Depressive
Symptoms
Robert
Ph.D.,
A. Stern,
and David
Objective: The primary to mood and vegetative
purpose of this study disturbance following
outpatients
single,
who
had
had
unilateral
Following
strokes
Stroke
L. Bachman,
M.D.
was to assess the relation of lesion location stroke. Method: Fifty-two inpatients and were
included.
Patients
with
past
CNS
or
psychiatric disorders were excluded. A modified Visual Analogue Dysphoria Scale was used to allow the inclusion of all but the most impaired aphasic patients. Sleep and eating disturbances were measured by using both self-report and nursing assessments. Location of lesions was determined by CT scan and classified according to three dimensions: right-left, dorsal-ventral, and frontal-nonfrontal. Results: On measures of dysphoric mood and sleep disturbance, results indicated significant three-way interactions among the three lesion dimensions. No differences were found with regard to eating disturbance. Greater dysphoria and sleep disturbance were found in subjects with left parietal/occipital, left inferior frontal, right superior frontal, and right temporal lesions than in subjects with lesions in other locations. Depressive symptoms were not associated with functional impairment as measured by activities of daily living, motor strength, or severity of aphasia. Conclusions: These results support the mixture of influences relation between the more complex than (AmJ Psychiatry
hypothesis that lesion location is a valid which may result in a dysphoric mood site ofthe lesion and subsequent depressive has been reported previously. 1991; 148:351-356)
M
ood disorders following stroke are common (1), can significantly reduce the speed and success of rehabilitation (2), and do not appear merely to mepresent an expected grief response to loss of function (3, 4). Robinson and colleagues (5, 6) suggested that patients with left anterior cerebral lesions have a higher incidence and severity of the psychiatric syndrome of major depression than do patients with lesions in other locations. A more recent report by Robinson’s group (7) and one by others (4), however, indicate that this finding may not be as definite as was first thought. A more precise assessment of intrahemisphenic lesion location may help to clarify the relation between lesion focus and depressive symptoms. On the basis of known neumoanatomical connections of cortical/sub-
Revised version of a paper presented at the I 7th annual meeting of the International Neuropsychological Society, Vancouver, B.C., Canada, Feb. 8-1 1, 1989. Received Sept. 15, 1989; revision received Aug. 16, 1990; accepted Sept. 25, 1990. From the Department of Neurology, Department of Veterans Affairs Medical Center, Boston, Mass. Address reprint requests to Dr. Stern, Department of Psychiatry, CB 7160, University of North Carolina School of Medicine, Chapel Hill, NC 27599-7160. The authors thank Drs. Michael Alexander, Vikan Babikian, Dana Penny, Patricia Raymond, and Dominic Valentino, Ms. Carole Palumbo, and Ms. Denise Stiassny-Eder for their assistance in data collection and analysis.
Am
J
Psychiatry
148:3,
March
1991
cortical
limbic
and significant factor in the state following stroke. The symptoms, however, may be
system
pathways,
it would
that differences in mood and behavior within frontal or posterior regions, whether a lesion is in a more ventral or (e.g., orbital frontal versus domsolatemal temporal versus panietal lobe) (8-10).
be expected
would exist depending on dorsal location frontal lobe or Despite the ex-
istence of these dorsal-ventral neuroanatomical differences, no previous study of poststroke mood disordens has systematically evaluated this aspect of lesion
localization. A major limitation to previous investigations of depressive symptoms following stroke involves the selection of subjects. Many studies have included patients with prestroke histories of psychiatric illness or previous strokes (1, 11, 12). This has led to potentially confounding results, since these variables may increase the risk of depressive symptoms following stroke (12, 13). In addition, the majority of investigations have excluded many aphasic patients with impaired comprehension because of the inability of these patients to respond accurately to standardized depression scales and interviews (4, 14). Although some aspects of depressive bances,
symptoms, such as sleeping and can be readily observed, internal
eating mood
disturstate is
difficult to assess in aphasic patients (iS, 16). Finally, most previous studies have relied on multifactonial measures of the syndrome of depression or on psychi-
3S1
DEPRESSIVE
SYMPTOMS
FOLLOWING
atmic diagnostic criteria (e.g., measures of specific symptoms.
The lation
present study was of specific lesion
STROKE
DSM-III)
rather
than
on
munication
was
Boston
undertaken locations
to
to assess internal
the memood
states and to two vegetative symptoms common in psychiatnic depressive disorders, i.e., sleeping and eating disturbances. In addition to night-left and frontal-non-
the
Dysphomic Visual
line with was left-side
METHOD
of dysphonic
Subjects
were
selected
from
among
consecutively
ad-
private
rehabilitation
the study subjects
and or
after
their
facility.
explanation representatives
Before
participation
of the procedures, gave
informed
in
sent. All subjects were might-handed men who had suffered a unilateral cemebrovasculam accident between 1 month and 4 years before participation (mean±SD time since stroke onset was 16.2±12.3 months). Only patients whose lesions could be documented by CT scan were included. All subjects had lesions involving the cerebral hemispheres; no subjects had isolated brainstem or cerebellan lesions. Patients who had had previous cemebrovasculam accidents or other disorders of the CNS were excluded, as were those with prestroke histories of psychiatric illness (including significant alcohol abuse). Aphasic patients were excluded only if their mean auditory comprehension scores on the Boston Diagnostic Aphasia Examination (17) were below the 10th percentile; only two aphasic patients were excluded by this criterion. Twenty-two aphasic patients participated. The total number of subjects was 52, and their mean±SD age was 65.8±7.9 years. Ninety percent (N=47) of the subjects were Caucasian; the remaining 10% were black. Thirty-three percent (N= 17) of the subjects had high school degrees, 38% 29%
(N=20) (N=1S)
had had
not had
completed high some education
school, beyond
and high
school. Each subject was assessed on one occasion. Self-report and interview ratings were conducted in the late morning or early afternoon to preclude effects of diurnal variation.
Functional impairment was measured with the Modified Bamthel Index of Mobility and Self Care (18) and a typical 6-point motor strength scale (19). The severity of the aphasic patients’ impairment in com-
352
severity
scone
of the
assessed by a modification Mood Scale (20) in which
a vertical
rather
used in order hemispatial
represented
by words
than
of a
a horizontal
to preclude bias due to neglect. The two poles
and
cartoon
faces
indicat-
mood,
such
as the MMPI
depression
scale
and the Profile of Mood States. A similar scale has been found useful in assessing mood in stroke patients because of its brevity and its limited linguistic requirements (21). Formal assessment of reliability and validity
for other
scales
has
The
vertically recently
Visual
presented been
for
Analogue
Depression
Checklist
All
battery the
and
of their aphasia, were to be reliable responses
Scale
mood
was
of mood Hamilton
the
subjects,
analogue
(22).
Dysphonia
(23)
(24).
visual
reported
tened as pant of a larger sion inventories, including
adminis-
and depresRating Scale
Depression
regardless
Adjective of the
able to provide what by using the Visual
severity
appeared Analogue
Dysphonia correlated measures
Scale. Furthermore, highly with scones of dysphonic mood
ratings on this scale on other standardized such as the Depression
Adjective
Checklist
in the
(n=0.81)
aphasic
patients
(25), thus providing additional evidence of concurrent validity for this new mood scale. The correlation between ratings on the Visual Analogue Dysphomia Scale and the Depression Adjective Checklist existed for the aphasic subjects with the most intact auditory cornprehension (r= 0.84; mean±SD Boston Diagnostic Aphasia Examination auditory comprehension percentile=75.3± 13.8) as well as for those with the most compromised auditory comprehension (r=0.64; auditory comprehension pencentile=29.7± 13.7). Ratings of vegetative symptoms were created by using a combination of interview response and observed
behavior. Sleep disturbance was assessed by a combination of the three insomnia items of the Hamilton depression
scale
and
a 5-point
rating
of early
morning
awakening completed by a nurse familiar with the patient. Eating disturbance was similarly assessed by a combination of the two eating-related items of the Hamilton scale and a nurse’s 5-point rating of diminished
Instruments
was
the
Examination.
ing “happy” and “sad.” The scone on this Visual Analogue Dysphonia Scale was the number of millimeters from the “happy” pole. We found this modified scale to be a valid measure of dysphonic mood in a series of 95 general neurology patients referred for neumopsychological assessment, with significant correlations between scones on it and on other standardized measures
all con-
mood Analogue
100-mm
were
by
Aphasia
presentation nightor
frontal dimensions, lesions were classified by using a ventral-dorsal dichotomy. Patients with previous neurological or psychiatric histories were excluded, and a modification of a self-report measure of dysphomic mood was used to allow the inclusion of all but the most severely impaired aphasic patients.
mitted inpatients and outpatients with neurology clinic appointments at the Boston Department of Veterans Affairs Medical Center and the Bnaintnee Hospital, a
assessed
Diagnostic
appetite.
meaningful
One
responses
patient
during
was
unable
the Hamilton
to
provide
depression
interview, and therefore the sleep and eating disturbance scales were composed solely of the nurse’s assessments. Ratings on each of these two scales were
transformed to standardized 50± 10) for the analyses.
Am
J
T scores
Psychiatry
148:3,
(mean
March
±
SD
1991
ROBERT
Lesion
MANOVA. formed by
Classification
All subjects independently
received
CT
by
members
two
scans,
which of
the
were CT
judged research
than this
least sample
A. STERN
AND
The MANOVAs using unweighted squares were
DAViD
L. BACHMAN
and ANOVAs were permeans analyses rather
analyses, since most likely due
unequal cell sizes in to chance sampling
group of the Boston Department of Veterans Affairs Medical Center Aphasia Research Center. Location of the lesions was assessed with a modification of the technique of Naeser and Hayward (26). Since all lesions were unilateral, the side of the lesion was easily
error
determined.
appropriate simple effects tests to assess the underlying two-way interactions or main effects. On the basis of previous research and theoretical arguments, an a priori decision was made to assess
Analysis
of intrahemisphenic
lesion
tion was performed on a region-by-region of the following six regions was rated scale
with
respect
the inferior and 43-47),
to the
frontal lobe the superior
percentage
on
loca-
basis. Each a 4-point
of infarcted
tissue:
(Brodmann’s areas 10-15, 25, frontal lobe (areas 4, 6, 8, 9,
24, 32, and 33), the temporal lobe (areas 20-22, 2630, 34-38, and 41-43), the inferior occipital lobe (ameas 17-19), the inferior panietal lobe (areas 23, 31, 39, and 40), and the superior pamietal and occipital lobes
(areas 1-3, 5, and 7). In order to provide intmahemispheric classifications, these ratings were then transformed using two dimensions: 1) dorsal (including the superior frontal and entire pamietal and occipital lobes) versus
ventral
(including
the inferior
frontal
and
entire
temporal lobes) and 2) frontal versus nonfrontal. A lesion was considered to be frontal if the mean rating for the two frontal regions was greaten than the mean rating for the remaining four regions. Similarly, a Icsion
was
considered
to
be
ventral
if the
mean
in
one
area
or
another,
rather
than
solely in that area. The inclusion of inferior occipital regions in the dorsal classification was based on known corticolimbic connections rather than on actual anatomical location. On the basis of these lesion classifications, 37% (N=19)
of the
subjects
had
night
hemisphere
lesions
(63%, N=33, left hemisphere), 65% (N=34) had predominantly dorsal lesions (35%, N 18, ventral), and 44% (N=23) had predominantly frontal lesions (56%, N=29, nonfrontal). In this classification scheme no distinction was made between involvement of cortical and subcortical lesions. Statistical
Analyses
The relation between lesion classifications and each of the three dependent variables (i.e., ratings of dysphomia, sleep disturbance, and eating disturbance) were first assessed with a multivamiate analysis of vanance (MANOVA) to protect against type I error and because of interconnelations among the dependent variables. A 2 x 2 x 2 design was used, with each of the three dichotomous lesion dimensions used as a grouping variable. Individual univamiate analyses of variance (ANOVAs) were performed subsequent to a significant
Am
J
Psychiatry
148:3,
March
1991
than
an expected
ity
of dispersion
Significant
the
matrices,
three-way
inequality
in the popu-
assumptions analyses (e.g.,
underlying homogene-
homogeneity
ANOVAs
of variance).
were
followed
up
by
might-left by fmontal-nonfrontal interactions at each of the two dorsal-ventral levels. For example, following a significant three-way interaction, two separate twoway ANOVAs were performed as simple effects tests (using appropriate error terms), one assessing the me-
lation
between
patients
with
side of the lesion predominantly
and
caudality
dorsal
among
lesions
and
the
other assessing this relation among patients with predominantly ventral lesions. These tests, if significant, were followed up by further simple main effects tests. The relations between the dysphomia ratings and the measures
of functional
impairment
were
assessed
with
two-tailed Pearson correlations, as were the relations among the three dependent variables. An alpha level of
0.05 was
used
throughout.
com-
bined rating for the inferior frontal and temporal lobe regions was greater than the mean combined rating for the superior frontal, inferior pamietal, inferior occipital, and superior panietal and occipital regions. The classifications, therefore, indicated whether the lesion was predominantly
rather
lation. All analyses met univaniate and multivaniate
RESULTS
There were no significant relations between dysphonic mood and measures of functional impairment. Specifically, rating on the Visual Analogue Dysphonia Scale was not significantly correlated with the Barthel index (n= -0.02, dfSO, p>O.OS), with the measure of motor strength (r0.01, df=S0, p>O.OS), or with the aphasia
severity
score
(m=-0.19,
df=18,
p>O.OS).
The MANOVA assessing the relation between lesion location and depressive symptoms yielded a significant three-way interaction (Wilks’ lambda=0.69; approximately
distributed
as
F=6.16,
df=3,
42,
p