Aggressive Behavior in Patients With Stroke: Association With Psychopathology and Results of Antidepressant Treatment on Aggression

Methods 

All patients (N=92) were enrolled in Iowa and gave their informed consent to participate in a 12-week, double-blind trial of treatment with antidepressants for recovery after stroke.13, 14 Originally, there were 104 patients in these studies, but 12 patients who were enrolled in Argentina were excluded from this study because some of the PSE items related to aggression were missing. The 92 patients were recruited from the Younkers Rehabilitation Center of the Iowa Methodist Medical Center in Des Moines (n=89), the University of Iowa Hospitals and Clinics in Iowa City (n=1), and the Veterans Affairs Medical Center in Iowa City (n=2). The patients were between the ages of 18 and 85 years, had had an acute stroke within the past 6 months, and had a diagnosis of major or minor depression, or no depression. Exclusion criteria were: (1) any significant medical condition that was life-threatening or would interfere with recovery from stroke, (2) severe comprehension deficit resulting from decreased consciousness, dementia, or aphasia, (3) history of previous head injury, and (4) history of any previously diagnosed brain disease other than stroke.

After the patients were given a full explanation of the study and their informed consent was obtained, they were evaluated with a version of the semi-structured interview, the PSE,15 which was modified to assess diagnostic criteria according to the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV).16 Patients were diagnosed as having depression resulting from stroke with major depressive-like episode, or minor depression (research criteria), based on DSM-IV criteria, or no depression, using symptoms elicited by the PSE. We17 have previously reported the reliability and validity of this instrument. We measured the severity of depression with the 17-item HAMD18 and the severity of anxiety with the Hamilton Anxiety Rating Scale (HAMA).19 We used the Mini-Mental State Examination (MMSE),20 an 11-item instrument, to assess cognitive impairment. Scores on the MMSE range from 0 to 30, with lower scores indicating greater impairment. The reliability and validity of these instruments in treating stroke patients has been demonstrated in other studies.21, 22 We measured impairment in activities of daily living (ADLs) with the Johns Hopkins Functioning Inventory (JHFI).23 This is a 10-item inventory and scores range from 0 to 27, with higher scores indicating greater impairment. Quantitative assessments of social functioning were made with the Social Ties Checklist, which assesses the quantity of available social support, with scores ranging from 0 to 10; higher scores indicate less support. These instruments have also been proven reliable and valid for use in stroke populations.22, 24

Blind to any knowledge of treatment outcome or other variables, patients were identified as being irritable and aggressive based on positive ratings on 1 or more of the following 5 items on the PSE: (1) the patient acknowledges showing anger by shouting or quarreling (1 point), or showing anger by hitting people or throwing or breaking things (2 points); (2) 1 discrete episode of violent behavior that had a catastrophic impact on others (eg, significant bodily injury) (2 points) or more than 1 discrete episode of violence (3 points); (3) the patient showed hostile behavior to examiner through anger, irritability, or overt aggression (1 point); (4) the patient showed agitation during the interview (1 point); or (5) the patient showed gross excitement or violence during the interview (1 point). These behaviors were reported by the patient or by a family member or by a hospital staff member. All patient reports were confirmed by a family member or by hospital personnel. The items on the PSE reflect 3 of the 4 categories of aggression described in the OAS,11 namely, verbal aggression, physical aggression against objects, and physical aggression against people. An “aggression subscore” was calculated from these 5 PSE items, with a maximum possible score of 8 points. A diagnosis of irritability and aggression required that 1 or more items be positive.

Computerized tomography scans taken after the acute stroke were obtained from the treating hospital. The scans were read by a neurologist who was blind to any of the psychiatric assessments. The size of the lesion was calculated as a percentage of total brain volume, using the ratio of the largest cross-sectional area of the lesion to the area of the brain slice that included the body of the lateral ventricles.25 The lesion’s anterior location (expressed as a percentage of the total anteroposterior [AP] diameter of the brain) was defined as the mean distance of the lesion’s anterior border from the frontal pole, for all slices in which the lesion was visible.23, 26 Linear measurements of brain atrophy were done according to the method described by Gomori et al27 on the side contralateral to the lesion. The bifrontal ratio was calculated as the distance between the tips of the frontal horns divided by the distance between the inner tables of the skull along the same line, measured on the cut best showing the caudate nuclei. The bicaudate ratio was the minimal distance between the caudate indentations of the frontal horns divided by the distance between the inner tables of the skull along the same line, measured on the same cut as the bifrontal ratio. The lateral ventricular brain ratio (VBR 2) was the multiple of the minimal coronal distance between the lateral walls of the lateral ventricles (at the ventricular waist) and the saggital distance between the extreme anterior and posterior extents of the lateral ventricles, divided by the multiple of the coronal inner table diameter (at the ventricular waist line) and the mid-saggital inner table diameter, measured on the cut showing the bodies of the lateral ventricles above the thalamus. The third ventricular brain ratio (VBR 3), measured on the cut showing most of the third ventricle, was the multiple of the maximal sagittal and coronal diameters of the third ventricle divided by the multiple of the transpinal and mid-sagittal inner table diameters. The reliability of their measurements has been demonstrated previously.28

Results 

Patient Characteristics 

Twenty-five percent (23/92) of the patients were irritable or aggressive, as defined by an aggression score of 1 or greater. Overall, the level of aggression was mild, with only 2 patients showing physical aggression against people or objects and the remainder showing verbal aggression (ie, the PSE item that asks if the patient “shows anger by shouting or quarreling” was answered positively). The demographic characteristics of all patients are shown in table 1. There were no significant differences between the 23 aggressive patients and the 69 nonaggressive subjects except for an increased frequency of major depression (56.5% aggressive vs 19.1% nonaggressive; χ12 test=11.03, P=.005), or generalized anxiety disorder (GAD) (34.8% aggressive vs 7.4% nonaggressive; Fisher exact test, P=.003) among the irritable patients but no difference in the frequency of minor depression (4.3% aggressive vs 13.2% nonaggressive; P=not significant [NS]).

Table 1. Demographic Characteristics and Psychiatric Symptoms of Aggressive and Nonaggressive Patients

	Characteristic	Aggressive (n=23)		Nonaggressive (n=69)
		Mean ± SD		Mean ± SD
	Age (y)	65.8±12.6		68.9±11.9
	Education (y)	11.5±2.9		12.2±2.5
	Days since stroke	58.6±14.4		40/1±7.9
		n	%	n	%
	Male sex	17	73.9	42	60.8
	White race	22	95.6	68	98.5
	Married	16	69.6	43	62.3
	Socioeconomic status Hollingshead class IV or V46	10	43.5	28	40.6
	Right handedness	22	95.6	67	97.1
	History of alcohol abuse	4	17.4	4	5.8
	Prior psychiatric history	6	26.1	8	11.6
	Family psychiatric history	7	30.4	13	18.8
	DSM-IV major depression	13	56.5	13	19.1⁎
	DSM-IV minor depression	1	4.3	9	13.2
	DSM-IV GAD	8	34.8	5⁎	7.4†

⁎ χ42 test=11.00, P=.001. † P=.003.

Neurologic and Radiologic Findings 

The study’s neurologic and radiologic findings are shown in table 2. There were no differences between aggressive and nonaggressive patients in the type of stroke, side of lesion, neurologic deficit, or degree of brain atrophy. Lesion volume, however, was larger in aggressive patients although this was not statistically significant (Mann-Whitney U test). The anterior edge of the lesion (expressed as % AP diameter of the brain) was significantly closer to the frontal pole of the brain in aggressive patients than in nonaggressive patients (17.5%±8.8% vs 35.4%±18.7%; t44=−2.93, P=.005). This finding held under nonparametric analysis (Mann-Whitney U test, χ12 test=8.27, P=.004).

Table 2. Stroke Characteristics, Neurologic Findings, and Radiologic Findings for Aggressive and Nonaggressive Patients

	Characteristic of Stroke	Aggressive (n=23)	Nonaggressive (n=69)
	Stroke type (%)
	Infarction	91.3	87.0
	Hemorrhage	8.7	13.0
	Location (%)
	Left hemisphere	34.8	34.8
	Right hemisphere	56.5	59.4
	Brainstem/other	8.7	5.8
	Neurologic deficit (%)	(n=22)	(n=60)
	Motor deficit	81.8	68.3
	Sensory deficit	27.3	31.7
	Visual field deficit	31.8	25.0
	Aphasia	9.1	16.7
	Mean linear brain measurements ± SD	(n=16)	(n=51)
	Lateral ventricle/brain ratio	.110±.044	.125±.044
	Third ventricle/brain ratio	.007±.003	.007±.003
	Bifrontal ratio	.326±.087	.334±.059
	Bicaudate ratio	.155±.058	.164±.051
	Mean stroke lesion measurements ± SD	(n=10)	(n=36)
	Lesion volume (% of total brain volume)	8.77±10.55	4.14±5.84
	Anterior location (% of AP diameter of brain)	17.47±8.82⁎	35.43±18.66

⁎ P=.005.

Psychometric Measures, Functional Measures, and Psychiatric Symptoms 

Irritable and nonirritable patients differed on various psychometric measures. Aggressive patients had significantly higher depression rating scores on the HAMD than did nonaggressive patients (table 3) and were also significantly more anxious, as measured on the HAMA. Cognitive impairment, measured on the MMSE, was significantly greater in the aggressive patients than in the nonaggressive patients (see table 3). The level of total psychopathology, as measured by the total PSE score, was also significantly greater among the aggressive than among the nonaggressive patients.

Table 3. Psychopathology and Impairment Scores of Aggressive and Nonaggressive Patients

	Measures	Aggressive (n=23)	Nonaggressive (n=69)	P
		Mean ± SD	Mean ± SD
	HAMD score	13.2±6.5	7.90±5.32	.000
	HAMA score	11.3±4.9	7.79±4.59	.002
	PSE score	19.9±13.3	9.09±7.28	.000
	JHFI score	6.96±5.24	5.57±3.88	.182
	MMSE score	23.0±7.0	26.2±4.7	.015
	PLACS score	2.33±3.83	2.17±3.76	.925
	Irritability/aggression score	2.00±0.97	0.00±0.00	.000

Treatment Response 

We evaluated the response of irritability to treatment with antidepressant medication in the 23 initially aggressive patients who entered the 12-week treatment trial. Among those patients, 6 were treated with fluoxetine titrated up from 10 to 40mg/d, 4 were treated with nortriptyline titrated up from 25 to 100mg/d, and 13 received a placebo. The 10 patients who were given active antidepressant medication did not differ significantly from the 13 placebo patients in demographic characteristics, psychometric measures, or prevalence of major depression or GAD at baseline. Actively treated patients had aggression scores of 2.00±0.87 at baseline and 0.20±0.45 at 3 months, while patients on placebo had scores of 2.00±1.00 and 0.08±0.28 at baseline and 3 months. Efficacy analysis that compared only the 20 subjects (13 placebo, 7 treated) who completed the study also failed to show any significant effect of treatment (repeated-measures, F1,18=.28, P=.60).

Because aggression was associated with major depression, we analyzed whether response to treatment was associated with reduced aggression. Repeated-measures ANOVA of the 7 aggressive patients with HAMD scores greater than 10 (the approximate mean score for patients with poststroke minor depression22) and who had a greater than 50% decline in HAMD scores (ie, the standard definition for treatment response) showed a significant time-by-treatment interaction (F1,14=5.61, P=.033) (fig 1) when compared with aggressive patients who did not have a 50% decline in their HAMD score. Of the 7 patients with a greater than 50% decline, 3 were treated with nortriptyline, 1 with fluoxetine, and 3 were given a placebo. Of the 9 patients without a decline, 1 was given nortriptyline, 4 were given fluoxetine, and 4 were given a placebo. The number of patients with no aggression after treatment was significantly greater in those who responded to treatment than in those who did not (Fisher exact test, P=.03). There were no statistically significant differences between groups in demographic characteristics or severity of impairment scores or baseline HAMD scores (HAMD scores: responders, 17.3±5.1; nonresponders, 16.4±6.1; t14=.91, P=NS). The groups also did not differ in the frequency of major, minor, or no depression.

View Large Image Download to PowerPoint

Fig 1. Change in aggression score as derived from the PSE at the beginning and the end of a 12-week, double-blind treatment trial comparing effectiveness of nortriptyline, fluoxetine, and placebo in treating aggression after stroke. Active versus placebo analysis did not show a significant time-by-treatment interaction. Patients who responded to treatment of depression, however, showed a significantly greater decline in aggression scores than did patients who did not respond to treatment (F1,14=5.61, P=.033).

We also analyzed whether patients with significant anxiety and irritability symptoms would be less irritable after 12 weeks of treatment with fluoxetine, nortriptyline, or placebo. There were 15 patients with HAMA scores of 10 or higher (ie, arbitrary cutoff equivalent to mild GAD). Repeated-measures ANOVA of placebo (n=6) versus nortriptyline or fluoxetine (n=9) showed no significant time-by-treatment effect on aggression score (ie, nortriptyline: initial, 2.60±1.34; final, 0.80±1.10; fluoxetine (n=4): initial, 1.25±0.96; final, 1.25±0.96; placebo: initial, 1.67±1.03; final, 0.67±1.63; F1,12=.89, P=.43). Similarly, when patients with HAMA scores of 10 or higher who responded to treatment (ie, had a >50% drop in HAMA scores) were compared with patients who did not (n=9), there was no significant time-by-treatment interaction (F1,11=2.66, P=.131). This finding suggests that a decrease in depressive symptoms has a greater impact on aggression severity than does a decrease in anxiety symptoms.

Discussion 

This study found that among patients with a recent stroke, irritability and aggression were common (25%). The irritability, however, was generally mild, consisting predominantly of verbal aggression. Aggression was associated with a higher frequency of major depression and GAD, higher depression and anxiety rating scores, greater cognitive impairment, and closer proximity of the lesion to the frontal pole. It is noteworthy that these results are consistent with our findings from the earlier study,1 despite the fact that the 2 studies involved vastly different samples with regard to socioeconomic status, racial makeup, and degree of aggression required to be included in the study. In addition, this study found that although irritability in itself did not respond specifically to antidepressants, improvement in depression symptoms may lead to significantly greater reduction in aggressive behavior compared with such behavior in patients who remain depressed.

The methodologic limitations of this study should be acknowledged. The patients were enrolled in 2 double-blind studies designed to assess the therapeutic effect of antidepressants on poststroke depression and the prevention of poststroke depression and recovery in cognitive function and ADLs. Thus, the study was not designed to specifically treat aggression after stroke and our findings are limited to the patients who were willing to participate in studies about treating or preventing poststroke depression. Furthermore, the identification of aggressive patients was based on positive ratings on self-reported items of the PSE that were confirmed by family members or staff members providing treatment. Hence, we did not assess severe forms of aggression. It seems unlikely, however, that patients and families would over-report aggressive behavior, so it is possible that the severity of some patients’ aggression was underrated. Finally, based on our findings of significant aggression scores in 7 patients who responded to depression treatment and 9 patients who did not, these findings must be viewed as preliminary and further studies are needed.

Given these limitations, how might these study findings be construed? First, the cross-sectional frequency of aggression was 25% (23/92), compared with 5.8% (18/308) in our earlier study.1 This difference undoubtedly resulted from the higher threshold for aggression that was required for inclusion in the previous study. In nursing home populations, however, the reported frequency of aggressive behavior has ranged from 48% to 64%.10, 29, 30 The fact that the clinical and neuropathologic correlates of mild aggression in this population replicated the findings in a much more severely aggressive population suggests that the factors that produce mild irritability are the same as those that produce more severe aggression. Perhaps premorbid personality or the number of additive risk factors (eg, low MMSE score plus a lesion closer to the frontal pole) may have led to more severe aggression.

This study was also consistent with studies of aggression in other neurological disorders. Kant et al,31 for example, administered 50 to 200mg of sertraline to 13 patients 2 years after TBI in an open trial and found significant reductions in irritability and aggressive outburst during 8 weeks of treatment. Similar positive effects of antidepressants on aggression have been reported in patients with Alzheimer’s disease32, 33 and Huntington’s disease.34

The aggressive patients in this study had lesions that were more frontal than in nonaggressive patients. We found this lesion location to be an independent predictor of aggression. Frontal lobe lesions may result in aggression by causing a personality change, which includes an increase in aggression, or by causing an exacerbation of a preexisting aggressive personality trait.4 Alternatively, aggressive behavior may result from disinhibition caused by frontal lobe lesions. Disinhibited behavior as noted on the PSE, however, was no more common among our aggressive patients than among our nonaggressive patients.

We also found that aggressive patients had a higher frequency of major depression and higher depressive rating scores than nonaggressive patients. Irritability is known to be part of the symptomatology of depression,16 so the association was not surprising. The reason for this association may be because of similarities in underlying neurotransmitter dysfunction. Both depression and aggression have been associated with low cerebrospinal fluid levels of 5-hydroxyindoleacetic acid, a metabolite of serotonin.35 This underlying neurotransmitter dysfunction may result in both disorders. Alternatively, Calvert et al36 noted that irritability in stroke patients was associated with “emotionalism,” described as an increase in frequency of crying or laughing caused by loss of control of emotional expression. To test this hypothesis, we compared aggressive and nonaggressive patients for mood lability as measured by the Pathological Laughter and Crying Scale (PLACS).37 We did not, however, find significantly higher PLACS scores in the aggressive patients when compared with the nonaggressive patients (see table 3).

This study also found that aggressive patients were more cognitively impaired, as measured by the MMSE. Studies of demented patients have also found a similar association of aggression with cognitive impairment.38, 39 Greater cognitive impairment in aggressive patients may be caused by the combination of lesion location (ie, frontal) and depression, both of which are associated with cognitive impairment.40 Impaired executive function and loss of inhibition of aggressive impulses may contribute to aggressive behavior. Other hypotheses regarding the role of cognitive impairment in aggression, such as frustration because of impaired ability to communicate, inappropriate response to stimuli such as pain or noise,41 and misinterpretation of a caretaker’s action39 might also be proposed. We looked, however, at distractibility on the PSE, which may reflect attention deficit and an inability to filter out excessive stimulation, and did not find an association with aggression.

Another finding was the association of anxiety with aggression.16 Anxiety has been conceptualized along the same continuum of behavior as aggression.42 In a study of aggression in the elderly, aggressive behavior was associated with an anxiety disorder in 31% of the patients,43 suggesting that anxiety accompanies, or may even cause, aggression in certain populations.

Finally, we believe this is the first study to show that aggression among patients with a recent stroke can be reduced by treatment of the accompanying depressive symptoms. Aggressive behavior is a common reason for referral to a psychiatric service or admission to an institution.8, 9 Aggressive patients may be treated with many medications, including antipsychotic agents, benzodiazepines, β-blockers, anticonvulsant agents, and serotonin-specific antidepressants,44 which may cause undesirable side effects. Cohen-Mansfield45 found that agitated patients received more medication and tended to have more falls than nonagitated patients. Also, if aggression is a common reason for patients to be admitted to an institution, patients who become aggressive after a recent stroke may be prematurely and unnecessarily institutionalized if caretakers are unaware that the depressed patient’s behavior may be improved with treatment with antidepressants.

Conclusions 

Irritability and aggression are common after stroke and are associated with major depression, generalized anxiety disorder, greater cognitive impairment, and proximity of the lesion to the frontal pole. The finding that antidepressants may reduce aggression among depressed patients with stroke is preliminary and speculative. It does, however, invite further research to assess whether early detection of anxiety and depression and consequent treatment with antidepressants will reduce the incidence or shorten the course of aggression after a recent stroke.