Abstract
Psychiatric morbidity following stroke includes a wide range of cognitive and emotional disturbances, of which depression has received the most attention [1]. The combined frequency of all types of post-stroke depression (PSD) varies from 25% to 79%; frequency for major depression (MD) ranges from 3% to 40%; and for minor depression from 10% to 40% [2, 3]. Post-stroke depression was found in 27% to 57% of first time stroke patients [4–6].
Reports on anxiety disorders other than generalized anxiety disorder (GAD) in stroke patients are rare [7]. The frequency of GAD without concurrent depression in stroke patients varies from 3% to 24%, with a mean value of 14.4% [8, 9]. The frequency of post-stroke mania or psychosis has not been extensively studied [1].
Although stroke is a major public health problem in Chinese populations [10], a MEDLINE search of the literature between 1966 and August 2001, assisted by manual search yielded only three papers [11–13]. In the first study [12], the rate of depression was 43.4% in 58 patients. The second surveyed stroke survivors in a Taiwanese rural community and found 62.2% of the subjects depressed, twice the rate of depression (33%) for non-stroke subjects of the same age group [13]. The third study [11] reported 55% as depressed 3 weeks after a stroke. There are no data on the prevalence of other post-stroke psychiatric disorders in Chinese patients.
The aim of the present study was to examine the frequency of post-stroke psychiatric morbidity in Chinese patients. We also examined the extent to which psychiatric problems were identified by the treating physicians.
Methods
Selection of subjects
Prince of Wales Hospital is a university-affiliated general hospital serving a population of about 800 000 in Hong Kong. The overwhelming majority of stroke patients from the catchment area are admitted to Prince of Wales Hospital for initial assessment and acute care. Patients who need active rehabilitation are transferred to a stroke rehabilitation unit (SRU) located in a nearby hospital.
In the SRU, a multidisciplinary team provides comprehensive medical and functional assessment and intensive physical and occupational therapy. Psychiatric consultation is requested only on an ad hoc basis. The length of stay for patients in the SRU varies between 2 and 4 weeks.
All patients suffering their first stroke, who were admitted to SRU between 1 June 1999 and 31 August 2000, and who satisfied inclusion and exclusion criteria, were asked to participate in the study.
Inclusion criteria were: (i) well-documented first acute stroke (typical clinical picture supplemented by brain CT scan) occurring within a maximum of 4 weeks prior to the patient's transfer to the SRU; (ii) an Abbreviated Mental Test (AMT) [14] score of 7 or more (AMT is a well-established screening instrument for dementia, with good sensitivity and specificity); (iii) ability and willingness to give informed consent to participate in the study; (iv) Chinese ethnicity and fluency in Cantonese dialect. Exclusion criteria: (i) diagnosis of transient ischaemic attack (TIA), subdural haematoma, or subarachnoidal haemorrhage, as these specific forms of cerebrovascular disease are usually excluded from post-stroke studies of psychiatric morbidity; (ii) history of neurological disease; (iii) severe deficit of comprehension, communication, or impaired consciousness that made psychiatric assessment either impossible or unreliable; (iv) length of stay in SRU less than 2 weeks.
Neurological and functional status evaluation
Subjects’ pre-stroke functioning and neurological and functional impairment were measured by the Rankin Scale [15], the National Institute of Health Stroke Scale (NIHSS) [16] and the Barthel Index [17]. Neurological evaluation including the reading of CT scans was carried out independently of the psychiatric assessment.
Psychiatric assessment
Subjects for the study were identified upon entering rehabilitation and assessed 2 weeks after their admission to the SRU, (i.e. approximately 4 weeks after their stroke). The principal author conducted all the interviews using the Chinese versions of the Mini-Mental State Examination (MMSE) [18] and the Structured Clinical Interview for DSM-III-R (SCID-DSM-III-R) [19]. The same investigator also interviewed available relatives to aid in establishing clinical diagnoses including psychotic, mood, anxiety, and adjustment disorders according to DSM-III-R criteria. In keeping with DSM-III-R, the group of depressive disorders included MD, dysthymia, and adjustment disorder with depressed mood. The diagnosis of depression was based on the SCID supplemented by all available information on the patients gleaned from interviews with relatives and case notes. The investigator did not attempt to judge whether individual somatic symptoms were due to depression or stroke. We shall touch upon the issue of diagnosis of PSD in the Discussion section.
Throughout the study, 22 subjects chosen according to computergenerated random numbers were assessed by another psychiatrist, who was blind to the first author's diagnoses, to establish interrater reliability with the aim of reducing the likelihood of idiosyncratic diagnostic practice.
All but two subjects with diagnosable psychiatric disorder were reexamined by the first author at a mean of 6.0 ± 3.9 (SD) months after the first assessment using the SCID-DSM-III-R.
The study protocol was approved by the Ethics Committee of the Faculty of Medicine, Chinese University of Hong Kong. All subjects signed the Consent Form.
Statistical analysis
Descriptive statistics were used to characterize the demographic and clinical profile of the whole study population. The demographic and clinical variables of different groups were compared using χ 2 test and t-test. Barthel scores on admission and at discharge were compared with paired sample t-test. Statistical significance was set at the p < 0.05 level.
Results
Demographic data
Of the 509 stroke patients admitted to the SRU during the study period, 352 (54.5% male) did not meet criteria. Their age ranged from 20 to 98 (mean = 74.3 ± 10.8 years). Main reasons for exclusion included recurrent stroke (n = 165, 46.9%); severe aphasia (n = 60, 17%); discharge within 2 weeks (n = 22, 15.6%); and AMT < 7 (n = 31, 8.8%). Other causes of exclusion were history of dementia, dialect other than Cantonese, physical frailty, death, refusal to participate, intracranial pathology other than stroke and incomplete information.
One hundred and fifty-seven patients were included. The sample had a mean age of 71 ± 10 years; 45% of subjects were male; 48.1% received no education, 29.8% only 1–6 years; 56% and 33% were married and widowed, respectively; 60.5% were retired and 19.1% were housewives. Excluded patients were older than study subjects (t = − 2.852, p = 0.005).
The mean length of stay in SRU was 30.6 ± 12.9 days. Most patients were discharged home (n = 112) or to nursing homes (n = 38). Seven patients were transferred back to acute care.
Neurological and functional status
Hypertension (HT) and ischaemic heart disease (IHD) were found in 61.7% and 10.8% of cases, respectively. Left hemiplegia was detected in 52.9%, while 32.5% had right hemiplegia. No hemiplegia was found in 14.6% who had vascular injuries of the cranial nerves, cerebellum, or other structures. The mean NIHSS score on admission was 5.8 ± 3.9 (r = 0–36) indicating mild to moderate neurological impairment. Subjects’ improvement during rehabilitation was reflected in a significant increase in their Barthel score from 8.9 + 3.9 to 13.4 + 4.5 (paired sample t-test; t = − 21.3; p = 0.000). (The Barthel score ranges form 0–20; with the higher score indicating better function.)
Prevalence of psychiatric morbidity
Psychiatric assessment was conducted 24.9 ± 8.3 days post-stroke, and 15.8 ± 3.8 days following admission to the SRU. Relatives were interviewed in 95 (60.3%) cases; 62.8% of these were female, while only 40.4% of subjects whose relatives were not interviewed were female (χ 2 = 6.946, p = 0.008). Only 6 subjects (3.8%) had a family history of psychiatric disorder; three patients had a family history of schizophrenia, and one patient each had a family history of mania, depression, and mental retardation.
Fourteen subjects (8.9%) had a past psychiatric history, which included MD (n = 4), dysthymia (n = 4), alcohol dependence syndrome (n = 3), GAD (n = 1), and alcohol abuse (n = 1). One patient had alcohol dependence syndrome combined with alcoholic hallucinosis. No patient had a past suicide attempt.
The MMSE scores in the rest of the sample ranged from 10 to 30, with a mean value of 22.8 ± 4.5.
The second rater assessed 22 subjects. Inter-rater reliability between the two raters diagnosing depression was kappa = 0.776 (p = 0.001).
Twenty-seven subjects were diagnosed with PSD, thus the frequency of all depressive disorders in our sample was 17.2%. Twelve subjects (7.6%) met DSM-III-R criteria for MD; two of them also had a history of dysthymia. In terms of severity of depression, 5 patients (3.2%) had a moderately severe MD and one (0.6%) had a severe MD with mood congruent delusions. Severity of depression was mild in 6 patients (3.8%). Thirteen (8.3%) subjects had adjustment disorder with depressive mood. Two patients (1.3%) were diagnosed with dysthymia. In the non-depressed group, 59.5% and 12.4% had HT and IHD, respectively; corresponding figures for the depressed group were 71.4% and 3.6%. There was no significant difference in the proportion of subjects with HT (χ 2 = 0.241, p = 0.287) or IHD (Fisher's exact test, p = 0.312) between the depressed and non-depressed groups.
Generalized anxiety disorder was diagnosed in one patient (0.6%). No cases of other types of anxiety disorder, mania, or psychosis were found.
None of the patients had received antidepressant treatment prior to their stroke. While in the SRU, three patients, two patients with a mild and a moderately severe MD, and another with GAD, were identified as having a possible psychiatric disorder and referred for psychiatric assessment. A fourth patient with a moderately severe MD received hypnotic medication only. No subject received antidepressant medication or any other form of psychiatric treatment or traditional Chinese medicine prior to the psychiatric interview.
Follow-up
Twenty-five (92.6%) of the subjects diagnosed with depression were followed up 3–6 months after the initial assessment. Eighteen (72%) subjects became symptom-free. Of the 11 subjects with MD, 8 (72.7%) were in remission at follow-up, while 9 out of 12 (75%) patients with adjustment disorder were in remission. Neither of the two patients with dysthymia was symptom-free. Only 5 subjects, all with MD, received antidepressant treatment during follow-up. (Psychiatric treatment was offered to every subject who was identified with a psychiatric illness.)
Discussion
There are several limitations of this study. The sample was recruited from a rehabilitation unit capturing mainly the moderately severe range of stroke cases. It is uncertain whether depressed patients were more or less likely to be referred to the SRU. Furthermore, 69% of the admissions were excluded for medical reasons. They were older with a trend towards more male patients excluded. The number of subjects suffering from PSD or other psychiatric disorder was too small to establish clinically meaningful statistical associations between post-stroke psychiatric illness and demographic and clinical variables. The small number of affected subjects also precluded separate analysis of MD, dysthymia and adjustment disorder with depressed mood.
The fact that only 60.3% had relatives available for interview may have impacted on detection of psychiatric morbidity, for two main reasons. First, subjects with and without available relatives had different sex distributions that might reflect a different level of social support or other psychosocial factors relevant to psychiatric morbidity for the two sexes. Second, reports of patients with stroke or dementia suggest that relatives may over- or underreport depressive symptoms and activities of daily living [20–22].
A potential bias, common to any investigation of depression in medically ill subjects, was the uncertainty in evaluating somatic symptoms of depression. Identifying depression in the medically ill is fraught with difficulties [23]. When making a psychiatric diagnosis in our study, in some instances it was impossible to judge accurately if a particular somatic symptom was due to stroke or depression. In such cases, the diagnosis of PSD was arrived at by taking into account the whole clinical picture, particularly the presence of other depressive symptoms like feelings of guilt, hopelessness or worthlessness against the background of premorbid personality; reaction to previous medical conditions; and the severity of stroke. The formal research diagnosis provided by the SCID interview and other available information were integrated in making the diagnosis of PSD. The inclusion of another experienced clinician in the inter-rater reliability exercise served the purpose of avoiding, or at least reducing, the possibility of idiosyncratic diagnostic practice. The satisfactory inter-rater reliability suggests that the diagnosis of PSD in our sample was clinically meaningful.
The strengths of this study include the assessment of subjects by a trained psychiatrist who achieved satisfactory inter-rater reliability with an experienced academic psychogeriatrician; the use of a structured interview supplemented by all available clinical information, and a personal follow-up of all subjects with psychiatric morbidity. Subjects came from an active stroke research centre so the diagnosis of stroke was likely to be accurate. Also, ongoing psychiatric treatment did not influence the detection of psychopathology.
In order to make the study population more homogenous, the investigation was restricted to first time stroke patients since the demographics, medical and neurological condition and level of disability differ significantly between patients with first and recurrent stroke. Previous stroke is an independent predictor of disability and as such may influence psychiatric morbidity [24]. In addition, the longer-term social and psychological consequences of stroke and their impact on psychiatric morbidity particularly in Chinese societies are not well understood.
Only 25 subjects (15.9%) had MD or an adjustment disorder with depressed mood. Studies conducted in rehabilitation hospitals involving Caucasian stroke patients, and employing a structured interview and DSM criteria, reported a prevalence of depression between 35% and 46%, with a mean of 39% [25–28], more than twice the frequency found in this study. It is unlikely that this discrepancy is accounted for by the difference between the prevalence of depression in Chinese and Caucasian elderly in general. The point prevalence of depression in elderly Caucasians varies widely, ranging from 1% to 20% [29–31], with similar figures reported in Chinese populations [10, 32, 33].
The frequency of all depressive disorders in this study was 17.2%, a figure much lower than the rates reported in three previous studies examining Chinese stroke patients [11–13]. It is difficult to compare our findings with those of other Chinese studies as the present investigation differs from all three studies in several aspects including time elapsed since stroke, inclusion/exclusion criteria, age distribution and diagnostic criteria for depression. All three studies have methodological flaws such as small sample size [12, 13], non-validated diagnostic instruments [12] and a mixture of first time and recurrent stroke patients [11–13]. None of the Chinese studies employed standardized assessment for diagnosing PD.
The frequency of PSD in first time Caucasian stroke patients varies between 27% and 57% [4–6], figures significantly higher than the 17.2% found in our study. From a methodological point of view, Paolucci et al.'s study [6] is the most similar to ours; these investigators examined first time stroke patients 6 weeks post-stroke in a rehabilitation facility and patients had similar levels of functional disability to our subjects. Post-stroke depression was diagnosed if the Hamilton Depression Rating Scale score was 18 or above, roughly corresponding to our MD group; they reported that 27.4% of their sample met criteria for depression, compared to our 7.2% for MD and 17.2% for all types of PSD. A recent study [34] assessing the subjects just two weeks following their first stroke found the prevalence of PSD in Finnish patients only 5.6%, a figure close to ours.
MEDLINE (1966–August 2001) and EMBASE (January 1980–August 2001) searches did not locate any papers on psychiatric morbidity in first time stroke patients from other Asian countries or from the rest of the developing world.
In the present study, not only was the frequency of depression low, but also 72% of depressed subjects were in remission at the follow-up assessment, the majority of them without any psychiatric treatment. Thus, the shortterm outcome of PSD in our sample seems to be better than that reported in the literature. Burvill et al. [7] found that only 43–50% of patients were in remission at 12-month follow-up. Similar findings have been reported by other investigators [8, 35]. One may argue that the self-limiting course of PSD in our sample supports the hypothesis that PSD is primarily a psychological reaction to stroke [36]. Although this is an appealing hypothesis that cannot be rejected, at least for a proportion of cases, another possibility remains that PSD is mainly biological in origin and it improves with time, along with other neurological deficits.
No subject was diagnosed with post-stroke psychosis or mania in this sample, which is not surprising as both conditions are rare. In a series of 309 patients, no case of psychosis and only two cases of mania were identified [1].
The frequency of GAD was 0.8% (n = 1) in the present survey. Studies in community samples of Caucasian stroke patients reported 1–2% prevalence of GAD [7, 37]. The frequency of GAD among stroke outpatients was also higher, ranging from 4 to 11% [7]. Morris et al. [8] found a 3% frequency of GAD in a rehabilitation hospital. Generalized anxiety disorder was more frequently identified in stroke units, ranging from 19 to 28% [38, 39]. It is uncertain whether the considerable discrepancy between our results and those reported in Caucasian patients was due to different investigative methods, or whether there are genuine differences in the prevalence of GAD in different stroke populations.
Possible methodological shortcomings notwithstanding, the low psychiatric morbidity found in our study is puzzling. If replicated, lower morbidity figures in Chinese stroke patients may generate neurobiological and psychosocial hypotheses about the aetiology and pathogenesis of post-stroke psychopathology.
Sociocultural factors have doubtlessly played a role in the expression and recognition of PSD in our study sample. However, in the lack of data, the magnitude or even the direction of the complex influence of factors related to Chinese ethnicity could not be gauged. We are cautious to offer any, even tentative, interpretation of our data based on specific Chinese cultural characteristics for several reasons. First, despite their methodological shortcomings, the above-mentioned earlier studies found a significant proportion of Chinese patients with PSD [11–13]. Second, recent epidemiological surveys have shown that the rate of depression among Chinese elderly approximates that of Caucasians [40]. Third, we are uncertain to what extent traditional Chinese values are applicable in the fast changing and fairly westernized Hong Kong society. Fourth, the study sample is composed of nearly two generations coexisting in the diverse social environment of Hong Kong, therefore a blanket statement about cultural traditions, values, attitudes to illness and disability and social relations characterizing our sample would be probably faulty. The unpredictable impact of sociocultural factors was evidenced by the findings of our companion study concerning the detection of PSD in Chinese patients [Tang et al. submitted]. We hypothesized that the usually reserved Chinese elderly would be more willing to disclose personal feelings to staff members well-known to them as opposed to a research assistant encountered only once. The results unequivocally refuted our expectations.
It is possible that the tendency among Chinese people to deny or somatize their depression [41] reduced the rate of PSD in our study. It could be assumed, that our sample was composed of mostly poorly educated or uneducated subjects, who may not have been articulate enough to verbalize their feelings or were too deferential and/or distrustful toward the psychiatrist to open up. So far we have no adequate answer to this issue. Concurring with previous observations [42] we had the distinct impression that our subjects, once given the opportunity to talk and asked direct questions, were ready to reveal their emotional state.
A finding of great practical importance in our study was the failure of treating physicians to identify the majority of patients with psychiatric illness. Closer collaboration between physicians and psychiatrists will be necessary to educate stroke rehabilitation teams about stroke-related psychiatric morbidity.
Footnotes
Acknowledgements
This work was supported by a grant from the Chinese University of Hong Kong. We thank the following staff of Shatin Hospital for their help: S.Y. Chan, L.F. Leung, S.K. Lam, K.W. Lau, S.W. Lam, S.H. Shiu, S.C. Lau, E. Shum, T. Chui and W. Lee.