Functional status refers to an individual’s ability to carry out different levels of activities to meet the needs of daily living.¹ Patients with heart failure (HF) commonly have poor functional status that is affected by a number of physical, psychological, social and behavioural risk factors. Poor functional status is associated with markedly impaired quality of life and increased rehospitalization and mortality rates.¹ The focus of previous studies has been a single predictor or a single dimension of related predictors of functional status (e.g. only demographics or clinical factors), whereas the reality is that functional status has multidimensional determinants. As such, little is known from a multidimensional perspective about factors associated with functional status in HF. In the absence of published models or frameworks of functional status, we developed the Multidimensional Model of Functional Status (MMFS) (Figure 1) based on a review of the literature in patients with HF. The purpose of this study was to determine factors associated with functional status proposed in the MMFS (Figure 1). Using the MMFS, we evaluated demographic, clinical, psychosocial and behavioural factors along with symptom burden for their association with functional status using structural equation modelling (SEM).

Methods

Design, sample and setting

This was a cross-sectional study among patients with both preserved and reduced ejection fraction HF. We used SEM to test the MMFS and selected patients from the Research and Interventions in Cardiovascular Health (RICH) Heart Program Database who had complete data on all variables included in the model.²

Institutional review board approval was obtained for the study. All patients gave written consent after receiving information about the study and having the opportunity to ask questions. Trained research assistants recruited community-dwelling patients with chronic HF from cardiologists’ offices. Inclusion criteria included the following:

• adults with a clinical diagnosis of chronic HF with preserved or reduced ejection fraction confirmed by physician diagnosis;

• on stable doses of medical therapy;

• no major cognitive impairment that precluded patients from giving informed consent; and

• no diagnosis (other than HF) that was associated with risk of death in the next 6 months.

Patients were excluded if they:

• had suffered an acute myocardial infarction or stroke in the previous 6 months;

• had HF secondary to valvular heart disease, acute myocarditis or pregnancy;

• had frank liver or renal failure; or

• were awaiting cardiac transplantation.

Conceptual framework

The Multidimensional Model of Functional Status

In the MMFS (Figure 1), the demographic variables include sex, age, education and marital status.^3–9 The clinical variables include comorbidity burden and body mass index (BMI).^7,10 The psychosocial and behavioural variables include depression, anxiety, social support and adherence to self-care behaviours.^4,9,11,12

Demographic variables in the model reflect previous work demonstrating that female patients with HF have worse functional status than males and that older patients have worse functional status than younger patients.^6,7 Socioeconomic demographics, such as educational level and marital status, are related to health status, a construct in which functional status is a major component.⁸

Greater comorbidity burden, higher BMI, depression, anxiety, an increased need for social support, poor adherence and excess symptom burden are common in patients with HF.^4,7,9,10,12 All of these factors may affect functional status. Higher BMI and greater comorbidity burden were determinants of severe HF symptoms and worse functional status.^7,10 Depression and anxiety have both been identified as predictors of symptom burden, whereas HF symptoms have been identified as strong determinants of functional status.^4,13,14 Song et al. demonstrated in a sample of Korean patients with HF that an increased number of physical symptoms were associated with worse functional status through its effect on depression.¹¹ Social support from family, friends or important others has been found to be associated with HF symptoms.¹⁵ Adherence to the prescribed HF regimen is associated with lower symptom burden; however, the association between adherence and functional status remains unclear.¹⁶

Measures

Demographic variables

Data on demographic variables were collected using a standardized demographic questionnaire and from patient interviews.

Clinical variables

Comorbidity burden

Comorbidity burden was measured using the interview format of the Charlson Comorbidity Index (CCI).¹⁷ All participants were queried about pre-existing diseases and received weighted points for the presence of certain comorbid conditions, with higher scores indicating a higher comorbidity burden. The CCI has well-established reliability and validity for the prediction of worse outcomes in patients with chronic illness.¹⁷

Body mass index

BMI was calculated using the participant’s weight and height measured by the investigators and the formula: BMI (kg/m²)=weight (kg)/height (m).²

Psychosocial and behavioural variables

Depression

Depression was measured using the depression subscale of the Brief Symptom Inventory (BSI).¹⁸ The depression subscale of the BSI is a six-item self-reported measure of depression. Participants rated each item on a 5-point scale (0=‘not at all’ to 4=‘extremely’), indicating the degree to which they are affected by the stated symptom. The scores of the six items were averaged, with values ranging from 0 to 4. Higher scores indicate higher levels of depression. The depression subscale of the BSI had well-established reliability and validity. The Cronbach’s α for this scale in the current study was 0.87.

Anxiety

Anxiety was measured by the anxiety subscale of the BSI.¹⁸ The anxiety subscale of the BSI is a six-item self-reported measure of the state of anxiety. Participants rated their symptoms of anxiety on a 5-point scale (0=‘not at all’ to 4=‘extremely’). The scores of the six items were averaged, with values ranging from 0 to 4. Higher scores indicate higher levels of anxiety. The reliability and validity of the anxiety subscale of the BSI were reported in multiple studies. The Cronbach’s α for this scale in the current study was 0.86.

Social support

Social support was measured with the MPSSS.¹⁹ The MPSSS consists of 12 items rated on a 7-point scale ranging from 1 (very strongly disagree) to 7 (very strongly agree). The total scores on this scale range from 12 to 84; higher scores indicate a higher level of perceived social support. The reliability and validity of the MPSSS were reported in prior studies.²⁰ Cronbach’s α for this scale in the current study was 0.94.

Adherence

Adherence to self-care behaviours was measured using the Medical Outcomes Study (MOS) Specific Adherence Scale.²¹ The MOS scale is an eight-item self-reported measure that reflects patients’ adherence to the following self-care recommendations: (1) following a low salt diet; (2) taking all medications as prescribed; (3) exercising regularly; (4) not smoking; (5) drinking alcohol only in moderation; (6) following a low-fat diet; (7) weighing daily and (8) monitoring symptoms. The item response range is 0 (none of the time) to 5 (all of the time). Higher scores indicate higher levels of adherence. The MOS Specific Adherence Scale has well-established reliability and validity in patients with HF.²¹ The Cronbach’s α for this scale in the current study was 0.72.

Symptom burden

Burden from HF symptoms was measured using the eight-item physical subscale of the Minnesota Living with Heart Failure Questionnaire (MLHFQ), an instrument with well-established reliability and validity.²² Items on the MLHFQ are self-rated on a 6-point scale (0=‘no’ to 5=‘very much’), indicating the degree to which patients with HF had higher symptom burden during the last month. In the current study, the Cronbach’s α was 0.92.

Functional status

Functional status was assessed using the Duke Activity Status Index (DASI).⁶ The DASI has been used in a variety of cardiac disease populations, including HF.⁶ The DASI consists of 12 items, and each item has four response options ranging from 1=‘can perform activity without difficulty’, 2=‘yes, with some difficulty’, 3=‘no, I can’t do this’, to 4=‘do not perform activity for other reasons’. Each item is weighted based on the metabolic equivalent (MET) associated with the activity for that item. For example, the weight for walking indoors is 1.75, while that for running is 8. Only items that are rated 1 by the respondent receive a score that is equivalent to the MET level associated with that item. Items that are rated 2, 3 or 4, indicating that the activity can only be performed with difficulty or cannot be done at all, are scored as zero. The total score is calculated by adding the weighted score for each item. The total score can range from 0 to 58.2, with higher scores indicating better functional status. The reliability and validity of the DASI have been demonstrated previously.⁶ Cronbach’s α in the current study was 0.85.

Data analysis

Data analysis began with a descriptive examination of all variables using SPSS^® (IBM^®, Armonk, NY, USA), version 29.0, including frequency distributions, means, standard deviations, medians and interquartile ranges, as appropriate to the level of measurement of the variables. An alpha of <0.01 was denoted a priori.

In this study, SEM was done using Amos version 29.0 (IBM^®, Armonk, NY, USA) to determine whether demographic, clinical, psychosocial and behavioural factors were associated with symptom burden and whether symptom burden was associated with functional status (Figure 1). The model parameters were estimated using maximum likelihood estimation.

Because SEM is inherently a large-sample technique, the ratio of cases to free parameters should be at least 10:1 (preferably 15:1 to 20:1).²³ With a sample of 520 and 19 measured variables, the ratio of participants to free parameters was 27:1. Thus, the power was sufficient for this analysis.

The chi-square test is commonly used to assess model fit. However, chi-square alone has limitations because of its sensitivity to sample size: the larger the sample size, the greater the chances of obtaining a statistically significant finding.²³ Thus, we included a chi-square test to assess model fit and also the following indices: the Confirmatory Fit Index (CFI), Normed Fit Index (NFI) and the Root Mean Square Error of Approximation (RMSEA). A model was considered to have a reasonable error if the approximations of the CFI and the NFI were close to 0.90. Values of RMSEA in the range of 0.05–0.08 were used to indicate an acceptable fit.

Results

Sample characteristics

A total of 520 patients with HF were enrolled in this study. The mean age of patients in the sample was 62 years (SD=12) (Table 1). About one-third of the sample was female. The majority of patients were married (63%). The average left ventricular ejection fraction was 36.5%. The mean score for functional status was 13.8 (± 13.6). The full characteristics of the sample are described in Table 1.

Structural equation modelling

The MMFS model included four exogenous variables (sex, age, education and marital status) and eight endogenous variables (comorbidity burden, BMI, depression, anxiety, social support, adherence, symptom burden and functional status). The model produced a significant chi-square of 2,755.9 (df=118, p<0.001), indicating a poor model fit. Again, due to the large sample size in this study, we had a significant chi-square. Other fit indices indicated an acceptable fit of the model to the data (NFI=0.89, CFI=0.89, RMSEA=0.069).

Relationships among demographic, clinical, psychosocial and behavioural variables

Sex was associated with comorbidity burden and anxiety (Table 2, Figure 2). Female patients were more likely than male patients to have a lower comorbidity burden and higher levels of anxiety. Age was associated with BMI, depression, anxiety, social support and adherence. Older patients were more likely to have lower BMI, lower levels of depression and anxiety, more social support and better adherence compared with younger patients. Education was associated with comorbidity burden, depression, anxiety, social support and adherence. Patients who had higher levels of education had less comorbidity burden, lower levels of depression and anxiety, more social support and better adherence than those with lower levels of education. Marital status was associated with depression and social support. Married patients were more likely to have more social support and lower levels of depression.

Determinants of symptom burden and functional status

The determinants of symptom burden were sex, age, education, comorbidity burden, BMI, depression, anxiety and social support (p<0.01; Table 2). Patients who were female, older, less educated, had greater comorbidity burden, higher BMI, higher levels of depression and anxiety and less social support had greater symptom burden.

Patients who were older, less educated, had greater comorbidity burden and had greater symptom burden had worse functional status (Table 2). Age, education, comorbidity burden and symptom burden were directly associated with functional status, while sex, BMI, depression, anxiety and social support were indirectly associated with functional status mediated via symptom burden (Figure 2). Being married was indirectly associated with better functional status via the pathways of more social support and fewer depressive symptoms through lower symptom burden.

Discussion

We tested the MMFS among patients with HF. The findings supported most, but not all, of the relationships hypothesized in the MMFS. Advanced age, low educational level and greater comorbidity burden were directly and indirectly associated with poorer functional status. Being female and married, having higher BMI, more depressive symptoms, higher levels of anxiety and poorer social support were indirectly associated with poorer functional status. Most importantly, among the multiple variables proposed in the MMFS, symptom burden was the mediator of various demographic, clinical, psychosocial and behavioural variables with functional status.

Changes in clinical management are often driven by changes in functional status. Yet, functional status in HF is a complex and poorly understood phenomenon. The mean score of functional status in our patient population measured by the DASI was only 13.8, which is much lower (worse) than other patient populations.²⁴ Adequate understanding of the factors associated with functional status is important so that appropriate interventions targeting modifiable factors can be developed. Earlier studies of functional status failed to use multivariate analytic methods to study functional status. Numerous investigators have examined the bivariate relationships between functional status and single variables, such as sex, age, depression, anxiety, social support, adherence and symptom burden.^{3–7,9,11,12} The current study differed from previous studies in that the relationships of multiple variables with functional status were studied based on a multidimensional conceptual model (i.e. MMFS model) using SEM. Such a model can propose mechanistic relationships linking predictors with an outcome, thus providing testable hypotheses for future studies about cause and effect.

Based on our MMFS model, we found that symptom burden was the most important mediator among demographic, clinical, psychosocial and behavioural variables and function status. Most patients with HF have multiple symptoms (e.g. fatigue, shortness of breath and swelling).¹⁶ Their symptoms negatively influence their ability to perform activities of daily living and cause limitations in their daily lives. Poor functional status is associated with worse quality of life and higher rates of rehospitalization and death.¹ Guidelines from the American Heart Association and the Heart Failure Association of the European Society of Cardiology recommend undertaking daily and regular exercise to stay physically active for better health outcomes.^25,26 In a meta-analysis of 44 trials in patients with HF, investigators found that exercise-based cardiac rehabilitation significantly reduced hospitalization and improved health-related quality of life, indicating the importance of improving patients’ functional status.²⁷

However, in a multicentre, randomized controlled trial (i.e. HF-ACTION [Heart Failure: A Controlled Trial Investigating Outcomes of Exercise Training; ClinicalTrials.gov identifier: NCT00047437]) involving 2,331 out-patients with HF, investigators tested the efficacy and safety of exercise training and they reported no significant difference in all-cause mortality or all-cause hospitalization and no difference in 6-minute walk distance at 12 months between treatment and usual care groups.²⁸ The mixed findings might be due to not targeting or considering the mediator of symptom burden. Patients with HF often experience fatigue and shortness of breath even from simple activities such as walking around their home or performing their morning care and therefore quit doing exercises.²⁹ Under these circumstances, it is not surprising to find that symptom burden lowered patients’ motivation to keep exercising or physically active, thereby compromising functional status. One important reason the HF-ACTION investigators commented on their null findings was the lack of compliance with the exercise regimen in the treatment group, which might be due to symptoms.²⁸ Researchers have emphasized the importance of monitoring HF symptoms and responding to the symptoms in a timely manner.³⁰ Therefore, targeting symptom burden should be an integral component of HF management programs or interventions to improve health outcomes among patients with HF. Management of symptom burden includes educating and training patients on when, what and how to monitor their HF symptoms, how to respond promptly when symptoms occur and how to prevent or reduce symptoms (e.g. how to slowly, gradually and cautiously increase exercise intensity and frequency).

Our results about the relationships among psychosocial and behavioural factors, symptom burden and functional status were consistent with previous studies. Prior investigators have reported that depression, anxiety and social support were associated with symptom burden, and physical symptoms have been identified as strong determinants of functional status.^{1,13,14,31–35} Higher depression and anxiety were associated with greater dyspnoea symptom burden.³¹ In a recent study of 347 patients with advanced HF, depressed patients reported greater symptom burden than non-depressed patients.¹³ In a cross-sectional study of 60 patients with HF, the investigators found that depression was associated with a greater number of symptoms that in turn was associated with poor health-related quality of life.³³ However, the results were not based on a mediation analysis. In our study, in addition to finding a relationship between depression/anxiety and symptom burden, using SEM, we further demonstrated that lower levels of depression and anxiety were indirectly associated with better functional status through lower symptom burden. In a sample of 231 Korean patients with HF, symptoms were associated with worse functional status through its effect on depression.¹¹ Therefore, our findings as well as those of others indicate the importance of developing and testing interventions to alleviate depressive symptoms/anxiety and symptom burden to improve functional status.

Our findings on the relationship between sex and functional status were consistent with current studies, indicating that female patients with HF have more symptoms and poorer functional status.^3,5 In terms of BMI, we found that a higher BMI was indirectly associated with poorer functional status through higher symptom burden. To the best of our knowledge, no study has identified symptom burden linked BMI with functional status. However, our findings might not be in line with our understanding of the relationship between BMI and outcomes in the current literature. Specifically, it is well known that a higher BMI is associated with lower mortality and morbidity in patients with HF, as known as the ‘obesity paradox’.³⁶ Investigators reported that a higher BMI was associated with worse HF symptoms and poorer functional status; yet, poorer functional status resulted in higher mortality and morbidity.^1,7,10,37 Therefore, whether a higher BMI is a ‘protective’ factor for better health outcomes (better functional status and lower morbidity and mortality) in patients with HF remains unclear and requires further investigation.

In addition, we found that advanced age, low educational level and higher comorbidity burden were directly associated with poorer functional status, which is consistent with prior studies.^6,9 It is well known that functional status declines due to advanced age and greater comorbidity burden.^6,9 Patients with low educational levels were more likely to have low health literacy, low HF knowledge, poor self-care behaviours, more symptoms and poor functional status.³⁸ Thus, interventions to improve functional status need to target high-risk patients having greater symptom burden and poorer functional status, such as those who are female, older, have lower education and higher comorbidity burden.

We hypothesized that better adherence was associated with less symptom burden and better functional status. However, we did not find a relationship between adherence behaviours and symptom burden or between adherence and functional status, which is inconsistent with a prior study demonstrating that adherence was associated with symptom burden.¹² It is not clear why adherence was not associated with symptom burden or functional status in our study. In a recent study, medication adherence measured by an objective Medication Event Monitoring System was significantly associated with symptom burden in 219 patients with HF.³⁹ Therefore, the inconsistent finding may be due to our measurement method, in which we used a self-reported measure of multiple adherence behaviours, which was subject to recall bias or social desirability that may not reflect patients’ actual adherence behaviours.⁴⁰ Another possibility is that adherence is not a strong driver of symptom burden and functional status; thus, the relationships disappear in the company of more powerful determinants of these outcomes.

Limitations

This study had limitations. First, the study sample was recruited from out-patient settings. It is possible that our findings may not be applicable to compromised hospitalized patients with acute exacerbation of symptoms. Therefore, prospective multicentre studies are needed to explore the use of the MMFS model in hospitalized patients across different institutions. Second, although the model fit is acceptable and described in ‘one population’ without external validation, our data were from the RICH Heart Program Database that included participants from multiple studies with similar inclusion and exclusion criteria from urban and rural areas. Third, in our model, we did not include physiologic variables, such as B-type natriuretic peptide (BNP) or medications that could affect functional status.^34,41,42 Fourth, self-reported data can be affected by recall bias.⁴⁰ Therefore, a prospective study that uses objective measures of physiologic variables (e.g. BNP), behavioural variables (e.g. medication adherence using an objective measure such as Medication Event Monitoring System) and outcome variables (e.g. functional status using 6-minute walk test) is needed to confirm our findings.^6,34,39

Conclusion

Although there is evidence that multiple variables may be associated with functional status, such evidence is limited to only a few studies in which investigators have not considered functional status in the context of multidimensional influences. The MMFS provides a framework from which to consider multiple potential determinants, as well as mechanisms between variables to provide targets for interventions to improve functional status in patients with HF.⁴³ Most relationships proposed in the MMFS model were supported. We found that symptom burden was a strong significant mediator between functional status and a number of demographic, clinical, psychosocial and behavioural variables. The findings may help researchers and clinicians identify patients at risk of poorer functional status and assist them by developing interventions to alleviate symptom burden and thus improve functional status. Such interventions could specifically target symptom burden, as well as psychosocial and behavioural factors (e.g. depression, anxiety and social support), to improve functional status in patients with HF.