Title. Predictors of poststroke quality of life in older Chinese adults.
Aim. This paper is a report of a study to identify the changes in poststroke quality of life and other clinical issues among older Chinese adults from 1 month to 6 months after stroke and the predictors of poststroke quality of life at 6 months.
Background. Stroke survivors are known to suffer from prolonged and multiple impairments leading to a compromised quality of life, but few studies report early predictors for quality of life among older Chinese adults after active rehabilitation has been undertaken during the first 6 months after stroke.
Method. A total of 214 patients with first-ever ischaemic stroke were interviewed by a research nurse at 1 month and 188 patients were interviewed again 6 months after hospital admission for stroke. Assessment of quality of life was done using the Modified Rankin Scale for Quality of Life. Changes in and relationships between quality of life and variables in five domains were explored: bio-anatomical, physical, emotional, cognitive, communicative and social support. The data were collected in 2004–2005.
Results. Quality of life among two-thirds of participants was unchanged or lower when scores at 1 month and 6 months after stroke were compared. Length of hospital stay after admission for stroke and other 1-month factors – level of worry over current health, cognitive and self-care deficits – were identified as having independent effects on quality of life at 6 months.
Conclusion. Clinicians need to observe for early signs of mild cognitive impairments and emotional needs of stroke survivors, as well as to consider longer-term interventions to enhance poststroke quality of life.
•Stroke survivors suffer compromised quality of life, with prolonged and multiple impairments.
•There is a knowledge gap concerning early predictors of quality of life at the chronic stage.
What this paper adds
•Quality of life among two-thirds of older Chinese stroke survivors remained unchanged or lower at 6 months.
•Cognitive function did not show statistically significant improvement when compared with physical, emotional and communicative domains at 6 months after stroke.
•Self-care and cognitive deficits, together with levels of worry about current health at 1 month after stroke, were important predictors of poststroke quality of life at 6 months for these Chinese participants.
Implications for practice and/or policy
•Clinicians should observe for early signs of mild cognitive impairments and emotional needs of stroke survivors.
•Longer-term interventions may be needed to enhance poststroke quality of life.
Previous researchers have reported frequent emotional and physical sequelae among stroke survivors (Johnston et al. 2004, Nannetti et al. 2005) leading to compromised quality of life (QoL) (Kauhanen et al. 2000, Suenkeler et al. 2002, Moon et al. 2004). However, few researchers report changes in poststroke quality of life (PSQoL) at different stages or suggest factors that can predict such changes early. Information of this sort is crucial to help healthcare workers detect any undesirable trends and devise effective interventions for stroke survivors.
Many populations are susceptible to strokes because of inherent risk factors: increasing age and obesity, smoking, high prevalence of hypertension, diabetes and cardiovascular diseases (Stewart et al. 2001). Many stroke survivors have to live with unresolved disability and residual symptoms despite rehabilitation programmes (Jonkman et al. 1998, Kaplan 2005). More importantly, it has been reported that long-term QoL among stroke survivors remains low (Kim et al. 1999, Suenkeler et al. 2002).
According to Lassey and Lassey (2001), p.13, ‘QoL is used to describe responses to the “instrinsic” characteristics of an individual and the “extrinsic” social, economic and environmental factors that affect well-being’. As reported previously, the correlates of PSQoL may fall into different domains: bio-anatomical, communicative and cognitive, physical, emotional or social support and others (Lassey & Lassey 2001).
Discussion of bio-anatomical factors can be very broad, covering features such as brain lesions, arterial supply and biochemical status. In fact, not many researchers have addressed the relationship between factors in the bio-anatomical domain and QoL after stroke. Some suggest that severe subcortical grey-matter lesions and depressive symptoms in the acute phase of stroke are important in predicting low QoL 2 months after stroke (Moon et al. 2004). Another study exploring the relationship between laterality of brain lesion and PSQoL produced no useful findings (Kauhanen et al. 2000).
The aim of the study was to identify the changes in PSQoL and other clinical issues among older Chinese adults from 1 month to 6 months after stroke and the predictors of PSQoL at 6 months.
This was a longitudinal study with personal interviews undertaken by a single research nurse using questionnaires at 1 month and 6 months after stroke admission.
First-ever ischaemic stroke survivors were recruited consecutively for the study between 1 June 2004 and 31 May 2005 (see Figure 1 for recruitment protocol). The patients were identified at the stroke unit of a regional hospital serving about 0·58 million residents in one of the 18 regions of Hong Kong (Department of Census and Statistics 2005). Annual hospital discharges (including death) of stroke survivors were about 23,000 in Hong Kong in 2004 (Centre of Health Protection 2005). On average, around 1200–1300 discharges occurred in each of the 18 district hospitals and the recruitment of an adequate number of first-ever ischaemic stroke survivors was therefore feasible in one of these hospitals.
Patients were included if they were: local residents aged 50 years and over, able to communicate in Chinese, with first-ever ischaemic stroke diagnosed by a neurologist. Patients were excluded if they had a history of any psychiatric disease. A total of 222 patients consented to participate in the study constituting 52·1% of all those eligible (see Figure 1). Only 214 participants eventually completed the interview at 1 month. The attrition rate at 6 months was 12% (26 out of 214) because of death (n = 11, 5·1%) or unavailability for the follow-up interview (6·9%). Four people were away from Hong Kong (1·9%), six had changed their telephone numbers (2·7%) and five refused to continue their participation in the study (2·3%).
The rationale for sample-size estimation was based on the main study to identify the incidence and predictors of depression among stroke survivors reported previously (Lee et al. 2007). Sample size was estimated using epi info software (Epi Info™, Centres for Disease Control and Prevention, United States Department of Health and Human Services [Available at http://www.cdc.gov/epiinfo/epiinfo.htm]). (Centres for Disease Control and Prevention 2005). Taking 5% as the level of statistical significance, 80% as the power and allowing a 3–5% margin of error to predict 10% incidence/prevalence (which is on the low side compared with previous studies), a sample size of 137 was needed. Taking into account dropouts, 50% more participants were added and the sample was rounded up to 210.
Personal interviews were undertaken by the same research nurse at 1 month and 6 months after stroke onset at a rehabilitation unit or outpatient clinic during follow-up. The outcome to be examined was the patient’s QoL, measured by a single-item Modified Rankin Scale for Quality of Life (MRSQoL). This is a six-point scale ranging from 0 to 5, with 5 being the lowest level QoL (Rankin 1957, Farrell et al. 1991). The MRSQoL ratings were based on participants’ self-assessment, with options given by the interviewer.
Socio-demographic data on age, gender, years of formal education, marital status and clinical and health history were captured from case notes. Changes in the four domain factors were examined as elaborated below.
Under this domain, factors studied include laterality (right vs. left hemisphere), area and structure (cortical vs. subcortical) and arterial system of the brain lesions (large vs. small vessel diseases). The correlation between QoL and bio-chemical status (blood-sugar level and lipid profile during acute stroke admission) was also examined.
In the physical domain, we measured self-care functions using the Barthel Index of Activities of Daily Living (BADL; 0–20, 20 being the highest functional level) (Collin et al. 1988, Dickenson 1992). Patients were also asked about their level of disturbance as a result of pain and bodily discomfort: headache, stomach pain, joint pain, low back pain and non-specific muscle pain and various kinds of physical discomfort such as nausea, vomiting, dizziness, constipation. Pain and somatic discomfort are vital elements compromising QoL among older adults, as documented in previous studies (Kauhanen et al. 2000, Kong et al. 2004).
The idea of probing about these common types of pain (Yen & Chan 2003, Hansson 2004) and bodily discomfort (Robain et al. 2002, de Coster et al. 2005) among stroke survivors is supported by previous studies. A list of these should capture most of the common complaints faced by stroke and older patients. Initially, a seven-point scale was developed to allow more accurate estimates of discomfort levels. After the pilot phase of the main study using this tool, it was found that older stroke survivors had a very short attention span during long questions/interviews and the seven-answer options had to be cut to four to differentiate adequately and to provide mutually-exclusive choices (such as 0 = none, 1 = mild, 2 = moderate, 3 = severe) for these aspects. The choices were well-received by the participants in the main study. The revised responses were frequently reported by stroke survivors in the pilot study (Lee et al. 2007) and face validity was assured by a panel of professors in psychiatry, gerontology and neurology.
Cognitive and communicative domain
Stroke survivors are known to suffer cognitive and communicative problems. The cognitive impact resulting from stroke was measured by the Abbreviated Mental Test, AMT (0–10, 10 being the highest level of cognitive functions). The AMT has been validated in Chinese and is commonly used in clinical settings (Hodkinson 1972, Chu et al. 1995). As validated Chinese tools are not readily available for reliable assessments of communicative deficits among older adults, our participants were asked to evaluate the level of disturbance because of the following problems: expressive aphasia, receptive aphasia, global aphasia, dysarthria, swallowing, visual impairment, hearing deficit and agnosia (Black & Hawks 2005). Zero indicated none at all, 1 mild or infrequent level of disturbance (about once or twice weekly), 2 moderate level (about three or four times weekly) and 3 severe level (five times or more a week).
This domain covered the assessment of depression, anxiety and the nature of worries. Depression was measured by the Geriatric Depression Scale short form (GDS), which consists of 15 items. Validation of the Chinese version has demonstrated good reliability and validity (Brink et al. 1982, Chiu et al. 1994, Lee et al. 1994). The GDS has been adopted to study both white (Agrell & Dehlin 1989, Johnson et al. 1995, Mast et al. 2004) and Chinese stroke survivors (Tang et al. 2004, Lee et al. 2007). Patients are considered to be depressed if they have a score of 8 or more of the 15 points. In addition to the standardized tools used for depression, ten additional items in this domain were assessed. Patients were asked about levels of disturbance caused by anxiety, worry about stroke rehabilitation progress, worry about disease deterioration, life stressors such as disturbance because of financial problem, losing a job, bodily pain and discomforts, poor family relationships and loss of loved ones. These were stressors frequently reported previously by Chinese stroke survivors (Lee et al. 2007). The constructs of anxiety (Bourgeois et al. 2004, Sturm et al. 2004), worries and chronic stresses (Kurlowicz 1997, Forsell 2000, Hwang et al. 2000, Fung 2003, Li et al. 2003, Leonard 2006) and depression (Beekman et al. 2000) have been studied specifically with white stroke survivors, but have been less addressed among Chinese. Validation was sought from the literature (Lee et al. 2007) and from a panel of researchers. Their inclusion in the study helped to give a comprehensive view in respect of emotional domain factors in PSQoL among Chinese participants.
Social support domain
The Lubben Social Network Scale in Chinese was used (Chi & Chou 2001). The scale is a 10-item 5-point (1–5) tool that can be analysed using three sub-categories: (1) family network, (2) friends network, (3) confidant relationship. Aggregate scores range from 0 to 50, with 50 indicating maximal perceived support.
The study was approved by the appropriate ethics committee. After receiving an explanation of the purposes of the study and its procedures, assurance was given to potential participants about confidentiality. Patients were given time to clarify any doubts before signing a consent form, a copy of which was then given to them. They were also reassured of their right to withdraw from the study at any time without any effect on their current medical and healthcare regimes, so as to ensure entirely voluntary participation.
Data were analysed using the Statistical Package for Social Science. Descriptive statistics were used to show participant demographics. The mean and standard deviation of scores from the various assessment tools were calculated for 1-month and 6-month data. To identify the pattern of change in QoL, the proportion of participants who scored at the same level or higher or lower was computed in comparing the 1-month and 6-month data. Changes in MRSQoL scores and domain factors were analysed using a paired sample t-test after checking the normality of the data using a P–P plot (Easton & McColl 1997). Stepwise regression analysis was carried out to identify predictors of PSQoL at 6 months after adjusting for age, gender and length of hospital stay. Collinearity analysis was undertaken by examining the tolerance of the variables. Model adequacy was examined using standardized residuals.
Of the 214 participants at baseline, 135 were male (63·1%) and 136 married (63·4%). The mean age of all participants was 72·3 years (sd = 9·4), the majority had received secondary or lower levels of formal education and more than 60% lived with family members, who also provided them financial support (see Table 1).
Table 1. Participant demographics (n = 214)
Occupation before stroke
Others working status
Both spouse & children
Nature of residence
Privately owned premises
Privately rented premises
Care and attention home/nursing home
Financial assistance from government
Old age allowance
Disability allowance/high disability allowance
Combined social security allowance
Financial support from
Both children & spouse
Age (years), mean ± sd (range)
72·3 ± 9·4 (50–94)
Length of hospital stay (days), mean ± sd (range)
17·1 ± 15·4 (3–89)
Formal education (years), mean ± sd (range)
5·5 ± 5·0 (0–23)
Changes in quality of life and other domain factors
There was a statistically significant change in mean QoL score when comparing 6-month and 1-month data (P = 0·002, see Table 2). With reference to the scores at these two points, the proportion of participants with improved or deteriorated or similar QoL levels could be computed. Slightly more than a third reported improved QoL when comparing the 6-month and 1-month scores (69 out of 188, 36·7%). Nearly two-thirds had an unchanged QoL score (82 out of 188, 43·3%) or this had further deteriorated (30 out of 188, 20·0%). Functioning in physical, emotional and cognitive domains had all improved, as shown in Table 2. In the emotional domain, depression score, display of upset facial expression or crying, level of disturbance caused by worrying over stroke rehabilitation progress and anxiety had all reduced statistically significantly at 6 months (P < 0·001, see Table 2). In the physical domain, among the self-care abilities measured by the Barthel Index, the number of somatic symptoms, including pains, was statistically significantly reduced. Similar positive changes were detected in the cognitive and communicative domain, with the exception of the AMT score, where the improvement did not reach a statistically significant level (P = 0·083).
Table 2. Changes in scores from 1 month to 6 months after stroke onset
Variables (Range of scores)
6M (n = 188)
1M (n = 214)
Change in score (6M–1M)
Paired Samples t-test was conducted for all the above variables; LOD, level of disturbance due to aN not equal to 188 due to missing data; sd, standard deviation; 95% CI, 95% confidence interval; 6M, six-month; 1M, one-month.
Modified Rankin Scale for Quality of Life (Range: 0–5)
Sum score of Geriatric Depression Scale (0–15)
Days in the last week with (0–3)
a happy facial expression (0–3)
an average or nothing special facial expression (0–3)
an upset facial expression or cry (0–3)
Perceived current health status (1–3)
LOD worry about stroke rehabilitation progress (0–3)
LOD worry of losing job (0–3)a
LOD anxiety (0–3)a
Barthel index – Activities of Daily living (0–20)
Number of pains (0–5)
Number of body discomfort (exclude pains) (0–5)
Total number of body discomfort and pains (0–10)
Abbreviated Mental Test, Total score (0–10)a
Communicative domain (0–3)
LOD expressive aphasia
LOD visual acuity
Predictors of quality of life at 6 months
To identify predictors of PSQoL at 6 months, five domain factors at 1 month were included in the regression model. The final model revealed that QoL at 6 months could be predicted by factors present at 1 month: BADL, score on the AMT and levels of worry about current health, together with length of hospital stay (see Table 3). A model with four factors falling into three domains was able to explain 55% of the variance in the PSQoL score at 6 months.
Table 3. Predictors of poststroke quality of life scores at 6 months
Model was adjusted for age, gender and length of hospital stay;
Factors entered into the regression model:
Number of pains; Number of body symptoms; Barthel Index-Activities of Daily Living (BADL); Level of disturbance due to expressive aphasia; Level of disturbance due to dysarthria; Level of disturbance due to swallowing; Level of disturbance due to visual acuity; Level of disturbance due to hearing; Level of disturbance due to memory; Level of worry about current health; Level of disturbance due to worry about stroke rehabilitation progress; Level of disturbance due to anxiety; Level of disturbance due to financial problem; Score of Abbreviated Mental Test (AMT); Score of Geriatric Depression Scale (GDS); Laterality of brain lesions marked by Computerized Tomography.
Length of hospital stay (days)
BADL scores at 1 month
AMT scores at 1 month
Level of worry about current health at 1 month
The QoL of the majority of stroke survivors did not improve at 6 months in relation to their scores at 1 month. This supports earlier findings on the compromised QoL among stroke survivors in short and longer terms (Moon et al. 2004, Sturm et al. 2004). Another study has actually shown a drop in PSQoL 12 months after disease onset (Suenkeler et al. 2002). The less-frequently addressed domains – bio-anatomical, cognitive and communicative – were included in this study to give a more comprehensive picture of the relative effects of the various domain factors on PSQoL. This may help to reduce the chance of overlooking a seemingly less important domain as well as to ensure that adequate attention is directed towards a truly important domain factor.
The role of bio-anatomical factors correlating with stroke outcome has remained uncertain (Carota et al. 2002) and few studies have included this domain to PSQoL. Our initial findings on the association between lesion features and PSQoL may suggest that the severity of stroke induced by or associated with large-vessel diseases and the cortical region may be a plausible explanation for reduced QoL, as previously reported (Moon et al. 2004). Despite such findings, the impact of lesion laterality is weak and less clear because of missing data that might indicate evidence confirming its effect on QoL.
Most communicative domain factors were commonly found at 1 month, but these symptoms showed observable and statistically significant improvement by 6 months and their role in predicting PSQoL were not identified. As far as cognitive functions were concerned, participants reported a relatively high level of function as reflected by scores in the AMT, having a mean score of 8·96 out of 10 (10 = full functioning) at 1 month, in contrast to previous reports of a high incidence of cognitive impairment among stroke survivors (Madureira et al. 2001). In addition, our participants showed only mild and non-statistically significant improvement in AMT scores at 6 months. Despite such findings, AMT was one of the predictors of PSQoL in the final regression model. This may alert clinicians to the fact that seemingly mild cognitive impairment (MCI) can have an impact on QoL because of altered dependency levels, as reported previously (Pohjasvaara et al. 2002).
In the present study, self-care dependency at an early stage had a persistent role in PSQoL over the long term. Such a result was not fully supported by another study, where functional independence correlated with QoL at 6 months only and not at 1 month (Robinson-Smith et al. 2000). Some reports have even boosted the role of self-care capability in stroke survival (Sonde & Viitanen 2001, Pohjasvaara et al. 2002). However, this view is contradicted when improved physical capability does not improve mortality (Lightbody & Baldwin 2002). The effect of somatic symptoms on PSQoL appears less certain in our study, partly because of the natural history of disease progress. As shown earlier, most of somatic symptoms are attenuated in both frequency and intensity after active rehabilitation. No somatic problems were found to predict PSQoL at the chronic stage and this is also supported in a previous study (Kong et al. 2004).
Emotional domain factors have an indisputable role in PSQoL. In the current study, we found that the level of worry over perceived health status at early stages after stroke helped to predict QoL at 6 months. However, we did not have enough evidence to support the notion that either anxiety or depression at 1 month may predict QoL at 6 months. Nevertheless, the role of depression and social-support factors at 6 months as important determinants of PSQoL at 6 months is in line with other reports (Jonkman et al. 1998, Kim et al. 1999, Kauhanen et al. 2000, Perry & McLaren 2004). Despite such findings, the growing role of emotional status in PSQoL after the acute stage of stroke has been demonstrated (Jonkman et al. 1998, Shimoda & Robinson 1998, Kim et al. 1999, Kauhanen et al. 2000, Perry & McLaren 2004). Furthermore, levels of worry about perceived health status at an early stage after stroke has been identified as a predictor of depression (Lee et al. 2007). Thus, assessment of mood, including worries, should be done early to prevent compromised QoL in the long run (Moon et al. 2004, Sturm et al. 2004). Clinicians working on the clinical front-line may need to monitor factors falling into the cognitive, emotional and physical domains, as these are modifiable for positive changes. In particular, interventions to foster self-care functions, alleviate worries that may be either valid or merely apparent and because of irrational thoughts or target early mild cognitive impairment might be indicated. Efforts in that regard have been supported in that perceptions of control 6 months after discharge have been found to add statistically significantly to the predictive equations for QoL (Johnston et al. 2004).
On fixed demographic and clinical factors, it has been reported that being married may affect PSQoL (Kim et al. 1999, Kauhanen et al. 2000), but this was not the case in current study. Instead, length of hospital stay was identified as a predictor of QoL suggesting that from the onset clinical states might have impacts on PSQoL. This finding may help to guide clinicians in considering the unmet needs of stroke survivors whose QoL is low at 1 month and remains static at 6 months, despite rehabilitation efforts. The absence of a baseline quality-of-life score before stroke may weaken the above proposition, although it is highly likely that stroke will add burdens to survivors, in particular those with diminishing coping resources in old age (Hellstrom et al. 2004).
Several limitations to the study should be identified. We used the Rankin Scale for QoL to measure PSQoL, which, being a single-item measure, may render the resulting assessment very crude, and a more comprehensive assessment tool for QoL may be required. We chose 6 months as the follow-up point. In quite a few studies, this has been extended to 1 year or longer, when more complete rehabilitation and adaptation may have taken place. We adopted a 6-month follow-up because of concerns over likely concomitant life/health episodes (Bush 1999), resulting in contamination and high dropout levels, as the local population is quite mobile. Selection bias is a possibility and the results may not be generalizable to other populations. After application of the recruitment criteria, seriously ill or aphasic stroke survivors were excluded. It might also be argued that the domain factors included were not exhaustive, as other individual factors – such as personality (Remer-Osborn 1998, Golden & Golden 2003) and concomitant life events including institutionalization and bereavement – might also compromise QoL. A structured list of life events might be used, but, in view of the long interview agenda and participants’ deteriorating attention span, we only focused on the core items set out earlier. Future studies may help to supplement the information in this unmapped area. We did not address the role of social domain factors in PSQoL, although previous studies have shown the role of family support in PSQoL (Swartzman et al. 1998, Kim et al. 1999, Jaracz & Kozubski 2003, Tang et al. 2005); this was because such data were only been available for the 6-month point.
Quality of life in patients with stroke has been a consistent challenge to health care. Older stroke survivors are known to be susceptible to multiple functional deficits, but to have limited coping resources. Improvement in QoL and other clinical variables is highly desirable, but not guaranteed. Static or even falling levels of QoL among stroke survivors are unacceptable in view of the technological advancement and escalating healthcare investment in stroke care. Active intervention to reverse such trends must be initiated early and in a sustainable manner. While saving life and reducing complications are the immediate concerns of acute stroke care, observable and detectable early signs and symptoms that are important predictors of PSQoL must not be missed. It is a worthwhile aim of nursing to add quality rather than mere days to life. Clinicians should address impaired poststroke psychosocial and cognitive functions that are likely to compromise QoL. Therefore, healthcare professionals should assess and intervene in patients’ lingering worries and cognitive impairment at an early stage, in addition to focusing on self-care rehabilitation. In a similar direction, further research efforts are needed to examine the effectiveness of active early interventions in PSQoL. In particular, those that can alleviate patients’ worries and concomitantly improve cognitive and self-care functioning should be a priority research area. More importantly, these interventions should be conducted early, sustained beyond the first few months and be relevant to the cultural setting.
We would like to express our gratitude to the stroke survivors and their families, and to all those clinical partners who actively participated in the study.
The project was supported by a seed funding grant from the University of Hong Kong.
ACKL and SWT were responsible for the study conception and design. ACKL and THT performed the data collection. ACKL performed the data analysis. ACKL was responsible for the drafting of the manuscript. SWT, THT and DYTF made critical revisions to the paper for important intellectual content. DYTF provided statistical expertise. ACKL obtained funding. ACKL, SWT, THT and GKKY provided administrative, technical or material support. SWT and GKKY supervised the study.