ArticlePDF Available

Impairment of Expressive Behavior in Pediatric HIV-Infected Patients with Evidence of CNS Disease

Authors:

Abstract

Rated observations of videotapes were made of 16 variables representing 5 behavioral domains (task orientation, positive social-emotional, motor skills, expressive speech, and activity) on a sample of 83 HIV-infected children. Comparisons were made on the rated behaviors between children classified as either encephalopathic or nonencephalopathic. Analyses were conducted separately for infants (M age = 1.80 years) and older children (M age = 5.15 years). The nonencephalopathic infants exhibited higher activity levels and were superior in motor and verbal skills and showed more social and emotional responsiveness than did the encephalopathic group. The older nonencephalopathic children functioned in a more adaptive and appropriate manner than did the encephalopathic children in all domains of behavior. Independently made Q-sort ratings of behaviors during developmental testing were highly correlated with conceptually congruent ratings of the videotaped behaviors.
Journal of Pedialric Psychology. Vol 21. No. 3. 1996. pp. 379-4O0
Impairment of Expressive Behavior in Pediatric
HIV-Infected Patients with Evidence of CNS Disease
1
Howard A. Moss
2
and Pamela L. Wolters
National Cancer Institute and Medical Illness Counseling Center
Pirn Brouwers
National Cancer Institute
Michael L. Hendricks
National Cancer Institute and Medical Illness Counseling Center
Philip A. Pizzo
National
Cancer
Institute
Received August 8. 1993. accepted January 9, 1995
Rated observations of videotapes were made of
16
variables representing 5 be-
havioral domains (task orientation, positive social-emotional, motor skills, ex-
pressive speech, and activity) on a sample of 83 HIV-infected children. Compari-
sons were made on the rated behaviors between children classified as either
encephalopathic or nonencephalopathic. Analyses were conducted separately for
infants (M age = 1.80 years) and older children (M age = 5.15 years). The
nonencephalopathic infants exhibited higher activity levels and were superior in
motor and verbal skills and showed more social and emotional responsiveness
than did the encephalopathic group. The older nonencephalopathic children
functioned in a more adaptive and appropriate manner than did the encephalo-
pathic children in all domains of behavior. Independently made Q-sort ratings of
'We express our appreciation lo Dcbra El-Amin, Renee Smith, and Eugene Tassone for their assis-
tance with this research, and to Betsy Spiropoulos and Casey Dcsautcls for their assistance in
preparing this manuscript. Research for this report was carried out in part by Research Contract
NCI-CM-17529, awarded to the Medical Illness Counseing Center, Chevy Chase, Maryland.
2
AII correspondence should be addressed to Howard A. Moss, Pediatric Branch, National Cancer
Institute, Building 10, Room I3N240, 9000 Rockville Pike, Bethesda, Maryland 20892.
379
0146-8693/96/0600-0379109 W0
<0
1996 Plenum Puhltsliiiij Corpomwi
by guest on July 13, 2011jpepsy.oxfordjournals.orgDownloaded from
380 Moss, Wolters, Brouwers, Hendricks, and Pino
behaviors during developmental testing were highly correlated with conceptually
congruent ratings of the videotaped behaviors.
KEY WORDS: pedialric HIV-infected; encephalopathy; behavioral ratings; impaired expressive
behavior.
HIV disease in children is frequently associated with mild to severe encephalo-
pathy that can result in profound and pervasive cognitive, behavioral, and per-
sonality changes (Brouwers, Moss, Wolters, & Schmitt, 1994). Individuals differ
as to the number of functions that are affected, although typically impairments
appear to occur across a broad range of behavioral modalities (Armstrong,
Seidel, & Swales, 1993; Brouwers et al., 1994). The neurological bases for these
changes have not been established conclusively although there is evidence sug-
gesting multiple factors underlying the etiology and course of HIV-related en-
cephalopathy (Armstrong et al., 1993; Bel man, Brouwers, & Moss, 1992).
Possible causes for disease-related behavioral changes are direct effects such as
impact of the virus on the central nervous system (CNS), the release of neurotox-
ins in the CNS as the system's reaction to the infection, and structural changes in
the brain (based on imaging studies). Some of these adverse CNS conditions may
be transitory whereas others reflect permanent changes, such as basal ganglia
calcifications. In this regard, the type of biological evidence of CNS disease that
is present may be indicative of disease stage and the permanence, generality, and
severity of behavioral compromise. For example, extensive cognitive deficits and
aberrant psychological functioning in HIV have been found to be associated with
quinolinic acid in the cerebrospinal fluid (CSF) (Brouwers et al., 1993) and with
CT findings (Brouwers et al., 1995).
Most of the evidence of adverse behavioral outcome from HIV-related CNS
disease is based on impaired performance on cognitive tests. Furthermore, data
show that lowered cognitive scores can be reversed after a course of antiretroviral
treatment (Pizzo et al., 1988), which suggests that under certain circumstances
lowered IQ scores may be the result of some condition(s) interfering with "ex-
pression" of abilities and not necessarily because of any permanent CNS dam-
age.
Thus, the plasticity of behavior in response to treatment may provide some
clues to the nature and stage of the CNS disease process.
Comprehensive sampling of the range of behavior that may be effected by
HIV is important in order to establish the generality of the observed impairments
and to facilitate our understanding of the extent and manner in which underlying
CNS structures are compromised. Yet, documentation of those functions that
might be affected by HIV tend to be limited to IQ test results, probably because
of their greater availability and the routine usage of standard measures of intel-
ligence. On the other hand, assessment of social and emotional behavioral char-
by guest on July 13, 2011jpepsy.oxfordjournals.orgDownloaded from
Behavioral Impairments in Pediatric HIV Patients 381
acteristics tend to be underrepresented in depicting the effects of HIV on the
CNS,
apparently because of the dearth of appropriate methodologies and the
greater difficulty in obtaining systematic and objective data on these areas of
functioning. Thus, the lack of objective information on social and emotional
behaviors in HIV-infected children creates a potentially biased picture that over-
emphasizes deficits in cognitive functions as characterizing HIV-related enceph-
alopathy in children. The restricted behavioral range that has been evaluated in
pediatric HIV-infected patients furthermore appears to hinder comprehensive and
systematic brain-behavior analyses.
The same bias in focusing on cognitive deficits and neglecting behavioral
characteristics exists in studies of other childhood illnesses where there is CNS
involvement. Numerous studies have evaluated cognitive decrements among
children after receiving cranial radiation treatment and/or CNS chemotherapy for
acute lymphocytic leukemia (ALL) with little consideration given to possible
treatment-related changes in other modalities of behavior (Butler & Copeland,
1993).
Thus, again by omission, these studies of pediatric ALL patients has
contributed to the impression that this CNS treatment effect is limited to cogni-
tive deficits.
A few studies have examined aspects of behavior, other than cognitive
functioning, that seem to be adversely effected by HIV (Moss et al., 1994;
Wolters, Brouwers, Moss, & Pizzo, 1995). In one study children with HIV were
rated on a Q-sort behavioral rating procedure and those classified as encepha-
lopathic obtained significantly higher scores on rating scales measuring apathy,
nonsocial behavior, depression, and autistic-like behavior than those classified as
nonencephalopathic. In a study carried out by Wolters et al. (1995), expressive
language was more compromised than receptive language abilities in both en-
cephalopathic and nonencephalopathic children with this deficit in expressive
language being significantly greater among encephalopathic children. Additional
studies are needed to further document and elucidate both the nature and range of
impairments of expressive functioning among these patients, and to obtain more
objective measures of these behavioral changes.
The research reported here was designed to investigate the effects of HIV
encephalopathy in a range of additional behaviors. Children with severe HIV-
related encephalopathy seem to exhibit an underlying inability to engage in
purposeful, expressive social, emotional, and goal-directed behavior which may
be pervasive and contribute to their apparent cognitive impairments. These im-
pairments can result in diminished performance—flattened affect, impoverished
interpersonal interactions, and either the loss or arrest of verbal and motor skills
(Brouwers et al., 1994). Behavioral changes similar to those seen in HIV-related
encephalopathy (Moss et al., 1994) have been noted in other infections of the
CNS.
For example, autistic-like behaviors have been observed in children and
adults with herpes simplex encephalitis (Ghaziuddin, Tsai, Eilers, & Ghaziud-
by guest on July 13, 2011jpepsy.oxfordjournals.orgDownloaded from
382 Moss, Wotters, Brouwers, Hendricks, and Pizzo
din, 1992; Gilberg, 1991). Furthermore,
a
recent review stresses the importance
of investigating
and
developing methods
of
assessing changes
in
emotional
be-
havior associated with different forms
of
brain injury (Sherman, Shaw,
&
Glid-
den, 1994).
The purpose
of
this study
was to
determine objectively
and
systematically
whether "expressive, responsive behaviors"
in
several behavioral domains were
more impaired
in
encephalopathic than
in
nonencephalopathic HIV-infected
pe-
diatric patients.
To
study this phenomenon, children infected with
HIV
were
videotaped while being presented with stimuli designed
to
elicit responses rele-
vant
to the
behavioral domains being investigated. These responses were rated
from
the
videotapes according
to a
series
of
behaviorally defined variables.
The
use
of
videotapes helps overcome some
of
the shortcomings
of
other methods
for
rating behavioral and/or personality type variables. Similar use
of
videotaping
has been successfully employed
in
other types
of
behavioral research (Lyons-
Ruth, Connell, Zoll,
&
Stahl, 1987; Polan
et al.,
1991; Wilson
&
Matheny,
1983).
The
videotaping
of
these behaviors offers
the
advantages
of
standardiza-
tion
of
observational data, facilitating objectivity
of
evaluation
by
using raters
blind
to
possible biasing information
on the
child,
and
providing
a
method
whereby reliability
of
evaluation
can be
achieved
by
having
two
independent
raters view
and
assess
the
same material.
3
METHOD
Subjects
Eighty-three children with HIV-infection were studied. Most
of
these
pa-
tients were consecutively enrolled
in
this research. However,
for
the latter stage
of
the
data collection, only children less than
3
years
of age or
those with
evidence
of
CNS disease were included
in
this sample. This shift
in
the basis
for
selecting cases
is
unrelated
to the
variables that were studied.
The
primary
consequence
of
this change
in
sampling
was to
help even
out the
numbers
of
children assigned to the different groups used in the analyses (younger group
< 2
years
of age,
older group
> 2
years
of
age; encephalopathic group, nonen-
cephalopathic group). Each child's behavior
was
evaluated from
a
videotaped
session that was filmed prior
to or
shortly after their
(<
1 week) starting
on an
antiretroviral treatment protocol (median
of
1
day
before treatment start date).
The sample was recruited from
all
regions
of
the country
and
consisted
of
children
who
participated
in
clinical protocols
at the
National Cancer Institute
(NCI).
Research nurses initially contacted
the
child's parent
or
guardian
by
The manual
for the
administration
of
videotape procedures
and
rating
of
behaviors may
be
obtained
by writing
to the
first author.
by guest on July 13, 2011jpepsy.oxfordjournals.orgDownloaded from
Behavioral Impairments in Pediatric HIV Patients 383
phone after
the
referral
to
discuss
the
study. Patients
and
their families that
appeared eligible
for
the studies then visited the NCI
to
complete initial evalua-
tion.
Of
these patients,
70
were vertically infected,
6
were infected through
transfusion
(5 at
birth),
and 7
were hemophiliacs
who
received contaminated
blood products. The HIV-disease status
of
these patients had progressed to the P2
level
(the
level
of the
CDC classification system indicating symptomatic
HIV
infection)
of
the CDC (Centers
for
Disease Control, 1987). However, none
had
an acute illness, were
in
medical distress,
or
were febrile
at the
time
of the
assessment.
The
mean CD4% levels
(the
percentage
of the
total number
of
lymphocytes that are T4 immune helper cells) for the patients with encephalopa-
thy
was
10%
and
without encephalopathy
was
22%.
The
mean CD4%
of the
overall sample was within
the
lower 5%
of
the population based on comparisons
with norms from
an
uninfected sample (European Collaborative Study, 1992).
Although CD4% levels are correlated with disease progression, there are excep-
tions
in
which children with
low
levels
do
well clinically. The characteristics
of
the sample
are
shown
in
Table
I.
Encephalopathy
Each child was classified
as
either encephalopathic
or
nonencephalopathic,
according
to
the presence
of
moderate
to
severe symptomatic neurologic disease
Table I. Sample Charactenstics
n
Mean
age
Route
of
infection
(n)
Vertical
Transfusion
Gender (male/female)
M
F
Mean years parental education
Nonenceph.
Enceph.
Encephalopathic
(n)
Mean
CD4 %"
Mean
IQ
Nonencephalopathic
)
Mean
CD4 %
Mean
IQ
Younger patients
(<2 years
of
age)
35
1.08
35
0
21
14
130
12.9
19
18.37
60
16
26.25
102
Older patients
(>2 years
of
age)
48
5.1
35
13
33
15
13.5
12.9
26
4.38
61
22
19
99
"A CD4 % of 39 is approximately at the 50th percentile of the normative
sample.
by guest on July 13, 2011jpepsy.oxfordjournals.orgDownloaded from
384 Moss, Wolters, Brouwers, Hendricks, and Pizzo
from an independent evaluation made by senior medical investigators (physicians
trained in pediatrics, infectious disease, and neurology and a neuropsychologist)
in the program. This was done as part of a clinical decision-making process
regarding what treatment protocol a child should be assigned to and for assessing
the efficacy of a treatment. Children who were diagnosed with encephalopathy
were more likely to be assigned to more aggressive treatments which were more
likely to penetrate the CNS. The following criteria were central in making this
decision. Unambiguous evidence and relatively severe symptoms were required
in classifying a child as encephalopathic. The criteria used for encephalopathy
are consistent with the recommendations of the Working Group of the American
Academy of Neurology AIDS Task Force (1991). Children were classified as
encephalopathic if they exhibited one or more of the following criteria: (a) Loss
or delays in the acquisition of developmental milestones. Availability of this
evidence was contingent on the child having reached an age where these skills
should have been established. Clear loss or significant delays in developmental
milestones were sufficient for classification of encephalopathy, but necessarily
co-occurred with lowered developmental scores. Loss of developmental mile-
stones consisted of marked regression or disappearance of established speech
and/or locomotion skills. Delays were defined by these functions emerging at
least 25% behind (in months) their expected developmental timetable, (b) IQ or
developmental scores well below expected levels based on norms and medical
and environmental history (e.g., parental education, SES). In general, this con-
sisted of scores greater than
2
SDs below 100 for children under 2'/2 years of age
and 1—2 SDs below 100 for older children. This was based on age-appropriate
standardized IQ tests administered at the initial evaluation (Bayley, McCarthy, &
WISC-R). Of those children classified as encephalopathic, 86% had IQs below
80,
62% had IQs less than 70, and 42% had IQs less than 55. An IQ below 70,
unless there was a history of any preexisting non-HIV-related condition, was
sufficient to warrant being classified as encephalopathic. (c) Evidence of signifi-
cant decline in abilities. This depended on the availability of historical informa-
tion on the child and consisted of comparison of the results of earlier with more
recent test scores and on reports from parents, schools, and referring profession-
als of significant loss in skills. Evidence for this criterion of encephalopathy also
tended to co-occur with lower than expected current developmental scores, but
was helpful in classifying children when their current scores were at a marginal
level and where it was ambiguous whether or not this reflected a drop in func-
tioning, (d) The presence of positive computerized tomography (CT) findings
(enlarged ventricles, loss of white matter, cortical atrophy, and calcifications)
was not used to classify encephalopathy by itself but as supplementary data to
help classify cases where the other information was not definitive. This situation
rarely arose, but when it did, the CT findings tended to confirm the diagnosis of
encephalopathy based on the above criteria rather than reverse a preliminary
decision.
by guest on July 13, 2011jpepsy.oxfordjournals.orgDownloaded from
Behavioral Impairments in Pediatric HIV Patients 385
Although there are certain commonalities in the presentation of encepha-
lopathy and interrelations among symptoms (Brouwers et al., in press), there is
also a great deal of individual variation as to how HIV-related CNS disease is
manifested. Low neurodevelopmental scores could be the sole basis for being
classified as encephalopathic, but this criterion usually was accompanied by loss
of developmental milestones (particularly for younger children) or by positive CT
findings. Evidence for decline in abilities was inconsistently available, but when
it was it typically served as supportive data and not as an independent basis for
classifying a child as encephalopathic. Although these criteria are used consis-
tently at the NCI program, they may differ somewhat in relative emphasis at
other institutions, but there tends to be unanimity as to the basic factors used for
classifying HIV-associated encephalopathy in children (Working Group of the
American Academy, 1991). The reliability of this classification procedure was
established by having two physicians independently and blindly classify a sample
of 40 HI V-infected pediatric patients on the basis of information described in the
above criteria of encephalopathy.
4
This reliability study resulted in a kappa
coefficient of .90 (Woolson, 1987).
Videotape Procedure
The videotape procedure was developed for this research as a method for
obtaining and objectively quantifying information relevant to the aberrant behav-
ioral patterns observed among symptomatic, HIV-infected pediatric patients.
This procedure took place in a brightly lit and simply furnished room, 15 feet x
20 feet, in the Clinical Center of the NCI where the child previously had been
administered developmental tests by the same examiner. Thus, the room and the
examiner were familiar to the child. Most children, particularly those under
4 years of age, were accompanied by a parent or caretaker during the session.
The child was seated, or held by the caretaker, at a table located in the center of
the room. A staff member operated the camera that was mounted on a tripod, in
as unobtrusive a manner as possible, from a corner of the room and obtained
close-up recordings of the child by zooming-in with a telephoto lens. Much of
the session involved close-ups of the child's face and body since expressiveness
was a primary interest of this research.
The videotape procedure consisted of
a
20- to 30-minute session in which the
child was engaged by an examiner in a series of age-appropriate structured tasks
designed to elicit responses relevant to the five a priori domains of behavior under
investigation. These domains are (a) task-oriented behavior, (b) interpersonal-
social behavior, (c) affect, (d) sensorimotor behavior, and (e) communicative
behavior. Several variables were rated for each of these behavioral domains. The
4
\Nfe thank Linda Lewis and Lauren Wood for classifying whether or not these patients are encepha-
lopathic according to the criteria described above
by guest on July 13, 2011jpepsy.oxfordjournals.orgDownloaded from
386 Moss, Wolters, Brouwers, Hendricks, and Pizzo
variables included in each domain to be rated are listed in Table II. A total of 16
variables were each rated on a 7-point scale after observing the videotape re-
cords.
A manual was designed that provides a detailed definition of each of these
variables and also defines the end and middle scale points. The definition of the
variable and the use of ratings taken into consideration the age level of the child.
The manual also describes the specific procedures to be administered to the
children that were designed to elicit behavior relevant to the respective domains.
Different tasks were used that were appropriate to the child's age and level of
functioning for the following age groups: 3-18 months; 18-30 months; 30-48
months; and for children over 4 years. For example, to assess and elicit task-
oriented and fine motor behaviors, children 3-18 months were asked to grasp a
ring and pick up and place a small pill in a bottle; children 18-30 months were
asked to stack blocks as high as they were able; children 18-30 months and
30
48 months were given a bead-stringing task, and those over 4 years of age were
asked to trace a drawing. Approximately 12—15 tasks were administered to each
child. Other representative tasks were social interaction with parent, frolic play,
sensory stimulation, age-appropriate fine and gross motor tasks (e.g., throwing
and catching ball, skipping, drawing, puppet play, response to jack-in-the-box,
"Simon Says," reaction to pictorial incongruities, conversation, and verbal con-
struction of events). The ratings were not necessarily restricted to assessment of
reactions to specific tasks, but were based on overall behavior.
As an illustration, annotated definitions from the manual of a few of the
variables used to rate the videotapes are described below (these variables are
rated on a 7-point scale):
Persistence. This variable measures a child's sustained and continuous ef-
fort to successfully complete a task while maintaining a high standard of perfor-
mance: 1 (low persistence)—The child is indifferent and nonresponsive to the
various tasks; 4 (moderate persistence)—the child shows some interest in accu-
racy and mastery, but may give up before completion, accepts errors or moder-
ately accurate performance; 7 (high level of persistence)—the child strives for
perfection, works very carefully and exhibits high standards in performing tasks,
and persists until task is completed.
Positive Social Behaviors. This variable measures the degree to which a
child both initiates positive social interactions and responds positively to social
interactions from others: 1 (no positive social behavior)—The child exhibits
little to no social behavior, the child gaze averts, stares vacantly, is expression-
less,
and does not smile or verbalize positively; 4 (moderate positive social
behaviors)—the child exhibits moderate or intermittent social behavior, may
show some social responsiveness after encouragement and "warm-up"; 7 (high
degree of positive social behavior)—the child frequently initiates social behav-
iors (smiles, vocalizations, and eye contact) and is highly responsive to social
overtures.
by guest on July 13, 2011jpepsy.oxfordjournals.orgDownloaded from
Table II. r-Test Comparison of Younger (<2 Years) and Older (>2 Years) Encephalopathic and Nonencephalopathic Patients
on the Videotaped Behavioral Ratings
Variables
Task-oriented behaviors
1 Persistence
2 Attention span
3 Impulsivity
4 Goal-directed behavior"
Social behaviors
5 Positive social behaviors
6 Attachment to caregivcr
7 Compliance-cooperativeness
Affective behaviors
8. Presence of affect
9. Irritability
10.
Positive affect
Sensorimotor behaviors
1 1
Activity level
12.
Reactivity
13.
Fine motor skill
14.
Gross motor skill
Verbal behaviors
15.
Use of verbal language''
16.
Quality of speech"
Enceph
mean
4 1
3 9
4 6
4 0
5.8
4 3
3.5
2 9
3.7
2 8
3.4
3.6
2.7
1.7
Younger patients (n
Nonenceph
mean
4 9
4 4
4.1
5 4
6 3
5.1
4.9
2.7
4.8
4 1
3.4
5 3
50
3 6
= 35)
Two-tailed p
ns
ns
ns
< 01
ns
ns
<.O1
ns
ns
<.00l
ns
<.00l
<.00l
< 01
Enceph.
mean
38
3.6
3 7
3.5
3.7
6.1
3.9
3.6
3.2
3.4
3.5
3.5
3.6
3.0
3.0
36
Older patients in
Nonenceph.
mean
5.6
5.0
4.6
5 4
6 1
6.0
5.7
5 5
1.5
5.6
5.1
39
6.0
6.0
59
4.7
-
48)
Two-tailed p
< 001
<
01
ns
<.00l
<
001
MS
<.00l
<.00l
<
001
<.00l
<
001
ns
<
001
<.00l
<.00l
<05
"Not rated
for
younger
patients.
ft
Not rated
for
children less than
12
months
of
age.
n = 18 for
younger patients
by guest on July 13, 2011jpepsy.oxfordjournals.orgDownloaded from
388 Moss, Wolters, Brouwers, Hendricks, and Pino
The videotapes of the 83 children were rated on the 16 variables defined
in the manual by a clinical psychology doctoral student, trained in the rat-
ing system, who was unfamiliar with the children and was blind to whether
they were classified as encephalopathic or nonencephalopathic. To establish
interrater reliabilities two psychologists independently also rated the videotapes
of 18 of those patients who were rated by the primary rater. The intraclass
correlations (Bartko, 1966) between the raters for the 16 variables ranged
from .79 to .99 with a median interrater reliability coefficient of .92. All of
the variables had sufficiently high reliabilities to warrant inclusion in the an-
alyses.
Analyses were computed separately for children less than 2 years (M -
1.08, N = 35) and those greater than 2 years (M age = 5.15, n = 48) of age.
This division into two age groups was done since 2 years of age is an important
transition point associated with major developmental changes (walking, lan-
guage, etc.) and several of the variables rated from the videotapes were defined
differently for younger and older children. Children less than 2 years were not
rated on Variable 4 (goal-directed behavior) and 16 (quality of speech) and
children less than 12 months were also not rated on Variable 15 (verbal lan-
guage).
The ratings of the videotaped behavioral domains (summary scores for
clusters of related variables) were correlated with a separate set of ratings made
by staff members of behaviors observed during developmental (Bayley, McCar-
thy, and W1SC-R) and neuropsychological testing (PPVT, Ravens, Visual Motor
Integration, etc.) of these patients. The ratings of the behaviors observed during
testing were made by psychologists using a Q-sort rating procedure (N1H Child
Q-sort) designed and validated to assess possible patterns of aberrant behaviors
that tend to characterize encephalopathy among HIV-positive pediatric patients
(Moss et al., 1994). This Q-sort procedure is based on 49 items of behaviors
listed in DSM-III-R (American Psychiatric Association, 1987) to describe behav-
ioral disorders in children. Scales were derived through factor analysis of ratings
made on a sample of 180 children infected with HIV. The NIH Q-sort scales,
similar to the video ratings, were developed separately for the same younger
(< 2 years of age) and older (>2 years of age) age groups. The respective
Q-sorts yield scores on five scales for the younger patients and scores for four
scales for the older patients. The five scales for the younger group are defined as
(a) Nonsocial, (b) apathy, (c) immature/poorly integrated, (d) flaccid, and (e)
Attentional deficit; and the four scales for the older group as (a) Depressed, (b)
Hyperactive/attentional deficit, (c) Autistic, and (d) Low frustration threshold.
The videotaping and neuropsychological testing were both completed within a
5-day
period. The ratings from the videotapes and the Q-sort were done indepen-
dently by different raters.
by guest on July 13, 2011jpepsy.oxfordjournals.orgDownloaded from
Behavioral Impairments in Pediatrk HIV Patients 389
RESULTS
Comparisons Between Encephalopathic and
Nonencephalopathic Groups on 16 Variables
Table II presents the means and the Mest significance levels for the differ-
ences between the encephalopathic and the nonencephalopathic patients for the
16 variables that were rated. The variables are organized and listed according to
the five behavioral domains described in the Method section: task oriented,
interpersonal, affective, sensorimotor, and verbal. The findings are presented
separately for the younger and the older patients.
Younger
Group.
The nonencephalopathic younger patients exhibited signifi-
cantly higher Activity levels and were superior in Fine and Gross Motor and
Verbal skills, and showed more responsiveness in "positive social behavior" and
greater "expression of affect" than did the encephalopathic group. In contrast,
encephalopathic and nonencephalopathic infants did not differ in any of the
"task-oriented behaviors" or on the ratings of several of the other Interpersonal
and Emotional variables. The greatest difference between the encephalopathic
and nonencephalopathic infants was found in the sensorimotor domain.
Older Group. In 13 out of the 16 comparisons, the older nonen-
cephalopathic children obtained significantly higher ratings, demonstrating more
mature, integrated, appropriate, and skilled behaviors than encephalopathic chil-
dren from this age group. The three variables that did not differentiate between
the encephalopathic and nonencephalopathic older groups included: Impulsivity,
Attachment, and Reactivity. Thus, older nonencephalopathic children functioned
in a more optimal manner than did the encephalopathic patients in the task-
oriented, interpersonal-social, affective, sensorimotor, and verbal behavioral do-
mains.
Construction of Summary Domains. The 16 variables were summarized into
five domain scores by identifying and pooling clusters of variables that were
conceptually cohesive and exhibited high intercorrelations based on the total
sample. This resulted in five revised domains that overlapped greatly with, but
still were somewhat different from, the five proposed a priori domains. In the
realignment of domains, several of the social and emotional variables (positive
social behaviors, compliance-cooperativeness, presence of affect, and positive
affect) were highly intercorrelated and thus were pooled to constitute the "posi-
tive social emotional domain," whereas others with low intercorrelations from
these groupings were deleted. Activity level represented an independent and
robust aspect of behavior and as a single variable was retained as a separate
domain. Fine and Gross Motor skills were highly correlated and scores for these
variables were summarized into the Motor Skills domain. Since the variable,
by guest on July 13, 2011jpepsy.oxfordjournals.orgDownloaded from
390 Moss, Wolters, Brouwers, Hendricks, and Pizzo
"quality of speech," was not rated for the younger group and only 18 out of 35 of
these patients (those > 12 months of age) were rated on "use of verbal lan-
guage," the Expressive speech domain was based on limited data for this group
and consequently this measure was not included in all analyses. The average
intercorrelations among the variables within the (a) task—oriented, (b) positive
social-emotional, (c) motor skill, and (d) expressive speech domains were .72,
.66,
.82, and .75, respectively.
Correlations Between Domains. Summary correlations between domains
were obtained by averaging all the correlations between variables from each
domain with variables from the other domains in order to demonstrate the degree
of overlap and independence between domains. The summary correlations are
presented separately in Table III for the younger and older children. For the
younger patients most of the correlations between domains were low to moderate
in magnitude. The two exceptions for this group were that expressive speech was
highly correlated with motor ability (r = .86) and with activity level (r = .70).
Most of the correlations between domains for the older group were in the moder-
ate range (rs ranged from .37 to .54) except, as was the case with the younger
patients, expressive speech and motor ability were highly correlated (r = .69).
Task-orientation and activity level were unrelated for the older group.
Comparisons Between Encephalopathic
and Nonencephalopathic Groups (ANOVA)
Scores for each of the domains were derived by obtaining the average rating
for each subject for all the variables that comprised the domain. These scores
were in turn compared for the encephalopathic and nonencephalopathic patients
for both younger and older children.
Table HI. Intercorrelations Between Domains: Intercorrelations of Younger Patients (n = 35)
Are Shown Above the Diagonal and Older Patients (n = 48) Are Shown Below the Diagonal
Domain
1 Task-oriented
2.
Social/emotional
3. Motor
4.
Speech"
5. Activity
1
.40''
A9
d
.46'
.08
2
.54'
.5\
J
.54
J
.53''
3
.17
.32
.69
rf
.51''
4
.24
.55'
.86--
.37''
5
.21
.51'
41*
.70'
"n = 18.
h
p < 05, two-tailed.
•p < .01. two-tailed.
J
p < .001, two-tailed.
by guest on July 13, 2011jpepsy.oxfordjournals.orgDownloaded from
Behavioral Impairments in Pediatric HIV Patients 391
A two-way repeated measure ANOVA (comparing the five domain scores
for patients classified as either encephalopathic or nonencephalopathic) for both
age groups was used with the Greenhouse-Geiser correction for correlations
between measures (Watemaux, 1991).
Younger
Group.
For comparisons involving the young group the expressive
speech domain score was not included since the small number of patients ob-
tained on this measure for this group disproportionately reduced the number of
cases in the overall analyses. An ANOVA indicated that in the younger group the
encephalopathic patients were rated significantly lower overall than the nonen-
cephalopathic patients on the domain scores, F(\, 33) = 11.76, p < .01. There
also was a significant interaction between domain scores and encephalopathy in
which motor skills and activity differentiated more between encephalopathic and
nonencephalopathic patients than did task-oriented and positive social-emotional
behaviors, F(3, 3) = 4.10, p < .01.
Older
Group.
An ANOVA comparing the older encephalopathic and nonen-
cephalopathic patients on the domain scores resulted in a significant effect,
F(l,
45) = 45.75, p < .001, showing that the older encephalopathic children
were impaired on all behavioral modalities. There was no interaction, F(4, 4) =
ns,
when comparing each of the domain scores for the encephalopathic and
nonencephalopathic patients. Thus, for this group, none of the domains differed
from one another in discriminating between encephalopathic and nonencephalo-
pathic patients. Table IV shows the means and p values for post-hoc F-test
comparisons between encephalopathic and nonencephalopathic patients on the
domain scores for the younger and older groups.
Post-Hoc F-Test Comparisons for
Younger
Group
For the younger group the motor skill, activity level, and positive social-
emotional scores showed significant differences between the encephalopathic
and nonencephalopathic patients, reflecting greater activity, motor skill, and
more appropriate behavior among the nonencephalopathic infants. Only the
Task-oriented domain did not statistically differ for the encephalopathic and non-
encephalopathic younger patients.
Post-Hoc F-Test Comparisons for Older Group
All five behavioral domains exhibited statistically significant differences in
the post-hoc comparisons between the older encephalopathic and nonencephalo-
pathic children.
by guest on July 13, 2011jpepsy.oxfordjournals.orgDownloaded from
Table IV. F-Test Comparisons of Younger (<2 Years) and Older (>2 Years) Encephalopathic
and Non-encephalopathic Patients on Clusters (Domains) of the Videotaped Behavioral Ratings
Clusters
Enceph.
mean
4.2
3.8
3 2
2.8
Younger patients (n
Nonenceph.
mean
4.5
4.9
5.2
4.1
= 35)
df
1,
33
1,
33
1,
33
1.
33
P<
ns
.02
.001
.001
Enceph.
mean
3.7
3.6
3.3
2.9
3.5
Older patients (n
Nonenceph.
mean
5.2
5.6
6.1
5.3
5.1
= 48)
df
1,45
1,45
1,45
1,
45
1.
45
P<
.001
.001
.001
.001
.001
1.
Tabk-oriented behavior (Variables I. 2. 3, 4)
2.
Positive social-emotional (Variables 5, 7, 8. 10)
3.
Motor skills (Variables 13, 14)
4 Expressive speech (Variables 15. 16)
5.
Activity (Variable II)
by guest on July 13, 2011jpepsy.oxfordjournals.orgDownloaded from
Behavioral Impairments in Pediatric HIV Patients 393
Correlations Between Behaviors Evaluated
from Videotape and Q-Sort Procedures
The intercorrelations between the videotaped behavior domain scores and
the Q-sort scale scores are shown in Table V for the younger patients and Table
VI for the older group.
Younger Group. The Q-sort scale that measures Nonsocial behavior was
significantly correlated with low positive social-emotional behavior and low
activity on the videotaped ratings; Apathy on the Q-sort was associated with poor
motor skills and low activity on the videotape ratings; and being evaluated as
Immature/poorly integrated on the Q-sort was related to low task-oriented and
low positive social-emotional functioning on the video ratings. The significant
correlations that occurred between scores on these procedures tended to be be-
tween conceptually related variables.
Older
Group.
The Q-sort scale measuring Depression was negatively corre-
lated with the videotape ratings on positive social-emotional behavior, expressive
speech, and activity level; the Q-sort ratings of Hyperactive/attentional deficit
was associated with low task-oriented behavior, but with high activity level
ratings from the videotapes; and the Q-sort rating of Autistic behavior was
inversely related to the ratings of all the videotaped behavioral domains, but
particularly for social-emotional behavior, motor skill, and expressive speech.
The Q-sort ratings on Low frustration threshold were unrelated to any of the
videotaped behavioral domains. Again, the significant correlations that emerged
between these two sets of behavioral ratings tended to be between conceptually
congruent dimensions.
Table V. Correlations Between the Videotape Behavioral Cluster Scores and the Q-Sort Scores
for the Younger Patients (<2 Years of Age)
Videotaped behavioral domains
Task-oriented (Variables 1, 2, 3, 4)
Positive social-emotional
(Variables 5, 7, 8, 10)
Motor skills (Variables 13, 14)
Expressive speech (Variable 15)
Activity (Variable 11)
"p < .05, two-tailed.
h
p < .01. two-tailed
7> < .001, two-tailed.
Nonsocial
-.32
-.50*
-.23
- 25
- 49*
Apathy
.07
-.16
- 51*
- 20
-.65-
Q-sort scales
Immature/poorly
integrated
-.40"
-.37°
- 23
-08
-.25
Flacid
.04
-.03
-.13
.08
-.13
Attention
deficit
.12
- II
.10
.07
.05
by guest on July 13, 2011jpepsy.oxfordjournals.orgDownloaded from
394
Moss,
Wolters, Brouwcrs, Hendricks, and Pizzo
Table VI. Correlations Between Videotaped Behavioral Cluster Scores and Q-Sort Scale Scores
for the Older Patients (>2 Years of Age)
Videotaped behavioral domains
Task-oriented (Variables
1, 2, 3
and
4)
Positive social-emotional
(Variables
5, 7, 8 and 10)
Motor skills (Variables
13 and
14)
Expressive speech (Variables
15
and
16)
Activity (Variable
11)
Depression
.09
-.36*
.27
-.35°
-.50'
Older patients
Hyperactive/
attentional
deficit
-.43*
.10
-.07
.14
.45'
Q-sort scales
Autistic
-.34"
-.51'
-.!&<•
-.65'-
-.44*
Low
frustration
threshold
.29
-.22
-.27
-.18
-.09
"p < .05, two-tailed.
h
p < .01, two-tailed.
c
p < .001. two-tailed.
DISCUSSION
Several statistically significant differences were obtained from comparisons
of encephalopathic
and
nonencephalopathic HIV-infected younger
and
older
groups
on
ratings
of
expressive behaviors, based
on
videotaped behavioral sam-
ples.
Both younger
and
older children
who
were nonencephalopathic demon-
strated more appropriate behavior than their same-age encephalopathic counter-
parts.
However, some
of
these differences were less clearly defined
in the
younger group, possibly because
of
methodological limitations
of
the assessment
procedure
or
developmental characteristics
of
this
age
group.
The
older
non-
encephalopathic children obtained significantly higher scores
on the
task-
oriented domain than
did the
encephalopathic patients from this
age
group.
A
similar difference
on
this domain
was not
observed
for the
younger group,
presumably because children
at
this
age
level
may not yet
have matured suffi-
ciently
for
behaviors such as persistence and attention span to be notably relevant
aspects
of
their functioning.
It is
also possible that
the
tasks
we
used
and our
methods
of
assessing these variables were
not
designed
and
adapted appro-
priately
to
this age group. A further example
of
how developmental factors could
have contributed
to the
different findings
for the
younger
and
older groups
for
these variables
is
that crying (irritability)
may be an
appropriate
and
adaptive
behavior
for
infants
(< 2
years
of
age)
in
order
to
communicate discomfort and
provoke relief responses from
the
caregiver. Thus, sick infants who may
not be
by guest on July 13, 2011jpepsy.oxfordjournals.orgDownloaded from
Behavioral Impairments in Pedialric HIV Patients 395
feeling well may exhibit elevated levels of irritable behavior, regardless of CNS
status.
Conversely, irritability is a more extreme and atypical reaction for older
children and therefore its presence may be a more sensitive indication of aberrant
functioning for this age group. The nonencephalopathic patients exhibited
signif-
icantly higher domain scores on activity levels and motor skills that the encepha-
lopathic patients for both the younger and older age groups. Expressive speech
was better in the nonencephalopathic patients in both age groups with these
differences again being stronger among the older children. Since expressive
speech is still emerging and not yet present in many of the younger infants (and
often not evaluable), this behavior may not be as sensitive in reflecting CNS
disease in this group as it is in older children. In any case, the more salient
differences observed in the older group should not be regarded as indicating that
encephalopathy is more severe among older children. Developmental theory and
previous research suggests that the CNS of younger children actually might be
more adversely affected than that of older children by the HIV. For example,
CNS prophylactic radiation for treating ALL resulted in greater intellectual im-
pairment among younger compared to older children (Moss, Nannis, & Poplack,
1981).
Thus, the less mature brain may be more vulnerable to deleterious events.
Unfortunately, our research was not designed to properly test for age effects.
Overall the ratings of the videotaped observations show that several modal-
ities of expressive behavior are impaired in pediatric patients with HIV-related
CNS disease. Children with severe HIV-related CNS disease seem less able to
take initiative and act in a purposeful way, may be unresponsive socially and
emotionally (may stare vacantly ahead with little change in facial expression),
have impaired motor functioning, are inactive, apathetic, and exhibit a decre-
ment or absence of verbal behavior. There is no indication that this lack of
expressiveness reflects any underlying motivation or emotional state, but rather
seems tied to behavioral impotence or loss of efferent functioning. Support for a
neurological explanation of this aberrant behavioral pattern comes from the find-
ing that normal behavior can often be at least partially restored after a course of
antiretroviral treatment for a drug that crosses the blood-brain barrier (Brouwers
et al., 1990; Pizzo et al., 1988). Also, it was found that more severe CT scan
abnormalities were significantly correlated with greater deficiency in expressive
compared to receptive language scores in encephalopathic children (Wolters et
al.,
in press). There is evidence from several sources that HIV infects the CNS
and results in a variety of neurological symptoms and impairments (Armstrong et
al.,
1993; Belman et al., 1985).
The loss and/or impairment of expressive functions often appears to happen
concurrently for several behavioral modalities. This may account for the correla-
tions among the domain scores. Many of the 16 variables are intercorrelated, as
was the case with the domain scores. Data were presented on each of the 16
variables, even though they are intercorrelated, because the various behaviors
by guest on July 13, 2011jpepsy.oxfordjournals.orgDownloaded from
396 Moss, Wolters, Brouwers, Hendricks, and Pizzo
that are represented often are regarded as conceptual entities in clinical evalua-
tions,
so that it seems useful to show how they are independently affected by
HIV-related CNS disease. There is a precedent in psychology, largely for con-
ceptual reasons, to retain individual variables as separate measures (indices) of
behavior, even when they are highly intercorrelated (Kagan & Moss, 1962;
McCarthy, 1972; Wechsler, 1974). It is difficult to ascertain whether the intercor-
relation among variables in the present case reflects a lack of conceptual differen-
tiation in the definitions of
the
variables or whether the CNS disease affects many
functions in a concurrent, generalized manner. The same rationale is used to
defend the use of the five derived behavioral domains which were also intercorre-
lated. The condensing of the variables into domains seems useful and appropriate
since the 16 variables were reduced to a statistically more manageable number of
summary measures while still retaining the original proposed conceptual organi-
zation.
Verbal
expression and motor skills are highly intercorrelated in this popu-
lation, yet they represent distinctly different behavioral systems. A hypotheses
we have considered, and are investigating currently, is that the high correlation
between these domains may be because HIV-related CNS disease in children
tends to affect these functions concurrently. These systems appear to be some-
what intercorrelated in noninfected children as well, as is reflected in the mean
correlation across different ages (r = .45) between the Verbal and Motor Scales
of the McCarthy Scales of Children's Abilities (McCarthy, 1972). There may be
some circularity between behavior on these two domains and the criteria for
encephalopathy, particularly for the younger group, since verbal and motor skill
are included as evidence of developmental status in the tests we used for assess-
ing this age group.
The Q-sort behavioral ratings, made from observations of these children
when they were administered IQ and neuropsychological tests, were highly
correlated with conceptually congruent domain scores from the ratings of the
videotapes. The fact that conceptually similar variables from these two proce-
dures were highly intercorrelated, whereas conceptually divergent variables were
unrelated, provides strong support for the validity of both procedures. These
findings not only corroborate the sensitivity of these two procedures in differen-
tiating encephalopathic and nonencephalopathic pediatric HIV patients but also
provide evidence of the validity (construct validity) of the conceptual structure of
these methods since similarly defined constructs were correlated across methods.
The cross-situational generality of our findings could have been further extended
if we had obtained parental ratings of comparable child behavior.
Since this research focused on behavioral manifestation of HIV-related
encephalopathy it seems unlikely that factors such as length of illness, route of
infection, or sampling or recruitment procedures may have influenced the ob-
tained findings. However, this is an assumption that cannot be concluded from
our data since length of illness and route of infection were not controlled for in
by guest on July 13, 2011jpepsy.oxfordjournals.orgDownloaded from
Behavioral Impairments in Pediatric HIV Patients 397
this investigation. This assumption could best be tested in the future through a
longitudinal design where these factors could be better controlled for and mon-
itored over the course of
the
illness. Moreover, our groups were heterogeneous as
to such factors as race, sex, SES, CD4 status, and so forth,