Health-related Quality of Life Using WHODAS 2.0 and Associated Factors 1 Year After Stroke in Korea: a Multi-center and Cross-sectional Study

Background: The self-perceived level of disability for stroke survivors in community is little known. We aimed to characterize HRQoL 1 year after stroke and to investigate how socio-demographic and stroke-related factors and medical adherence explain the self-perceived level of disability among a Korean stroke population. Methods: This study was a multi-center and cross-sectional study. A total of 426 ischemic stroke survivors at 1 year after onset recruited from 11 university hospitals underwent a one-session assessment including: socioeconomic variables, the modied Rankin Scale (mRS), various neurological sequelae, the modifed Morisky scale (MMS), and the World Health Organization Disability Assessment Schedule 2.0 (WHODAS 2.0) 36-items. The relationship between disability and different variables was analyzed using ordinal logistic regression. Results: The prevalence of disability by global WHODAS 2.0 score was 62.6% (41.6% for mild; 16.0% for moderate; 5.0% for severe) in subjects. Prevalence of severe disability was higher in Participation in society (16.8%) and Getting around (11.8%) than in the other domains. Low MMS motivation was the only factor determining the signicant association between all six domains of disability after adjustment. Different predictors for specic domains were age, mRS, dysarthria, trouble seeing, cognition problem and MMS-motivation for Understanding and communicating; age, recurrent stroke, mRS, hemiplegia, facial palsy, general weakness and MMS-motivation for Getting around; age, education, mRS, hemiplegia and MMS-motivation for Self-care; age, education, recurrent stroke, hemiplegia, dysarthria and MMS-motivation for Getting along with people; age, education, income, mRS, hemiplegia, dysarthria, MMS-knowledge and MMS-motivation for Life activities; living without spouse, recurrent stroke, mRS, hemiplegia, dysarthria, trouble seeing, cognition problem, general weakness and MMS-motivation for Participation in society. year after stroke. Each domain of disability increased with various associated factors. Interventions promoting medical adherence of motivation seem to help high HRQoL in all domains.


Introduction
Stroke is common and serious non-communicable health problem. It is the second cause of mortality [1] and the third cause of disability-adjusted life years (DALYs) [2] in the world. In Korea, the Epidemiologic Research Council of the Korean Stroke Society reported an age-and sex-standardized incidence of rst ever stroke of 92.2 per 100,000 populations in 2013, an age-standardized prevalence of stroke of 1.37% in Korean adults aged over 19 years in 2014, and an age-standardized stroke mortality of 29.6 per 100,000 populations in 2015 [3]. Stroke was the third leading cause of DALYs in Korea, following diabetes mellitus and low back pain in 2012 [4].
When stroke survivors suffer from becoming disabled related with neurological sequelae, outcome assessment of acute stroke traditionally focuses on prevention of deaths, alleviation of symptoms, impairments, and restoration of function [5]. However, health related quality of life (HRQoL) measures may capture patients' perception on disability better than traditional ways. Not only because they are multidimensional instruments which comprise functional, physical, cognitive, psychological and social elements [6] but also because the impact of limitation following stroke on well-being may differ by each patient [7]. Furthermore, they re ect health from patients' own perspectives [8]. WHODAS 2.0 is a generic instrument of HRQoL for measuring function and disability in major life domains linked to the International Classi cation of Functioning, Disability and Health (ICF). It is reliable and applicable across cultures in adult populations [9][10][11][12].
In Korea, the level of disability for stroke survivors in community is little known. We aimed to characterize HRQoL 1 year after stroke using WHODAS 2.0 and to investigate how socio-demographic, stroke-related factors and medical adherence explain the self-perceived level of disability among a Korean stroke population.

Study design and population
This study was a multi-center and cross-sectional study conducted across the period, December 2015 -March 2016. A total of 426 participants were recruited from the neurology outpatient clinics from the 11 university hospitals designated as Regional Cardiocerebrovascular Centers (RCCs) in Korea (Daegu-Gyeongbuk, Gangwon, Jeju, Chungbuk, Gwangju-Jeonnam, Gyeongnam, Daejeon-Chungnam, Jeonbuk, Busan-Ulsan, Inchoen, and Gyeongi RCC) [13]. Participants were stroke survivors who had been admitted to one of the RCC hospitals due to acute ischemic stroke occurred 12 to 15 months before the interview and were willing to be informants. A one-on-one interview was conducted by trained nurses at the 11 hospitals using a structured questionnaire. Patients with di culties in verbal communication were excluded. Written informed consent was obtained from all participants. The study protocol was approved by the institutional review board of Kangwon National University Hospital.

Measurement
Socio-demographic factors and stroke-related data Data on socio-demographic and stroke-related characteristics were collected. The common sociodemographic variables on general characteristics were sex, age, live with spouse or not, highest education quali cation (elementary school /middle school /high school/college and above) and monthly household income (1 and less/1 to 2/more than 2 million Korea won; 1.2 million Korea won≒1,000 USD). And, the stroke-related variables were recurrent or rst-ever stroke, modi ed Rankin Score (mRS), Complications after stroke (hemiplegia, dysarthria, facial palsy, trouble seeing, paresthesia, cognition problem, general weakness) [14]. The mRS is robust and the most commonly recommended functional measure in acute stroke research [5,7,14]. We categorized mRS into 'normal to mild' with score range from 0 to 2 and 'moderate to severe' from 3 to 5.

Self-reported Medication adherence
The Morisky Scale is self-reported measure of medication adherence. It had been originally developed to predict the adherence of outpatients to antihypertensive medications with four items in the mid-1980s [15]. Modi ed Morisky Scale (MMS) has 6 items measuring two domains of adherence (knowledge and motivation). Three items as 'When you feel better do you sometimes stop taking your medicine?', 'Sometimes if you feel worse when you take your medicine, do you stop taking it?', 'Do you know the longterm bene t of taking your medicine as told to you by your doctor or pharmacist?' are for knowledge and the other 3 items as 'Do you ever forget to take your medicine?', 'Are you careless at times about taking your medicine?', 'Sometimes do you forget to re ll your prescription medicine on time?' are for motivation.
Each item has the score of 0 or 1 and a higher score indicates high adherence, and MMS score can be categorized into 'low' with score range from 0 to 1 and 'high' from 2 to 3 for each subdomain [16].
Health-related quality of life (HRQoL) We measured HRQoL of 12 to 15 month post-ischemic stroke patients with the World Health Organization Disability Assessment Schedule 2.0 (WHODAS 2.0), a standardized cross-cultural measurement of disability [9]. The WHODAS 2.0 questionnaire has several forms according to number of items, administration, and respondent. We used the WHODAS 2.0, 36-items covering six domains of functioning: understanding and communicating (UAC, 6 items), getting around (GAR, 5 items), self-care (SAC, 4 items), getting along with people (GAP, 5 items), life activities (LAC, 8 items), and participation in society (PSO, 8 items) [10]. We computed 6 domain-speci c scores using 36-item complex scoring. Score ranged from 0 to 100, a higher score indicates greater disability like lower QoL [10]. WHODAS 2.0 domain-speci c and global scores were originally categorized as 5 grades: no problem (0-4%), mild disability (5-24%), moderate disability (25-49%), severe disability (50-95%), and extreme disability (96-100%). In fact, there were few subjects with extreme disability in this study, so the 5 groups were collapsed into 4: no, mild, moderate, and severe disability. Reliability and validity of Korean version has been established [17].

Statistical analysis
We analyzed data from 382 participants who had completed all the assessments. For descriptive purposes, absolute numbers and percentages were calculated for categorical variables and means ± SDs for continuous variable. The 6 domain-speci c scores of WHODAS 2.0 were separately treated as dependent variables. As we had categories for the dependent variable that were ordered, ordinal logistic regression was used. The link function used for model tting was logit. The evaluating overall model t to the data was done through the Model Fitting Information. It is determined that a model exhibits good t to the data when a signi cant improvement in t of the nal model containing full set of independent variables over the null model. A parallel line test con rmed that the proportional odds assumption was satis ed for every model for ordinal logistic regression. Every estimated ordinal logistic regression coe cient was transformed into odds ratio, as the exponential of a particular coe cient is an estimate of the odds ratio. Data analyses were performed using SPSS version 24.0 (SPSS Inc., Chicago, IL, USA) and p-value under 0.05 is considered as statistically signi cant. Table 1 showed that the participants were relatively elderly (mean age ± SD, 65.7 ± 12.2 years). Among 382 participants, 138 (36.1%) were female. Most of the participants (272, 71.2%) were living with spouse. As for their highest academic quali cation, 144 (37.7%) were elementary school, 68 (17.8%) were middle school, 105 (27.5%) were high school and 65 (17.0%) were college and above. Additionally, 148 (38.7%) had a monthly family income of less than 1,000,000 Korean won.  Domain-speci c levels of WHODAS and associated factors Among 382 participants, prevalence by WHODAS 2.0 level was 37.4% for no (disability-free), 41.6% for mild, 16.0% for moderate, and 5.0% for severe disability in Global scores. The breakdown by domain also shows that prevalence decreased with severity. People with no disability was relatively common in SCA (63.6%) and GAP (51.6%), whereas the prevalence for severe disability was higher in PSO (16.8%) and GAR (11.8%) than in the other domains of WHODAS 2.0 ( Fig. 1). Table 3 showed associations between different variables and disability in domain-speci c WHODAS 2.0 scores. The aORs obtained from ordinal logistic regression models for different variables represent disability in the index group compared with that in the reference group. The results for domain-speci c scores were adjusted for the 5 demographic, 9 stroke-related and 2 medication adherence variables.

Participants' characteristics
Age, mRS, dysarthria, trouble seeing, cognition problem and MMS-motivation were signi cant positive predictors of the disability of UAC in the model. For every one-year increase on age, the odds of being in more severe category on UAC was 1.03 times higher (p = 0.01). This indicated that a participant aged older was more likely to indicate greater disability of UAC. The odds of being in more severe level on UAC was 4.04 times higher for those who had the level of moderate to severe as compared to those who had the level of normal to mild in mRS (p < 0.001). And the odds of being in more severe level on UAC were 1.88, 2.86 and 5.59 times higher when a participant had the complication in dysarthria, trouble seeing and cognition problem, respectively (p = 0.01; 0.03; 0.007). In addition, the odds of being in more severe level on UAC was 3.12 times higher for those who showed low level of adherence to medication as compared to those who showed high in MMS-motivation (p < 0.001).
Age, recurrent stroke, mRS, hemiplegia, facial palsy, general weakness and MMS-motivation were signi cant positive predictors of the disability of GAR in the model. For every one-year increase on age, the odds of being in more severe category on GAR was 1.06 times higher (p < 0.001). This indicated that a participant aged older was more likely to indicate greater disability of GAR. The odds of being in more severe level on GAR was 1.88 times higher for those who experienced recurrent stroke as compared to those who had experienced stroke attack only once (p = 0.024). The odds of being in more severe level on GAR was 8.27 times higher for those who had the level of moderate to severe as compared to those who had the level of normal to mild in mRS (p < 0.001). And the odds of being in a higher level on GAR were 3.86, 4.85 and 3.19 times higher when a participant had the complication in hemiplegia, facial palsy, and general weakness, respectively (p < 0.001; 0.007; 0.027). In addition, the odds of being in more severe level on GAR was 3.22 times higher for those who showed low level of adherence to medication as compared to those who showed high in MMS-motivation (p < 0.001).
Age, highest academic quali cation, mRS, hemiplegia and MMS-motivation were signi cant positive predictors of the disability of SCA in the model. For every one-year increase on age, the odds of being in more severe category on SCA was 1.07 times higher (p < 0.001). This indicated that a participant aged older was more likely to indicate greater disability of SCA. The odds of being in more severe level on SCA were 2.92 and 2.66 times higher for those who had their highest academic quali cation as high and middle school, respectively as compared to those who had college and above (p = 0.018; 0.045). The odds of being in more severe level on SCA was 11.6 times higher for those who had the level of moderate to severe as compared to those who had the level of normal to mild in mRS (p < 0.001). And the odds of being in more severe level on SCA was 5.32 times higher when a participant had the complication in hemiplegia (p < 0.001). In addition, the odds of being in more severe level on SCA was 2.88 times higher for those who showed low level of adherence to medication as compared to those who showed high in MMS-motivation (p = 0.001).
Age, highest academic quali cation, recurrent stroke, hemiplegia, dysarthria and MMS-motivation were signi cant positive predictors of the disability of GAP in the model. For every one-year increase on age, the odds of being in more severe category on GAP was 1.02 times higher (p = 0.051). This indicated that a participant aged higher was more likely to indicate greater disability of GAP. The odds of being in more severe level on GAP was 2.17 times higher for those who had their highest academic quali cation as middle school as compared to those who had college and above (p = 0.048). The odds of being in more severe level on GAP was 1.72 times higher for those who experienced recurrent stroke as compared to those who had experienced stroke attack only once (p = 0.049). And the odds of being in more severe level on GAP were 2.72 and 1.82 times higher when a participant had the complication in hemiplegia, and dysarthria, respectively (p < 0.001; 0.015). In addition, the odds of being in more severe level on GAP was 3.83 times higher for those who showed low level of adherence to medication as compared to those who showed high in MMS-motivation (p < 0.001).
Age, highest academic quali cation, monthly family income, mRS, hemiplegia, dysarthria, MMSknowledge and MMS-motivation were signi cant positive predictors of the disability of LAC in the model. For every one-year increase on age, the odds of being in more severe category on LAC was 1.05 times higher (p < 0.001). This indicated that a participant aged older was more likely to indicate greater disability of LAC. The odds of being in more severe level on LAC was 2.23 times higher for those who had their highest academic quali cation as middle school as compared to those who had college and above (p = 0.051). The odds of being in more severe level on LAC was 1.77 times higher for those who had their monthly family income as one million and less Korean won as compared to those who had more than two million and less Korean won (p = 0.048). The odds of being in more severe level on LAC was 10.17 times higher for those who had the level of moderate to severe as compared to those who had the level of normal to mild in mRS (p < 0.001). And the odds of being in more severe level on LAC were 6.23 and 1.87 times higher when a participant had the complication in hemiplegia, and dysarthria, respectively (p < 0.001; 0.013). In addition, the odds of being in more severe level on LAC was 4.35 and 3.83 times higher for those who showed low level of adherence to medication as compared to those who showed high in MMS-knowledge and MMS-motivation, respectively (p = 0.016; <0.001).
Living without spouse, recurrent stroke, mRS, hemiplegia, dysarthria, trouble seeing, cognition problem, general weakness and MMS-motivation were signi cant positive predictors of the disability of PSO in the model. For living without spouse, the odds of being in more severe category on PSO was 1.76 times higher (p = 0.017). The odds of being in more severe level on PSO was 1.73 times higher for those who experienced recurrent stroke as compared to those who had experienced stroke attack only once (p = 0.050). The odds of being in more severe level on PSO was 12.48 times higher for those who had the level of moderate to severe as compared to those who had the level of normal to mild in mRS (p < 0.001).
And the odds of being in more severe level on PSO were 3.87, 1.94, 5.45, 6.06 and 3.88 times higher when a participant had the complication in hemiplegia, dysarthria, trouble seeing, cognition problem and general weakness, respectively (p < 0.001; 0.008; 0.001; 0.007; 0.010). In addition, the odds of being in more severe level on PSO was 2.59 times higher for those who showed low level of adherence to medication as compared to those who showed high in MMS-motivation (p = 0.002).

Discussion
To our knowledge, this is the rst detailed and nationwide disability prevalence survey on ischemic stroke patients at 1 year after onset in Korea. The study shows that prevalence of disability based on the WHODAS 2.0 is 62.6% which is almost double compared to hemiplegia (33.8%), one of the most common neurological sequelae 1 year after stroke. And the prevalence of severe disability (the WHODAS 2.0 of 50 100%) is higher in PSO (16.8%) and GAR (11.8%) than in the other domains. It also demonstrates that each domain of disability increases with various associated factors. Particularly, age, recurrent stroke, moderate to severe mRS, hemiplegia and dysarthria are generally related to different domains of disability and low MMS motivation is the only modi able factor determining the signi cant association between all six domains of disability after adjustment.
Concerning the personal background, age is associated with disability like as previous studies using WHODAS 2.0 [18-20]. Greater disability tends to increase as age goes higher. The elderly are more vulnerable to age-related comorbidity related with physical health problems [21]. However, even though the adjusted odds of being in a higher category on each domain except SCA is higher (aOR of 1.13 ~ 1.42) for female compared to male, these sex-related differences in WHODAS disability measurements are not signi cant. A Korean study previously has reported that male elderly stroke patients seem to be more vulnerable to self-care because of Korean tradition of the passive domestic role of male [19]. The Framingham study has reported that female with ischemic stroke is not functionally more disabled than male [21]. PSO is particularly limited by the most variables such as living without spouse, recurrent stroke, moderate to severe mRS, hemiplegia, dysarthria, trouble seeing, cognition problem, general weakness, and low MMS motivation. However, both LAC and GAP are associated only with hemiplegia and dysarthria among seven variables of neurological sequelae. This indicates PSO is not only about getting along with people either not only doing daily life. A prior study considers PSO as most problematic and important because this domain has involved the usage of complex skills and navigation in daily life [20].
Each neurological sequela is associated with different domains of WHODAS 2.0. For example, hemiplegia is associated with ve domains except UAC; dysarthria with UAC, GAP, LAC, and PSO; trouble seeing with UAC and PSO; and general weakness with GAR and PSO. Therefore, a tailored support can be shaped such as home visiting, a comprehensive type for hemiplegia, and companion going out, a simpler type for general weakness. It would be reasonable to manage these supports according to periodically assessed HRQoL.
It is of interest and importance that low MMS motivation is signi cantly associated with all six domains of disability after adjustment (OR of 2.59 ~ 3.83) because this variable is modi able and essential to prevention from repeating event. Medication adherence is usually known as the proportion of days covered (PDC), the percentage of medication actually taken of the prescribed doses [22], at 1 year after stroke. The Epidemiologic Research Council of the Korean Stroke Society reports a much lower adherence compared to a previous study from the US [23] (75% vs. 91% for lipid-lowering drugs, 74% vs. 91% for antidiabetic drugs, and 82% vs. 92% for antihypertensive drugs) [3]. Moreover, unlike MMS knowledge, MMS motivation is also associated with the adherence to lifestyle modi cation for risk reduction [24]. Such evidences imply that there is a substantial room of improvement on HRQoL for stroke survivors. It is necessary for stroke survivors to provide with interventions to improve MMS motivation by the speci c methods such as a tailored education, computer-based education, mobile phone reminders.
This study has several limitations. Our participants having regular outpatient follow-up at a particular university hospital are regarded as persons from higher socioeconomic status in Korean context. Thus, it is possible they demonstrated better level of stroke-related factors as well as adherence to their medication compared with stroke survivors in the general population. There is also a possibility of selection bias from excluding the stroke survivor 1 year after event due to di culties in the interview in spite that we tried to ensure stroke survivors with eligibility participated in the study unless they did not agree to participate. In addition, the WHODAS 2.0 covers mainly the activities and participation domains of the ICF, so there has been a need to be addressed for bodily impairments and environmental factors [9]. However, this study choose several bodily impairments related factors to be investigated such as hemiplegia, dysarthria, facial palsy and so on.

Conclusions
Self-perceived disability by the WHODAS 2.0 had almost double the prevalence compared to hemiplegia, one of the most common neurological sequelae 1 year after stroke. Each domain of disability increased with various associated factors. Interventions promoting medical adherence of motivation seem to help high HRQoL in all domains. Declarations JKS. All authors contributed to the discussion as well as the process of review and approved the nal manuscript.

Funding
This work was supported by the Ministry of Health and Welfare of Korea.

Availability of data and materials
The datasets generated and/or analyzed during the current study are not publicly available because an informed consent was not obtained from the participants during enrollment, but are available from the corresponding author at songj@jejunu.ac.kr on reasonable request.

Ethics approval and consent to participate
The study was approved by the institutional review board of Kangwon National University Hospital.

Consent for publication
Not applicable.