Plain English summary

Health-related quality of life (QoL) is poor after stroke, but may be improved with a comprehensive care plan. We determined the effects of an individualized care plan on QoL in people with stroke, and characteristics of survivors associated with poor QoL. We found that an individualized management program, comprising a tailored care plan and stroke-specific education in addition to usual care, did not significantly improve QoL. However, female sex, older age, having a lower educational attainment, and having greater handicap, anxiety and depression, were associated with poor QoL. Greater focus on managing handicap, anxiety and depression may be required to make more meaningful impacts on QoL in people with stroke.

Introduction

Stroke adversely affects health-related quality of life (QoL), a multidimensional concept that includes physical, mental and social wellbeing [1]. Evidence for using individualized care plans to improve QoL of survivors of stroke is inconsistent, with some comprehensive and multidisciplinary care approaches found to be associated with improvements in QoL [2,3,4,5], while others have not been [6,7,8]. There have been several studies on factors associated with QoL in people with stroke [9], but relatively few have been conducted to identify longitudinal relationships [10,11,12,13,14,15,16].

In a cluster randomized controlled trial among community-dwelling survivors of stroke or transient ischemic attack (TIA), we have previously evaluated the effects of a tailored management program comprising an individualized management plan developed by multidisciplinary team and nurse-led stroke-specific education to manage cardiovascular risk [17,18,19,20]. While the intervention improved the control of lipids and knowledge of risk factors at 12 months after baseline [17, 18] there were no detectable effects on absolute cardiovascular risk or knowledge of secondary prevention medications [19, 20]. It remains unclear whether QoL can be improved with this intervention. In the present investigation, we aimed to: (1) determine whether an individualized management program improved the QoL in the 24 months following stroke or TIA; (2) describe changes in QoL over time; and (3) identify variables associated with better QoL in survivors of stroke/TIA over 24 months.

Methods

Study design

The Shared Team Approach between Nurses and Doctors For Improved Risk factor Management (STANDFIRM) study was a multicenter, cluster randomized controlled trial with blinded assessment of outcomes and intention-to-treat analysis. It was designed to determine the effectiveness of an individualized management program in patients with stroke or TIA compared with usual care. The trial was registered in the Australian New Zealand Clinical Trials Registry (ACTRN12608000166370) and had ethical approval from the Human Research Ethics Committee at each participating hospital. The study protocol and statistical analysis plan including sample size calculation have been published previously [21, 22].

Participants

Patients hospitalized with stroke or TIA were recruited from four hospitals in Melbourne, Australia. Eligible participants were aged ≥ 18 years and lived within 50 km of the closest recruitment hospital. Patients were excluded if they were participating in another clinical trial, discharged to a nursing home, or were not expected to survive the duration of the study due to a rapidly deteriorating condition or disease. Written informed consent was obtained from participants prior to their participation in the trial [23]. When participants required help to complete assessments (e.g., because of aphasia), the person assisting was also consented.

Randomization

Participants were randomly assigned to receive the intervention or usual care. A computer-generated block design, with blocks assigned within a hospital, enabled equal allocation of participants from hospitals to each group, thereby eliminating potential bias from variation in treatment during their hospitalization. Randomization was clustered by general practice to reduce contamination between groups. Study participants, outcome assessors and stroke specialists were blinded to group allocation.

Intervention

The intervention group received a tailored management program in addition to usual care, following their baseline, 3-month and 12-month assessments. This management program comprised:

  1. 1.

    An individualized chronic disease management (CDM) plan that was developed by an unblinded intervention nurse and overseen by an independent stroke specialist (neurologist or geriatrician). Recommendations within CDM plans were in accordance with guidelines for stroke care and included evidence-based health targets that were specific to each patient. The CDM plan covered stroke therapy (antihypertensive, antiplatelet, anticoagulant and antihypercholesterolemic prescriptions), management of comorbidities, mood, self-care and risk factors (smoking, physical activity, alcohol consumption and healthy diet). The CDM plan, which was in a format that was familiar to general practitioners (GPs) for treating chronic conditions, was sent to the patient’s GP. GPs could modify the CDM plan in consultation with the patient. GPs who use these CDM plans receive greater remuneration than they would receive for time-based consultations under the Australian universal healthcare insurance scheme (Medicare).

  2. 2.

    Individualized stroke-specific education provided at an in-home visit by the unblinded intervention nurse who had also prepared the CDM plan. During this visit, the intervention nurse also made an appointment for the participant to visit their GP for further discussion of their CDM plan.

At 6 and 18 months, the intervention nurse prepared a review of the CDM plan based on information obtained during an assessment conducted over the telephone, and provided these reviews to the GP. Recommendations from earlier plans were altered when circumstances changed. Patients in the control group received usual care as per the standard arrangement of the hospital and usual GP practice.

Baseline and outcome assessments

Details of the stroke or TIA, and other clinical details, were obtained from hospital medical records. Demographic data collected included age, sex, country of birth, educational attainment, occupation category, and marital and living status. The QoL of participants was assessed using the Assessment of Quality of Life (AQoL-4D) instrument [24]. Baseline (after discharge to home) and outcome assessments (3, 12, and 24 months after randomization) were conducted face-to-face by nurses blinded to group allocation. When a participant was unable to respond, proxies (usually the carer) completed the questionnaires. Data on anxiety, depression, handicap and living status were collected at each time point. The Hospital Anxiety and Depression Scale (HADS) [25] was used to determine anxiety and depression, with scores of ≥ 8 in each domains considered abnormal. Disability was measured using the London Handicap Scale (LHS) [26], ranging from 0 (the most disability) to 1 (no disability).

Outcome

The AQoL-4D is a generic multi-attribute utility instrument that has been validated in the stroke population [27]. Version 3 of this scale included 15 items in 5 dimensions: illness (prescribed medicines, medication and aids, medical treatment), independent living (self-care, household tasks, mobility), social relationships (friendships, isolation and family role), physical senses (seeing, hearing and communication), and mental health (sleeping, worrying and pain) [24]. A study participant was instructed to choose one of 4 possible responses for each item that best described him or her over the preceding week.

A pre-defined algorithm (https://www.aqol.com.au/index.php/scoring-algorithms) was used to convert the raw responses on the questionnaire to individual scores for four dimensions (the illness dimension was not scored) and an overall instrument utility score [28]. When a response was missing, the algorithm imputed this dimension from the mean of the other 2 responses in the dimension. When more than one response was missing in a dimension, no utility score was calculated for that dimension. The AQoL-4D utility score ranged from − 0.04 (worse than death) to 0 (equivalent of death) to 1 (full health). Participants who had died during the study were assigned a utility score of zero.

Statistical analysis

Descriptive statistics were used to compare participant characteristics at baseline by group allocation, and to summarize AQoL-4D utility scores by participant characteristics. Categorical variables were summarized as frequencies and percentages and continuous variables as median and as 25th and 75th percentiles.

To assess between-group differences in total AQoL-4D utility scores and in each of the four AQoL-4D dimensions at each time point, we used median regression adjusted for age and sex. Median regression was also used to determine variables associated with QoL at each time point because median regression is more robust than ordinary least squares regression in using highly skewed data distributions. Our pre-determined list of variables assessed in univariable median regression analyses comprised those that were associated with QoL in prior studies [29,30,31,32,33]. All variables that appeared to be associated with QoL in univariable analyses (P value ≤ 0.05) were initially included in multivariable analyses. Variables with P values > 0.1 were then eliminated in a stepwise fashion, with age group and sex forced into the model.

To assess the longitudinal effects of variables on changes in QoL, we used multivariable, multilevel mixed-effects linear regression. This technique handles missing observations and allows the use of unequal time intervals [34]. All available variables were included in a mixed-effects model with time being a level-one variable and participant being a level-two variable. Then demographic and clinical variables with P values > 0.1 were excluded in a stepwise fashion. The final mixed-effects model provides unstandardized coefficients.

Description of QoL estimates at each time point were subsequently described with stratification by the factors found to be associated with QoL in the multivariable regression analyses. The analyses were performed using STATA version 16.0 [35]. Two-sided P values ≤ 0.05 were considered statistically significant.

Results

Of the 563 participants recruited between January 2010 and December 2013, 283 were randomized to the intervention and 280 to usual care. The mean age was 68.4 years (SD 13.2), 64% were male and baseline characteristics were balanced between groups (Table 1 and Supplementary Table 1). The mean time interval between hospital discharge and the baseline assessment for participants in the trial was 11 weeks with no difference between the study groups (P = 0.95). AQoL-4D utility scores were available for 502 participants (89%) at 24 months after baseline, slightly fewer than at the other time points (Supplementary Fig. 1). This included those who died and were assigned zero score. Of the 61 participants with missing data on AQoL-4D at 24 months, 57.4% were male and the mean age was 70.2 years (SD 13.0). Apart from stroke type and living status at baseline, there were no statistically significant differences between participants with and without missing data (Supplementary Table 2). Proxies provided AQoL-4D data for < 1% of participants at each time point.

Table 1 Baseline characteristics of study participants
Full size table

Effects of the intervention on quality of life

The pattern of the distribution of AQoL-4D utility scores was similar between study groups at 24 months (Fig. 1) and at other time points (data not shown). Using median regression, we were unable to detect differences in the total AQoL-4D utility scores between treatment groups at any time point (Table 2). At 24 months, participants in the intervention group had greater scores for the dimension of physical senses than those in the usual care group (P = 0.044). This between-group difference was not observed at earlier time points. No other between-group differences were detected in any dimension or at any time point.

Fig. 1
figure 1

Distribution of Assessment of Quality of Life 4D (AQoL-4D) utility scores in the intervention and usual care groups at 24 months

Full size image
Table 2 Difference in utility scores of AQoL-4D dimension between usual care and the intervention groups, adjusted for age and sex, at 3, 12, and 24 months
Full size table

Differences in quality of life over time

Multilevel, mixed-effects linear regression model revealed that AQoL-4D utility scores were greater at 3 and 12 months than at baseline (Table 3). However, the improvement in the scores between baseline and two years was not significant (P = 0.140). The mean AQoL-4D utility score (SD) was 0.649 (0.256) at baseline, 0.687 (0.262) at 3 months, 0.696 (0.269) at 12 months and 0.663 (0.291) at 24 months.

Table 3 Multilevel mixed-effects linear regression model of variables associated with AQoL-4D utility scores with stepwise backward elimination in all participants (intervention and usual care combined)
Full size table

Variables associated with quality of life

In the longitudinal, multivariable, multilevel mixed-effects linear regression analyses older age, females, lower educational attainment, greater handicap, presence of anxiety and depression were associated with lower AQoL-4D utility scores over two years (Table 4).

Table 4 Multivariable median regression† of factors associated with quality of life at 3, 12, and 24 months
Full size table

In cross-sectional multivariable median regression analyses older age, presence of handicap and depression were consistently associated with lower AQoL-4D utility scores at each time point (Table 5). However, being single and anxious was associated with lower AQoL-4D utility scores at 24 months only. Results of the univariable median regression analyses that underpin these results are provided in Supplementary Table 3. In all participants, the median AQoL-4D utility scores differed according to age, sex, and other characteristics (Table 5).

Table 5 Median AQoL-4D utility scores with quartiles according to participant characteristics and time of follow-up
Full size table

Discussion

People with stroke or TIA who received the STANDFIRM intervention did not appear to have better QoL over a 2-year period following baseline, compared to those who received usual care. QoL improved from baseline to one year, but not at two years. A major finding of our study was the identification of factors associated with QoL that were independent of time over this 2-year time period: older age, females, lower educational attainment, greater handicap and being anxious and depressed. In our cross-sectional analysis, marital status and anxiety were associated with QoL at 24 months follow-up, but not at other time points.

Our inability to detect a difference in AQoL-4D utility scores between the intervention and usual care groups is likely to be because our intervention was not specifically targeted at improving QoL, although it covered some aspects of QoL. In a similar study to ours, no effect of monthly nurse specialist-led education and tailored advice about risk factors was observed at 3 months [6], or at 40 months [8]. Similarly, others also reported no effect of a disease management plan with stroke-specific education on QoL at 6 months [7]. In contrast, 1-year QoL among patients with stroke was found to be better in a community-based self-directed rehabilitation intervention that was specifically designed to cover issues related to physical abilities, communication skills, emotions and mood, information and financial needs, and social relations [4, 36].

In analyses of all participants, we noted two patterns in QoL. First, in line with previous studies, QoL among survivors of stroke improved over the first year after baseline but not at 2 years [12, 15, 37]. In contrast, QoL did not improve longitudinally over the first year in a community-dwelling sample of patients with stroke in Australia that were older and more often female than patients in our study [29]. Second, the overall mean and distribution of AQoL-4D utility scores in our study was considerably greater than in two other Australian or international studies (0.47 in the North East Melbourne Stroke Incidence Study (NEMESIS) at 2 years and 0.47 in A Very Early Rehabilitation Trial (AVERT) at 1 year) [30, 38]. Compared to these previous studies, our cohort included relatively minor stroke and TIA and baseline assessments occurred on average 11 weeks after discharge from hospital by which time most recovery is thought to typically occur [39]. This is further borne out by the negatively skewed distribution of AQoL-4D utility scores that we observed, a shape more aligned to the general population [40], compared with the U-shaped distribution in NEMESIS and AVERT. Furthermore, in our study < 1% of people had proxy responses, compared with 23% in NEMESIS. This is an important disparity as proxies have been shown to be more likely to underestimate QoL than patients themselves [41]. Therefore, we provide reliable information on QoL to estimate quality adjusted life years in economic evaluations of secondary stroke interventions that are applicable for community-dwelling patients with mild to moderate stroke or TIA.

In this study, we undertook a comprehensive investigation of factors associated with long-term QoL. Our findings are consistent with those of previous studies, in which associations were reported between QoL and older age [12, 38, 42, 43], being female [12, 43], having lower educational attainment [30], and having greater anxiety [12, 30, 38, 42], depression [10,11,12] or handicap [11, 38]. Our findings reinforce the importance of understanding the differential long-term impacts of stroke whereby certain subgroups of survivors of stroke or TIA may require greater support. In a registry cohort of patients with stroke and TIA, the poorer QoL observed in older women was in part explained by depression and social circumstances [44]. By managing their handicap, anxiety and depression, there may be potential to improve their QoL. We have identified subgroups of survivors that could benefit from interventions aimed at improving the QoL after stroke. Therefore, our findings provide further impetus for clinicians to target these subgroups of patients. Greater support for patients who live alone may also be warranted, particularly in the longer term after stroke. Our observation that being single or separated was associated with poorer QoL at 24 months, but not longitudinally, may indicate the benefits of informal care in the longer term after stroke, the deterioration of QoL in those without informal care, or both. There is evidence from a study conducted in Canada that being married is positively associated with QoL at 1 to 3 years post-stroke [32], while in Finland, marital status was associated with poorer QoL at 1 year [33]. Further research into these associations and establishing effective interventions to improve QoL in these vulnerable groups are required.

Our study has several limitations. First, there was a potential risk of contamination of the usual care with some components of the intervention, as general practitioners were incentivized to use CDM plans. This limitation can be addressed by linking the trial data with administrative datasets on disease management plan use in future studies. Secondly, eleven percent of patients had missing QoL data at 24 months. This could affect only the results of the cross-sectional analyses as missing values were well handled with our longitudinal mixed model approach. Those with missing data more often had intracerebral hemorrhage than those without missing data. Because intracerebral hemorrhage is a more devastating form of stroke, those missing may have had poorer QoL [37]. So, it is likely that we have overestimated the QoL of our participants in our cross-sectional analyses only. Finally, as we did not collect data on stroke severity while the person was in hospital, we were unable to adjust our QoL estimates for initial stroke severity. Overall our study participants had milder stroke or TIA than studies such as AVERT (appr. 2% died at 12 months in this study vs. 12% at 12 months in AVERT) [30], a factor that may have limited the generalizability of our findings.

A major strength of this study is its design and enhanced statistical analyses which enabled us to analyze longitudinal changes in QoL on top of cross-sectional analyses. Literature on longitudinal analyses of QoL after stroke beyond 12 months is limited. We managed to identify variables associated with QoL after stroke over the period of two years, thereby extending findings of previous studies, most of which were cross-sectional or only 12 months in duration. This is a considerable addition to the current literature. Next, with 563 patients, our study had a relatively larger sample size compared to other studies [10,11,12,13,14,15, 29, 31,32,33, 38, 45]. Furthermore, our comprehensive outcome assessments, which included the LHS and HADS, enabled us to consider repeated measures of covariates, such as handicap, anxiety, and depression. These advantages allowed us to derive more reliable estimates with comparatively greater precision and power for patients with minor stroke and TIA. Finally, the low rate of proxy responses (< 1%) in this study, compared to other studies [30, 38], enabled us to minimize underestimation of QoL [41, 46, 47].

Conclusion

A comprehensive, multidisciplinary team care management program, incorporating an individualized CDM plan and education, did not affect QoL in the first 2 years following a stroke or TIA. Among community-dwelling survivors of stroke or TIA, older age, being female, having lower educational attainment, and having more anxiety, depression and handicap was associated with poorer QoL. These subgroups of survivors of stroke or TIA may require greater support.