Testing of a Health-Related Quality of Life Model in Patients with Heart Failure: A Cross-Sectional, Correlational Study

Abstract

Psychological symptoms, physical symptoms, and behavioral factors can affect health-related quality of life (HRQOL) through different pathways, but the relationships have not been fully tested in prior theoretical models. The purpose of this study was to examine direct and indirect relationships of demographic (age), biological/physiological (comorbidity), psychological (depressive symptoms), social (social support), physical (physical symptoms and functional status), and behavioral (dietary sodium adherence) factors to HRQOL. Data from 358 patients with heart failure were analyzed using structural equation modeling. There was a good model fit: Chi-square = 5.488, p = .241, RMSEA = .032, CFI = .998, TLI = .985, and SRMR = .018. Psychological symptoms, physical symptoms, and demographic factors were directly and indirectly associated, while behavioral and biological/physiological factors were indirectly associated with HRQOL through different pathways. Behavioral factors need to be included, and psychological factors and physical factors need to be separated in theoretical models of HRQOL

Introduction

More than 50% of the population in the United States has at least one chronic illness, and the prevalence increases with advanced age.¹ Health-related quality of life (HRQOL) in people with chronic illnesses is considerably impaired.^2–4 Chronic heart failure (HF) is a serious chronic illness and is especially prevalent in individuals of advanced age.⁵ Health-related quality of life in people with HF is considerably poorer than that in healthy populations.^{2, 6} People with HF value good HRQOL as much as longer survival.⁷ Good HRQOL in older people, including people with HF, can be maintained until near death with appropriate management.⁸ Thus, a major focus on care in patients with chronic illnesses, such as HF, is enhancing HRQOL.

Health-related quality of life is commonly defined as an individual’s subjective perception of overall quality of daily life related to health,⁹ and is influenced by multidimensional factors.^10–12 To enhance HRQOL, healthcare providers and researchers need to understand the multidimensional factors affecting HRQOL and their inter-relationships. In patients with chronic illnesses, the relationships of health perceptions, physical factors, behavioral factors, psychological factors, socio-spiritual factors, biological/physiological factors, and demographic factors to HRQOL have been examined. Health perception is defined as a person’s subjective perceptions of their overall health status and is associated with HRQOL.¹⁰.¹² Physical factors include perceptions of both physical symptoms and functional status¹¹ and are associated with health status and HRQOL.^{9, 12} Behavioral factors are defined as any activity that an individual takes to promote, protect, and/or maintain health,¹³ and they are associated with physical factors.¹⁴ Psychological factors are defined as individuals’ subjective perceptions of the state of own cognition and affect.¹¹ Psychological factors, such as depressive symptoms or somatic-affective symptoms, have been associated with biological/physiological factors,¹⁵ behavioral factors,^16–18 physical factors,^{19, 20} health perception, and HRQOL.²¹.^22–24 Socio-spiritual factors are defined as individuals’ subjective perceptions of the state of support from others and own spiritual.¹¹ Socio-spiritual factors are associated with biological/physiological,²⁵ psychological factors,²⁶ behavioral factors,^{17, 27} physical factors,²⁶ health perception, and HRQOL.^{26, 28, 29} Biological/physiological factors, such as comorbidities, left ventricular ejection fraction, inflammation, and body mass index, are related to psychological factors,³⁰ behavioral factors³¹ physical factors,^32–34 and HRQOL.^{35, 36} Demographic factors, such as age,^{24, 37, 38} gender,^{24, 38, 39} race,³⁹ and education level,³⁹ are commonly associated with biological/physiological factors,⁴⁰ socio-spiritual factors,⁴¹ psychological factors,^{19, 42} physical factors,³⁴ and HRQOL. However, to our knowledge, prior HRQOL models have not included the direct and indirect relationships of these factors to HRQOL for patients with chronic illness.^10–12

In the prior HRQOL models,^10–12 behavioral factors and spiritual factors were not considered as factors affecting HRQOL, and physical and psychological factors were not clearly separated. Inclusion of behavioral factors in HRQOL models can be important because they can affect HRQOL through the relationships to physical factors,¹⁴ which strongly affect HRQOL.⁴³ Inclusion of spiritual factors can be important because they can directly and indirectly affect HRQOL through the relationships to psychological and behavioral factors.^{27, 44, 45} Separation of physical and psychological factors is important for three reasons. First, the prevalence of physical symptoms (present in approximately 80% - 90% of patients) differ from that of psychological symptoms (present in approximately 22% - 44% of patients) in patients with HF.^{19, 46–48} Second, to examine the unique relationship of physical symptoms with HRQOL and to manage physical symptoms appropriately, it is critical to assess physical symptoms separately from psychological symptoms. Third, the relationships and directions of behavioral and biological/physiological factors to physical symptoms differ from those to psychological symptoms.^{14, 33, 49, 50} For example, psychological symptoms commonly affect behavioral factors,⁵⁰ while behavioral factors commonly affect physical symptoms.¹⁴ In a prior HRQOL model for people with HF, health perception, symptom status, functional status, social status, biological/physiological status were included in the model to determine factors associated with HRQOL and the direct and indirect relationships.¹² However, behavioral factors and spiritual factors were not included, and physical and psychological symptoms were treated as a latent variable in the model.¹² The prior HRQOL model explained 58% of the variance in HRQOL of patients with HF. The findings imply that the addition of behavioral factors and spiritual factors and study of the indirect associations of physical and psychological symptoms to HRQOL by separating physical symptoms from psychological symptoms may increase our ability to understand HRQOL in people with HF.

We developed Heo-Moser Health-Related Quality of Life Model based on findings in the literature to increase the explanatory power in people with chronic illness by including behavioral factors and spiritual factors and by separating physical and psychological symptoms. The purpose of this study, therefore, was to introduce the model and pilot test the model, including the relationships of demographic (age), biological/physiological (comorbidity), psychological (depressive symptoms), social (social support), physical (physical symptoms and functional status), and behavioral (dietary sodium adherence) factors to HRQOL, in people with HF. We hypothesized that demographic, biological/physiological, psychological, social, and physical factors would be directly and indirectly associated with HRQOL, while behavioral factors would be indirectly associated with HRQOL. In addition, we also hypothesized that physical symptoms would be associated with HRQOL indirectly through health perception, while psychological symptoms would indirectly be associated through biological/physiological factors, behavioral factors, physical factors, and health perception.

Materials and Methods

Design

This was a correlational study using cross-sectional baseline data from two prospective, observational studies in which research participants were recruited from outpatient clinics and inpatient units of university-affiliated and community hospitals in four Midwestern and Southern cities in the United States (2002–2009; and 2008–2012).^{14, 32, 51}

Sample

The inclusion criteria for research participants were having a confirmed diagnosis of HF with reduced or preserved ejection fraction, taking stable doses of medications for 2 consecutive clinic visits or a 3-month period, and having the ability to read and speak English. The exclusion criteria were HF due to rheumatic disease, valvular heart disease, or pregnancy; diagnosis of stroke or myocardial infarction within the prior 3 months, fatal comorbidities (e.g., end stage of renal failure), severe cognitive issues (e.g., dementia), or severe psychiatric issues (e.g., schizophrenia). Using structural equation modeling for this secondary analysis, sample size was calculated based on number of parameters in the model (38) × 10 participants = 380.⁵² In the parent studies, 472 patients were enrolled. Among them, 11 were excluded because of withdrawal or being lost to follow-up, and 103 were excluded because of missing data. Thus, data from 358 patients were used in this study.

Procedure

The approvals of the two studies were obtained from relevant Institutional Review Boards (131581, 0712–65, R2008–064, 09–0579-P6H, and 04–0485-F2L). Research participants were recruited from outpatient clinics and/or inpatient units in community hospitals or academic medical centers in four Midwestern and Southern cities in the U.S. All research participants gave written informed consent. A research team member collected baseline data at each participant’s home, a clinic, or a place that both the research team member and the research participant agreed upon. The majority of research participants filled out the questionnaires using paper and pen. For a small number of participants, trained research nurses read the questionnaires for the research participants and filled out all or part of the questionnaires.

Measures

Sample characteristics included gender, race/ethnicity, etiology of HF, and left ventricular ejection fraction. Data on gender and race/ethnicity were collected using a standardized Sociodemographic Questionnaire. Data on etiology of HF and left ventricular ejection fraction were collected using a standardized Clinical Questionnaire.

Health-related quality of life was assessed using the Minnesota Living with Heart Failure Questionnaire.⁵³ The validity and reliability of this questionnaire have been well supported.⁵³ Higher scores indicate poorer HRQOL (0–105). Cronbach’s α in this study was .94.

Physical factors were physical symptoms and functional status. Physical symptoms were assessed using the Symptom Status Questionnaire-Heart Failure, consisting of 7 combined items of dyspnea when lying down, dyspnea during daytime, chest pain, fatigue, sleeping difficulty, edema, and dizziness.³² The validity and reliability have been supported well.³² Higher scores indicate more severe HF symptoms (range 0–84). Cronbach’s α of the combined 7 items in this study was .81. Functional status was defined as New York Heart Association functional class and assessed based on New York Heart Association Functional Classification System⁵⁴ by trained research nurses using in-depth interview.

Psychological factors were reflected by depressive symptoms. Depressive symptoms were assessed using the Beck Depression Inventory II.⁴² The validity and reliability have been well supported.⁴² Higher scores indicate more severe depressive symptoms (range 0–63). Cronbach’s α in this study was .91. Socio-spiritual factors were reflected by social support. Social support was assessed using the Multidimensional Scale of Perceived Social Support.⁵⁵ The validity and reliability have been supported in patients with HF.⁵⁶ Higher scores indicate higher levels of support (12–84). Cronbach’s α in this study was .95.

Behavioral factors were reflected by adherence to the low sodium diet. Dietary sodium adherence was defined as the amount of sodium excreted over 24 consecutive hours.⁵⁷ Patients collected urine in a container or containers and recorded the time of starting 24-hour urine collection, time and volume of each urination, and end time of 24-hour urine collection in a diary.

The biological/physiological factor used was comorbidity burden because comorbidities are a common physiological factor that affects physical factors and HRQOL.^{58, 59} Comorbidity burden was assessed by the Charlson Comorbidity Index.⁶⁰

Age was selected to reflect demographic factors because age is consistently associated with HRQOL²⁴ or factors affecting HRQOL, such as physical factors, psychological factors, and biological/physiological factors.^{34, 40, 42} Age was measured by self-report using a standard questionnaire.

Data analysis

Descriptive statistics were used to describe sample characteristics using IBM SPSS Statistics for Windows (Version 25.0 Armonk, NY).⁶¹ Structural equation modeling was used to analyze the relationships among demographic (i.e., age), biological/physiological (i.e., comorbidity), behavioral (i.e., sodium intake), psychological (i.e., depressive symptoms), social (i.e., social support), and physical (i.e., physical symptoms and functional status) factors with HRQOL using Mplus.⁶² In structural equation modeling, the criteria for good model fit were: Chi-square test (non-significant p value), root mean square error of approximation < .07, Tucker-Lewis Index > .95, comparative fit index > .95, and standard root mean square residual < .08.⁶³ Two tailed tests and p value < .05 were used for all the analyses.

Results

Table 1 presents sample characteristics. The mean age of the sample (N = 358) was 60.44 ± 12.51 years. The majority were male (61%), White (68%) and at New York Heart Association functional class II (45%) or III (36%). The mean score of HRQOL assessed by the Minnesota Living with Heart Failure Questionnaire was 43.35 ± 24.68.

Table 1.

Sample Characteristics (N = 358)

Characteristic	Mean	Standard Deviation
Age, years	60.44	12.51
Comorbidity	3.07	1.88
Left ventricular ejection fraction, %	34.42	13.36
Education, years	13.91	3.47
Health-related quality of life	43.35	24.68
Physical symptoms	25.79	16.96
Depressive symptoms	10.97	8.69
Social support	66.23	18.07
Dietary sodium adherence, g/day	3.82	2.18
	n	%
Gender (male)	218	60.9
Marital status (married)	179	50.0
Race
White race	244	68.2
African American race	109	30.4
Other races	5	1.4
New York Heart Association functional class
Class I	25	7.0
Class II	162	45.3
Class III	129	36.0
Class IV	42	11.7
Etiology of heart failure
Ischemic	148	41.3
Idiopathic	65	18.2
Hypertension	46	12.8
Other or unknown	99	27.7

Health-related quality of life was assessed by the Minnesota Living with Heart Failure Questionnaire. Physical symptoms were assessed by the Symptom Status Questionnaire-Heart Failure. Depressive symptoms were assessed by the Beck Depression Inventory II. Social support was assessed by the Multidimensional Scale of Perceived Social Support. Dietary sodium adherence was assessed by 24-urine sodium excretion.

In structural equation modeling, good model fit was evident by the following: 1) Chi-square = 5.488, p = .241 (acceptable level: non-significant p value); 2) root mean square error of approximation = .032 with 95% confidence interval = 0.00, 0.09 (acceptable level: < .07; 3) comparative fit index = .998 (acceptable level: > .95); 4) Tucker-Lewis Index = .985 (acceptable level: > .95); and 5) standard root mean square residual = .018 (acceptable level: < .08). All the factors in the Heo-Moser model were directly and/or indirectly associated with HRQOL (Figure 2). The model explained 63.7% of the variance in the model.

Figure 2. Factors Associated with Health-Related Quality of Life: Structural Equation Modeling.

Note: Significant relationships are presented using thick arrows, while non-significant relationships using thin arrows. Solid lines indicate direct relationships of the variables to, while dotted lines indicate indirect relationships of those variables to health-related quality of life.

Direct relationships with HRQOL.

Age (demographic factor) ([standardized] estimate = −0.26 ± standard error [SE] = 0.07, p < .001), depressive symptoms (psychological factor) (0.76 ± 0.12, p < .001), physical symptoms and functional status (physical factors) (0.76 ± 0.06, p < .001 for physical symptoms = 2.72 ± 1.13, p = .016 for functional status) were directly related to HRQOL.

Indirect relationships with HRQOL.

Age was also related to HRQOL indirectly through its relationships to biological/physiological, psychological, and socio-spiritual factors. Age was significantly associated with comorbidity (estimate = 0.03, SE = 0.01, p = .004), depressive symptoms (−0.16 ± 0.04, p < .001), and social support (0.23 ± 0.08, p = .002). Comorbidity was only indirectly related to HRQOL through the relationships to physical factors. Comorbidity was significantly associated with physical symptoms (1.79 ± 0.38, p < .001) and functional status (0.09 ± 0.02, p < .001). Depressive symptoms were also related to HRQOL indirectly through physical symptoms (0.99 ± 0.09, p < .001). Social support was only indirectly related to HRQOL through depressive symptoms (−0.14 ± 0.02, p < .001). Dietary sodium adherence was only indirectly associated with HRQOL through the relationship to physical symptoms (1.27 ± 0.33, p < .001). Physical symptoms were also significantly related to HRQOL indirectly through functional status (0.01 ± < 0.01, p < .001).

Comorbidity, depressive symptoms, and social support were not related to dietary sodium adherence; depressive symptoms and comorbidity were not related to each other; age and social support were not related to physical symptoms; depressive symptoms and social support were not related to functional status; and comorbidity and social support were not directly related to HRQOL.

Discussion

All factors in the Heo-Moser model were directly and/or indirectly related to HRQOL in people with HF, although not all the hypothesized relationships were supported. The results of model fit tests indicate good model fit. These findings indicate that all the variables in the model should be considered factors affecting HRQOL. In addition, the findings of this study supported the hypothesized indirect relationship of the behavioral factor to the HRQOL and different pathways in the relationships of physical factors and psychological factors to HRQOL. Thus, the findings of this study show that multidimensional factors affect HRQOL, and the relationships are complex. In addition, the findings of this study demonstrate the need to consider behavioral factors, physical factors, psychological factors separately in HRQOL models. Therefore, the findings of this study contribute to the body of knowledge regarding factors affecting HRQOL.

In the model, age, depressive symptoms, and physical symptoms were directly and indirectly, functional status only directly, and social support, comorbidity, and dietary sodium adherence only indirectly associated with HRQOL. The model explained more of the variance of HRQOL than a prior HRQOL model.¹² The prior HRQOL model of people with HF did not include behavioral factors and used physical and psychological symptoms as a latent variable. In our analysis, spiritual factors and health perception were not included. Inclusion of these two variables may further increase the explanatory power. The significant direct relationships of more severe depressive symptoms, physical symptoms, and functional impairment and younger age to poorer HRQOL in this study were consistent with those in prior studies.^{9, 22, 24, 43} Direct relationships of social support and comorbidity to HRQOL were not supported in this study, which is inconsistent with findings in prior studies.^{26, 35} In a prior study of HF,²⁶ the significant direct relationship between social support and HRQOL disappeared when depressive symptoms and physical symptoms were added into the model. Combined, these results imply that, when physical and psychological symptoms are considered, social support may be associated only indirectly with HRQOL. In a prior study of HF,³⁵ comorbidity was directly associated with HRQOL.³⁵ In that study,³⁵ the relationship was tested without physical symptoms, which are strong factors associated with HRQOL in people with HF.⁴³ Further studies are needed to examine the direct relationships of comorbidity and other biological/physiological factors to HRQOL considering other multidimensional factors affecting HRQOL. Overall, the findings of this current study partially supported the direct relationships of demographic, biological/physiological, psychological, socio-spiritual, and physical factors to HRQOL.

Among the hypothesized indirect relationships, the findings of this study supported the indirect relationship between behavioral factors and HRQOL through physical symptoms, indicating the mediator role of physical symptoms in the relationship. These findings were consistent with those in prior studies.^{9, 14} In a cross-sectional study,¹⁴ dietary sodium adherence was significantly associated with physical symptoms, and physical symptoms were significantly associated with HRQOL. However, in the study, dietary sodium adherence was not independently associated with HRQOL. In a recent longitudinal study,⁹ both baseline medication adherence and dietary sodium adherence predicted physical symptoms at 12 months, but baseline dietary sodium adherence did not predict HRQOL at 12 months. Thus, it is logical to hypothesize that behavioral factors can affect HRQOL through the effects on physical symptoms. On the other hand, in the prior study,⁹ baseline medication adherence predicted HRQOL at 12 months. All the findings of the current study and prior studies suggest that behavioral factors can affect HRQOL indirectly and in some cases also directly. These findings support the inclusion of behavioral factors in HRQOL models. Further studies are needed to determine whether behavioral factors only indirectly or also directly affect HRQOL in people with chronic illness, considering different behavioral factors.

The findings of this study demonstrate that physical symptoms and psychological symptoms can be associated with HRQOL through different pathways. In addition to the direct effects on HRQOL, physical symptoms indirectly affected HRQOL through the effects on functional status and also mediated the relationships of comorbidity, depressive symptoms, and dietary sodium adherence to HRQOL. That is, comorbidity, depressive symptoms, and dietary sodium adherence were directly associated with physical symptoms. In prior studies, biological/physiological factors (e.g., comorbidity and blood pressure),^{34, 58} psychological factors (e.g., depressive symptoms),^{9, 14} and behavioral factors (e.g., dietary sodium adherence and medication adherence)^{9, 14} were also significantly associated with physical symptoms. In addition, some prior studies suggested the possible mediating roles of physical symptoms in the relationships of behavioral factors or psychological factors to HRQOL.^{9, 14}

In contrast, we hypothesized the relationships of depressive symptoms to biological/physiological factors and behavioral factors differently from the relationships of physical symptoms to those factors. Physical symptoms were hypothesized as mediators of the relationships between those factors and HRQOL. On the other hand, depressive symptoms were hypothesized as a predictor of behavioral factors or a predictor and mediator of biological/physiological factors, although the hypothesized relationships were not supported in this study. In the relationships of psychological factors to behavioral factors, the predictor role was hypothesized based on the direct relationships of depressive symptoms, self-efficacy, or lack of knowledge to self-care or dietary sodium adherence.^16–18 In the relationships of depressive symptoms to biological/physiological factors, the predictor and mediator roles were hypothesized based the direct relationships of inflammation to depressive symptoms and depressive symptoms to inflammation.^{15, 30} One possible reason for the inconsistent findings in this study and in the prior studies may be that those relationships in the prior studies were examined using different variables without considering both direct and indirect relationships. Even though the hypothesized relationships of depressive symptoms to biological/physiological factors and behavioral factors were not supported in the current study, the findings of the current study and the prior studies demonstrate different pathways in the relationships of physical and psychological symptoms to HRQOL. Therefore, previous findings and our findings warrant and suggest that future researchers separate physical symptoms from psychological symptoms in HRQOL models.

The findings of this study suggest for those developing intervention to improve HRQOL to consider HRQOL as comprehensive, multidimensional construct, that includes demographic, biological/physiological, psychological, socio-spiritual, behavioral, and physical factors. In several meta-analyses of patients with HF,^64–66 the effects of different types of interventions on improvement in HRQOL were inconsistent. Several types of mobile phone interventions did not improve HRQOL.⁶⁴ Exercise-based interventions improved HRQOL based on the Kansas City Cardiomyopathy questionnaire, but not based on the Minnesota Living with Heart Failure Questionnaire.⁶⁵ Collaborative interventions, including components of psychosocial and palliative strategies, improved HRQOL, anxiety, and functional status.⁶⁶ However, none of the interventions assessed all relevant factors that can affect HRQOL. Thus, in further studies, it may be beneficial to consider all relevant factors in designing interventions to improve HRQOL.

There are some limitations of this study. This was a secondary data analysis. Thus, several factors that can affect HRQOL, such as spiritual factors, biological factors (e.g., inflammatory biomarkers), cognitive aspects, and health perception, could not be included, which could change the direct and indirect relationships in the model. We used only baseline data, which prevented us from examining the causal relationships. The mean age of the sample was relatively young, which could limit the generalization of the findings, but included balanced gender and New York Heart Association functional class ratios.

Conclusions

We proposed a HRQOL model and tested key variables of the model. All demographic, biological/physiological, psychological, socio-spiritual, behavioral, and physical factors were directly and/or indirectly associated with HRQOL, indicating the complex and dynamic relationships among those factors and HRQOL. These findings also demonstrate that all hypothesized variables in the model should be included in HRQOL models. Further research is needed to test the causal relationships in the full model.

Figure 1.

Heo-Moser Health-Related Quality of Life Model

Highlights.

• Multidimensional factors were related to health-related quality of life (HRQOL).

• Physical and psychological symptoms were related to HRQOL by different pathways.

• Behavioral factors were indirectly related to HRQOL through physical symptoms.