Health Related Quality of Life Among Heart Failure Patients Attending an Outpatient Clinic at University of Gondar Comprehensive Specialized Hospital Northwest Ethiopia 2020: Structural Equation Modeling Approach

Background: Chronic heart failure (CHF) is one of the most important public health concerns in the industrialized and developing world having increasing incidence and prevalence. Measuring quality of life using rigorous statistical method may be helpful to provide input for decision makers, policy makers and development of guidelines for the Ethiopia. The aim of this study was to determine health-related quality of life and its associated factors among heart failure patients attending University of Gondar comprehensive specialized hospital. Methods: A cross-sectional study design was employed to select 469 heart failure patients who has follow up at the University of Gondar comprehensive specialized hospital consecutively from March 1 to 30, 2020. Data were entered to Epi-info7 and exported to STATA 14 and Amos for further analysis. The four Quality of life domains were measured with Standardized World Health Organization Quality of Life BREF. Structural equation modeling was employed to estimate the relationships among exogenous, mediating, and endogenous variables simultaneously. Results: Chronic heart failure patients had a signicant lower mean score in all domains of health-related quality of life (p-value< 0.0001). Age had a direct positive effect on all domains of health-related quality of life and a positive total effect on overall health related quality of life. Residency also had a direct negative effect on both physical and environmental health related quality of life domain. Duration of chronic heart failure had a direct negative effect on psychological health. There was strong correlation among the four domains of health-related quality of life.

World Health Statistics in 2012 showed that CHF has created an economic burden of 180 million dollars in the health system (5). Different physical and mental complications such as fatigue, depression, anxiety, edema, shortness of breath due to the chronic and prolonged disease course, and therapeutic processes have a serious and negative impact on the health related quality of life (HRQOL) (5,6). Lower HRQOLs correlate with increased hospitalization occupancy and mortality rates, and higher costs imposed on health systems, families, and patients (7)(8)(9).
Health-related quality of life is in uenced by numerous physical, emotional, and social factors and uniquely perceived by each individual. The structured assessment of HRQOL is considered important in promoting patient centric care. It puts the patient's perspective at the forefront and can identify areas of speci c need. This helps to facilitate shared decision-making and ensure that the preferences of the patient are used to guide management (4).
The assessment of symptoms and functional status, using the New York Heart Association (NYHA) classi cation, is a standard practice in the care of heart failure patients. It is recommended in guidelines as a measure of heart failure severity and is integrated into treatment decision algorithms, despite its shortcomings in reproducibility (10)(11)(12). The use of validated instruments to assess HRQOL remains largely limited to clinical trials, with minimal guidance on the practical assessment of HRQOL outside of this setting (13,14).
Maximizing HRQOL among treating CHF is still a major challenge (15)(16)(17)(18). In primary care, CHF patients are in need of new concepts adapted to the severity of their disease that consider life expectancy and HRQOL. Patients with CHF mostly experience symptoms such as dyspnea, fatigue, sleep disorders, and ankle edema (19).
Health-related quality of life is a broad and multidimensional concept that subjectively evaluates the physical, psychological, and social health status of individuals and in uenced by their understanding, experiences and expectations (20), but most of previous studies on health-related quality of life were considering as observed variable but HRQOL is a multidimensional concept that is better evaluated by a number of latent constructs.
Measuring HRQOL will help in monitoring treatment guidelines and improving patients' HRQOL. Analysis of HRQOL can identify groups with poor HRQOL, and this could guide interventions that will improve their situation and avert more serious consequences, allocate limited resources based on unmet need, guide strategic plan, and monitor the intervention given. More over limited studies were conducted on healthrelated quality of life among heart failure patients in Africa speci cally in Ethiopia.
In response to these identi ed gaps, we conducted a study based on su cient sample size with appropriate statistical analysis (multivariate statistical analysis). The main objectives of the study were (a) to determine health-related quality of life and associated factors among heart failure patients, and (b) to examine the association among socio-demographic and economic, clinical related variables, and among the domain of HRQOL variables. To account for the interdependency of various factors and health related variables, structural equation modeling was employed.

Study setting and period
An institution based Cross-sectional study was employed in University of Gondar comprehensive specialized hospital (UoGCSH) to take a snapshot of population of adult heart failure patients from March 01 to 30/2020. The Gondar city is 727 km North West of capital city of the country, Addis Ababa.
UoGCSH is located the center of Gondar city. Currently, the hospital has a catchment population about 7 million serving as a referral hospital for nearby hospitals. The hospital runs several medical outpatient services including heart failure follow up clinic.

Population
Patients diagnosed with CHF who had follow up for at least 6 month and age greater than 18 years who visit UoGCSH during the study period were selected as study participants. Patients were excluded from the study if they experienced a concurrent diagnosis of other life-threatening diseases (e.g., cancer), or a chronic severe psychiatric condition (e.g., psychosis).

Sample size determination
A general rule of thumb is that the minimum sample size should not be less than where k is number of observed variables (22). According to the foregoing rule, the minimum sample required were 465 since we have 30 observed variables.

Sampling technique and procedures
Most of patients in the follow up clinic have one follow up per each month, and the required sample was taken by enumerating all patients during the study period consecutively until the required sample size was secured. In the case of patients who had more than one follow up appointments during the data collection period, their appointment dates were checked and they were excluded from the study.

Variables and measurement
Data were collected with face to face interview using structured questionnaires. The questionnaire consists of socio-demographic characteristics, clinical related questions and the WHO-QOL tool. Data related to underlying causes of HF and medication regimen were obtained by reviewing the chart, and variables like age, sex, duration of disease, marital status, educational status, occupation, monthly income, enrolment in community-based health insurance and lastly all 26 item variables were obtained by interviewer administered face to face interview.
The Health related quality measurement domain questionnaire was adopted from WHOQOL which is also validated in Ethiopia and other part of the countries in the world (23). It has 4 domains that denote an individual's perception of quality of life in each particular domain. The WHOQOL-BREF is a 26-item instrument consisting of four domains: physical health domain (7 items), psychological health domain (6 items), social relationships domain (3 items), and environmental health domain (8 items); it also contains the overall perception of QOL and general health (2 items).
Initially questionnaire was prepared in English version, then translated into Amharic (local language) and translated back in to English by another person to check the consistence. Data processing, model building, and analysis The lled questionnaires were checked manually for completeness. Data were coded and entered into Epi-Info version 7 statistical packages and then exported to STATA version 14 and Amos version 25 for further analysis. Descriptive and summary statistics were done using gures and tables. Reliability was also be assessed for each domain of WHO-QoL -Brief using the Cronbach's α coe cient and values of 0.7 or higher were considered satisfactory. The score of each domain of WHO-QOL -Brief was obtained by averaging their corresponding items for each participant. Then the scores were transformed linearly to a 0-100-scale as described by author (4) The Structural Equation Modeling (SEM) was employed to examine the relationship between various exogenous and endogenous or mediating variables. Because HRQOL and its domain were latent variables which constitute items with ordered responses, their measurement model was analyzed using SEM because multivariate normality assumption was satis ed after the items parceling (24). The analysis was started with the hypothesized model ( Fig. 1), and modi cations were performed iteratively by adding path links or including mediator variables, if theoretically supported, and comparing the Root Mean Square Error of Approximation (RMSEA), Comparative Fit Index (CFI) of each model tted. Finally, an over identi ed model with value RMSEA < 0.05 and CFI > = 0.95 was retained. Diagrammatically, the effect of each exogenous or mediating variable on the respective dependent variable was indicated by the path coe cient along with single headed arrow, and the correlation among disturbances (residual errors that re ect the unexplained variances in the latent endogenous variables due to all unmeasured causes) was indicated by double arrows. When mediation of effects was present, the direct, indirect, and total effects were determined using the nonlinear combination of estimator technique.
The nal model (Fig. 5) that tted the data well and appeared theoretically meaningful was built by analyzing the hypothesized model ( Fig. 1) and inspecting iteratively the statistical signi cance of path coe cients and the relevance of relationships in the model.
The mean age of patients was 53 (18.88 SD) years (Table 1).    Perceived Health satisfaction and self-rating of HRQOL of patients Study participants were asked to give their perception on their quality of life and health satisfaction.
Based on their response; about one third 165 (36%) study participants reported that their quality of life was neither good nor poor, while 142 (31%) of them had poor QOL. Regarding health satisfaction, 202 (43.10%) of them were very dissatis ed with their health and only 3.40% of them were satis ed with their health (Fig. 3 and Fig. 4).

Health-related quality of life and other associated factors
The nal model containing both the structural part (relationships among latent or observed variables) and measurement part (relationship between a latent variable and its indicators or items) is shown in Fig. 4 and Table 5. The tted model was relatively parsimonious and good tted with RMSER = 0.04 and CFI = 0.96. Variables like sex, education, type of job, current medication and Utilization of health insurance were excluded from the nal model as their contributions were not statistically signi cant at an alpha level of 0.05. Table 5 The direct, indirect and total effect of socio-demographical and clinical factor on HRQOL domains among patients with CHF attending at UoG, 2020. This estimated structural equation model indicates that environment health factor had the most substantial causal effect on HRQOL, which was larger than the causal effects of psychological QOL, physical health and social relationship factors. Physical health was statistically signi cantly associated with the age (p < 0.0001), residency (p < 0.0001) and underlying cause of CHF(p = 0.002), psychological health domain was signi cantly associated with duration of CHF(P = 0.05), income (0.001), marital status (p = 0.022) and age (p < 0.0001), social relation was statistically signi cantly associated with age (p < 0.0001) and education (p = 0.006) and Regarding to environmental health domain, it was associated with age (p = 0.03), residency (p < 0.0001) occupation (p = 0.04), income (p < 0.0001)and health insurance (p = 0.007) (Fig. 4).

Testing the structural model of the HRQOL of patients with CHF
Feasibility assessment for the hypothetical model We conducted a con rmatory factor analysis of the measurement model in step 1. The con rmatory factor analysis was performed with demographic factors, social support, disease-related factors, behavioral factors, and HRQQOL, whereas we excluded insigni cant variable (p > 0.05) from the nal model.

Test of the goodness of t of the hypothetical model
The Effectiveness analysis of the hypothetical model The direct, indirect, and total effects of the factors associated with the HRQOL of the patients with CHF are presented in Table 5. The physical health domain had the greatest direct effect on the HRQOL with a score of 0.502. The environmental health factor had a direct effect on the HRQOL with a path coe cient of 0.485, and psychological health factor had a direct effect on HRQOL with a path coe cient 0.359.
Social relation had a direct effect of 0.167 but statistically not signi cant (p-value = 0.280 > 0.05) on the HRQOL. Monthly income had direct effect on HRQOL with a path coe cient of 0.01, and a total effect of 0.02 when added to the indirect effect of environmental health factor (0.001) ( Table 5).

Discussion
In this study, we aimed to construct a hypothetical model and verify the signi cance of the direct/indirect paths and the goodness of t of the model under the theoretical assumption that demographic factors, personal related factors, social relation, environmental factor, physical factor, disease-related factors, and behavioral factors, including depression, anxiety, fatigue, pain, sexual activity, and body image, determine the HRQOL of patients with CHF directly and indirectly.
In this study, we found that patients with CHF had lower quality of life in all domains of the WHOQOL-BREF especially in environmental health domain (mean score of 38.56) and an overall quality life, which indicate that they know that their heath is poor and affecting their HRQOL. This nding is congruent with other previous study (25). This consistence result might be heart failure is a serious condition and leads to poor quality of life when the condition is not managed with heart failure management's.
From structural equation model, we found that environmental health factors like physical security, nancial resources and health care facility had the most substantial causal effect on HRQOL of patients with CHF with path coe cient of 0.53 (95%CI, 0.32, 0.75), which was larger than the causal effects psychological health, and physical health, which were in turn larger than the causal effect of social relation domain. This nding is Inconsistent with other study (26). This inconsistency may be due to the research was conducted in developed country and the environmental health may not had larger cause on HRQOL than the other factors. In Environmental health factor physical security, information and skill, and health facility (with loading of 0.80) are the highly in ucial items on HRQOL as compared with physical environment, nancial resource and recreation items (with loading 0.6). and transportation and home environment items (with loading 0.58). This result might be, heart failure often leads to the development of physical disabilities that, in turn, can have a detrimental effect on a patient's quality of life.
Our nding demonstrated that, among domains of HRQOL, the physical health domain was most affected domain for HRQOL next to environmental health domain. This nding is In line with a couple of studies (26)(27)(28). This consistency could be defensible by CHF has more physical than mental (psychological) manifestations and social relations.
Similarly, psychological health and social relation was the least affected domain among the CHF patients. This nding is in line with other studies that was conducted southwest (29) and northwest (30) Ethiopia among DM patients. This consistence result in social relation and psychological health might be their social-culture that gives support for diseased individuals with DM and CHF. Patients manifest more physically than mentally (psychologically).
Our results revealed that age had signi cant association with all domains of HRQOL and had both a positive direct and indirect effect that resulted in a total positive effect on overall HRQOL of CHF patients.
Aged 60 and above years had worse mental state, physical health, social relation and environmental health. This nding is lined with previous studies conducted in different research setting (31,32). Based on the knowledge that CHF incidence increases with age, researchers would anticipate that older patients who experience several limitations such as cognitive impairment, loss of personal autonomy, or anxiety and depression may have poor quality of life (31).
A study conducted in Greece (31)  There is a study's (31) which had consistence nding with our study. This congruent nding might be patients' lives in rural area are more likely low in income, the physical environment may not be good, low health care accessibility to get medical treatment and most of them are uneducated so have the low awareness about CHF. Being married was another socio-demographic factor that had a positive effect on environmental health and social relation among CHF patients and this is of course in congruent with other reports (31). This positive effect may be due to, support from others can facilitate recovery from physical illness and enhance the ability to cope with and adapt to the consequences of chronic illnesses.
Also associated with both physical and mental (psychological) health were the years of suffering from the disease (duration of CHF) which may re ect symptoms' severity. Patients often experience loss of functional independence in daily activities such as feeding, dressing, housekeeping, bathing, and walking (33).
It is noteworthy that evaluation is needed of all the changes that take place through years and that may exacerbate HF patients' quality of life such as inability to ful ll their prior role (social, professional, and family), diminished self-esteem, and distorted picture of themselves.
The nding of the present study also showed that income had positive direct effect on physical health, psychological health and environmental health domain of HRQOL. These ndings were consistent with previous studies (34). We know that nancial situation is are important determinants of health, negatively affecting health outcomes and contributing to health inequities. Patients with low income had low quality of life because most of CHF patients are medical treatment dependence and unable to afford treatment costs.

Strength And Limitation
In this study, HRQOL was assessed based on standardized tool that is validated for both developed and developing country (WHOQOL_BREF). The current study also used multivariate analysis (SEM) that enables a simultaneous analysis of the impact of multiple independent variables on several dependent variables and the subsequent direct comparison of the respective impact of the independent variables on the dependent variables. This also enables to incorporate the correlation between latent variable and to determine direct, indirect and total effect when mediation effect present simultaneously. However, this study is not without limitations, the data were collected through face to face interview by considering the different educational level of respondents and this might prone to social desirability bias and could overestimate the result. Moreover, the present study has been conducted in single center which limits the generalizability of the nding in Ethiopia; further multicenter studies are needed to address this issue.

Conclusion
The nding of the study indicated that a moderate to poor HRQOL in the physical dimension, moderately poor overall HRQOL and a moderate to high HRQOL in the psychological health domain. Sociodemographic factors (age, residence and marital status), clinical factor (duration of CHF) were factors associated with HRQOL among CHF. The results indicated that the importance of QOL assessment at appropriate time periods, determining the exact treatment dimensions required, and implementing comprehensive HRQOL promotion programs in all physical, mental, environmental and social relation dimensions.
Based on this result, approaches should be developed or effectively managing physical factors, psychological, and environmental factors to improve the QOL of patients with CHF. Developing and providing intervention programs to enhance social support can lead to improved quality of life for patients, because of the long-term and chronic illness of the patient.    Perceived self-rated QOL of chronic heart failure patients attending at University of Gondar Hospital,2020.