Self-rated health after stroke: a systematic review of the literature

Background Self-rated health (SRH) allows for comparison and identification of the health status of various populations. The aim of this study was to conduct a systematic review of the literature to expand the understanding of SRH after stroke. Methods This systematic review was registered with PROSPERO (CRD42017056194) and conducted according to PRISMA guidelines. Studies published until December 2018 that evaluated the SRH of adults with stroke were included. Results Of the 2132 identified studies, 51 were included. Only four studies had experimental designs (7.8%). In 60.7% of the studies, SRH was assessed by variations on direct questions (i.e., general and comparative SRH). Analog visual scales and quality of life instruments were also used to evaluate SRH, but there is no consensus regarding whether they are appropriate for this purpose. The results of cross-sectional and longitudinal studies revealed significant associations between poor SRH and stroke as well as between SRH, function, and disability. The power of SRH to predict stroke mortality is still uncertain. Two interventions (a home-based psychoeducational program concerning stroke health care and family involvement in functional rehabilitation) effectively improved SRH. Conclusions Direct questions are the most common method of evaluating SRH after stroke. Studies reported significant associations between the SRH of individuals with stroke and several relevant health outcomes. However, few experimental studies have evaluated SRH after stroke. Interventions involving health education and family involvement had a significant impact on SRH.


Background
Self-rated health (SRH) is a simple measure of how individuals evaluate their own health status [1,2] recommended by the World Health Organization (WHO) [3]. The most common type of SRH evaluation, which asks simple and direct questions [4,5], is useful within both clinical and research contexts [1] because it features reduced observation bias and lower costs related to measurement and data collection. This facilitates data analysis [6] and aligns with the principles of client-centered practice [6,7]. SRH evaluations measure individuals' perceptions of their own health and are therefore dependent on individuals' pre-existing concept of health and context [4]. However, it is believed that people can synthesize a large amount of information about themselves in response to direct questions as part of an SRH evaluation [8].
Health status, quality of life (QoL), and health-related quality of life (HRQoL) are often used interchangeably [9]. According to the WHO, QoL refers to "individuals' perceptions of their positions in life within the context of their culture and value systems in which they live, and in relation to their goals, expectations, standards, and concerns," and it is "the product of the interplay between social, health, economic, and environmental conditions, which affect human and social development" [9]. In addition, the WHO defines health as "a state of a complete physical, mental, and social well-being, and not merely the absence of disease or infirmity" [9]. Therefore, compared to SRH, which is a measure of health status, QoL is a more comprehensive construct that covers all aspects of life. HRQoL is not clearly defined, but it is related to the way health affects QoL [10]. Despite their differences, these three constructs have similar characteristics; they are multidimensional, selfreported, and involve physical, mental, and social aspects of individuals' lives [9,10].
SRH has been considered a valuable outcome in studies with various objectives and populations. It has been demonstrated to have a significant association with the risk of diseases, such as depression [2] and type 2 diabetes [11], and/or decreased activity performance [1] in both adults [2,11] and the elderly [1]. Recently, there have been several systematic reviews of the literature examining SRH with the elderly [12,13], indigenous people [14], and adults in general [15]. Systematic reviews of the associations between SRH and relevant outcomes, such as mortality [16,17], can also be found. However, no systematic literature reviews focus on the SRH of individuals with important, complex health problems, such as stroke.
Stroke is the leading cause of serious long-term disability, and it accounts for most of the global burden of disease [18]. Globally, the lifetime risk of stroke is 24.9% [18]. An estimated 7 million people suffered from a stroke worldwide between 2013 and 2016, and of those, about 1,806,000 were left with some type of disability [19]. One year after a stroke, 57% of people need assistance with daily living activities [20]. Therefore, health indicators of subjects with stroke, such as SRH, are of great clinical utility.
SRH has been associated with demographic, psychological, physical, and social factors in stroke patients [21] as well as increased risk of death [22] and the development of stroke in older adults without history of the disease [22]. Among elderly people who suffered from a stroke, poor SRH is related to reduced social interactions and limited mobility outside the home [21].
Due to the importance of SRH evaluations, the general aim of this study was to conduct a systematic literature review to expand the understanding of SRH after stroke. The specific aims were to (a) describe how SRH has been assessed and used, (b) synthesize previously reported results, and (c) verify the effects of interventions on the SRH of individuals who suffered from a stroke.

Methods
This systematic review followed the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) guidelines [23,24], and it was registered in the International Prospective Register of Systematic Reviews (PROSPERO; CRD42017056194). All steps were performed by two independent examiners, and a third examiner was involved in cases of lack of consensus.
This review included studies that assessed SRH in individuals ≥18 years of age who suffered from a stroke. The WHO's definition of a stroke was adopted [25]. Of the studies with mixed populations, only those that separately reported the SRH results of individuals with stroke were included. All studies published until December 2018 in any language, except for theses or dissertations, case series, or case studies, were analyzed.
Searches were conducted in the following electronic databases: Medical Literature Analysis and Retrieval System Online (MEDLINE), Physiotherapy Evidence Database (PEDro), Latin American and Caribbean Health Sciences Literature (LILACS), and Scientific Electronic Library Online (SCIELO). An initial search strategy was created for MEDLINE and then was adapted to the other databases. The search strategy for stroke published in a recent systematic review of the Cochrane Database of Systematic Reviews [26] was elaborated upon by the authors with terms that were used in previous systematic reviews of the same outcome of interest (i.e., SRH; see Appendix for the search terms) [3,17,22,27].
All studies found in the electronic databases were screened based on their titles and abstracts. Those that clearly did not meet the eligibility criteria were excluded. Then, the full texts of the remaining studies were analyzed to determine whether they met the eligibility criteria. A manual search in the references section was also performed in the included studies. The methodological quality of the clinical trials was evaluated using the PE-Dro scale [28], and the risk of bias in quasi-experimental studies was evaluated by the Transparent Reporting of Evaluations with Nonrandomized Designs (TREND) [29]. Information that was relevant to the objectives of this systematic review was extracted using a structured form developed based in a prior study that contained the following information: author/year, study design, objective, sample characteristics, instrument/question used to evaluate SRH, response items, SRH operationalization, statistical analysis, and conclusions regarding SRH.

Results and conclusions of the longitudinal observational studies
In the six longitudinal observational studies (23.1%) [5,22,30,39,40,60], poor SRH was associated with stroke. One study (3.8%) [39] reported this relationship specifically for individuals who reported poststroke visual impairments. The predictive power of SRH for stroke mortality could not be confirmed since significant results were observed in two studies [43,66] but not in two others [22,64]. SRH was associated with morbidity, especially after a stroke [31,53,69], and with return to work and post-stroke stability [44]. A combination of diabetes and stroke was strongly associated with poor SRH [58]. Furthermore, improvements in balance self-efficacy were associated with improvements in functional walking capacity, which in turn led to increased SRH [61] (Table 3).

Results and conclusions of the longitudinal experimental studies
According to the PEDro scale, scores of the methodological quality of three experimental longitudinal trials [34,62,65] ranged from six [34] to eight [65], which is considered good [78]. The quasi-experimental study [67] scored 13 points out of 22 on the TREND scale (Table 4).
Only the study of Ostwald et al. [62], which compared the efficacy of a home-based psychoeducational program concerning stroke health care to mailed information on stroke prevention, showed that improvements in SRH favored the experimental group [62].
The quasi-experimental study [67], which investigated the effects of family involvement in functional rehabilitation performed by a physiotherapist and a nurse at a rehabilitation center for post-stroke elderly patients, found significant improvements in SRH after the end of the intervention.

Discussion
The present study performed a systematic review of the literature on SRH after stroke. Most of the studies employed general and comparative direct questions to assess SRH. The cross-sectional and longitudinal observational studies revealed significant associations between poor SRH, stroke, and other important health outcomes. In addition, SRH was reported to predict the occurrence of stroke. However, the association between SRH and stroke mortality remains unclear. To date, few studies have evaluated the effects of interventions on SRH. However, those that do exist found two types of interventions that effectively improve the SRH of individuals with stroke. SRH was mostly evaluated by direct questions, although there was great variability in the structure of these questions and the response items. Jürges et al. [79] reviewed two versions of response items, the one recommended by the WHO (Very Good, Good, Fair, Bad, and Very Bad) and the European version (Excellent, Very Good, Good, Fair, and Poor). Although some differences were found, both versions were highly correlated after the items were resized to allow for comparison [79]. Comparative questions mainly had three options, making direct comparison with general SRH evaluations difficult [4].
Studies investigating elderly people compared general and comparative direct questions but reported different results [80,81]. One study, which investigated whether the reference point (i.e., people of the same age) would be a good predictor of mortality in the elderly, showed that comparative questions better predicted mortality in men [80]. However, the other study, which compared general and comparative questions among elderly individuals of the same age, found that general questions were better since the comparative questions are influenced by age [81]. To our knowledge, no study has compared general and comparative questions among stroke patients, and therefore there is no scientific information that can be used to determine the best method of assessing SRH. The total scores of QoL/HRQoL instruments, such as the SF-36 and the SF-12, were used to evaluate SRH [35,47,49,62,65]. Although QoL, HRQoL, and health status have some similarities, they use different constructs [9]. Additionally, if the aim is to measure health status, specific questions should be used. Some QoL/HRQoL instruments have some SRH-specific questions, such as the first and second items of SF-36 [41]. However, the total scores of QoL/HRQoL instruments cannot be used as SRH measures. Therefore, the results of studies that assess SRH based on total scores [35,47,49,62,65] should be considered with caution.
The most common analog visual scale for SRH assessment is the EQ5D [82]. This scale is traditionally used to measure subjective phenomena, and it is easy and quick to apply, with excellent properties for measuring pain and QoL/HRQoL [83]. However, it is necessary to investigate whether its measurement properties are adequate for SRH evaluation [84]. Therefore, the results of the studies that use analog visual scales [33,45,59,61] should also be interpreted with caution.  Wilcoxon test There was no significant difference in SRH between stroke survivors in acute phase (16 months) and in a long term (10 years) Dong et al., 2018 [69] Cox proportional hazards model General and age comparative SRH were significantly associated with an increased risk of first-ever stroke and recurrent stroke in Chinese adults Vogelsang, 2017 [55] Logistic regression model Stroke is associated with improvement in comparative SRH but not with retrospectively reported SRH Mavaddat et al., 2016 [22] Cox proportional hazards model There is a small but significant independent relationship between poor SRH and stroke incidence. However there is no relationship between SRH and stroke mortality in the short or longer term in the older population. In older people with a history of stroke, there is no relationship between SRH and stroke outcomes Larsen et al., 2016 [44] Logistic regression model SRH 3 months post-stroke and stroke severity were found to be strongly associated with return to work and subsequent work stability after stroke Univariate and multivariate logistic regression model Self-rated ill-health independently increases the risk of stroke, specifically for men, and that the interaction effect between SRH and biomedical risk factor load is greater for men than for women Hillen et al., 2003 [5] Wilcoxon test, Logistic regression model Patients reporting a health transition to "much worse" 3 months after stroke have an increased risk of disability at 1 year and decreased chances to survive free of stroke recurrence over the next 5 years Otiniano et al., 2003 [58] Chi square test, Logistic regression model Diabetes and stroke in combination is strongly associated with a higher risk of disabilities, poor SRH, and higher 5-year mortality rates than persons without these diseases, regardless of the presence of other conditions Muntner et al., 2002 [60] Not done Self-reported "health in general" was worse among those with a history of stroke compared with those without a history of stroke for all three time periods In most studies, poor SRH was associated with stroke. Since SRH is a measure and indicator of health status [3], this finding may reflect the health care needs of this population [75]. According to this review, studies have found strong associations between poor SRH and visual impairments [39], balance impairments [61], worse affective-emotional state [46], mobility deficits [63], limitations in ADL [63], worse function [37], reduced social life [46] and inability to return to work [44]. Additionally, using the Barthel index, Hillen et al. [5] reported that functional independence appeared to have a greater impact on SRH than other outcomes and was a strong predictor of health status in individuals with stroke [5].
Some studies did not find any association between poor SRH and stroke. Tuomilehto et al. [48] observed good or satisfactory SRH in individuals 14 years after a stroke episode. These findings could be explained by the duration of the disease; longer periods of time since the onset of stroke were associated with better acceptance of disability [85]. Most of the individuals investigated in prior studies with better SRH had mild disabilities [48] and higher survival rates [86]. In their qualitative study, Welin, et al., 2010 Anderson et al.,  Mavaddat et al. [42] reported that the severity of physical limitations alone did not influence perceptions of SRH; even individuals with severe morbidities could report good SRH. Similarly, Varela et al. [76] found an association between good SRH and patients with chronic obstructive pulmonary disease (COPD) who had suffered a stroke. Many studies have demonstrated the predictive power of SRH for mortality and morbidities among various populations [8,36,69]. However, two studies employing a logistic regression model adjusted for sociodemographic factors and morbidities found that SRH did not predict mortality after stroke. Adjusting sociodemographic factors does not reduce the predictive ability of SRH measures, but the presence of morbidities, especially among respondents with poor SRH, tends to reduce predictive power [16].
Only two of the four experimental longitudinal studies found improvements in SRH [62,67]. This outcome could be changed only by interventions targeting factors that observational studies found were related to SRH. The improvements in SRH produced by a home-based psychoeducational program concerning health care after stroke may be related to the educational level of the individuals [87]; understanding of the disease is an important factor affecting SRH [62]. Family involvement in functional rehabilitation also improved the SRH of individuals with stroke [67], and it is recognized to have strong psychological effects on therapy through physical and emotional support [4,88].
SRH was used as an outcome less often in experimental longitudinal studies than in observational studies. Future experimental studies should examine SRH evaluation due to its importance and informative capacity for individuals' health [3]. In addition, healthcare professionals should routinely assess SRH using both general and comparative simple and quick questions [4] in order to identify and summarize the health status of patients with stroke. Poor SRH may be used to identify priority patients since it is commonly related to disabilities, and it can be used to monitor patients who have not had a stroke but are at risk. Home-based psychoeducational programs as well as family involvement in functional rehabilitation at rehabilitation center for post-stroke may be used as interventions to improve SRH.

Conclusions
Direct questions were the most commonly mentioned and recommended way to measure SRH in subjects with stroke. It is unclear whether a certain type of question is superior to others, but general SRH is most commonly used. Cross-sectional and longitudinal studies have shown significant associations between SRH and several relevant health outcomes, while few experimental studies have evaluated the efficacy of interventions for improving SRH after stroke.