Background and Purpose This prospective study examined the prevalence and longitudinal course of generalized anxiety disorder (GAD) after stroke and its comorbidity with major depression over time. The contributions of lesion characteristics, functional impairment, and psychosocial factors to the development of GAD after stroke were studied.
Methods In a population-based cohort of 80 patients with acute stroke, we assessed GAD and comorbid major depression, functional ability, and social network at regular time points over 3 years. Cerebral atrophy and brain lesion parameters were determined from CT scans performed at the acute stage and after 3 years.
Results The prevalence of GAD after stroke was 28% in the acute stage, and there was no significant decrease through the 3 years of follow-up. At 1 year, only 23% of the patients with early GAD (0 to 3 months) had recovered; those not recovered at this follow-up had a high risk of a chronic development of the anxiety disorder. Comorbidity with major depression was high and seemed to impair the prognosis of depression. At the acute stage after stroke, GAD plus depression was associated with left hemispheric lesion, whereas anxiety alone was associated with right hemispheric lesion. Cerebral atrophy was associated with both depression and anxiety disorder late but not early after stroke. Dependence in activities of daily living and reduced social network were associated with GAD at all follow-up periods except at the acute stage.
Conclusions GAD after stroke is a common and long-lasting affliction that interferes substantially with social life and functional recovery. There is a differentiation of factors implicated in its development based on the period of time since the stroke event.
Over the past 10 years, there has been a growing interest in psychiatric sequelae of stroke. Our study1 and those of others (for summary see References 2 and 3) have shown that depressive disorders often follow a stroke. Other common psychiatric disorders such as anxiety disorders have rarely been studied in this population. Clinically, anxiety can be potentially serious and disabling with manifold adverse consequences on a patient’s daily functioning, interpersonal relationships, and quality of life. Considerable symptom overlap exists between anxiety and depressive disorders. Community studies have shown a high comorbidity of anxiety disorders with major depression.4 From the Oxfordshire Community Stroke Project, House et al5 reported prevalences of 3% for agoraphobia and 1% for GAD at 6 months after stroke. In a study by Starkstein et al,6 GAD was found in 24% of patients with acute stroke. Most of these patients also had a diagnosis of major depression. GAD alone was found in 6%. In a study by Castillo et al,7 GAD occurred in 11% of nondepressed stroke patients. This high frequency of anxiety disorder in patients after stroke justifies continued investigation. We have thoroughly followed a population-based cohort of stroke patients with repeated assessments of anxiety and depressive disorders, lesion variables, functional ability, and social network at regular time intervals for a period of 3 years.
The aim of this study was to (1) determine the prevalence of GAD during the immediate poststroke period and after 3 months, 1 year, 2 years, and 3 years; (2) examine the relationship of this anxiety disorder to major depression over time; (3) identify factors related to GAD at the various time points; and (4) examine the longitudinal course of GAD after stroke.
Subjects and Methods
All patients admitted to the stroke unit of the Department of Medicine, Umeå University Hospital (Sweden), during 1 year were considered for the study (n=98). This stroke unit and criteria for admission have been described in detail.8 Patients are admitted to the unit directly from the emergency department and are representative of all patients admitted to the hospital for acute stroke within a well-defined population in northern Sweden.8 This hospital is the only one serving this population. In the consecutive series of 98 patients, 16 patients died early after admission, 1 patient was excluded due to congenital mental handicap, and 1 refused to participate. The remaining 80 patients were the subjects of this study. Mean age was 73 years (range, 44 to 100 years); 49 were men and 31 were women. The majority (80%) suffered from their first stroke. The stroke diagnoses were cerebral infarction 79%, intracerebral hemorrhage 5%, and TIA 16%.
At follow-up after 3 months, 3 patients had died. One patient refused to participate; all other survivors were seen (n=76). After 1 year, another 3 patients had died. Two patients had a recurrent stroke and were excluded. One patient refused to participate; all others were seen (n=70). Two years after the stroke, 11 more patients had died. All survivors were seen except 1 patient who had moved out of the region (n=58). At the 3-year follow-up, another 4 patients had died. Two patients had a recurrence of stroke and were excluded. Two patients and/or their relatives refused to participate; 1 partly refused and was excluded. All other survivors were seen (n=49).
ADL was recorded according to the criteria of Katz et al.9 The psychiatric interviews were performed by the same psychiatrist (M.Å.) without knowledge of the radiological and neurological assessments. GAD was diagnosed based on DSM-III-R symptom criteria,10 which were slightly modified. In addition to the obligate criterion of excessive anxiety and worry about a variety of everyday events, DSM-III-R requires the presence of at least 6 of a list of 18 anxiety symptoms. Of these, as many as 9 symptoms belong to autonomic hyperactivity (eg, shortness of breath, tachycardia, nausea, and dizziness) and could not be systematically examined and separated in this group of medically ill patients. Instead, these symptoms were brought together under one heading called “autonomic hyperactivity.” In addition to that item, we considered 3 of 9 (4/10; 40%) symptoms of motor tension, scanning, and vigilance comparable to 6 of 18 (33%) symptoms traditionally required by DSM-III-R. This is in accordance with Castillo et al7 in their Baltimore studies. To get a view of comorbidity, a nonhierarchical approach was used (DSM-III-R criteria C and E were disregarded). In these patients, the DSM-III-R diagnosis of “organic anxiety syndrome” and “adjustment disorder with anxious mood” could have been fulfilled also. However, we chose DSM-III-R GAD criteria because they constitute a well-defined syndrome (see “Discussion”) free from preconceived assumptions of etiology.
The time constraint of 6 months or longer could not be met because patients initially were assessed for the presence of psychiatric symptoms during the interval between the acute stroke and the psychiatric examination. Patients were assessed 4 to 5 days after admission, after 10 days, and at discharge as a control for stability of symptoms. The significance of the modified time constraint is further elucidated in “Discussion.”
The diagnosis of a major depression was made on the basis of DSM-III-R symptom criteria10 with the physical disorder (stroke) on axis III.1 When appropriate (eg, in dysphasic patients), informants’ accounts and information from the staff were used to supplement patient interviews. In the acute stage, 9 patients were not assessable regarding GAD because of severe comprehension deficits; they could not reliably answer questions with affirmative or negative answers (71 patients were assessed). At 3 months 70 patients could be assessed, at 1 year 66, at 2 years 57, and at 3 years 49.
A questionnaire on living conditions and psychosocial function was administered at the same time points. We have reported our findings in detail elsewhere for this group of patients and compared them with a general population of elderly persons.11 12 Concerning the social network, it was shown that most of the patients had children (85%), and contact with children did not change after the stroke. Contact with other relatives, neighbors, and friends was more likely to change, and this was used in our study as an indicator of the social network as well as the factor of “living together” with someone or not (spouse or grown-up child).
CT scan of the brain was performed within the first 2 days after admission and also after 3 years. All CT scans were analyzed by the same neuroradiologist, who did not have access to the clinical assessments. Methods for analyzing the structural brain measurements have been presented in detail elsewhere.1
During the follow-up period, 3 patients were treated with antidepressants in low doses (1 patient took 25 mg/d nortriptyline, and 2 patients took 10 to 20 mg/d clomipramine). A few more patients had been treated with antidepressants for short periods (less than 3 weeks) but the treatment was canceled because of side effects or poor compliance. Fifteen patients were treated with benzodiazepines regularly early after stroke; the usage later in the course was not possible to document reliably. Benzodiazepines were given mostly as hypnotics at night and probably did not influence our results.
Informed consent was obtained from all subjects and/or their relatives, and the study was approved by the ethics committee of Umeå University.
Statistical analyses were performed using Systat.13 In the case of continuous measures, comparisons between groups were made by means of Student’s t test. For categorical data, χ2 tests were used as follows: for two-way square tables in which data represent paired comparisons, McNemar symmetry χ2 test was used; otherwise Pearson χ2 test or, when appropriate, Fisher’s exact test was used. All tests were two-tailed. A value of P<.05 was considered significant.
Characteristics of Patients Before Stroke
We have previously shown that demographic characteristics as well as prestroke social network and psychiatric morbidity did not differ from the general elderly population.11 The majority of the group lived at home (91%), while the remainder (9%) lived in homes for the aged. Of the 73 patients living at home, 45 (62%) lived with someone else (spouse and/or adult child). Fifty-one percent of the patients were married, 13% were single or divorced, and 36% were widowed. Thirteen patients (16%) had a personal history of psychiatric disorder. On the basis of retrospective assessments through interviews with the patient and informants who knew the patient well, none of the subjects fulfilled the criteria for GAD or major depression the month before the index stroke. Sixteen percent reported experiencing the symptom “nervousness/anxiety” and 6% “sadness” before the stroke. These frequencies do not differ from the general elderly population in Sweden.11 By the Katz criteria,9 12% of the prestroke patients were dependent on others in their ADL.1
At discharge, 39% of the patients were dependent on others in their ADL according to the Katz criteria.9 In this patient group, we have previously shown that this measure of functional impairment as well as paresis and other measures of functional ability differed significantly compared with prestroke and also compared with a general population of elderly persons.11 12 The majority were discharged to their own homes (66%), 8% to homes for the aged, and 26% to a geriatric hospital. Of the 71 patients who could be assessed, 20 (28%) fulfilled the criteria for GAD in the acute stage (Figure⇓). Of those, more than half (11/20) also had major depression. Thus, GAD alone was found in 9 patients (13%). Patients who lived alone at the time of their stroke were significantly more likely to develop GAD than patients living with another person (12/27 versus 8/44; P=.028 by Fisher’s exact test).
Ten patients (12%) had dysphasia and another 12% had mild dysphasic problems, 8% of which were transient during the first days after the stroke. In the following analyses, only the first-mentioned 10 patients were classified as dysphasic. Of patients with dysphasia, only 3 could be reliably assessed regarding GAD, and all of these patients had GAD (P=.020 by Fisher’s exact test). There was no significant association between the prestroke symptom of “nervousness/anxiety” and the complete syndrome of GAD. At this acute stage after stroke, no association between GAD and other social network variables, personal history of psychiatric disorder, age, sex, stroke diagnosis, or dependence in ADL by the Katz criteria9 was found. These results did not change when TIA patients were excluded from the analyses.
Of patients who could be assessed for GAD, recent lesions were visualized on the CT scan in 43 patients (61%). In the 39 patients with cerebral hemispheric lesions, 17 lesions were localized in the left hemisphere and 22 in the right hemisphere. In the 15 patients with GAD who had a CT-verified cerebral hemispheric lesion, there was no significant difference in frequency with regard to side of lesion (7 right, 8 left). But breaking down GAD cases by comorbidity with depression showed that GAD alone was significantly more often associated with right hemispheric lesion, whereas GAD comorbidity with depression was significantly associated with left hemispheric lesion (right, 5/7 versus 2/7; left, 1/8 versus 7/8; P=.041 by Fisher’s exact test). Thus, when the lesion was in the left hemisphere, comorbidity with depression was significantly higher than with a lesion in the right hemisphere. Patients with “pure” GAD had lesions in the right hemisphere significantly more often. Only 1 patient with pure GAD had the lesion localized in the left hemisphere. Three patients had pure major depression; all had lesions in the left hemisphere. There were no associations between GAD and frontality of lesion, lesion volume, subcortical versus cortical lesion, or cerebral atrophy.
Three Months After Stroke
Of the 70 patients assessable at 3 months, 22 were diagnosed as having GAD (31%). Of these patients, 15 also had major depression. Thus, GAD alone was found in 7 patients (13%). Significantly more GAD patients were dependent on others in their primary ADL than were patients without GAD (10/22 versus 9/48, P=.040 by Fisher’s exact test). Whether GAD was comorbid with depression did not change the result. The prior association with dysphasia remained (P<.031 by Fisher’s exact test). In regard to social network, the variable of “living alone” was no longer significant, whereas “few social contacts” outside the immediate family was then associated with GAD. Of the 22 patients with GAD, 6 had met a friend or relative other than a spouse or child in the last week as opposed to 26 of the 48 non-GAD patients (27% versus 54%, P=.042 by Fisher’s exact test). The result was the same whether GAD was comorbid with depression or not. No associations were found with age, sex, personal history of psychiatric disorder, living in a geriatric hospital or not, stroke diagnosis, or the CT scan parameters. Thus, the prior association with hemispheric side was not significant at this follow-up.
One Year After Stroke
At 1 year, the frequency of GAD was 24% (16/66). The decrease during the first year was not significant (P=.414 by McNemar symmetry χ2 test). Of patients with GAD, 9 had comorbid major depression; 7 had GAD alone. Patients with GAD had significantly fewer contacts with friends and relatives except their children (P=.004 by Fisher’s exact test). The prior association with capacity in ADL remained (P=.039 by Fisher’s exact test). No other significant associations were found.
Two Years After Stroke
Approximately the same proportion of patients was diagnosed as having GAD (14/57; 25%). Eight patients also had major depression. Thus, 6 patients had GAD alone. The association between GAD and fewer social contacts remained (P=.002 by Fisher’s exact test) and so did the prior association with capacity in ADL (P=.021 by Fisher’s exact test). No other significant associations were found.
Three Years After Stroke
There was no significant change in the proportion of patients diagnosed as having GAD (9/48; 19%). Seven patients also had major depression; 2 patients had GAD alone. The association between GAD and few social contacts was also strong at this follow-up (P=.002 by Fisher’s exact test). Of the 9 GAD patients, no one had met a friend or relative other than their children during the last week as opposed to 24 of the 39 patients without GAD (P=.002 by Fisher’s exact test). This last proportion does not differ from the general elderly population in Sweden, where approximately half had met a friend or relative at least once a week.1 The prior association with capacity in ADL remained (P=.021 by Fisher’s exact test). No associations were found with other social network variables or living conditions, age, sex, personal history of psychiatric disorder, dysphasia, stroke diagnosis, or the initial lesion parameters.
Another CT scan was performed at 3 years (n=46; in 2 patients CT scan was not obtained). Cortical atrophy at 3 years significantly differed between patients who had GAD and those who did not (7/7 versus 19/39, P=.014 by Fisher’s exact test). Of patients who had had a CT scan at 3 years, there were only 2 patients with GAD alone (not comorbid with depression). Both patients had cortical atrophy. Also, subcortical atrophy (as measured by the anterior horn index) showed a significant difference between patients with and without GAD (mean±SD, 29.3±3.1 versus 25.9±4.8; P=.032, two-tailed t test).
To control for the possibility that mortality could explain the results concerning changes in frequencies and associated factors, a subsequent analysis was limited to the group of patients who had survived up to 3 years. The frequencies of GAD for these long-term survivors of stroke are shown in the Figure⇑ and did not significantly differ from the cross-sectional frequencies. The results concerning associated factors did not change; the same factors as in the cross-sectional analyses were associated with GAD at the different time points.
Longitudinal Course of GAD After Stroke
Of the 24 patients with early GAD (ie, onset within 3 months after stroke), 21 were reassessed at 1 year. Of these patients, 5 (23%) had recovered from the anxiety disorder within the first year after stroke.
At 3 years, 13 patients who had GAD early after stroke had survived and were reassessed. Restricting the analysis to these survivors, 4/13 (31%) had recovered from GAD at 1 year. At 3 years, only 1 more patient had recovered (38%). Hence, of patients with early GAD, 62% (8/13) had such a long-lasting affliction that they did not recover during the following 3 years.
Of these 13 patients with early GAD who were followed up for 3 years, only 2 patients had GAD without comorbid depression at any period. Thus, the majority (11/13) also fulfilled the criteria for major depression at some period during these 3 years. In 5 of these patients, GAD was preexisting and there was a superimposition of a major depression. The other cases began with comorbidity. Due to the high frequency of comorbidity, it was not possible to compare the course of GAD with and without comorbid depression.
In an attempt to study the influence of this comorbidity on the course of depression, patients with pure early depression (n=8) and early depression comorbid with GAD (n=14) were compared. In the group with pure depression, the majority (6/7, 86%) had recovered from depression at 1 year, whereas in the comorbid group half as many (6/14, 43%) had recovered from depression. However, this difference was not significant (P=.159 by Fisher’s exact test).
This prospective study demonstrates the high prevalence of GAD after stroke. The prevalence remained high throughout the 3 years of follow-up. The longitudinal course and the poor prognosis of anxiety disorder after stroke have not been demonstrated previously. Comorbidity with major depression was high at all time points and seemed to impair the prognosis of depression. Our prospective data also provide information on changes over time in biological as well as psychosocial correlates of anxiety disorder after stroke. Before proceeding with a discussion of these findings, some methodological issues need to be pointed out.
The problem of spurious associations in studies of hospital inpatients14 may be less severe in the Scandinavian countries than in other medical settings because more than 90% of stroke patients (excluding those with TIA) are admitted to the hospital.15 The method for patient selection ascertains a representative sample of patients admitted to the hospital for stroke.8 We have avoided sampling bias by studying a consecutive series of patients during 1 year. This population-based cohort was followed up almost completely at the different time intervals up to 3 years. Thus, our study group should permit generalization of these results to other stroke patients.
Since the whole spectrum of cerebrovascular disease was covered, the patient group included TIA as well as recurrent stroke. There were no major differences between either of the subgroups and the rest of the study group in GAD, depression, and the reported associated variables. The small number of patients in the subgroups may partially explain this, but it seems likely that lower functional ability already present before the stroke11 (also in the TIA group) contributes to reduced differences between subgroups.
Some methodological difficulties relating to diagnosis of anxiety disorders in medically ill patients need to be discussed. For example, we found that panic disorder was not possible to assess with any accuracy in the acute stage after stroke because of substantial overlap with symptoms of physical illness. Also, phobic behavior was difficult to diagnose in a stringent manner in this patient group, since the majority of patients described fear of being alone outside the home, of having another stroke, of being unable to handle a situation because of disabilities, etc. These fears often were realistic, and we decided not to register them as agoraphobia or social phobia. These adjustment reactions need further study, especially in a longitudinal perspective. Normally, they fade out with the passage of time. Nevertheless, they seem clinically to have a considerable negative influence on psychosocial life and relationships.
In this study, we exclusively applied the DSM-III-R criteria of GAD10 because they constitute a well-defined syndrome for research purposes. It is used by others7 with similar modifications. Because our study is prospective and longitudinal with a very low rate of missing subjects, it was possible to evaluate the effects of our modified time criterion (in DSM-III-R, a duration of 6 months is required). Intraindividual comparisons showed that the frequency of GAD did not decrease significantly between 3 and 12 months (illustrated in the Figure⇑; line graph shows long-term survivors). Thus, the modified time constraint did not seem to lead to overdiagnosis of the prevalence of GAD.
The prevalence of GAD was high in stroke patients; in fact, 1 patient in 4 suffered from that severe anxiety state in the acute stage after stroke. Since all dysphasic patients were not possible to assess and GAD was more frequent in dysphasic patients, the prevalence of GAD might have been even higher. The prevalence is in accordance with that seen by Castillo et al,7 who found that GAD was present in 27% of stroke patients. Starkstein et al6 found 24% GAD using modified DSM-III criteria.
Our study is the first prospective longitudinal study extended to 3 years after stroke, and it gives further information about the longitudinal course of the disorder. Only 1 patient in 4 with early GAD (0 to 3 months) had recovered at 1 year. If the patient had not recovered by 1 year, there was a high risk of development of a chronic anxiety state. If chronicity is defined by a criterion of 2 years’ duration (an established limit in psychiatric research), 74% of GAD after stroke had a chronic course.
In a previous study,1 we have shown that major depression after stroke had a more favorable prognosis. At 1 year, 60% of the patients with early depression had recovered, a more than doubled 1-year outcome compared with GAD (23% 1-year outcome). Our results suggest that comorbidity with GAD seems to impair the prognosis of major depression after stroke. We have shown that GAD after stroke has a very high comorbidity with depression; actually, 85% of GAD patients also fulfilled the criteria for major depression at some time during the 3-year follow-up period. This is higher than among patients without demonstrable brain lesions. In the Epidemiologic Catchment Area study,4 it was estimated that during a 6-month period depression occurred in 21% of adults with an anxiety disorder. In a long-term care facility, Parmelee et al16 found that 60% of patients with GAD also had major depression.
In general psychiatry, the question of anxiety and depression as separate entities or as a spectrum disease remains controversial (for review, see Reference 17). The elaboration of these concepts is not only of academic relevance but has implications for treatment that will be discussed later. With a few exceptions,2 6 7 18 research into poststroke psychiatric disorders has given this question limited attention.
The prevalence of GAD without depression was 13% early after stroke. This is in line with the work of Castillo et al,7 who reported 11%, whereas others have reported lower frequencies.2 3 19 In our study, the low frequency late after stroke (4% at 3 years) was not a significant decrease but an aggravation over time into comorbidity with depression. In the general elderly population, the prevalence of GAD varies depending on the criteria used for case definition (for review, see Reference 20). Blazer et al21 found that when cases of major depression were excluded, the 6-month prevalence of GAD in patients over 65 years was 1.9%. Thus, the prevalence of GAD in stroke patients is considerably higher.
This study has provided evidence of a differentiation of factors likely to be implicated in the development of anxiety disorder after stroke based on the period of time since the stroke event. In the acute poststroke period there was an association with lesion location: pure GAD was associated with right hemispheric lesion, whereas comorbid anxiety/depression and depression1 was significantly associated with left hemispheric lesions, thus confirming results from the Robinson group.6 7 22 Our findings have clarified that the influence from lesion location was significant only in the acute period after stroke.
Besides this biological correlate of GAD, the psychosocial parameter of living alone and dysphasia were correlated with the development of GAD in the acute stage after stroke. This is in accordance with our previous findings on poststroke depression.1 Under the stressful condition of an acute stroke, being without the social support from living with a family seems to promote the development of both anxiety disorder and depression after stroke. From 3 months and onward, patients with GAD had fewer social contacts outside the family than had nonanxious patients. This indicates that GAD may result in a reduced social network or that patients with a poor social network are at greater risk for developing an anxiety disorder. In this system of continuous interaction, it is not possible or meaningful to isolate dependent and independent variables. Most important is the clinical implication that early and active treatment efforts, both pharmacological and psychosocial, are necessary to break this vicious circle of negative interactions. Otherwise, there is a very high risk that the anxiety will take a chronic course.
At all time periods after discharge from the hospital, dependence in ADL was associated with generalized anxiety (Table⇓). Thus, anxiety was not an immediate response to loss of function. The fact of being dependent on others in daily life will probably be most obvious after discharge from the hospital and return to everyday life. However, there were only a few new cases of GAD at 3 months, so the most plausible interpretation seems to be that generalized anxiety contributes to poor functional ability.
Depression after stroke is associated with cerebral atrophy,1 a result in line with earlier findings of an association of depression and degenerative changes in the brain in old age.23 24 Late but not early after stroke, GAD also was associated with cerebral atrophy. In fact, all patients with GAD at 3 years after stroke had cortical atrophy. Subcortical atrophy also was associated with GAD. As there were no new cases of GAD occurring at 3 years, our group of GAD patients at this follow-up consisted of patients with a chronic course. Thus, the main influence of cerebral atrophy on GAD seems to be on the maintenance and not onset of the disorder.
GAD after stroke is common and long-lasting. Contrary to the traditional view that GAD is a mild disorder, we found a substantial interference with social relationships and functional recovery. This will make heavy demands on early and effective treatment. However, data on treatment of patients with severe anxiety and comorbid depression are limited, especially for the elderly, and unfortunately are quite lacking for the stroke population. There are findings from patients without brain lesion that azapirones such as buspirone, a 5HT1 antagonist, are effective for patients with GAD and coexisting depressive symptoms.25 Also of concern are findings from a community study26 that none of the comorbid depressed patients had been given antidepressant medication. Instead, they were inappropriately treated with only benzodiazepines, which will not alleviate the core depressive symptoms. There is a definite need to educate physicians further about the benefit of antidepressant therapy for this patient group.
Selected Abbreviations and Acronyms
|ADL||=||activities of daily living|
|DSM-III-R||=||Diagnostic and Statistical Manual of Mental Disorders, edition 3, revised|
|GAD||=||generalized anxiety disorder|
|TIA||=||transient ischemic attack|
This study was supported by grants from the Swedish Society of Medicine, the Foundation for Stroke Research (1987), the Joint Committee of the North Region, and the Fund for Medical Research, Umeå University (Sweden).
- Received August 18, 1995.
- Revision received November 20, 1995.
- Accepted November 20, 1995.
- Copyright © 1996 by American Heart Association
Åström M, Adofsson R, Asplund K. Major depression in stroke patients: a 3-year longitudinal study. Stroke. 1993;24:976-982.
House A. Mood disorders after stroke: a review of the evidence. Int J Geriatr Psychiatry. 1987;2:211-221.
Starkstein SE, Robinson RG. Affective disorders and cerebral vascular disease. Br J Psychiatry. 1989;154:170-182.
Regier DA, Narrow WE, Rae DS. The epidemiology of anxiety disorders: the Epidemiologic Catchment Area (ECA) experience. J Psychiatr Res. 1990;24:3-14.
House A, Dennis M, Moridge L, Warlow C, Hawton K, Jones L. Mood disorders in the year after first stroke. Br J Psychiatry. 1991;158:83-92.
Strand T, Asplund K, Eriksson S, Hägg E, Lithner F, Wester PO. A non-intensive stroke unit reduces functional disability and the need for long-term hospitalization. Stroke. 1985;16:29-34.
Katz S, Ford AB, Moskowitz RW, Jackson BA, Jaffe MW. Studies of illness in the aged. The index of ADL: a standardized measure of biological and psychosocial function. JAMA. 1963;185:914-919.
American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 3rd ed, rev. Washington, DC: American Psychiatric Association; 1987.
Åström M, Asplund K, Åström T. Psychosocial function and life satisfaction after stroke. Stroke. 1992;23:527-531.
Wilkinson L. SYSTAT: The System for Statistics. Evanston, Ill: SYSTAT Inc; 1988.
House A, Dennis M, Warlow C, Hawton K, Molyneux A. Mood disorders after stroke and their relation to lesion location: a CT scan study. Brain. 1990;113:1113-1129.
Stavrakaki C, Vargo B. The relationship of anxiety and depression: a review of the literature. Br J Psychiatry. 1986;149:7-16.
Johnson GA. Research into psychiatric disorder after stroke: the need for further studies. Aust N Z J Psychiatry. 1991;25:358-370.
Blazer D, George LK, Hughes D. The epidemiology of anxiety disorders: an age comparison. In: Salzman C, Lebowitz B, eds. Anxiety in the Elderly: Treatment and Research. New York, NY: Springer; 1991.
Robinson RG, Starr LB, Lipsey JR, Rao K, Price TR. A two-year longitudinal study of post-stroke mood disorders: findings during the initial evaluation. Stroke. 1983;14:736-744.
Jacoby RJ, Dolan RJ, Levy R, Baldy R. Quantitative computed tomography in elderly depressed patients. Br J Psychiatry. 1983;143:124-127.
Roth M. Depression and affective disorder in later life. In: Angst J, ed. The Origins of Depression: Current Concepts and Approaches. Dahlem Konferenzen 1983. Berlin, Germany: Springer-Verlag; 1986:39-75.
Pecknold JC. Mixed anxiety and depressive disorder. Curr Can Psychiatry. 1993;1:5-9.
Lydiard RB. Coexisting depression and anxiety: special diagnostic and treatment issues. J Clin Psychiatry. 1991;52(June suppl):48-54.