eat and eating). Provided words are examined to

help provide insight into a search strategy and for errors (e.g., perseverations, intrusions,

and paraphasias) to elucidate evidence towards the diagnosis of certain types of disorders.

Inter-rater reliability estimates range from r = .70 to .88 (Spreen & Strauss, 1998) for this

measure.

Stroop. The Stroop is a measure of selective attention and cognitive flexibility

that also taps into executive fiinctioning (Spreen & Strauss, 1998). The Stroop has three

parts that require the individual to name the color of X's (Stroop Color), name the color

of non-color words (e.g., "when," "hard," and "over" printed in blue, red, yellow, or

green; Stroop Word), and name the color of color words (e.g., the words "blue," "red,"

15

"yellow," or "green" printed in a color other then the one the word describes; Stroop

Color-Word). Time to completion and number of errors are recorded for each of these

parts. The Stroop is based on the pretense that it will take longer to name the color of

color words, due to interference, than it takes to name the color of X's. Spreen and

Strauss (1998) reported test-retest reHability estimates of r = .90, .83, and .91 for Stroop

Color, Stroop Word, and Stroop Color-Word, respectively.

Clock Face Drawing Test. Clock face drawing tests require visuospatial,

graphomotor, and executive functioning abilities for successful completion. However,

Royal, Cordes, and Polk (1998) have devised an executive clock drawing task, the

CLOX, with a scoring system that reflects the specific contribution of executive

functioning over visuospatial praxis abilities. The CLOX scoring system has two parts, a

free-drawing condition (CLOXl), and a copy condition where the examiner provides a

model (CL0X2). CLOXl inherently requires more executive functioning abilities, and

the subtraction of CLOXl from CL0X2 reflects the specific contribution of executive

control (Royall et al , 1998). The CLOX scoring system has high intemal consistency

(Cronbach alpha = .82). Interrater reliability on the two conditions CLOXl and CL0X2

are r = .94 and r = .93 respectively (Royall et al, 1998).

Trail Making Test TTMT-A; TMT-B). The Trail Making Test taps into

information processing speed, knowledge of numerical and alphabetical sequencing,

visual search/scan abilities, motor functioning, and mental flexibility. Part A (TMT-A)

requires the participant to connect by pencil, in order, twenty-five encircled numbers that

are randomly distributed on a page. Part B (TMT-B) requires the participant to correctly

16

connect by pencil, in correct sequence, twenty-five encircled numbers and letters in

alternating order. Time to completion and number of errors committed is determined.

Comparison of a participants' TMT-A versus TMT-B competed protocols yields

information regarding their mental flexibility. Therefore, this comparison is considered

to provide insight into an individual's executive functioning abilities because TMT-A and

B tap into the same domains minus the ability to alternate between numbers and letters.

Interrater reliability varies by study, but Spreen and Strauss (1998) reported it as .94 for

TMT-A and .90 for TMT-B.

17

CHAPTER III

RESULTS

Demographics

Demographic characteristics by sample are presented in Table 3.1. A one-way

three-group analysis of variance (ANOVA) revealed no group differences for education

[F (2, 48) = .078, E = .925], or time since injury [F (2, 45)= 2.03 e = -143], but a

significant age effect [F (2, 48) = 5.56, p = .007] was observed between the CHI-C, CHI-

NC, and MN groups. Additional one-way ANOVAs revealed that age approached

significance between the CHI-C and CHI-NC groups [F (1,24) = 3.73, p = .065], was

significantly different between the CHI-C and MN groups [F (1, 40) = 11.42, p = .002],

and was not statistically different between the CHI-NC and MN groups [F (1, 32) = .04, p

= .851]. Mean time since injury until assessment was 167 days (SD = 308) for the CHI-C

group, 76 days (SD = 109) for the CHI-NC group, and 44 days (SD = 46) for the MN

group. The relations between GCS/PTA and the demographic variables of age,

education, and occupation were examined and no significant correlations were observed

for the CHI-C and CHI-NC groups (see Table 3.2). However, age was significantly

related to the GCS for the MN group while education and occupation were not.

Correlations between PTA and age, education, and occupation were not statistically

significant for the MN group. However, sample size may have reduced the ability to

identify relationships.

18

Table 3.1: Demographic Characteristics of Sample by Percentage.

Gender Males Females

Race Caucasian African-American Latin-American Other

Marital Status Married Single Divorced Widow/er

Occupation Professional/Technical Managers/Administrator/

ClericaVSales Craftsman/Foreman Operators/Service/

Domestic/Farmers Laborers Unemployed

Retired Yes No

CHI-C

17.6(52.9) 15.7(47.1)

15.7(47.0) 3.9(11.8) 9.8 (29.4) 3.9(11.8)

17.6(52.9) 11.8(35.3) 0.0 (00.0) 3.9(11.8)

5.9(17.6)

3.9(11.8) 2.0 (05.9)

11.8(35.3) 7.8 (25.3) 2.0 (05.9)

5.9(17.6) 27.5 (82.4)

CHI-NC

9.8 (56.6) 7.8 (44.4)

13.7(77.8) 0.0 (00.0) 3.9 (22.2) 0.0 (00.0)

11.8(66.7) 2.0(11.1) 0.0 (00.0) 3.9 (22.2)

5.9(33.3)

3.9 (22.2) 0.0 (00.0)

2.0(11.1) 5.9 (33.3) 0.0 (00.0)

11.8(66.7) 5.9(33.3)

MN

23.5 (48.0) 25.5 (52.0)

37.3 (76.0) 0.0 (00.0)

11.8(24.0) 0.0 (00.0)

29.4 (60.0) 2.0 (04.0) 2.0 (04.0)

15.7(32.0)

7.8(16.0)

9.8 (20.0) 2.0 (04.0)

19.6(40.0) 9.8 (20.0) 0.0 (00.0)

27.5 (56.0) 21.6(44.0)

Note: CHI-C = Closed Head Injury-Clean; CHI-NC = Closed Head Injury-Not Clean; MN = Mixed Neurologic. Scores in parentheses represent percentages within group.

19

Table 3.2: Correlations between GCS/PTA and demographic variables by group.

CHI-C (n=17)

Age Education Occupation Time since injury

CHI-NC (n = 9)

Age Education Occupation Time since injury

MN (n = 25)

Age Education Occupation Time since injury

GCS

.398

.127 -.264 -.024

.538 -.165 -.280 .099

.496 -.107 .043

-.094

p-value

.114

.628

.307

.928

.135

.671

.466

.801

.019*

.636

.848

.677

PTA

-.334 -.371 .433 .381

-.446 -.382 .612

-.049

-.119 -.403 -.079 .095

p-value

.190

.142

.083

.131

.228

.311

.080

.901

.589

.057

.721

.684

Note.- CHI-C = Closed Head Injury-Clean; CHI-NC = Closed Head Injury-Not Clean; MN = Mixed Neurologic. •Denotes statistically significant correlations.

20

No correlations were calculated between GCS/PTA and history of alcoholism,

drug abuse, neurologic condition, and/or psychiatric history for the CHI-C group because,

by definition, this group was excluded for these conditions. However, as can be seen in

Table 3.3, within the CHI-NC group GCS performance was significantly correlated with

a history of drug abuse (r = -.881, p = .001) and neurologic condition (r = .621, p = .037),

but not alcoholism (r = -.438, p = .094) or psychiatric condition (r = .027, p = .473).

Duration of PTA for the CHI-NC group was significantly correlated with history of drug

abuse (r = .807, p = .004), but was not significantly correlated with alcoholism (r = .395,

P = .114), neurologic condition (r = -.485, p = .093), or psychiatric history (r = -.381, p =

.156). However, small sample size in the CHI-NC group may have obscured the

relationships between the GCS/PTA and alcoholism, neurologic condition, and

psychiatric history for this group.

As seen in Table 3.3, history of alcoholism, drug abuse, neurologic condition, and

psychiatric problems were not significantly related to GCS or PTA for the MN group

despite the greater power available with the increased sample size of the MN group. It

should be noted that if the MN group (n = 25) was broken down into "clean" and "not

clean" subgroups it would have 12 "clean" and 13 "not clean" participants. Removal of

the MN "clean" participants and re-computation of the correlations between pre-morbid

history variables (e.g., history of alcoholism) and GCS/PTA also did not yield any

significant relationships. Overall, the significant correlations between drug abuse and

GCS/PTA plus the magnitude of correlations between alcoholism, neurologic condition

and psychiatric condition and GCS/PTA for the CHI-NC group suggest that the

21

Table 3.3: Correlations between GCS/PTA and pre-morbid history variables.

GCS

CHI-NC (n = 9)

History of: Alcoholism -.438 Drug Abuse -.881 Neurologic Condition .621 Psychiatric Condition .027

MN (n = 25)

History of: Alcoholism -.305 Drug Abuse* Neurologic Condition .139 Psychiatric Condition .237

p-value

.094

.001

.037

.473

.084

.269

.144

PTA

.395

.807 -.485 -.381

-.288

-.092 -.195

p-value

.114

.004

.093

.156

.091

.338

.187

Note: CHI-NC = Closed Head Injury-Not Clean; MN = Mixed Neurologic •There was no reported drug abuse for any of the mixed neurologic participants.

22

segregation of the closed head injury group for history of drug abuse, alcoholism,

neurologic problem, and psychiatric condition was a valid procedure.

Glasgow Coma Scale and Posttraumatic Amnesia Analyses

A one-way analysis of covariance (ANCOVA) for all three groups, witii age as

the covariate, found no statistically significant differences between groups for the GCS [F

(2, 43) = 1.21, p = .305] or PTA [F (2,44) = .668, p = .518] variables. Three MN

participants were excluded from the GCS comparisons between group because of missing

GCS scores, whereas two MN participants were excluded from the PTA comparisons

because their PTA duration was indeterminable. The CHI-C group had a mean GCS

score of 9.41 (SD = 5.34) and mean PTA duration of 365.82 hours (SD = 506.22). The

CHI-NC group mean GCS score was 13.56 (SD = 2.79) and 105.67 hours (SD = 185.09)

mean PTA duration. The MN group had a mean GCS score of 12.91 (SD = 3.29) and

mean PTA duration of 302.35 hours (SD = 426.88). According to the GCS and PTA

severity classification systems from Lezak (1995; Appendix B), the GCS severity level

by group was as follows: the CHI-C group had 7 mild, 2 moderate, and 8 severe

participants; the CHI-NC group had 7 mild, 1 moderate, and 1 severe participant; and the

MN group had 17 mild, 1 moderate, and 4 severe participants. The severity levels by

group according to the PTA severity classification system (Lezak, 1995; Appendix B)

were as follows: the CHI-C group had 5 mild, 2 moderate, and 10 severe participants; the

CHI-NC group had 3 mild, 3 moderate, and 3 severe participants; and the MN had 8

mild, 2 moderate, and 13 severe participants. Thus, the CHI-C group had more severely

23

impaired individuals according to the GCS and duration of PTA severity classification

systems, although this was not reflected as a significant difference between mean scores.

Patterns of correlations between GCS/PTA and cognitive outcome variables

Inspection of the pattern of correlations between the GCS/PTA severity measures

and the cognitive outcome variables is instructive with respect to the following issues.

First, the magnitude of correlations can be compared between the log transformed and

non-log transformed GCS and duration of PTA and cognitive outcome variables.

Second, the magnitude and significance of correlations can be compared between the

CHI-C, CHI-NC, and Closed Head Injury-Combined (CHI-Comb) groups.

Inspection of the distributions of the GCS and the PTA variables found that both

scales had a skewed distribution of scores due to the number of individuals with a score

of zero for PTA duration and GCS score of fifteen. Furthermore, as a group, individuals

with very mild head injury scores showed a wide range of performance on the cognitive

measures compared to a more circumscribed performance observed when scores were

above zero on the PTA and below fifteen on the GCS (see Figures 3.1 & 3.2).

Consequently, the PTA and GCS variables were transformed to logarithms (Log x lo + 1

for PTA and Log x lo for GCS) for purpose of computing linear correlation (Howell,

1992).

Table 3.4 illustrates that, prior to logarithmic transformations, the CHI-C group

had a total of 14 significant correlations between GCS as well as 14 significant

correlations between PTA and the 48 cognitive outcome variables examined. Following

24

CO

Cm-C: RELATIONSHIP BETWEEN

PTA DURATION AND FSIQ

0 200 400 600 800 1000 1200 1400 1600

Time since injury in hours.

Figure 3.1 The relationship between PTA duration and FSIQ scores for the CHI-C group.

25

c/ I—I CO

CHI-C: RELATIONSHIP BETWEEN

THE GCS AND FSIQ

5 6 8 9 10 11 12 13 14 15 16

GLASGOW COMA SCALE SCORE

Figure 3.2 The relationship between the GCS and FSIQ for the CHI-C group.

26

Table 3.4: Significant correlations by group between GCS/PTA and cognitive outcome variables (CHI-C).

GCS p-value PTA p-value

CHI-C (n= 17)

WAIS-R Arithmetic Scaled Block Design Comprehension Comprehension Scaled Digit Symbol Scaled Similarities Scaled Vocabulary Vocabulary Scaled

FDI VCI PIQ VIQRAW VIQ FSIQ

COWA S

Stroop Color in seconds

.424

.458

.497

.577

.424

.456

.569

.606

.397

.557

.387

.422

.525

.481

.415

-.567

.045

.037

.021

.008

.045

.033

.005

.005

.057*

.010

.069*

.046

.015

.030

.049

.014

,475 ,468 ,489 ,532 ,545 ,451 ,564 579

540 ,528 ,471 444 ,516 540

.027

.027

.023

.014

.012

.035

.018

.007

.015

.015

.033

.037

.017

.015

.277 .14r

.140 .309^

NUMBER OF SIGNIFICANT CORRELATIONS FOR CHI-C GROUP BY: GCS = 14/48 (29%) PTA = 14/48 (29%) GRAND TOTAL = 28/96 (29%)

Note: CHI-C = Closed Head Injury-Clean. The total number of cognitive outcome variables examined from the WAIS-R, WMS/WMS-R, Stroop, Trail Making Test, and COWA was 48. •These correlations did not reach statistical significance, but were included for comparison to the other severity indicator (i.e., GCS or PTA).

27

the logarithmic transformation, the number of significant correlations changed to a total

of 7 significant correlations between the GCS and the 48 cognitive variables and 23

significant correlations for PTA with these same cognitive outcome variables (see Table

3.5). The average magnitude of the significant correlations, post-transformation, between

the GCS and PTA were r = .35 and r = .54 respectively for the CHI-C group. Before

logarithmic transformation, the CHI-NC group had a total of 6 significant correlations

between GCS and 2 significant correlations between PTA and the 48 cognitive outcome

variables (see Table 3.6). Logarithmic transformation resulted in 9 significant

correlations between GCS and the 48 cognitive outcome variables, and 1 for PTA and

these same variables (see Table 3.7). The average magnitude of the significant

correlations, post-transformation, between the GCS and PTA were r = .65 and r = .36

respectively for the CHI-NC group.

When the CHI-C and CHI-NC groups were combined to form the CHI-Comb

group this resulted in only 2 significant correlations between the GCS and the 48

cognitive outcome variables and 8 for PTA and these same variables pre-logarithmic

transformation (see Table 3.8). Following logarithmic transformation, 2 significant

correlations were found between GCS and the 48 cognitive outcome variables, and 16

between PTA and these same variables (see Table 3.9). Furthermore, despite the

increased sample size, the magnitude of the correlations for the CHI-Comb group were

equal to or smaller than those for the equivalent variables in the CHI-C group. Overall,

the average magnitude of the significant correlations, post-transformation, between the

GCS and PTA were r = .29 and r = .44 respectively for the CHI-Combined group.

28

Table 3.5: Significant correlations by group between transformed GCS/PTA and cognitive outcome variables (CHI-C).

GCSLOG

CHI-C (n= 17)

WAIS-R Arithmetic .253 Arithmetic Scaled .362 Block Design .098 Comprehension .416 Comprehension Scaled .493 Digit Symbol Scaled .378 Information .197 Information Scaled .255 Picture Arrangement Scaled .328 Similarities Scaled .399 Vocabulary .511 Vocabulary Scaled .541

FDI .329 POI .273 VCI .469 PIQRAW .031 PIQ .328 VIQRAW .340 VIQ .436 FSIQRAW .178 FSIQ .394

Stroop Color in seconds -.548 Color-Word in seconds -.313

p-value

.163*

.076*

.359*

.048

.022

.067*

.225*

.162*

.107*

.056*

.018

.012

.099*

.153*

.029

.455*

.107*

.091*

.040

.254*

.066*

.017

.138*

PTALOG p-value

-.485 -.604 -.549 -.605 -.633 -.591 -.496 .501 .508 .516 .610 .638

.495

.416

.634

.279

.542

.561

.612

.454

.615

540 509

.025

.005

.014

.005

.003

.006

.022

.020

.023

.017

.005

.003

.022

.054

.003

.148

.015

.010

.005

.039

.006

.019

.032

NUMBER OF SIGNIFICANT CORRELATIONS FOR CHI-C GROUP AFTER TRANSFORMATION BY: GCS = 7/48 (15%) PTA = 23/48 (48%) GRAND TOTAL = 30/96 (31%)

Note: CHI-C = Closed Head Injury-Clean. The total number of cognitive outcome variables examined from the WAIS-R, WMS/WMS-R, Stroop, Trail Making Test, and COWA was 48. *These correlations did not reach statistical significance, but were included for comparison to the other severity indicator (i.e., GCS or PTA).

29

Table 3.6: Significant correlations by group between GCS/PTA and cognitive outcome variables (CHI-NC).

GCS p-value PTA p-value

CHI-NC (n = 9)

WAIS-R Digit Symbol Scaled Similarities Scaled

WMS-R Logical Memory I Logical Memory II

COWA F A S Total Score

-.670 -.595

-.642 -.665

-.592 -.713 -.684 -.737

.035

.046

.060*

.051*

.047

.049

.021

.012

082 167

748 855

248 208 178 235

.423*

.333*

.026

.007

.260*

.141*

.323*

.271*

NUMBER OF SIGNIFICANT CORRELATIONS FOR CHI-NC GROUP BY: GCS = 6/48 (15%) PTA = 2/48 (4%) GRAND TOTAL = 8/96 (8%)

Note: CHI-NC = Closed Head Injury-Not Clean. The total number of cognitive outcome variables examined from the WAIS-R, WMS/WMS-R, Stroop, Trail Making Test, and COWA was 48. *These correlations did not reach statistical significance, but were included for comparison to the other severity indicator (i.e., GCS or PTA).

30

Table 3.7: Significant correlations by group between transformed GCS/PTA and cognitive outcome variables (CHI-NC).

GCSLOG p-value PTALOG p-value

CHI-NC (n = 9)

WAIS-R Digit Symbol Scaled Similarities Scaled

WMS-R Logical Memory I Logical Memory II Mental Control

-.641 -.604

-.677 -.715 -.632

.043

.042

.048

.035

.047

059 .267

508 606 ,275

.445*

.243*

.123*

.075*

.255*

Verbal Memory Index -.797 .053* .410 .247'

-.750 .043 Stroop

Color in

COWA F A S

seconds

Total Score

.425

-.596 -.713 -.666 -.730

.20 r

.045

.016

.025

.013

283 001 090 156

.230*

.999*

.410*

.345*

NUMBER OF SIGNIFICANT CORRELATIONS FOR CHI-NC GROUP AFTER TRANSFORMATION BY: GCS = 9/48 (19%) PTA = 1 /48 (2%) GRAND TOTAL = 10/96 (10%)

Note: CHI-NC = Closed Head Injury-Not Clean. The total number of cognitive outcome variables examined from the WAIS-R, WMS/WMS-R, Stroop, Trail Making Test, and COWA was 48. *These correlations did not reach statistical significance, but were included for comparison to the other severity indicator (i.e., GCS or PTA).

31

Table 3.8: Significant correlations by group between GCS/PTA and cognitive outcome variables (CHI-Combined).

GCS p-value PTA p-value

CHI-Combined (n = 26)

WAIS-R Arithmetic Scaled Comprehension Scaled Vocabulary Vocabulary Scaled

FDI VCI VIQ FSIQ

.312

.317

.322

.329

.274

.295

.331

.235

.061*

.057*

.054*

.050

.088*

.071*

.049

.141*

,383 ,334 417 407

331 ,373 401 394

.027

.048

.017

.020

.049

.030

.021

.031

NUMBER OF SIGNIFICANT CORRELATIONS FOR THE CHI-COMBINED GROUP BY: GCS = 2/48 (4%) PTA = 8/48 (17%) GRAND TOTAL = 10/96 (10%)

Note: The total number of cognitive outcome variables examined from the WAIS-R, WMS/WMS-R, Stroop, Trail Making Test, and COWA was 48. *These correlations did not reach statistical significance, but were included for comparison to the other severity indicator (i.e., GCS or PTA).

32

Table 3.9: Significant correlations by group between transformed GCS/PTA and cognitive outcome variables (CHI-Combined).

GCSLOG p-value PTALOG p-value

CHI-Combined (n = 26)

WAIS-R Arithmetic Scaled Comprehension Comprehension Scaled Digit Symbol Scaled Information Information Scaled Picture Arrangement Scaled Similarities Scaled Vocabulary Vocabulary Scaled

FDI VCI PIQ VIQRAW VIQ FSIQ

WMS-R Information Visual Paired Associates II

Stroop Word in seconds

.278

.196

.291

.094

.139

.208

.131

.136

.310

.320

.238

.271

.068

.210

.296

.202

.181 -.420

-.347

.084*

.168*

.075*

.327*

.250*

.154*

.272*

.254*

.061*

.056*

.120*

.090*

.379*

.152*

.071*

.178*

.194*

.041

.049

,450 .386 .420 .353 .522 .516 .365 .389 .548 .545

.383

.523

.374

.447

.522

.509

.330 322

.011

.026

.017

.042

.003

.004

.040

.023

.002

.002

.027

.003

.039

.011

.003

.066*

.054*

.097*

.459 .024

NUMBER OF SIGNIFICANT CORRELATIONS FOR THE CHI-COMBINED GROUP AFTER TRANSFORMATION BY: GCS = 2/48 (4%) PTA = 16/48 (33%) GRAND TOTAL = 18/96 (19%)

Note: The total number of cognitive outcome variables examined from the WAIS-R, WMS/WMS-R, Stroop, Trail Making Test, and COWA was 48. *These correlations did not reach statistical significance, but were included for comparison to the other severity indicator (i.e., GCS or PTA).

33

Before logarithmic transformation, the MN group had 13 significant correlations

between the GCS and the 48 cognitive outcome variables and 15 between PTA and these

same variables (see Table 3.10). Following logarithmic transformation tiie number of

significant correlations was 13 between the GCS and the 48 cognitive outcome variables,

and 28 for PTA these same variables (see Table 3.11). The average magnitude of

significant correlations, post-transformation, between the GCS and PTA for the MN

group were r = .36 and r = .47, respectively.

Based upon the number of significant correlations, following logarithmic

transformation, across groups the GCS had a total of 31 significant correlations with

cognitive outcome variables as compared to 68 between PTA duration and these same

variables. Therefore, in response to the issue concerning the reliability of PTA duration

versus the GCS in predicting cognitive outcome, it appeared that PTA duration was

generally the more reliable predictor. Furthermore, greater predictability resulted when

PTA duration was log transformed to overcome the skewness in the distribution of PTA

scores across groups. This skewness was due to the wide range of performance on

cognitive outcome variables when there was no PTA (PTA = 0). To assess whether

including both the GCS and or PTA in a prediction equation would improve

predictability above the GCS or PTA duration alone, stepwise multiple linear regressions

were computed with each cognitive outcome variable. The results of these analyses

indicated that adding both of these severity indicators into a regression equation did not

improve predictability above that of either one of them alone.

34

Table 3.10: Significant correlations by group between GCS/PTA and cognitive outcome variables (MN).

GCSLOG p-value PTALOG p-value

MN (n = 25)

WAIS-R Comprehension Comprehension Scaled Digit Span Digit Span Scaled Information Information Scaled Picture Arrange. Scaled Vocabulary Vocabulary Scaled

FDI VCI VIQRAW VIQ FSIQ

WMS-R Information Logical Memory II Attention/Concentration

Index

Stroop Word in seconds Word # of errors Color-Word # of errors

.370

.397

.288

.336

.478

.523

.498

.352

.383

.335

.423

.423

.417

.363

.147 .041

.298

-.627 -.642 -.483

.050

.037

.097*

.063*

.014

.007

.013

.059*

.043

.069*

.028

.028

.030

.058*

.257*

.431*

293 298 360 356 371 354 261 482 ,513

,387 401 302 ,444 ,370

,458 ,367

.093*

.089*

.046

.048

.045

.053*

.126*

.011

.007

.037

.032

.086*

.019

.050

.014

.046

.174* -.497 .035

.001

.001

.034

Trail Making Test (TMT-A; TMT-B) TMT-A # of errors -.170 .243* TMT-B # of errors -.526 .039

NUMBER OF SIGNIFICANT CORRELATIONS FOR THE MN GROUP BY: GCS = 13/48 (27%) PTA = 15/48 (31%) GRAND TOTAL = 28/96 (29%)

.314

.485

.569

.522 -.085

.077*

.011

.011

.009

.391*

Note: MN = Mixed Neurologic. The total number of cognitive outcome variables examined from the WAIS-R, WMSAVMS-R, Stroop, Trail Making Test, and COWA was 48. *These correlations did not reach statistical significance, but were included for companson to the other severity indicator (i.e., GCS or PTA).

35

Table 3.11: Significant correlations by group between transformed GCS/PTA and cognitive outcome variables (MN).

GCSLOG p-value PTALOG p-value

MN (n = 25)

WAIS-R Arithmetic Arithmetic Scaled Comprehension Comprehension Scaled Digit Span Digit Span Scaled Information Information Scaled Picture Arrangement Picture Arrange. Scaled Picture Completion Picture Completion Scaled Vocabulary Vocabulary Scaled

FDI VCI PIQ VIQRAW VIQ FSIQRAW FSIQ

WMS-R Information Attention/Concen. Index Visual Reproduction I

Stroop Word in seconds Word # of errors Color-Word # of errors

.178

.235

.395

.413

.300

.346

.474

.512

.279

.497

.128

.330

.353

.389

.336

.427

.272

.423

.419

.237

.358

.187

.323

.149

-.662 -.651 -.491

Trail Making Test (TMT-A; TMT-B) TMT-A # of errors .340 TMT-B # of errors -.502

.220*

.152*

.038

.031

.088*

.058*

.015

.009

.117*

.013

.291*

.072*

.058*

.041

.068*

.027

.123*

.028

.029

.157*

.061*

.203*

.153*

.284*

.001

.001

.032

.077*

.048

.363

.385

.387

.430

.480

.507

.502

.502

.369

.367

.537

.562

.494

.502

.534

.495

.445

.582

.582

.542

.582

•.427 -.573 .484

.369

.376

.630

.472

.269

.049

.039

.038

.023

.010

.007

.009

.009

.050

.051*

.005

.004

.010

.009

.005

.010

.022

.002

.002

.006

.003

.021

.016

.021

.046

.043

.005

.018

.188*

36

Table 3.11: Continued.

GCSLOG p-value PTALOG p-value

COWA S -.343 .059* -.462 .013

NUMBER OF SIGNIFICANT CORRELATIONS FOR THE MN GROUP AFTER TRANSFORMATION BY: GCS = 13/48 (27%) PTA = 28/48 (58%) GRAND TOTAL = 41/96 (43%)

Note: MN = Mixed Neurologic The total number of cognitive outcome variables examined from the WAIS-R, WMS/WMS-R, Stroop, Trail Making Test, and COWA was 48. *These correlations did not reach statistical significance, but were included for comparison to the other severity indicator (i.e., GCS or PTA)

37

Correlational patterns across domains

Significant first-order Pearson correlations between the cognitive outcome

variables and the transformed severity indicators can be seen in Tables 3.5, 3.7, 3.9, and

3.11. Examination of these various groups' tables revealed some trends. The GCS

consistently had significant positive relations with the WAIS-R subtests (cf, Paniak et

al., 1992) thought to be more resistant to the effects of a head injury (e.g.. Vocabulary)

for the CHI-C and MN groups (Farr, Greene, Fischer-White, 1986). Consequently,

significant positive relations between the GCS and the VCI and VIQ for the CHI-C and

MN groups resulted. For the CHI-NC group, the GCS was significantly and inversely

related to the WAIS-R Digit Symbol and Similarities subtests, but no other significant

relationships with any indices or summary scores resulted. When the CHI-C and CHI-

NC groups were combined into the CHI-Comb group, all significant relations between

the WAIS-R variables and the GCS disappeared as shown in Table 3.9. Finally, with the

exception of the CHI-NC group, duration of PTA was consistently and inversely related

across a variety of WAIS-R variables, indices, and summary scores at a statistically

significant level.

No consistently significant relationships were observed between the GCS or PTA

duration and the WMS-R memory variables across groups (see Tables 3.5, 3.7, 3.9, and

3.11). Likewise, the GCS and duration of PTA were not consistently and significantly

related across the CHI-C, CHI-NC, and MN groups for the Stroop, Trail Making Test,

CLOX, or COWA variables.

38

Comparison bv group on cognitive outcome variables

Although the groups did not differ significantly on the severity indices, all

cognitive variables were submitted to one-way ANCOVAs with age as a covariate and

CHI-C, CHI-NC, and MN as the groups. Despite the multiple analyses, no significant

group differences were observed. Means and standard deviations for the cognitive

measures by group can be seen in Tables 3.12-3.14. It should be noted that the

ANCOVA for a VIQ/PIQ discrepancy by group just failed to obtain statistical

significance [F (2, 42) = 3.01, p = .059]. The VIQ/PIQ mean discrepancies by group

were CHI-C = 6.13 (SD = 10.22), CHI-NC = 18.14 (SD = 15.54), and MN = 4.52 (SD =

14.52). However, the low sample size (n = 9) in the CHI-NC group may have resulted in

lowered power to detect group differences.

Comparison bv group: Pre/Post Intellectual Functioning

The calculation of OPIE pre-morbid IQ scores permitted a unique analysis for the

IQ measures. Although the CHI-C and CHI-NC groups did not significantly differ in

level of initial severity, it would be expected that the CHI-NC would be more vulnerable

to the effects of a closed head injury than the CHI-C group (Satz, 1993). A CHI-C by

CHI-NC Group by Pre-Post analysis was conducted with the OPIE scores as pre-morbid

IQ scores and the FSIQ, VIQ, and PIQ scores as the post-measures. In this analysis the

group by pre-post interaction is the same as the main effect for a change score analysis

given that the correlation between pre- and post scores is about r = .90 or greater, which it

39

Table 3.12: Means and Standard Deviations by group across WAIS-R subtests and indices.

VERBAL SUBTESTS Arithmetic Comprehension Digit Span Information Similarities Vocabulary

CHI-C (n=17)

9.35 (4.46) 8.59 (3.04) 8.88 (4.34) 7.76 (2.88) 8.41 (2.83) 8.65 (3.35)

PERFORMANCE SUBTESTS Block Design Digit Symbol Object Assembly Picture Arrangement Picture Completion

INDICES FDI POI VCI VIQ PIQ FSIQ VIQ/PIQ discrepancy

7.88 (2.66) 6.82 (2.60) 7.19(2.64) 7.75 (2.24) 7.82 (3.09)

90.93 (19.24) 87.38(14.15) 90.78 (15.46) 89.77 (16.02) 84.63 (13.76) 87.38 (14.85) 6.13(10.22)

CHI-NC (n = 9)

8.33 (2.40) 8.78 (3.63) 8.89(2.14) 8.67 (2.69) 7.78 (3.38) 9.33 (4.47)

4.71 (2.93) 5.25(2.31) 5.00 (3.79) 7.25 (3.77) 7.67 (3.57)

90.91 ( 9.54) 74.71 (20.51) 92.38(17.81) 91.89(14.23) 75.57 (17.37) 85.00(15.58) 18.14(15.54)

MN (n = 25)

7.42 (2.43) 8.21 (3.09) 7.04 (3.25) 8.21 (3.05) 7.88 (2.36) 8.25 (3.03)

6.95 (2.77) 6.48(2.91) 6.78 (3.61) 7.35 (2.60) 7.42 (2.65)

83.37(14.30) 83.61 (15.72) 89.56(14.05) 87.21 (14.01) 83.04(12.67) 85.13(11.67) 7.70(15.63)

Note: CHI-C = Closed Head Injury-Clean; CHI-NC = Closed Head Injury-Not Clean; MN = Mixed Neurologic. Means with standard deviations in parentheses.

40

Table 3.13: Means and Standard Deviations by group across WMS/WMS-R subtests and indices, Stroop, and Trail Making Test.

CHI-C (n=17)

CHI-NC (n = 9)

MN (n = 25)

WMS-R* Figural Memory Logical Memory I Logical Memory II Mental Control Verbal Paired Assoc. I** Verbal Paired Assoc, n Visual Memory Scan Visual Paired Assoc. I Visual Paired Assoc. II Visual Reproduction I Visual Reproduction n

WMS-R INDICES Attention/Concentration Delayed Recall General Memory Verbal Memory Visual Memory

STROOP Color time in sees. Color errors Word time in sees. Word errors Color-Word time in sees. Color-Word errors

-.07 ( .93) -.14 ( .74) -.06 ( .90) .33 ( .96) .18(1.01) .11(1.10)

-.15 ( .89) -.08 ( .95) .03 ( .96) .48 ( .98) .34(1.07)

92.23 (25.63) 76.77 (19.58) 71.31 (16.08) 74.54(15.95) 80.54(17.98)

21.93 ( 8.84) .47 ( .92)

23.06(12.82) 1.38 ( 2.33)

50.14(22.69) 3.93 ( 5.74)

TRAIL MAKING TEST (TMT-A; TMT-B) TMT-A time in sees. 51.47 (16.24) TMT-A errors .06 ( .24) TMT-B time in sees. 167.18 (92.37) TMT-B errors 1.71 ( 3.05)

.21 (1.26) -.10(1.13) .29(1.35)

-.04 ( .71) -.13 ( .95) -.08(1.00) .38(1.27) .20(1.23)

-.08(1.21) -.26(1.48) .03(1.40)

89.00 (19.56) 87.25 (24.58) 77.75 (32.48) 78.80 (20.96) 86.50 (35.59)

23.17(12.47) 1.83 ( 3.25)

18.88 ( 7.59) .13 ( .35)

24.60 ( 8.85) .40 ( .89)

128.67 (97.80) .22 ( .44)

241.23(92.38) .67 ( .58)

-.15(1.10) .12(1.12)

-.05 ( .96) -.22(1.07) -.09(1.01) -.06 ( .94) -.08 ( .91) -.05 ( .81) .04 ( .99)

-.31 ( .67) -.35 ( .63)

84.64(16.35) 76.67(15.01) 80.25 (17.90) 83.36 (20.50) 82.00(11.98)

32.09 (16.25) 2.00 ( 3.26)

32.17(33.29) 1.00 ( 1.87)

51.83(23.26) 7.61 ( 8.35)

78.19(66.69) 100.27 (78.44) 250.89 (97.48)

1.53 ( 2.79)

Note: CHI-C = Closed Head Injury-Clean; CHI-NC = Closed Head hijury-Not Clean; MN = Mixed Neurologic. *A11 Wechsler Memory Scale subtest data is presented as a z-score, which has a mean of 0 and a standard deviation of 1. **Assoc. = Associates. Means with standard deviations in parentheses.

41

Table 3.14: Means and Standard Deviations by group for the Clock Face Drawing Test and COWA.

CHI-C (n=17)

CLOCK FACE DRAWING TEST CLOXl CLOX2 Fraction Score

COWA F A S FAS Total

11.36( 2.73) 13.50 ( .94)

.55 ( . 27)

8.47 ( 5.59) 7.47 ( 5.14) 8.59 ( 4.54)

24.94 (14.91)

CHI-NC (n = 9)

9.40 (3.97) 11.20(4.09)

.54 ( .23)

7.56 (3.47) 4.33(2.18) 7.00 (4.06)

18.89(8.67)

MN (n = 25)

9.74 ( 2.86) 11.90( 2.38)

.55 ( .68)

6.32 ( 4.21) 6.39 ( 4.56) 6.04 ( 4.13)

16.88(11.07)

Note: CHI-C = Closed Head Injury-Clean; CHI-NC = Closed Head Injury-Not Clean; MN = Mixed Neurologic. Means with standard deviations in parentheses.

42

is in this study Huck & McLean (1975). The correlations between pre- and post-FSIQ,

VIQ and PIQ were r = .91, r = .90, and r = .78, respectively.

One-way ANOVA comparisons of the various OPIE pre-morbid indices (i.e.,

OPIE-Best, OPIE-FSIQ, OPIE-VIQ, and OPIE-PIQ) by group (i.e., CHI-C, CHI-NC, and

MN) were not statistically different by group as expected (refer to Table 3.15 for means

and standard deviations by group). This lack of difference in the pre-injury OPIE scores

is why the main effect for group, which is a combination of pre- and post-injury scores, is

underestimated in this design and not interpretable (Huck & McLean, 1975). Also, as

expected, significant differences were found within each group between their obtained

and predicted mean FSIQ scores, using the OPIE Best as the estimate of pre-morbid

abilities (see Table 3.16), indicating that head injury had a negative influence on

cognitive performance.

When the CHI-C and CHI-NC groups were compared for the age-standardized

post-injury IQ scores only, there were no significant differences between CHI-C and

CHI-NC for the FSIQ [F (1, 21) = .121, p = .731], VIQ [F (1, 21) = 1.802, p = .742], or

PIQ [F (1, 24) = 111, p = .194] summary scores. However, the VIQ/PIQ discrepancy was

significantly different [F (1, 22) = 4.90, p = .038] between the CHI-C and CHI-NC

groups with the CHI-C group demonstrating less discrepancy. The average VIQ/PIQ

discrepancy for the CHI-C group was 6.13 (SD = 10.22), while the CHI-NC was 18.14

(SD = 15.54), see Figure 3.3.

43

Table 3.15: ANOVA comparison of OPIE formulas by group.

CHI-C (n=17)

CHI-NC (n = 9)

MN (n = 25)

p-value

OPIE BEST OPIE VIQ OPIE PIQ

97.11 (14.88) 93.31(15.86) 92.51(13.54)

99.58(16.66) 95.80(18.55) 94.17(15.74)

97.51 (12.30) 92.90(14.51) 93.21 (12.48)

.9067

.8920

.9556

MN = Mixed Neurologic. OPIE BEST = FSIQ estimate calculated with the best of either the participant's Vocabulary or Picture Completion raw score.

OPIE VIQ = VIQ estimate calculated with the Vocabulary raw score. OPIE PIQ = PIQ estimate calculated with the Picture Completion raw score.

44

Table 3.16: Mean Difference Between Obtained and Predicted IQ's Using the OPIE Best Formula Presented by Diagnostic Category and the Total Sample.

Group N FSIQ Obtained M(SD)

FSIQ Best M(SD) Difference p-value

CHI-C CHI-NC MN TOTAL

17 9

25 51

87.38 (14.85) 85.00(15.58) 86.65 (14.76) 85.89(13.18)

97.11(14.88) 99.58 (16.66) 97.51 (12.30) 97.75 (13.76)

9.73 14.58 10.86 11.86

19.45 20.52 61.15 95.26

.001

.004

.000

.000

Note: CHI-C = Closed Head Injury-Clean; CHI-NC = Closed Head Injury-Not Clean; MN = Mixed Neurologic.

Means with standard deviations in parentheses.

45

WMS-R SuMlAi: CHI-C Vt. CHI-NC

1

^ / / / / / / / / / /

/

WAI8-R SuUwts

Figure 3.3. Mean WAIS-R subtest scores for the CHI-C and CHI-NC groups.

46

Table 3.17 shows the means and standard deviations for the pre-morbid OPIE

scores and the post-morbid FSIQ, VIQ, and PIQ scores. The group by pre-post

interaction for the FSIQ scores was non-significant [F (1,21) = 1.01, p = .32]. The group

by pre-post interaction for the VIQ scores was also non-significant [F (1,24) = .01, p =

.90]. The group by pre-post interaction for the PIQ scores was significant [F (1,21) =

4.56, p = .04] indicating that the estimated 18.60 point reduction fi-om pre- to post-injury

in the CHI-NC group was significantly greater than the estimated 8.68 point loss in the

CHI-C group. An analysis of covariance with the pre-morbid scores as the covariate,

showed the same significant outcome.

47

Table 3.17: Means and Standard Deviations by group for Predicted versus Obtained FSIQ, VIQ, and PIQ.

CHI-C CHI-NC MN (n=17) (n = 9) (n = 25)

Predicted / Obtained Predicted / Obtained Predicted / Obtained

FSIQ 97.11 (14.88)/87.38 (14.85) 99.58 (16.66)/85.00 (15.58) 97.51 (12.30)/85.13 (11.67) VIQ 93.31 (15.86)/89.77 (16.02) 95.80 (18.55)/91.89 (14.23) 92.90 (14.51)/87.21 (14.01) PIQ 92.51 (13.54)/84.63 (13.76) 94.17 (15.74)/75.57 (17.37) 93.21 (12.48)/83.04 (12.67)

Note: CHI-C = Closed Head Injury-Clean; CHI-NC = Closed Head Injury-Not Clean; MN = Mixed Neurologic.

Means with standard deviations in parentheses.

48

CHAPTER W

DISCUSSION AND CONCLUSIONS

The primary hypothesis of this investigation was that refining the CHI group by

breaking the group into CHI-C and CHI-NC groups would improve the prediction of

cognitive outcomes fi-om the PTA and GCS severity indices. Both before and after

adjusting the PTA and GCS indices for the skewed distribution caused by the number of

zero and fifteen scores respectively, the majority of evidence supports this hypothesis.

The first evidence that separating head injury patients into CHI-C and CHI-NC

groups would lead to better prediction between severity indices and cognitive outcomes

was the presence of significant correlations between history of alcoholism, drug abuse,

neurologic condition, and psychiatric condition and the severity indicators. As seen in

Table 3.3, for the CHI-NC group the correlation between a history positive for drug abuse

and both severity indicators was extremely robust. Moreover, for the CHI-NC group,

there was a significant correlation between a history of previous neurologic insult and

GCS scores. This same relationship approached significance for duration of PTA.

Finally, for the CHI-NC group, a history positive for alcoholism approached significance

with the GCS.

Further examination of correlation Table 3.4 revealed that pre-log transformation

the CHI-C group had a total of 28 significant correlations between the GCS and PTA

duration and cognitive outcome variables. In comparison, looking at Table 3.6, the CHI-

NC group only had a total of 8 significant correlations between the GCS and PTA

49

duration and cognitive outcome variables pre-log transformation. Log transformation of

the GCS and PTA duration scores reduced the skew of the distributions and the ensuing

correlations fiirther illustrated the point that separating the groups is necessary to yield

improved prediction between severity indicators and cognitive outcome. Overall, 30

significant correlations were found between both GCS and PTA duration and cognitive

outcome variables for the CHI-C group (see Table 3.5), as compared to only 10 for the

CHI-NC group (see Table 3.7). It is acknowledged that a Bonferroni may have been

appropriate given the large number of correlations computed. However, a Bonferoni

correction was not completed because doing this across groups would have uniformly

lowered the number of significant correlations, and because a main goal of this

investigation was to find out if separating a CHI group into subgroups was a usefiil

procedure or not.

A number of the reviewed investigations had larger closed head injury samples

than this investigation (cf Alexandre et al., 1983; Brooks et al., 1980; Paniak et al.,

1992). Nevertheless, larger samples still did not translate into consistent and positive

findings across these investigations and they did not break their groups down into

subgroups. For illustrative purposes in this study, the CHI-C and CHI-NC groups were

combined into a CHI-Comb group with increased sample size (n = 26). Thus, the

probability of finding a greater number of significant correlations between severity

indicators and cognitive outcome variables could have been increased. However, as

predicted, the summation of the CHI-C and CHI-NC into a CHI-Comb group only

obscured findings as illustrated by the decreased number of significant correlations

50

between the severity indicators and cognitive outcome variables. In addition to

decreasing the number of significant correlations, the magnitude of the resuhing

correlations was also generally lower for the CHI-Comb group as compared to the CHI-C

group.

In general, it appeared that duration of PTA was a better choice as a predictor of

cognitive outcome. The magnitude of the correlations between duration of PTA and pre­

morbid history variables was consistently less than that between these same variables and

the GCS indicating that the PTA may be less influenced by a history of previous head

injury, neurologic problems, mental illness, alcoholism or drug abuse. Furthermore, log

transformed PTA duration appeared to be consistently and significantly related to general

intellectual fimctioning (i.e., WAIS-R subtests, indices, and IQ summary scores) for the

CHI-C and MN group whereas the GCS was not as consistently related to these same

variables. Duration of PTA was also significantly related to a larger number of cognitive

outcome variables across groups compared to the transformed GCS. The combination of

PTA duration and the GCS were examined to determine if using them in combination

would contribute significant individual variance and improve prediction of cognitive

outcome. However, the use of both PTA and the GCS in a multiple regression equation

did not add to the predictability above that of PTA duration alone. This was likely due to

the shared variance between the GCS and PTA, and the fact that the PTA scores were

more meaningfiil for post-injury predictions. The predictability of cognitive outcomes

fi-om PTA duration was not as consistent in the other cognitive domains including

memory fimctioning, attention/concentration, executive fimctioning, or verbal fluency.

51

Overall, a primary purpose of this investigation was to examine whether separating a

closed head injury group into "clean" and "not clean" groups would improve prediction

of cognitive outcome fi-om PTA duration and/or the GCS, and it was.

Comparing the CHI-C, CHI-NC, and MN groups across the various cognitive

outcome measures with one-way ANCOVAs, covarying for the age effect, did not result

in any significant group mean differences. Significant correlations between PTA

duration and the WAIS-R variables were uniformly found for both the CHI-C and MN

group. Therefore, in this investigation, the inclusion of a MN group to illustrate what

persistent cognitive deficits were unique to the closed head injury versus a generalized

brain insult were not supported by mean differences across groups or correlational

analyses.

Few studies have attempted to adjust for the non-linear distribution of PTA and

GCS scores. This lack of linearity is due principally to the wide variability in cognitive

performance for those whose PTA duration was zero and/or had a GCS score of fifteen.

This results in a skewed distribution of scores at one end of the PTA and GCS measures.

The literature shows that approximately 90% of mild CHI patients demonstrate little or

no deficits three-months post-injury (King, 1996; Satz et al., 1999). However, between

5-15% (Alexander, 1995) report deficits (e.g., memory problems, decreased attention,

mood lability, headache, dizziness, fatigue, and diplopia) that are equal to or worse than

that of severe CHI patients several months-to-years post-injury. Therefore, inclusion of

these mild head injury patients in any investigation examining severity indices and

cognitive outcome variables can result in distributional problems. However, only a few

52

authors have attempted to correct for this difficulty (Haslam et al., 1994; Haslam et al.,

1995) by transforming their severity indices with a square-root transformation. In the

case of Haslam et al. (1994 & 1995), the square root transformation only pulled in the

outliers and did not affect the bunching of scores for the most mildly head injured

participants. Therefore, despite the transformation, the skewed score issue still remained

in these investigations. A log transformation is more appropriate when skewness is due

to a bunching of scores at one end of a distribution as well as a range of scores at the

other end of the distribution (Howell, 1992).

The use of the OPIE scores as pre-estimates of IQ fimctioning added an extra

dimension to this investigation. The OPIE was chosen as an estimate of pre-morbid

intellectual fimctioning because previous research had demonstrated that it was a reliable

estimate of pre-injury intellectual fimctioning for normals (Krull et al., 1995), and clinical

populations (Ropacki & Elias, 1999; Scott, Krull, Williamson, Adams, & Iverson, 1997).

Furthermore, the OPIE Best, OPIE VIQ, and OPIE PIQ formulas were chosen because

these prediction equations were found superior to the original OPIE FSIQ formula, which

uses both Vocabulary and Picture Completion raw scores (Scott et al., 1997), and because

these formulas most closely approximate an expected mean of 100 and standard deviation

of 15 (Ropacki & EUas, 1999; Scott et al.,1997) The use of the OPIE as a pre-morbid

indicator for comparison to post-injury performance in a pre-post design had not been

previously validated. Finally, the use of the pre-post design was possible because of the

extremely high and comparable correlations between predicted OPIE IQ scores and

obtained WAIS-R IQ scores (Huck & McLean, 1975).

53

As expected, no significant differences were found between the CHI-C, CHI-NC,

and MN groups on their predicted pre-morbid abilities, and all groups showed a

significant reduction fi-om pre-morbid estimated to post-morbid assessment. It has been

proposed that VIQ is a more stable following brain insult whereas PIQ is more sensitive

to the effects of neurologic insuh (Crosson, Greene, Roth, Farr, & Adams, 1990; Farr et

al., 1986). Therefore, it is not surprising that any difference between the mean

performance of the CHI-C and CHI-NC groups would be more likely to occur for the PIQ

comparison and the VIQ-PIQ discrepancy. This finding fits with what is referred to as

the "brain reserve capacity and threshold theory" (Satz, 1993).

Satz (1993) completed a comprehensive literature review on cognitive reserve

theory (a.k.a. brain reserve capacity or threshold theory) that is helpfiil to understanding

the CHI-C and CHI-NC group differences. Basically, according to this theory, previous

neurologic insult increases vulnerability to the fixture effects of neurologic problems such

as dementing illnesses, whereas higher education and larger head circumference serve as

buffers to the development of fiiture neurologic problems. Review of all the support for

this theory is beyond the scope of this investigation, but is well reviewed in Satz, 1993.

Since 1993, fiirther support has been found for the cognitive reserve theory in the areas of

Parkinson's disease (e.g., Glatt et al., 1996; Schmand, Smit, Geeriing, & Lindeboom,

1997), Alzheimer's disease (e.g., Alexander, Gurey et al., 1997; Graves et al., 1996),

Schizophrenia (e.g., Dwork et al., 1998), HIV (e.g., Pereda et al., 2000; Stem, Silva,

Chaisson, & Evans, 1996), but no investigations were found that examined the effects of

closed head injury on cognitive reserve. In the current investigation, there were less

54

severe head injuries in the CHI-NC group, but greater pre-post differences for PIQ and a

significantly larger VIQ/PIQ discrepancy for the CHI-NC group as compared to the CHI-

C group. Furthermore, the significant PIQ difference and VIQ/PIQ discrepancy was

found despite nearly equivalent pre-injury education levels between CHI-C and CHI-NC

groups. Examination of Figure 3.3 reveals that these significant discrepancies are largely

resultant fi-om the divergence of group performance across the performance subtests, but

especially two WAIS-R subtests that comprise the POI (i.e.. Block Design and Object

Assembly). Moreover, the subtests that make up the POI and PIQ are thought to measure

"fluid" intelligence, which is believed to be more sensitive to the effects of brain insult.

In comparison, the subtests that comprise the VIQ and reportedly reflect "crystallized"

intelligence, which is thought to be relatively stable following brain injury (Kaufinan,

1990). Overall, it appeared that the CHI-NC group was more sensitive to the effects of

the head injury than those in the CHI-C group supporting the cognitive reserve theory

with closed head injury. Moreover, these findings provide fiirther support for the need to

clean data for CHI investigations and suggest a direction for fiiture research.

Summarv and suggestions for fiiture research

In summary, this investigation contributed to the closed head injury literature in a

number of ways. First, it demonstrated that separating those with a positive pre-morbid

history of drug and alcohol abuse, neurologic condition, and psychiatric problems from

those without such a history results in increased predictability between severity indicators

and cognitive outcome variables. Duration of PTA was found to be better predictor of

55

cognitive outcome following a closed head injury in this investigation. Furthermore, this

study showed that fiiture closed head injury research should be alert to the effects of

including those with mild CHI into their samples because of potential issues of non-

linearity of data. This investigation did not find the unique effects of closed head injury

above that of other brain insuhs. That is, no mean differences were observed between the

CHI groups and the MN group. Nevertheless, it was shown that the CHI-NC was more

sensitive to the effects of head injury than the CHI-C group when pre-morbid estimates of

fimction were estimated. This latter finding is in support of the cognitive reserve theory

(Satz, 1993). Finally, this investigation has important implications not only for fiiture

research, but also for clinical practice because of the fiirther vaUdation of the OPIE with

CHI patients, and the demonstrated importance of obtaining a thorough history and

background in regards to pre-morbid history for a more accurate assessment and

diagnosis.

Future research can focus on some of this investigation's shortcomings.

Specifically, generalizability was limited in this study by the inclusion of only individuals

referred to rehabilitation following insuh. A great number of individuals are not referred

to rehabilitation following injury because they either do not seek initial medical attention,

or because lack of resources. Such individuals are often seen throughout the acute phase

of their injury and then discharged because of lack of resources (i.e., insurance). Also,

many individuals with continuing problems have a higher probability to be referred for

rehabilitation. Future investigations could somewhat ameliorate this problem by pre-

screening all emergency room admissions following closed head injury, with follow-up

56

neuropsychological assessments regardless of whether they are admitted and treated

and/or transferred to rehabilitation.

Another problem, that appears to be unavoidable, is the time since injury versus

severity of injury issue. Specifically, in this investigation the time since injury was

greater in the CHI-C group than the CHI-NC or MN groups, but not at a statistically

significant level. The CHI-C group also had a larger number of severe head injuries as

compared to the CHI-NC group, but not at a statistically significant level. It seems

logical that the more severe the injury, the longer time from injury until assessment and

that these two issues will be hard to separate in research studies. Although group

differences in times since injury and severity of injury were not significant problems for

this investigation, fiiture studies could match groups to include a more equivalent number

of patients of each severity level.

Future studies can also improve upon this investigation's findings, support the

significant findings of this study, and help fiirther delineate the effects of head injury

versus other variables. For instance, a fiiture large-scale investigation could include

multiple comparison groups including a CHI-C litigating, CHI-C non-litigating, CHI-NC

litigating, CHI-NC non-litigating, other injury group (01) excluded for pre-morbid

history variables, psychiatric group, and non-impaired group excluded for pre-morbid

history variables (e.g., those referred for neuropsychological evaluation prior to job entry

such as police officer, pilots, et cetera). Through multiple comparisons, this investigation

could help delineate what deficits are attributable to the head injury versus other factors

even if the CHI groups sustained bodily injury. Only three investigations similar to this

57

proposed investigation have been completed (Asamow et al., 1995; Bijur, Haslum, &

Goldfing, 1990; Dikmen, Machamer, Winn, & Temkin, 1995), and two of these focused

on children (Arsanow et al., 1995; Bijur et al., 1990), while none examined the effects of

litigation.

58

REFERENCES

Alexander, G. E., Furey, M. L., Grady, C. L., Pietrini, P., Brady, D. R., Mentis, M. J., & Schapiro, M. B. (1997). Association of premorid intellectual fimction with cerebral metabolism in Alzheimer's disease: implications for the cognitive reserve hypothesis. American Journal of Psvchiatrv. 154. 165-172.

Alexander, M. P. (1995). Mild traumatic brain injury: Pathophysiology, natural history, and clinical management. Neurology. 46.1253-1260.

Alexandre, A., Colombo, F., Nertempi, P., & Benedetti, A. (1983). Cognitive outcome and early indices of severity of head injury. Journal of Neurosurgerv. 59. 751-761.

Alfano, D. P., Neilson, P. M., & Fink, M. P. (1993). Long-term psychosocial adjustment following head or spinal cord injury. Neuropsvchiatrv. Neuropsvchologv. and Behavioral Neurology. 6. 117-125.

Arsanow, R. F., Satz, P., Light, R., Zaucha, K., Lewis, R., & McCleary, C. (1995). The UCLA study of mild closed-head injury in children and adolescents. In S. Broman & M. E. Michel (Eds.), Traumatic brain injury in children (pp. 117-146). New York: Oxford University Press.

Bachman, D. L. (1992). The diagnosis and management of common neurologic sequelae of closed head injury. Journal of Head Trauma Rehabilitation. 7. 50-59.

Barth, J. T., Macciochi, S. N., Giordani, B., Rimel, R., Jane, J. A., & Boll, T. J. (1983). Neuropsychological sequelae of minor head injury. Neurosurgerv 13. 529-533.

Bennett-Levy, J. M. (1984). Long-term effects of severe closed head injury on memory: evidence from a consecutive series of young adults. Acta Neurologica Scandinavica. 70. 258-298.

Bijur, P., Haslum, M., & Golding, J. (1990). Cognitive and behavioral sequelae of mild head injury in children. Pediatrics. 86. 336-344.

Bohnen, N., & Jolles, J. (1992). Neurobehavioral aspects of postconcussive symptoms after mild head injury. Journal of Nervous and Mental Disease. 180. 683-692.

Brooks, D. N. (1974). Recognition memory, and head injury. Journal of Neurology. Neurosurgerv. and Psychiatry. 37. 794-801.

59

Brooks, D. N. (1976). Wechsler Memory Scale performance and its relationship to brain damage after severe closed head injury. Journal of Neurology. Neurosurgery, and Psychiatry. 39. 593-601.

Brooks, D. N., Aughton, M. E., Bond, M. R., Jones, P., & Rizvi, S. (1980). Cognitive sequelae in relationship to early indices of severity of brain damage after severe blunt head injury. Journal of Neurology. Neurosurgery, and Psychiatry. 43. 529-534.

BuUard, J. A., & Cutshaw, R. (1991). Vocational evaluation of the closed head injury population: a challenge of the 1990's. Vocational Evaluation and Work Adjustment Bulletin. 24. 15-19.

Coppens, P. (1995). Subpopulations in closed-head injury: preliminary results. Brain Injury. 9. 195-208.

Crosson, B., Greene, R. L., Roth, D. L., Farr, S. P., & Adams, R. L. (1990). WAIS-R pattern clusters after blunt-head injury. The Clinical Neuropsychologist. 4. 253-262.

Crosson, B., Novack, T. A., Trenerry, M. R., & Craig, P. L. (1988). California Verbal Learning Test (CVLT) performance in severely head-injured and neurologically normal adult males. Journal of Clinical and Experimental Neuropsychology. 10. 754-768.

Dikmen, S. S., Machamer, J. E., Winn, H. R., & Temkin, N. R. (1995). Neuropsychological outcome at 1-year post head injury. Neuropsychology. 9. 80-90.

Diller, L. (1994). Finding the right treatment combinations: changes in rehabihtation over the past five years. In A. Christensen & B. Uzzell (Eds.), Brain injury and neuropsychological rehabilitation: International perspectives (pp. 1-16). Hillsdale, NJ: Erlbaum.

Dwork, A. J., Susser, E. S., Keilp, J., Waniek, C, Liu, D., Kaufinan, M., Zemishlany, Z., & Prohovnik, I. (1998). Senile degeneration and cognitive impairment in chronic schizophrenia. American Journal of Psychiatry. 155. 1536-1543.

Esselman, P. C, & Uomoto, J. M. (1995). Classification of the spectrum of mild traumatic brain injury. Brain Injury. 9. 417-424.

60

Farr, S. P., Greene, R. L., & Fisher-White, S. P. (1986). Disease process, onset, and course and their relationship to neuropsychological performance. In S. Filskov & T. Boll (Eds.), Handbook of Clinical Neuropsychology. Volume 2 (pp. 213-253). New York: John Wiley and Sons.

Garcia, J. M., Garrett, K., Stetz, M., Emanuel, L., & Brandt, J. (1990). Early behavioral responses in severe head injury. Cognitive Rehabilitation. 8. 30-33.

Gass, C. S., & Apple, C. (1997). Cognitive complaints in closed-head injury: relationship to memory test performance and emotional disturbance. Journal of Clinical and Experimental Neuropsychology. 19. 290-299.

Geffen, G. M., Butterworth, P., Forrester, G. M., & Geffen, L. B. (1994). Auditory Verbal Learning Test components as measures of the severity of closed-head injury. Brain Injury. 8. 405-411.

Glatt, S. L., Hubble, J. P., Lyons, K., Paolo, A., Troster, A. I., Hassanein, R. E., & Koller, W. C. (1996). Risk factors for dementia in Parkinson's disease: effect of education. Neuroepidemiology. 15. 20-25.

Graves, A. B., Mortimer, J. A., Larson, E. B., Wenzlow, A., Bowen, J. D., & McCormick, W. C. (1996). Head circumference as a measure of cognitive reserve. Association with severity of impairment in Alzheimer's disease. British Journal of Psychiatry. 169. 86-92.

Gronwall, D., & Wrightson, D. G. (1981). Memory and information processing capacity after closed head injury. Journal of Neurology. Neurosurgery, and Psychiatry. 44. 889-895.

Gupta, A., & Ghai, D. (1996). Memory and information processing capacity after closed head injury. Indian Journal of Clinical Psychology. 23. 170-177.

Haslam, C, Batchelor, J., Feamside, M. R., Haslam, S. A., Hawkins, S., & Kenway, E. (1994). Post-coma disturbance and post-traumatic amnesia as nonlinear predictors of cognitive outcome following severe closed head injury: findings from the Westmead Head Injury Project. Brain Injury. 8. 519-528.

Haslam, C, Batchelor, J., Feamside, M. R., Haslam, S. A., & Hawkins, S. (1995). Further examination of post-traumatic amnesia and post-coma disturbance as non­linear predictors of outcome after head injury. Neuropsychology. 9, 599-605.

Haut, M. W., & Shutty, M. S. (1992). Pattems of verbal learning after closed head injury. Neuropsychology. 6, 51-58.

61

Howell, D.C. (1992). Statistical methods for Psychology (3"^ edX Boston: PWS-Kent Publishing.

Huck, S. & McLean, R. (1975). Using a repeated measures ANOVA to analyze the data from a pretest-posttest design: A potentially confiising task. Psychological Bulletin. 82(41311-318.

King, N. S. (1996). Emotional, neuropsychological, and organic factors: their use in the prediction of persisting postconcussion symptoms after moderate and mild head injuries. Joumal of Neurology. Neurosurgery, and Psychiatry. 61. 75-81.

Kaufinan, A. S. (1990). Assessing adolescent and aduh intelligence. Boston, MA: Allyn and Bacon, Inc.

Katz, D. I. (1992). Neuropathology and neurobehavioral recovery from closed head injury. Joumal of Head Trauma Rehabilitation. 7. 1-15.

Kmll, K., Scott, J., & Sherer, M. (1995). Estimation of premorbid intelligence from combined performance and demographic variables. The Clinical Neuropsychologist. 9. 83-87.

Lees-Haley, P. R., Smith, H. H., Williams, C. W., & Dunn, J. T. (1996). Forensic neuropsychological test usage: an empirical survey. Archives of Clinical Neuropsychology. 11.45-51.

Levine, M. J., Van Hom, K. R., & Curtis, A. B. (1993). Developmental models of social cognition in assessing psychosocial adjustments in head injury. Brain Injury. 7. 153-167-

Levin, H. S., Gary, H. E. Jr., Eisenberg, H. M., Ruff, R. M., Barth, J. T., Kreutzer, J., High, W. M., Portman, S., Foulkes, M. A., Jane, J. A., Marmarou, A., & Marshall, L. F. (1990). Neurobehavioral outcome 1 year after severe head injury. Joumal of Neurosurgery. 73. 699-709.

Levin, H. S., Grossman, R. G., Rose, J. E., & Teasdale, G. (1979). Long-term neuropsychological outcome of closed head injury. Joumal of Neurosurgerv. 50. 412-422.

Levin, H. S., High, W. M., Goethe, K. E., Sisson, R. A., Overall, J. E., Rhoades, H. M., Eisenberg, H. M., Kalisky, Z., & Gary, H. E. (1987). The neurobehavioural rating scale: assessment of the behavioural sequelae of head injury by the clinician. Joumal of Neurology. Neurosurgery, and Psychiatry. 50. 183-193.

62

Levin, H. S., WilUams, D. H., Eisenberg, H. M., High, W. M. Jr., & Guinto, F. C. Jr. (1992). Serial MRI and neurobehavioural findings after mild to moderate closed head injury. Joumal of Neurology. Neurosurgery, and Psychiatry. 55. 255-262.

Lezak, M.D. (1995). Neuropsychological Assessment. (3'^ edX New York: Oxford University Press.

Loken, W. J., Thorton, A. E., Otto, R. L., & Long, C. J. (1995). Sustained attention after severe closed head injury. Neuropsychology. 9 (4). 592-598.

Mandleberg, I. A. (1976). Cognitive recovery after severe injury: 3. WAIS Verbal and Performance IQs as a fimction of post-traumatic amnesia duration and time from injury. Joumal of Neurology. Neurosurgerv. and Psychiatry. 39. 1001-1007.

Mandleberg, I. A., & Brooks, D. N. (1975). Cognitive recovery after severe head injury: 1. Serial testing on the Wechsler AduU Intelligence Scale. Joumal of Neurology. Neurosurgery, and Psychiatry. 38. 1121-1126.

Man, D. W. K. (1996). Remediation of cognitive deficits following closed head injury through occupational therapy procedures: a multiple baseline across behaviours design. In E. Bonke, E. Benno, & J. G. Bovill (Eds.), Memory and awareness in anesthesia III (pp. 151-163). Assen, Netherlands: Van Gorcum & Co.

McClelland, R. J. (1988). Psychosocial sequelae of head injury: anatomy of a relationship. British Joumal of Psychiatry. 153. 141-146.

McMillan, T. M., Jongen, E. L. M., Greenwood, R. J. (1996). Assessment of post­traumatic amnesia after severe closed head injury: retrospective or prospective? Joumal of Neurology. Neurosurgery, and Psychiatry. 60. 422-427.

Nelson, L. D., Adams, K. M. (1997). Challenges for neuropsychology in the treatment and rehabilitation of brain-injured patients. Psychological Assessment. 9. 368-373.

Oder, W., Goldenberg, G., Spatt, J., Podreka, I., Binder, H., & Deecke, L. (1992). Behavioural and psychosocial sequelae of severe closed head injury and regional cerebral blood flow: a SPECT study. Joumal of Neurology. Neurosurgery and Psychiatry. 55. 475-480.

Paniak, C. E., Silver, K., Finlayson, M., A, & Tuff, L. P. (1992). How usefiil is the WAIS-R in closed head injury assessment? Joumal of Clinical Psychology. 48, 219-225.

63

Pereda, M., Ayuso-Mateos, J. L., Gomez Del Barrio, A., Echevarria, S., Farina, M. C, Garcia Palomo, D., Gonzalez Macias, J., & Vazquez-Barquero, J. L. (2000). Factors associated with neuropsychological performance in HFV-seropositive subjects without AIDS. Psychological Medicine. 30. 205-217.

Ponsford, J., & Kinsella, G. (1992). Attentional deficits following closed-head injury. Joumal of Clinical and Experimental Neuropsychology. 14. 822-838.

Reid, D. B., & Kelly, M. P. (1993). Wechsler Memory Scale-Revised in closed head injury. Joumal of Clinical Psychology. 49. 245-254.

Richardson, J. (1990). Clinical and neuropsychological aspects of closed head injury. London: Taylor.

Ropacki, M. T., Elias, J. W. (1999). Further validation of the Oklahoma Premorbid Intelligence Estimate with closed head injury and mixed neurologic patients. Archives of Clinical Neuropsychology. 14 (8). 788.

Royall, D. R., Cordes, J. A., & Polk, M. (1998). CLOX: an executive clock drawing task. Joumal of Neurology. Neurosurgery and Psychiatry. 64. 588-594.

Satz, P. (1993). Brain reserve capacity on symptom onset after brain injury: A formulation and review of evidence for threshold theory. Neuropsychology. 7 (3). 273-295.

Satz, P., Alfano, M. S., Light, R., Morgenstem, H., Zaucha, K., Asamow, R. F., & Newton, S. (1999). Persistent post-concussive syndrome: A proposed methodology and literature review to determine the effects, if any, of mild head and other bodily injury. Joumal of Clinical and Experimental Neuropsychology. 2L 620-628.

Schmand, B., Smit, J. H., Geerlings, M. I., & Lindeboom, J. (1997). The effects of intelligence and education on the development of dementia. A test of the brain reserve hypothesis. Psychological Medicine. 27. 1337-1344.

Scott, J. G., Kmll, K. R., WiUiamson, D. J. G., Adams, R. L., & Iverson, G. L. (1997). Oklahoma Premorbid Intelligence Estimation (OPIE): Utilization in clinical samples. The Clinical Neuropsychologist. 11 (2). 146-154.

Scott, J. G., Sherer, M., & Adams, R. L. (1995). Clinical utihty of WAIS-R factor-derived standard scores in assessing brain injury. The Clinical Neurologist, 9(1), 93-97.

64

Smith-Seemiller, L., Lovell, M. R., Franzen, M. D., Smith, S. S., & Townsend, R. N. (1997). Neuropsychological fimctioning in restrained versus unrestrained motor vehicle occupants who suffer closed head injury. Brain Injury. 11. 735-742.

Spreen, O., & Strauss, E. (1998). A compendium of neuropsychological tests (3""̂ ed.). New York: Oxford.

Stambrook, M., Moore, A. D., Lubusko, A. A., Peters, L. C , & Blumenschein, S. (1993). Altematives to the Glasgow Coma Scale as a quality of life predictor following traumatic brain injury. Archives of Clinical Neuropsychology. 8. 95-103.

Stem, R. A., Silva, S. G., Chaisson, N., & Evans, D. L. (1996). Influence of cognitive reserve on neuropsychological fimctioning in asymptomatic human immunodeficiency vims-1 infection. Archives of Neurology. 53. 148-153.

Stuss, D. T., Binns, M. A., Carmth, F. G., Levine, B., Brandys, C. E., Moulton, R. J., Snow, W. G., & Schwartz, M. L. (1999). The acute period of recovery from traumatic brain injury: posttraumatic amnesia or posttraumatic confiisional state? Joumal of Neurosurgery. 90. 635-643.

Teasdale, G., & Jennett, B. (1974). Assessment of coma and impaired consciousness: a practical scale. Lancet. 2. 81-84.

Tremont, G., Hoffinan, R. G., Scott, J. G., Adams, R. L., & Nadolne, M. J. (1997). Clinical Utility of Wechsler Memory Scale-Revised and predicted IQ discrepancies in closed head injury. Archives of Clinical Neuropsychology. 12. 757-762.

Vilkki, J., Ahola, K., Hoist, P., Ohman, J., Servo, A., & Heiskanen, O. (1994). Prediction of psychosocial recovery after head injury with cognitive tests and neurobehavioral ratings. Joumal of Clinical and Experimental Neuropsychology. 16,325-338.

Vilkki, J., Poropudas, K., & Servo, A. (1988). Memory disorder related to coma duration after head injury. Joumal of Neurology. Neurosurgery, and Psychiatry, 51. 1452-1454.

Wechsler, D. (1945). A standardized memory scale for clinical use. Joumal of Psychology. 19. 67-70.

Wechsler, D. (1955). Manual for the Wechsler Adult Intelligence Scale. New York: Psychological Corporation.

65

Wechsler, D. (1981). Manual for the Wechsler Aduh Intelligence Scale-Revised. San Antonio, TX: Psychological Corporation.

Wechsler, D. (1987). Manual for the Wechsler Memory Scale-Revised. San Antonio, TX: Psychological Corporation.

Wilson, J. T., Hadley, D. M., Wiedmann, K. D., & Teasdale, G. M. (1992). Intercorrelation of lesions detected by magnetic resonance imaging after closed head injury. Brain Injury. 6. 391-399.

66

APPENDIX A

EXTENDED LITERATURE REVIEW

67

Introduction

Closed head injury (CHI) has been called the "silent epidemic" of our time

(Bullard & Cutshaw, 1991; Man, 1996; McClelland, 1988). Closed head injuries are

those head injuries where the skull remains intact and the brain is not exposed (Lezak,

1995). This excludes head injury from missiles or other penetrating objects. Lezak

(1995) indicated that the incidence of head injuries in the Untied States ranges from

500,000 to more than 1.9 million per year with 400,000 to 500,000 necessitating

hospitahzation or resulting in death. Esselman and Uomoto (1995), citing a citing a

National Health Interview Survey, reported that only 89% of those sustaining a head

injury consuU a physician, and 16% were admitted to a hospital. McClelland (1988)

proposed that head injury is "the" major cause of death for those 35 years-old and

younger. In summary, it is clear that CHI has become an increasingly prevalent malady

in the United States making it a major, important, and costly sociohealth problem facing

westem society today (Alfano et al., 1993; Bullard & Cutshaw, 1991).

Assessment of Closed Head Injury Severity

Glasgow Coma Scale. The GCS (Teasdale & Jennette, 1974) is a brief

assessment technique designed to assess severity, describe the posttraumatic states of

altered consciousness, and predict outcome in head injuries (see Appendix B). A

patient's total GCS score is based on standardized cumulative scores for eye-opening,

verbal performance, and motor responsiveness. For example, for eye opening, if a patient

68

opens their eyes on their own this is scored as a four, a three is assigned if the patient

opens their eyes to a loud voice, a two is assigned if the patient opens their eyes to

painfiil stimulus administration, and if the patient does not open their eyes to painftil

stimulus administration they receive a one. Overall, scores are summed across tiiree

domains: (1) eye opening, (2) verbal response, and (3) motor responsiveness to yield a

total GCS score. The GCS total score can range from three to fifteen, and once

determined the GCS total score is compared to established severity classifications (see

Appendix B). Traditionally, patients with a GCS greater than 13 are considered to have a

mild CHI, between 9-12 a moderate CHI, and less than 8 is indicative of a severe CHI.

The GCS was designed for use by emergency medical technicians, nurses, and doctors.

Lezak (1995) stated that the GCS's "greatest virtue" was its ability to predict outcome,

and that it has been generally accepted as the standard measure for determining severity

of injury in patients with compromised consciousness. However, not everyone agrees

that the GCS is the standard measure for determining injury severity or predicting

prognosis (cf, McMillan et al., 1996; Stambrook, et al., 1993; Vilkki et al., 1994).

Although many clinicians and investigators accept and use the GCS in their work

(e.g., Esselman & Uomoto, 1995; Levin et al., 1992; McMillan et al., 1996; Smith-

Seemiller et al., 1997; Wilson, Hadley, Wiedmann, & Teasdale, 1992), there is

considerable disagreement on how well the GCS can determine severity and predict

outcome. For instance, Garcia, Garrrett, Stetz, Emanuel, and Brandt (1990) explained

that the GCS presents a range of patient behaviors grossly categorized within a restricted

number of descriptive ratings. Furthermore, the GCS has a restricted range (3-15)

69

additionally compromising its predictive utility. The utihty of the GCS in predicting

outcome of moderate and mild CHI has been questioned because it was not designed to

make sharp distinctions of lesser severity (Stambrook et al., 1993). Richardson (1990)

stated that when the GCS is used as it was designed to be used, that is in the first hours to

days post-insuh, it often fails to correctly classify unusual cases. For example, when the

patient is admitted intoxicated or under the influence of a controlled substance, their

admission GCS score is not valid. Unreliable GCS scores are often produced by

intoxicated individuals, or individuals under the influence of a controlled substance.

Some of these individuals may have overdosed on either or a combination of both. Under

these conditions, GCS scores may be incorrectly attributed to head trauma severity, or

head trauma may be erroneously attributed to substance abuse (Stambrook et al., 1993).

Furthermore, some trauma patients are lucid for a short period following CHI, but then

deteriorate (Lezak, 1995). Hence, these individuals' GCS scores would be inaccurate.

Moreover, some CHI patients are intubated or anesthetized prior to emergency room

assessment with the GCS rendering their GCS assessment results suspect. Another

problem involves individuals who experience multiple trauma as often is found in CHI

patients involved in vehicular accidents. Specifically, one or more examination

modalities (e.g., their ability to speak, open their eyes due to facial trauma, or paralysis

and fractures compromising their motor abilities) may not be assessable due to injury

thereby limiting the utility of the GCS (Reid & Kelly, 1993; Stambrook et al., 1993).

Coppens (1995) reported that the GCS severity ratings are often unreliable as evidenced

by the variability in patients' outcome despite initial ratings of severe CHI. Although the

70

GCS has demonstrated reliability in predicting mortality, its ability to predict cognitive

and psychosocial outcome has been criticized (Coppens, 1995; Garcia et al., 1990; Haut

& Shutty, 1992; Levin et al., 1987; Reid & Kelly, 1993).

Duration of Posttraumatic Amnesia. According to McMillan et al. (1996), since

Russell's proposition in 1932, tiiat the sum of the comatose and confiisional periods was

the most usefiil predictor of outcome, "many studies" have demonstrated that the duration

of PTA provides the "best yardstick" for predicting outcome that exists. Duration of

PTA has been defined as the interval from the time of injury until the patient can form

new memories and continuously recalls day-to-day events (Oder et al., 1992).

Traditionally, PTA has been retrospectively assessed during clinical interview by

questioning the patient. Specifically, retrospective assessment typically involves the

clinician (i.e., physician, psychologist, or nurse) asking the patient questions regarding

their accident, their memory surrounding the accident, and other questions to determine if

the patient has formed memories for routine daily events. Posttraumatic amnesia

duration can range from minutes to days, and once determined it is compared to an

established severity classification system (see Appendix B). For example, PTA duration

of five minutes to one hour is considered a mild CHI whereas PTA duration of one-to-

seven days is indicative of a severe CHI. Many investigators agree that PTA duration has

superior utility in assessing severity and predicting outcome (Gupta & Ghai, 1996; Katz,

1992; McClelland, 1988). For instance, Katz (1992) reported that PTA's relationship to

outcome was more reliable than that of the GCS. McMillan and colleagues (1996)

asserted that the duration of PTA is a better indicator than depth or duration of coma and

71

MRI changes. Nonetheless, not all researchers are as steadfast in their convictions

regarding the reHability of PTA duration to classify severity and predict outcome (cf,

Coppens, 1995; Vilkki et al., 1994).

Although the duration of PTA is commonly used to determine severity and predict

CHI outcome, investigators have pointed out the duration of PTA's limitations (Alfano et

al., 1993; Barth et al., 1983; Bohnen & Jolles, 1992; Coppens, 1995; Gass & Apple,

1997; Levin et al., 1987; Lezak, 1995; Vilkki et al., 1994). For instance, Coppens (1995)

stated that PTA's relationship to recovery of language and "cognitive functioning" is less

than adequate. Coppens (1995) blamed the heterogeneous nature of CHI for the lack of

correlation between initial severity ratings and recovery. Moreover, difficulties

determining the duration of PTA make its usefuhiess questionable. Specifically, by

definition, the endpoint of PTA occurs when the patient can continuously register

memories of daily events. However, many of these authors believe that, in practice, this

determination is a highly subjective decision for the clinician because determining the

point at which continuous memory retums is often difficult to specify. Lezak (1995)

pointed out that the clinician's decision is even more clouded by the fact that many CHI

patients experience transient periods of confusion while others are aphasic.

Another limitation of PTA duration is its usage with milder cases of CHI.

According to Bohnen and Jolles (1992), the reliability of PTA to assess severity and

predict outcome decreases if PTA duration is less than one hour. Moreover, many CHI

individuals have a lack of insight into their deficits rendering their reports questionable,

and family members' memories are often unreliable due to the stressfulness of the

72

situation (Alfano et al., 1993; Gass & Apple, 1997; Levin et al., 1987; Lezak, 1995;

McMillan et al., 1996). Dmg use and alcohol consumption also limits the usefiihiess of

PTA duration to assess severity and predict outcome. Alcohol consumption and/or dmg

usage often makes the estimation of PTA inaccurate and severity classification

complicated (Bohnen & Jolles, 1992). Finally, adding to duration of PTA's reliability

problems are the fact that many CHI individuals sustain mild injuries and are eitiier not

admitted or discharged before their PTA has ended.

McMillan and colleagues (1996) attempted to address the limitations of using

PTA duration to determine severity and predict outcome. Furthermore, these

investigators wanted to determine if assessing PTA duration prospectively with the

Galveston Orientation and Amnesia Test (GOAT) was statistically more accurate than

determining PTA duration retrospectively through questioning and patient interview.

The prospective assessment of PTA duration involves asking the patient questions daily

such as their name, the day, date, month, and what they recall occurring before and

following the injury. Their performance is compared to objective data (e.g., the family's

reports and the actual date), then scored. Once the patient obtains a score in the "normal"

range on the GOAT, defined as three consecutive performances of 75% correct or better,

they are considered to have cleared from their PTA. Therefore, the duration of the

patient's PTA is considered the interval between the accident and the time when three

consecutive scores of 75% or higher are obtained on the GOAT.

McMillan et al. (1996) attempted to re-contact (N = 126) consecutive patients that

sustained a severe closed head injury. Of the original patients, 79 were traceable (73%).

73

Specifically, 3.5 to 5.0 years post-discharge, an interviewer blind to their original hospital

records recontacted these CHI patients. By interviewing and retrospective questioning

this interviewer attempted to determine the duration of PTA for each former patient.

Following this interview, the interviewer then attempted to determine each patient's

severity classification (see Appendix B). McMillan and colleagues determined tiiat both

prospective and retrospective approaches for determining PTA duration were reliable and

vahd techniques for assessing injury severity and predicting outcome. Specifically, their

results did not support the notion that retrospective assessment of PTA was unreliable or

that prospective assessment was more accurate. These authors also reported "highly

significant correlations" between the GCS, days of lost consciousness, and both methods

for assessing PTA duration. Finally, McMillan and colleagues reported that both

methods for determining duration of PTA (i.e., retrospective questioning and prospective

questioning with the GOAT) were superior to the other outcome measures (i.e., GCS) as

evidenced by these patients' psychosocial outcome, making PTA duration the "gold

standard" in severity and outcome assessment.

Severity assessment and outcome prediction with the GCS and PTA duration have

some of the same limitations. For instance, a patient's waxing and waning cognitive

status is problematic for both as are dmg and/or alcohol usage. However, when

determining PTA duration if the patient is under the influence of a substance, more time

can pass before a severity determination and outcome prediction has to be made. On the

other hand, the GCS does not require patient or family reports, which are often biased

due to lack of insight into deficits or the stress of the situation. Nevertheless, the GCS is

74

typically completed at emergency room admission, hence, one or more assessment

modalities may be compromised due to injury. The GCS and PTA duration have some

overiapping limitations, but mostly their problems are unique. Hence, the GCS and

duration of PTA could be used to complement one another, but research comparing their

cumulative abilities for assessing severity and predicting outcome is needed.

Studies examining the relationship between CHI severity and cognitive outcome

Mandleberg and Brooks (1975) investigated the "natural course of recovery" of

cognitive functions as measured by the WAIS (Wechsler, 1955). These investigators

conducted serial testing with (n = 48) head-injured patients and compared their results to

"comparison individuals" (n = 40) serial testing results. Mandleberg and Brooks' CHI

participants were unrepresentative of the CHI population because they were admitted to

the Institute of Neurological Sciences, Southem General Hospital, Glasgow after being

preselected for possible neurosurgical intervention. Specifically, forty out of forty-eight

CHI participants were positive for skull fracture, twenty-nine had a motor abnormality,

and seventeen hematoma. This limits the generalizability of the findings to only severe

CHI individuals with similar medical complications. In addition, Mandleberg and

Brooks' "comparison individuals" included (n = 13) individuals referred for vocational

guidance and (n = 27) "neurotics" referred for treatment. Mandleberg and Brooks

reported that these "comparison individuals" had histories of alcoholism, dmg abuse,

and/or psychiatric problems. However, the authors did not state whether the head-injured

participants were screened for these problems or history of head trauma and/or

75

neurological problems. This lack of control of the participants used in both groups

potentially adds error variance, obscures findings, and renders any found results

questionable because these conditions are known to affect performance on

neuropsychological assessment.

Mandleberg and Brooks reported that practice effects did not affect the three-year

WAIS scores from their head-injured group. These authors also found that Performance

subtests were more deteriorated during the earlier stages of recovery and take longer to

recover as compared to the Verbal subtests. However, they added that "the cognitive

abilities of the head-injured patients as a group eventually retumed to normal levels,

despite the severity of their injuries as measured by duration of PTA" (p. 1125). Overall,

these authors concluded that the duration of PTA does not predict long-term cognitive

outcome.

In a two-study investigation, Mandleberg (1976) looked at the relationship

between PTA duration and cognitive functioning with the time of injury treated as "an

important independent variable." In the beginning of the article, Mandleberg admitted

that his subjects were unrepresentative of the CHI population because they were admitted

to the Institute of Neurological Sciences, Southem General Hospital, Glasgow after being

preselected for neurosurgical intervention. His participants were more severely impaired

and they had a "disproportionate number with intracranial hematoma." This limits the

generalizability of the findings to only severe CHI individuals with similar medical

complications. Moreover, inclusion/exclusion criteria regarding a history of previous

head trauma, alcoholism, dmg abuse, psychiatric problems, and/or neurological problems

76

was lacking for both study one and two. This lack of control over the participants used in

both studies potentially adds error variance, obscures findings, and renders any found

results questionable. Nevertheless, study one included (n = 51) CHI participants,

subdivided by their PTA duration into subgroups that completed serial testing on the

WAIS. Study two included (n = 98) "non-systematic retumees" whose earliest WAIS

data was analyzed to control for practice effects.

In both studies, Mandleberg found that VIQ abilities were related to PTA duration

only at three months, while PIQ was related to PTA duration at three and six montiis. In

other words, performance abilities were slower to recover. Mandleberg reported that

despite PTA level, all participants' IQ levels were indistinguishable after six months. He

also indicated that practice effects were not found in the second investigation. Finally,

Mandleberg concluded that PTA was not related to long-term cognitive outcome and

added that these findings were consistent with his earlier findings.

Levin et al. (1979) studied the relationship between indices of severity and

outcome. Their subjects (N = 27) were selected from over 700 CHI cases referred for

neurosurgery to find the most severe cases that had recovered to a testable level.

Obviously, these participants are a small subset of the CHI population and this limits the

generalizability of any findings. Participants were excluded with a history of head injury,

alcoholism, dmg abuse, or neuropsychiatric disorder, and if they were under 16 or over

50 years-old. Severity of injury was determined by GCS and duration of coma, and only

patients with GCS scores of eight or less were included in the study. Measures used in

this investigation included the GCS, Glasgow Outcome Scale (GOS), Brief Psychiatric

77

Ratmg Scale, WAIS, a "selective reminding technique" described as an unconventional

method to assess memory, Multilingual Aphasia Examination, Controlled Word

Association, Token Test, Auditory Comprehension of Words and Phrases, Reading

Comprehension, and selected subtests from the Neurosensory Center Comprehensive

Examination for Aphasia.

Levin et al. concluded that the relationship between severity level and outcome

was "not invariable." Nevertheless, level of outcome, as determined by the GOS, was

significantly related to level of cognitive deficit. These authors added that, compared to

Mandleberg and Brooks (1975), their patients demonstrated heterogeneity of intellectual

recovery following CHI. Specifically, they found that marked intellectual impairment, as

measured by the WAIS, was present after intervals as long as three years. In summary,

they stated that neuropsychological assessments should be conducted at numerous points

over an extended period to evaluate treatment protocols for CHI patients.

Brooks et al. (1980) acknowledged the discrepancies in the research literature

regarding measures of severity level and cognitive outcome. Therefore, they hoped to

clear up these discrepancies by examining a number of severity indices in relation to

cognitive outcome across the domains of intellectual/cognitive functioning, memory,

language, and visuospatial/constmctive tasks. Their participants (N = 89) were those

referred to the Neurosurgical Unit of the Institute of Neurological Sciences, Glascow for

"operable lesions such as intracranial haematoma." This is a small subset of all CHI

patients. Hence, the generalizability of any findings is limited to only severe CHI

patients with similar medical maladies. Moreover, inclusion/exclusion criteria regarding

78

a history of head trauma, alcoholism, dmg abuse, psychiatric problems, and/or neurologic

problems were lacking. This lack of control over the participants used in the study

potentially adds error variance, obscures findings, and renders any found results

questionable. Brooks et al.'s severity measures of brain damage included coma duration

as measured by the GCS, duration of PTA, presence of hematoma, skull fracture, and age

(p.530). All participants' inteUigence was reportedly measured using the Raven's

Progressive Matrices and the Mill Hill Vocabulary Scale. Other measures included part

of the Logical Memory subtest, Story 1 of Form 1, from the WMS, the Inglis Paired

Associate Learning test, Rey Picture, "three measures of word fluency," Part 5 of the De-

Renzi and Vignolo Token Test, a copy condition of the Rey Picture, and the WAIS Block

Design Subtest.

Brooks et al. reported that all one-way ANOVAS between severity level, as

determined by the GCS, and cognitive outcome measures were insignificant.

Specifically, they stated that "there was no statistically significant influence of coma

duration on later cognitive outcome" (p. 531). However, increased PTA duration was

significantly related to the part of the Logical Memory subtest investigated. Associate

Leaming (only the delayed portion), one word fluency condition out of three (the letter j),

and the Rey Picture (immediate and delay conditions only). Overall, PTA duration was

the only measure of brain severity that had a consistent relationship on a small subset of

the neuropsychological measures. Finally, it should be noted that these authors

completed at least seventy ANOVAS, but no mention was made for correction of

79

familywise error rate. This could result in spuriously finding results that are not tmly

statistically significant.

Alexandre et al. (1983) investigated the possibility of predicting cognitive

outcome in individual patients. Their participants (N = 100) were unrepresentative of the

entire range of CHI individuals because they were all referred for neurosurgical

interventions for the removal of extradural clot, acute subdural hematoma, or

combination of hematoma, cerebral contusion, and laceration. This limits the

generaUzability of their fmdings to only severe CHI patients with similar neurological

problems; these individuals are only a subset of the possible range of all CHI sufferers.

Moreover, inclusion/exclusion criteria regarding a history of previous head trauma,

alcoholism, dmg abuse, psychiatric problems, and/or neurologic problems were lacking.

This lack of control over the participants used in the study potentially adds error variance,

obscures findings, and renders any found resuhs questionable. All participants were

evaluated with the GCS, and PTA duration was determined for each patient. Participants

were divided into groups using an unorthodox system based upon initial performance on

the neuropsychological measures of this experiment to determine severity level.

Specifically, participants completed the Wechsler-Bellevue Form 1, Benton's test of

visual memory, Rey's test for verbal leaming, Corsi's test for spatial memory, Kimura's

test for recognition memory, and the Token test. Assessments were conducted once

participants cleared from PTA, and then their severity level was determined by their

performance on the above measures. All participants were assessed again at one and two

year intervals.

80

The GCS was reliable in predicting the outcome of survival, but the GCS was not

accurate for patients with initial GCS scores of five, six, or seven. Moreover, no

significant relationships were found between PTA duration and cognitive performance.

However, the failure to find significance could be related to the unusual severity level

classification system employed. Overall, using such a system limits the ability to

compare this investigation's results to other research and places question on their

findings.

Paniak et al. (1992) wanted to discover the generalizability of the WAIS findings

to the WAIS-R. Specifically, they wanted to discover if PIQ was more sensitive than

VIQ to CHI effects, if left-hemisphere mass lesions resulted in lower VIQ than PIQ, if

right-hemisphere mass lesions resulted in lower PIQ than VIQ or bilateral mass lesions.

Additionally, they wanted to investigate whether severity was related to WAIS-R

performance, and if CHI subjects produced greater intersubtest scatter than normals.

Their participants were (N = 71) all right-handed aduhs, several months to years post­

insult that were more severely impaired than usual given the "clinical nature" of their

referrals. Using an unrepresentative sample limits the generalizability of these findings

to only those individuals with similar circumstances. Duration of PTA was available for

some subjects (n = 63) as was CT scan data (n = 57). However, these authors

acknowledged that CT scanning often fails to detect lesions that MRI might find.

Paniak et al. found that Digit Symbol was the only WAIS-R subtest that

correlated with severity of injury, as determined by PTA duration. These authors also

reported that there is little practical applicability for VIQ/PIQ discrepancies for individual

81

CHI patients. Only 27% produced a VIQ that exceeded a PIQ by more tiian 10 points

although these were all severe CHI patients. Furthermore, they stated that contrary to

"clinical lore," there was no evidence of greater WAIS-R subtest scatter in the CHI

sample versus the WAIS-R standardization sample. These authors also failed to find

VIQ/PIQ or Vocabulary-Block Design discrepancies in CHI patients with lateralized

mass lesion. Overall, Paniak et al. concluded that only some WAIS findings can be

generalized to the WAIS-R.

Paniak et al. admitted that CT scan procedures, with limited resolution, might

have missed diffuse lesions that generally characterize CHI. If this occurred, then

participants would be incorrectly classified by severity level and this would add error

variance to the results and obscure findings. No GCS scores were reported, some PTA

data was missing, as was some CT scan data all used to document the CHI severity. This

again could result in incorrect classifications and cause problems as discussed above.

Moreover, the average time from injury to assessment was 504 days, but more

surprisingly, the standard deviation was almost three years. Hence, there was wide

variability in recovery time before the first assessment; this was not a study designed to

longitudinally look at outcome. Furthermore, some investigators have reported that

longer time from injury until IQ assessment resuhs in recovery of cognitive deficits to

normal levels regardless of injury severity (Mandleberg, 1976; Mandleberg & Brooks,

1975). Paniak et al.'s assessment of CHI participants' so long after injury could have

contributed to their failure to find significant results between severity level and cognitive

outcome. These authors also did not list any inclusion/exclusion criteria screening for

82

history of alcohol or dmg abuse, previous head trauma, psychiatric problems, or

neurologic conditions. Missing data, large variability between insult and first assessment,

lack of exclusion for conditions known to effect test performance all confound

assessment results and places questions on any significant findings.

Although it is well established that CHI patients have deficits on a wide range of

cognitive tasks, difficulties with memory and attention/concentration are two of the most

commonly reported problems following CHI (Bennett-Levy, 1984; Geffen, Butterworth,

Forrester, & Geffen, 1994; Ponsford and Kinsella, 1992). Therefore, it seems clear that a

literature review on the relationship between severity level and cognitive deficits would

be incomplete if an examination of the research regarding attention/concentration and/or

memory, and severity level was not conducted.

In order to ascertain the relationship between severity level and recognition

memory following CHI, Brooks (1974) tested (n = 34) aduUs on a recognition memory

procedure that involved the identification of eight recurring shapes out of a series of 160.

In addition, he used a control group consisting of orthopedic outpatients undergoing

rehabilitation for injuries of the lower limbs. Brooks did not report whether any

participants were excluded for a history of previous head trauma, dmg abuse, alcoholism,

neurological, and/or neuropsychiatric problems. Failure to exclude these individuals

from either group can potentially add error variance to the results and obscure findings.

All head-injured participants were tested after transfer to the Division of Neurosurgery at

the Institute of Neurological Sciences, Glasgow. Therefore, these individuals are a

distinct subgroup of the total CHI population and the results of this investigation have

83

limited generalizability. Brooks reported that all participants were tested only after they

were fully out of PTA and had received a neurological examination before memory

testing. The Continuous Recognition Test by Kimura (as cited in Brooks, 1974) was

given to all participants to examine the amount of correct responses, false positives, and

false negatives by group.

According to Brooks, significant differences were found between the CHI group

and the control group on the total amount correct, number of false negatives, but not on

the amount of false positives. Furthermore, duration of PTA was correlated significantly

with the amount correct and the number of false positives. This is surprising because no

significant correlation was found between amount of false negatives and severity,

although the amount of false negatives was significant by group. Brooks' resuhs were

also unexpected since there was not a significant difference by group on false positives.

Nevertheless, an examination within the head-injured group for false positives (moderate-

severe neurologically impaired participants versus mild neurologically impaired

participants) was also significant. Specifically, the more severely injured participants

provided more false positives. However, Brooks wamed that when age is taken into

account, the relationship between PTA and memory deficits becomes less clear because

PTA duration bears a stronger relationship with memory in older (over 30 years-old) than

younger (15-30 years-old) participants. Overall, Brooks concluded that there was no

association between memory and persisting neurological signs or skull fracture, but

added that fijrther work was needed to clarify these findings.

84

Brooks (1976) attempted to further his previous work and clarify tiie relationship

between severity of injury, severity of persisting neurological signs, skull fracture, time

post-injury, age, and memory performance. The focus here will be on injury severity,

age, and memory performance. His CHI group consisted of (N = 82) patients that had

PTA duration of at least two days. Part of his CHI group (n = 30) were neurosurgical

cases while the others (n = 52) were referred for examination of cognitive recovery

following CHI. The author stated that these two subgroups did not significantly differ by

memory test performance. Hence, their data was combined. Control participants were (n

= 34) orthopedic patients with "primarily fractures of the lower limbs." Brooks clearly

stated that no member of the control group had suffered a head injury. Nevertheless, the

author did not indicate whether CHI participants were excluded for previous head injury,

dmg abuse history, alcoholism, neurological conditions, and/or psychiatric problems.

Moreover, Brooks did not report whether the control participants were excluded for

history of dmg abuse, alcoholism, neurologic conditions, and/or psychiatric problems.

Failure to exclude these individuals from either group can potentially add error variance

to the results and obscure findings. Brooks indicated that memory was tested on the

WMS "a scale that is far from ideal." Specifically, he did not agree that the Memory

Quotient produced by the WMS was efficacious, but he added that few clinical batteries

of memory tests were available.

Brooks reported that CHI participants performed significantly worse than controls

on six of the eight subtests. Specifically, the groups did not differ on Mental Control

(errors) and Digits Forward. To examine the effects of severity level on WMS

85

performance, Brooks subdivided the CHI group into subgroups based upon duration of

PTA. Specifically, his groups were PTA of (1) seven days or less (n = 14), (2) PTA

between eight and fourteen days (n = 14), (3) PTA of fifteen to twenty-eight days (n =

22), and (4) PTA greater than twenty-nine days (n = 32). Brooks did not find

significance between the CHI subgroups on Information, Orientation, Mental Control

(errors and time), Digits Forward, Digits Backwards, and Visual Reproduction.

However, CHI subgroup 1 performed significantly better than subgroups 3 and 4 on

Logical Memory (LM). Therefore, Brooks reported a "significant association" between

LM and PTA duration with a "threshold" above PTA of approximately four weeks, after

which PTA duration was less important. On Associate Leaming, (AL) CHI subgroup 1

significantly differed from 4. Brooks' resuhs did not support that longer PTA duration

detrimentally affects older participants' memory performance significantly more than

younger participants'. Overall, Brooks' findings were supported that severe CHI

individuals demonstrate marked memory deficits many months post-injury.

Gronwall and Wrightson (1981) examined the relationship between the PTA

duration and performance on memory and attention tests in two experiments.

Specifically, (N = 71) participants for experiment one were between 17 and 30 years-old.

They were administered the WMS, the PASAT, and the Quick Test (QT). Participants

were excluded for skull fracture, intracranial hematoma, "localising neurological signs, or

other complications." However, it was not reported whether participants were excluded

for a history of previous head trauma, alcoholism, dmg abuse, or psychiatric problems.

Failure to exclude these conditions can add additional error variance and obscure results

86

because these conditions can detrimentally affect neuropsychological testing

performance. Length of PTA and PASAT performance was used to determine

participants' severity level. Specifically, CHI subgroups were formed for PTA duration

of less than one hour (n = 20) [Mild]; see Appendix B), PTA duration of one to twenty-

four hours (n = 38 [Moderate]), and PTA of greater than one day (n = 13 [Severe]). The

classification of participants by time on the PASAT and PTA duration is unusual but was

based upon "grades." Specifically, Gronwall and Wrightson assigned participants into

three "grades" based upon their time to completion on the PASAT. Specifically, grade 1

was comprised of individuals who performed the task in less than 3.5 seconds (PTA < 1

hour), grade 2 between 3.6 and 5.5 seconds (PTA 1-24 hours), and grade 3 greater tiian

5.6 seconds (PTA > Iday). For further information on his classification strategy, the

interested reader is referred to Gronwall and Wrightson (1981) page 890.

Gronwall and Wrightson failed to find significance on WMS Mental Quotient

(MQ) between the mild CHI group versus moderate CHI group, and mild CHI group

versus severe CHI group. However, the MQ difference between the moderate CHI group

versus severe CHI group was significant. Overall, these authors indicated that the MQ

had obscured findings. They also submitted all moderate and severe CHI participants' (n

= 51) assessment data to a "Varimax rotated factor solution" and three factors emerged.

Factor 1 reportedly loaded on the PASAT and Mental Control (MC) subtest of the WMS,

Factor 2 on the WMS Associate Leaming (AL) subtest, and Factor 3 on the QT as well as

Information and Orientation subtests from the WMS. The duration of PTA's loadings on

all three factors was reportedly low. Overall, PTA duration was moderately, but

87

significanfly correlated with performance on the PASAT (L=.30), MC (L=.35) , and AL (r

=.28). Overall, Gronwall and Wrightson stated that he was surprised with the low

loading of PTA on Factor 2 (reportedly a leaming and memory factor), hence, he ran a

second study to further examine this relationship with additional memory assessment

devices.

Gronwall and Wrightson's second investigation included CHI participants (n =

20) whose CHI was in the moderate to severe range as determined by duration of PTA

(range 2-56 hours) and that met the same age and injury inclusion criteria of experiment

one. These investigators also included a control group (n = 30) in the second

investigation comprised of hospital staff and university students. No inclusion/exclusion

criteria were reported. Therefore, they did not screen either group for a history of

previous head injury, dmg abuse, alcoholism, neurological conditions, or psychiatric

disorders. Again, these conditions could have added error variance to his findings as

discussed in experiment one, and obscured differences by group. In the second study, all

participants completed the PASAT, the Selective Reminding Task, and the Visual

Sequential Memory subtest of the Illinois Test of Psycholinguistic Abilities (a full

description of these instmments is provided in the article).

Grownwall and Wrightson found significant correlations between the PASAT and

PTA duration, but severity level was not associated with performance on the Visual

Sequential Memory test. These authors reported that the Selective Reminding Task

yields three measures, total number correct, cumulative number of words in memory

storage, and consistent retrieval. Severity level, as measured by PTA duration, was

88

significantly related to the total amount correct for the CHI participants. No statistically

significant group mean differences were found between the control group and the CHI

group on the cumulative number of words in memory storage. Nevertheless, PTA

duration was correlated significantly with cumulative number of words in memory

storage for the CHI group. Severity level, as measured by PTA duration, was not

significantly related to consistent retrieval for the CHI participants. Based upon tiie

results of both investigations, these investigators stated that it appeared tiiat PTA duration

did not predict degree of impairment on memory tasks. However, these authors indicated

that the ability to place information into long-term memory store was related to either

PTA duration or time between injury and testing, and that PTA duration did predict

impairment in the ability to retrieve information from memory once it was stored.

Bennett-Levy (1984) noted that investigations examining the relationship between

severity level and cognitive outcome had produced varying results. Therefore, he wanted

to examine the relationship between CHI severity and "memory and related functions"

two to five years post-insult. Specifically, he attempted to discover if there was a

"threshold of brain damage" near PTA duration of three weeks, above which patients

would incur long-term cognitive deficits. Head-injured participants (n = 39) were

individuals that sustained head-injuries between 1976-1978. They were all between 17

and 35 years-old, had PTA duration of greater than one week "as determined by the

neurosurgeons," and an estimated pre-morbid IQ of 80 or greater. Selecting only CHI

participants between the age of 17 and 35 limits this investigation's generalizability to

only young CHI individuals. Head-injured participants were excluded for a history of

89

alcoholism, dmg abuse, or a neurological disorder. Nevertheless, Bennett-Levy did not

exclude participants for a history of previous head trauma or neuropsychiatric problems.

Control participants (n = 32) were orthopedic patients "under assessment for surgical

intervention from limb injuries." Control participants were excluded if they had a history

of head trauma or epilepsy, but the authors did not report excluding these individuals for

a history of alcoholism, dmg abuse, or psychiatric problems. Failure to exclude

participants from either group with conditions known to detrimentally affect

neuropsychological assessment performance could add error variance to the results and

obscure findings.

The control group was assessed in the hospital while the CHI participants were

tested between two to five years post-injury. Head-injured participants were divided into

two subgroups based upon their PTA duration. The first group (CHIl) was comprised of

participants whose PTA duration was between one and three weeks, the second group

(CHI2) had PTA duration of more than three weeks. All participants' IQ was estimated

with the National Adult Reading Test (NART) and the Schonell Graded Word Reading

Test. Furthermore, all participants completed the Logical Memory (LM) subtest of the

WMS, Face-Naming Leaming Test, Word Recognition (an admittedly unpublished test).

Keeping Track, Famous Personalities Test, Figure-Ground Test of Discrimination,

Mooney Faces, Recurring Faces Test, Rey-Osterrieth Complex Figure Test, Speed of

Information Processing test adapted from the British Abilities Scale, and the Subjective

Memory Questionnaire (these relatively unknown instmments are fully described in the

article pp. 288-290).

90

Bennett-Levy reported that CHI2 group performed significantly worse on the LM

subtest and the Keeping Track test than the control group, whereas tiie CHIl group did

not. Significant differences by severity level, as determined by PTA duration, were only

evident for LM (delayed recall & percent forgetting conditions), the Keeping Track test,

and the Rey-Osterrieth (recall condition). No other significant differences were obtained

between CHIl versus CHI2 on any other assessment devices or conditions. In other

words, severity level was not related to performance on other measures except the ones

noted above. Moreover, no significant differences were produced by group (CHIl, CHI2,

and control) on Word Recognition, Face-Name Leaming Test, and the Famous

Personalities Test. Furthermore, he stated that severity of injury, as determined by PTA

duration, was not related to performance on the Speed of Processing Task across any of

the five conditions. Bennett-Levy concluded that "the increasing task demands in no way

differentially affected the head-injured patients since both the patient groups' and the

controls' mean times increased by constant amounts" (p. 293). Surprisingly, age was

only significantly correlated with one test, the Famous Personality Test, and the older

patients performed better. Therefore, because of the lack of significant correlations

between age and Bennett-Levy's measures combined with the finding that older subjects

performed better on the Famous Personality Test, this author concluded that, within the

age range of this investigation, there was no evidence that older patients are affected

more by head injury than younger patients. In closing, this Bennett-Levy stated that the

results of his investigation do not support that individuals classified as "very severe" by

PTA duration suffer permanent cognitive impairment.

91

Vilkki et al. (1988) noted that PTA duration was found to be related to memory

performance in some investigations, whereas coma duration had even less proven

consistency in relation to the severity of residual memory deficits. Therefore, these

investigators wanted to demonstrate that CHI creates a memory deficit that is not

secondary to perceptual or conceptual analysis of the to be remembered material and that

is related to coma duration in patients without intracranial hematoma. Their participants

(N = 51) were patients admitted to the Neurosurgical Clinic of Helsinki following CHI

that had also been comatose immediately after injury for a measurable period. They

excluded participants who were confused, disoriented, or unable to cooperate sufficiently

during the psychological examination as well as participants who underwent operation

for intracranial hematoma. Vilkki et al. admitted that their group of participants was a

"biased sample." Specifically, participants were only those referred to the neurosurgical

unit and who had experienced a loss of consciousness (LOC); not all CHI patients

experience a LOC. Furthermore, these authors did not report that the participants were

out of the acute or post-acute phase before testing ensued. In fact, they stated that some

participants were given a neuropsychological assessment only five days following CHI

with LOC. These investigators also did not exclude CHI participants with a history of

previous head trauma, alcoholism, dmg abuse, neurological conditions, or psychiatric

problems. Overall, because of the limited generalizability of their findings, lack of

documentation that their participants were out of the acute or post-acute phases, and

failure to exclude participants with conditions known to detrimentally affect

neuropsychological test performance, their results should be interpreted cautiously.

92

Vilkki et al.'s participants were subdivided into two groups. The Short Coma

Group (SCG; n = 25) consisted of participants who experienced a LOC of less tiian six

hours. The Long Coma Group (LCG; n = 26) was comprised of individuals whose LOC

was greater than six hours. All participants completed the WAIS subtests of Digit Span,

Similarities, and Block Design as well as free and cued recall conditions of Similarities.

Furthermore, all participants completed the BVRT with homogeneous interference.

Coma duration was not found to be related to performance on the WAIS subtests

or the Immediate Reproduction of the BVRT. This was evident by the lack of

significance between subgroups on these measures. However, the LCG was significantly

inferior to the SCG on the free and cued recall condition of Similarities, and the

forgetting percentage of the BVRT. Overall, Vilkki et al. concluded that the effects of

CHI on verbal and visual memory was related to coma duration, but not secondary to

deficient conceptual or perceptual analysis of the material to be remembered.

Crosson et al. (1988) investigated CHI patients' memory performance on the

CVLT because it provides insight into the qualitative aspects of memory impairment.

Moreover, they were concemed with examining the relationship between severity of

injury, as determined by coma duration and length of PTA, and performance on the

CVLT. Crosson et al.'s CHI participants (n = 33) were individuals whose medical

records indicated that had experienced a "blunt-head trauma," had a documented period

of unconsciousness and/or CT scan indicating brain damage, and that had completed the

CVLT. Nevertheless, these authors stated that they were unable to establish an estimate

of coma or PTA duration for only one case, but that they had established this for all of the

93

other twenty-four subjects (p. 756). Therefore, it is unclear whether their CHI group was

comprised of thirty-three or twenty-four CHI participants. All CHI participants in this

investigation were reportedly in the severely injured range. Head-injured participants

were excluded if they were abusing dmgs or alcohol at the time of the investigation

and/or if they had experienced a penetrating head wound.

The control group was made up of "neurologically normal adults" that were paid

participants (except for five participants) obtained through advertisement. However,

Crosson et al. did not report the sample size of the control group in the body of tiie

article, but stated that the size was thirty-three in the abstt-act. Crosson et al.'s only

exclusion criterion for the control group was history of a central nervous system disorder

before admission. Therefore, they did not exclude control participants with a history of

alcoholism, dmg abuse, or psychiatric problems. Failure to exclude individuals from

either group for conditions known to detrimentally affect neuropsychological assessment

performance potentially adds error variance to the results, obscures findings, and places

question on any significant results.

Crosson and colleagues found that CHI participants demonstrated lower scores

than controls across leaming trials. Moreover, there were significant differences between

groups and leaming trials for correct responses. Furthermore, they found that head-

injured participants produced more intmsions and a lower level of semantic cluster

responses. Crosson et al. noted that previous investigations had demonstrated

relationships between indices of injury severity (e.g., length of coma, or length of PTA)

and performance on verbal memory measures. Overall, these authors reported using

94

Spearman's rho and finding a consistent absence of significant relationship between

CVLT performance and severity level.

Haut and Shutty (1992) reported that recent advances in cognitive psychology

made it possible to examine specific memory processes, and successftilly delineate

deficits. They added that although previous sttidies have demonstt-ated deficits in the rate

of leaming of normal contt-ols versus CHI patients, they "are aware of no other sttidies

that have explored within-group differences in leaming and memory after CHI" (p. 56).

Hence, the purposes of their investigation was to further explore subgroup differences in

verbal leaming following a CHI, and to explore differences in memory performance

according to indices of prior functioning and severity level. Their subjects (N = 70) were

men (n = 54) and women (n = 16) who were referred for a neuropsychological

evaluation. They classified these participants' severity level with the GCS. No other

inclusion/exclusion criteria were reported. Finally, all participants completed the WAIS-

R, WMS-R Logical Memory subtest, and CVLT.

Their results revealed three distinct pattems of verbal leaming and memory on the

CVLT following CHI. Haut and Shutty reported that these pattems differed by leaming

rate, amount leamed, amount recalled following a delay, and retroactive interference.

Moreover, they indicated that the groups demonstrated "clear differences" on concurrent

cognitive measures (e.g., LM subtest). Specifically, they reported that poor CVLT

performance was associated with lower IQ, attention, and semantic memory. However,

they added that they found a lack of differentiation across subgroups on variables

believed to be associated with severity level. For instance, they failed to find a

95

significant relationship between severity of injury, as determined by tiie GCS, and CVLT

or WAIS-R performance. Therefore, they concluded that the GCS was not sophisticated

enough to differentiate "long-term higher level variations in cognitive performance." In

other words, despite different severity levels as determined by the GCS, it did not

discriminate between subgroups' WAIS-R or CVLT performances. Overall, these

authors called for fiittire research that combines the GCS with other variables, such as the

duration of PTA, to sttidy CHI severity and pattems of memory performance.

This study was important at identifying leaming pattems in CHI sufferers.

However, it had some methodological difficulties. First, Haut and Shutty did not report

whether participants were excluded with a history of alcoholism, dmg abuse, previous

head trauma, psychiatric problem, or neurological disorders. These conditions are known

to adversely affect neuropsychological assessment performance, especially memory

testing. Hence, by not excluding participants for these conditions can potentially add

error variance and obscure findings. Haut and Shutty also did not classify their subjects

according to the standardized and accepted GCS criteria. Specifically, they classified

individuals as moderate if they had a GCS score of 13, but GCS scores of 13 are

considered mild head injuries according to standard GCS criteria (see Appendix B;

Lezak, 1995). These authors stated in the conclusion that the GCS severity ratings failed

to differentiate CVLT performance, but they did not follow standard classification

criteria. Therefore, this statement seems premature. Overall, by failing to follow

standardized classification criteria their findings cannot be compared to other

investigations.

96

Reid and Kelly (1993) assessed the vahdity of the WMS-R in a group of CHI

patients. Specifically, they examined the relationship between WMS-R performance and

CHI, severity of injury, and day-to-day memory performance. Their participants

included (n = 20) inpatients from a rehabilitation unit and "neurologically normal"

volunteer participants (n = 20). Inclusion criteria specified that the participants were

between 16-74 years-old, with an eighth grade education, no history of a diagnosed

leaming disability, psychiatric problems, and had reportedly not abused alcohol or dmgs

for at least one month pre-testing. No significant differences were found between the

groups on demographic variables. Their admission GCS score and amount of PTA as

measured by the GOAT determined all CHI patients' severity levels. Staff occupational

therapists completed the Inpatient Memory Impairment Scale (IMIS) for each CHI

participant to assess their day-to-day memory functioning. All participants completed a

full WMS-R. The authors hypothesized that CHI individuals would perform more poorly

on the WMS-R than controls. Moreover, they postulated that CHI patients would

evidence more deficits on measures of long-term memory (LTM) than controls. Finally,

they also expected that more severe injuries as determined by the GCS and duration of

PTA would perform more poorly on the WMS-R indices.

Reid and Kelly's (1993) results supported their contentions that contt-ols would

outperform CHI patients on overall memory performance and LTM. In particular, CHI

patients had difficulty with the delayed memory component of the WMS-R. However,

severity level, as determined by both the GCS and duration of PTA, reportedly was not

significantly related to WMS-R performance. Overall, Reid and Kelly found support for

97

the construct validity of the WMS-R by comparing scores on its indices to the MIS

ratings.

Reid and Kelly acknowledged that their sample sizes were small by group (n =

20) and that the generalizability of their findings was limited. These investigators also

included individuals in their sttidy if they had abused substances (i.e., alcohol and dmgs)

if it had been more than one month since they "abused" these substances. Overall, a

history of dmg and/or alcohol abuse has been associated with impaired performance on

neuropsychological assessment devices. Therefore, by not excluding all individuals with

a history of substance abuse and/or alcoholism these authors may be adding error

variance to their data and obscuring findings on the WMS-R overall and specifically in

relation to severity level.

Geffen et al. (1994) wanted to examine how the RAVLT performance varied by

severity level, as determined by the PTA duration, and compare the differences between

matched controls and CHI participants on individual RAVLT items. Their participants

included (n = 18) CHI patients referred for neuropsychological assessment and (n = 23)

matched controls. Specifically, they were matched on age, estimated IQ, education, and

occupational level. Estimates of full-scale IQ were computed with the NART for most of

the participants. No inclusion/exclusion criteria were provided. Failure to exclude

participants with a history of head trauma, alcoholism, dmg abuse, neurological

disorders, or psychiatric problems is problematic because these conditions are known to

affect neuropsychological test performance. Consequently, this may add error variance

to results and occlude findings. Ten of the CHI participants and all controls were tested

98

individually on Form 1 of the RAVLT, while the remaining CHI participants were

examined with an altemative form of the RAVLT, Form 4. It should also be noted that

the delayed-recall trial, trial eight, of the RAVLT was not administered to most of tiie

CHI participants. Therefore, this ttial was excluded from analyses.

Geffen et al. reported that a two-way ANOVA, Group by Trial, revealed

significant main effects for Group and Trial, but the interaction was not significant. They

stated that this indicated that the two groups leamed at a similar rate despite poorer

overall performance by CHI participants. A series of one-way ANOVAs for total recall

and the first seven trials, and the recognition measures revealed that the control group

significantly performed at a superior level. Moreover, the CHI participants demonsttated

significantly more retroactive interference, but did not differ from the conttols on

proactive interference. Pearson correlation coefficients were used to assess the effects of

injury severity on recall and recognition performance. Only trial seven was significantly

correlated with severity of injury, as determined by PTA duration. In other words, no

other RAVLT recall or recognition measures were correlated with CHI severity.

Haslam et al. (1994) examined the relationship between early GCS scores,

duration of coma or unconsciousness, PTA duration, the presence and evacuation of

hematoma, post-coma disturbance (PCD) on cognitive outcome. They defined PCD as

the period between a patient's emergence from coma and the regaining of continuous

day-to-day memory. These authors stated that there is considerable variation in

researchers' definitions of cognitive outcome. Therefore, they attempted to predict

outcome on two specific areas of cognitive deficit, recent memory functioning and speed

99

of information processing. Their sample (N = 57) was comprised of individuals over the

age of 15 with documented severe head injury by the GCS. Nevertheless, Haslam et al.

did not report excluding individuals with a history of previous head ttauma, alcoholism,

dmg abuse, neurological disease, or neuropsychiattic disttirbance. Failure to exclude

participants with these conditions known to detrimentally affect neuropsychological

assessment performance may add error variance to the resuhs and obscure findings.

Furthermore, they admittedly chose individuals for inclusion that had made good or

moderate outcomes, as measured by their GOS score, to complete cognitive assessment.

This inclusion hurts their generalizability because they cannot generalize results to those

who do not make do not have good or moderate recovery.

Following admission all subjects were reportedly assessed with these neurological

indices the GCS, Coma duration, CT scan. Surgery, Length of PTA, PCD, and nature of

the trauma (pp. 522-523). Twelve months following admission, these participants

underwent a cognitive assessment that included the PASAT, SDMT, RAVLT, and the

Vocabulary subtest of the WAIS-R as a measure of pre-morbid intelligence. Haslam et

al. conducted a principal-components analysis to reduce these cognitive assessment

variables to discrete factors. The first factor, PSYl, was reported to be primarily a

measure of recent memory while factor two, PSY2, was reported to be an index of speed

of information processing. Next, they transformed the PCD and PTA by a square root

transformation. Then, they conducted a stepwise regression to determine which

neurologic variable or combination of variables best-predicted scores on the cognitive

factors.

100

Haslam et al. reported that PTA and transformed PTA were significant predictors

of performance on PSYl (recent memory). However, the combination of ttansformed

PCD and presence of subarachnoid hemorrhage reportedly best predicted performance on

this factor. They also indicated that only two factors, tt-ansformed PTA and PCD were

significant predictors for performance on PSY2 (speed of information processing).

Overall, these authors concluded that the predictive power of PTA and PCD could be

significantly improved by controlling for their non-linearity.

Haslam et al. (1995) reported that previous investigations have varied

considerably in their ability to demonstrate consistent relationships between neurological

indices (e.g., PTA duration and the GCS) and cognitive variables for CHI patients.

Nevertheless, they investigated the relationship between early neurologic variables and

cognitive outcome in a sample of head-injured patients two years post-insult. Their head-

injured sample (N = 77) included forty participants from their original study (Haslam et

al., 1994). All participants had a documented history of severe head injury by the GCS.

However, in this investigation the authors excluded individuals with a history of previous

head trauma, psychiatric disturbance, or alcohol abuse. Haslam et al. admittedly had a

sample skewed towards those judged to have made a good recovery because they only

chose individuals for inclusion that had made good or moderate outcomes as determined

by the GOS to participate. This exclusion hurts their generalizability because they cannot

generalize results to those who do not make do not make this type of recovery.

Following hospital admission, all subjects were reportedly assessed with these

neurological indices the GCS, Coma duration, CT scan. Surgery, Length of PTA, PCD,

101

and nattire of the tt-auma. Twenty-four montiis post-admission, these participants

underwent a cognitive assessment that included the PASAT, SDMT, RAVLT, and the

Vocabulary subtest of the WAIS-R as a measure of pre-morbid intelligence. Haslam et

al. also performed a principal component factor analysis on the cognitive measures and a

square root tt-ansformation on two of the neurological indices, PTA duration and PCD.

These authors obtained three factors PSYFACl, reportedly a recent memory factor,

PSYFAC2, reportedly a factor tapping into information processing speed, and PSYFAC3,

reportedly a measure of vocabulary and education.

Haslam et al. reported that five neurological variables, tt-ansformed PCD, PTA,

ttansformed PTA, PCD, and coma duration were significantly related to PSYIFACI

(recent memory). Transformed PTA, ttansformed PCD, coma duration, and surgical

evacuation of exttadural hematoma were reported as significant predictors for

performance on the second factor PSYFAC2 (information processing speed). No

significant correlations were obtained for the third factor PSYFAC3 (vocabulary).

Overall, reportedly 20% more variance in prediction of cognitive outcome was accounted

for with ttansformed PTA and PCD above what was found with these same

untransformed variables.

Gupta and Ghai (1996) investigated memory and information processing capacity

impairment following CHI. Specifically, they examined the effect of severity, as

determined by PTA duration, on information processing and memory performance. Their

participants (N = 48) were all males between 17-50 years-old. They noted that all CHI

patients (n = 24) that participated were alert, oriented, and unequivocally out of PTA at

102

the time of testing as judged by the attending physician and psychologist. Duration of

PTA was determined by rettospective questioning. All patients reportedly suffered from

"simple" closed head injury, that is, none evidenced inttacranial hematoma, skull

fracture, or any other neurological complications. Exclusionary criteria for CHI patients

included history of alcoholism, cerebral disease, psychiattic ilhiess, and motor

disturbance or communication problem. A matched sample (n = 24) was obtained, but

the inclusion/exclusion criteria were not reported. Reportedly, information processing

was assessed by the PASAT, while verbal and non-verbal memory was tested three

subtests of the WMS, Digit Span, Associate Leaming, and Visual Reproduction. They

added that these WMS subtests were selected for their abilities to assess attention, verbal,

and visual memory respectively. All materials were in Hindi and presented in the same

order for all participants.

These authors reported that the CHI group performed significantly poorer across

all measures. They added that "deficient performance" by the CHI group on Digit Span

and on the PASAT might have been due to reduced encoding capacity, poor

attention/concenttation, or centtal executive dysfunctioning. To address whether severity

of injury played a role in compromised performance, the authors divided their subjects

into two groups, mild head injury (n = 12) and severe head injury (n = 12). Finally,

Gupta and Ghai found no significant differences on severity by group except for Visual

Reproduction in which the severe CHI group performed worse.

Although Gupta and Ghai covered their bases in regards to inclusion/exclusion

criteria with the CHI group, they did not exclude control participants for a history of head

103

ttauma, alcoholism, dmg abuse, neurological conditions, or psychiattic problems. Failure

to exclude individuals with conditions known to adversely affect neuropsychological

assessment performance potentially adds error variance, obscures findings, and makes

any found significant results questionable. Moreover, they only demonsttated what was

already known, that is that CHI individuals perform significantly worse than matched

controls on cognitive measures. These authors also stated the obvious in their

conclusion, that many different independent systems may be involved in visual and

verbal memory, attention/concenttation, and executive fiinctioning. Finally, they failed

to demonstrate a relationship between severity level and these "independent systems" that

affect cognitive outcome.

Loken et al. (1995) assessed severe CHI patients' ability to sustain attention on a

visual continuous performance task "of sufficient ease and duration to assess stability of

performance over time" (p. 593) and compared these performances to severity of injury

as measured by duration of PTA. These investigators used two groups, a severe CHI

group (n = 20) referred from a rehabilitation hospital once they were out of the acute

phase of injury (mean time since injury = 2.9 months), and a conttol group (n = 20)

recmited from the university subject pool and matched for age, education, and gender.

Duration of PTA and severity of injury was determined with the GOAT for the CHI

group. Head injury participants were excluded for history of former head injury or

neurologic insult, but not for alcoholism, substance abuse, or psychiatric history. Control

participants were excluded for previous neurologic insult or psychiatric illness, but not

104

for dmg abuse or alcoholism. All participants were administered a computerized version

of a visual continuous performance task.

Loken et al. (1995) reported that their investigation demonsttated that attentional

performance following a severe CHI is impaired in terms of "sustained allocation of

attention over time" and overall vigilance. However, the CHI and conttol groups did not

differ on the amount of errors of commission or omission. These authors also did not

find a significant relationship between CPT performance and severity of injury as

determined by duration of PTA. However, these investigators did not exclude

participants across groups with the same exclusion criteria. Moreover, they included a

healthy conttol group comprised of currently enrolled college students. This is

potentially problematic because of non-equivalence of groups. In other words, the two

groups are qualitatively and perhaps quantitatively different. Overall, failure to follow

standard exclusionary criteria and potential non-equivalence of groups is problematic

because of the potential to add error variance to findings and obscures results.

In summary, review of this body of literature has revealed great inconsistencies

between investigations' findings. The GCS and PTA duration have been related to

measures of cognitive outcome in some investigations, but not others. The lack of

consistent resuhs could be due to the fact that most investigations reviewed failed to

exclude individuals for conditions known to adversely affect neuropsychological test

performance (Alexandre et al., 1983; Bennett-Levy, 1984; Brooks, 1974; Brooks et al.,

1980; Crosson et al, 1988; Geffen et al., 1994; Gupta & Ghai, 1996; Haslam et al., 1994;

Haut & Shutty, 1992; Loken et al, 1995; Mandleberg, 1976; Mandleberg & Brooks,

105

1975; Paniak et al, 1992; Reid & Kelly, 1993; Vilkki et al, 1988). Only one reviewed

investigation (Haslam et al. 1995) implemented exclusion criteria for a history of

previous head injury, neurological condition, psychiattic problems, alcoholism, and dmg

abuse. It is also difficuh to obtain consistent and comparable when there is wide

variation of measures used to assess cognitive outcome across studies (Levin et al,

1990). Another potential contributor to the variability in findings may be attributable to

some investigations' failure to use standardized severity classification criteria (e.g.,

Alexandre et al, 1983; Bennett-Levy, 1984; Haut & Shutty, 1992; Wrightson, 1981).

Failure to use standardized classification methods puts any obtained significance at

question, and limits the usefuhiess on any investigation because the results cannot be

compared across investigations. Finally, many reviewed studies had low sample size per

group (e.g., Gupta & Ghai, 1996), therefore, low power to detect relationships that may

have also contributed to the failure to obtain consistent results.

106

APPENDIX B

SEVERITY CLASSIFICATION CRITERIA

107

Posttraumatic Amnesia-Severitv Classification Criteria

PTA duration

<5 minutes 5-60 minutes 1-24 hours 1-7 days 1-4 weeks More than 4 weeks

Severity Classification

Very mild Mild Moderate Severe Very severe Exttemely severe

Glascow Coma Scale-Severity Classification Criteria

GCS Coma duration Severity Classification

>13 or <20 minutes Mild

9-12 or <6 hours of admission Moderate

<8 or >6 hours of admission Severe

Both the Posttraumatic Amnesia-Severity Classification Criteria and the Glascow Coma Scale-Severity Classification Criteria were adapted from Lezak (1995).

Lezak, M.D. (1995). Neuroosvcholopical Assessment. (3'^ ed.). New York: Oxford University Press.

108