Exploring the Evolution of Heart Failure Symptoms After Hospitalization and Associations With Social Determinants of Health and Demographic Characteristics

Abstract

Background:

There is little data on how heart failure (HF) symptoms change after hospitalization for HF (HFH) or with social determinants of health (SDOH). Such gaps in knowledge limit the ability of clinicians to intervene earlier to reduce symptoms and prevent rehospitalization.

Objective:

The aim of this study was to explore HF symptoms from HFH through 6 months postdischarge and explore associations of SDOH and demographic characteristics with symptom trajectories.

Methods:

We used an observational longitudinal design from HFH through 6 months postdischarge in 49 patients with New York Heart Association class III HF. Dyspnea (HF Somatic Perception Dyspnea Scale), fatigue (PROMIS-Fatigue), sleep quality (Pittsburg Sleep Quality Index), pain (PROMIS-Pain Interference), depressive symptoms (8-item Patient Health Questionnaire), anxiety (PROMIS-Anxiety), and HF-specific health-related quality of life (Kansas City Cardiomyopathy Questionnaire Short Version) measures were collected during HFH and at 1, 3, and 6 months postdischarge. Latent growth curve modeling and multivariate regression were used to characterize trajectories in symptoms and to determine associations of growth model estimates with SDOH and demographic characteristics.

Results:

Participants (N = 49) were 68 ± 10 years, 53% women, with an average ejection fraction of 44 ± 20%. Fatigue, dyspnea, anxiety, and HF-specific health-related quality of life improved significantly from HFH through 6 months posthospitalization (P < .05). Sleep quality, pain interference, and depressive symptoms did not change significantly from HFH through 6 months postdischarge. Novel associations between HF symptom trajectories and race, sex, education, and income were identified.

Conclusions:

Symptoms associated with congestion improved to 6 months after HFH while other less frequently assessed symptoms did not improve after discharge. Associations between SDOH, demographic characteristics, and HF symptoms provide new insights into social factors that may influence HF symptoms.

Heart failure (HF) is an increasingly common condition, affecting 6.7 million Americans over 20 years old.¹ Heart failure is the most common reason older adults are admitted to the hospital, and hospitalizations for HF are increasing.^1,2 Hospitalization for the treatment of HF is a key indicator in the progression of the condition.³ After the first hospitalization for HF, up to 60% of patients are readmitted within 1 year.⁴ Repeated hospitalization for HF is associated with higher mortality and poor quality of life.^3,5,6 After discharge for a HF hospitalization (HFH), 20%–25% of patients will be readmitted within 30 days, typically due to symptoms of cardiopulmonary congestion, primarily dyspnea, and fatigue.^7,8 In addition to physical HF symptoms, affective symptoms, including anxiety and depression, are prevalent in people with HF.^9,10 Worse affective symptoms have been associated with less engagement in HF self-management behaviors and lower quality of life.^11,12 Despite the important role of symptoms in the diagnosis and prognosis of HF, little is known about how HF symptoms evolve after HFH.

Social determinants of health (SDOH) are additional factors affecting health outcomes for people with HF. Social determinants of health are conditions in which people live that significantly contribute to health and well-being,¹³ including income, education, social support, and access to healthcare. Some SDOH have important associations with health outcomes for individuals with HF.^14,15 For example, those in households with lower income are less likely to have a cardiologist involved in their care when admitted to the hospital for HF, which may contribute to further disparities in care and health outcomes. People with HF who rate social support higher describe a better quality of life and less depression and anxiety.¹⁶ Demographic characteristics, such as age, sex, and racial identity, have also been associated with differences in HF care and the experience of HF symptoms. Black individuals with HF tend to report greater symptom burden compared with White individuals, and Black men were less often diagnosed clinically with depression compared with White men.^17,18 Older adults often describe less symptom distress than younger people, while women with HF typically experience more severe symptom burden compared with men.^19–22

Managing HF symptoms effectively is a primary goal for clinicians to prevent hospital admissions and to minimize the progression of HF and the impact on patient health-related quality of life (HRQOL) and mortality.^3,23 The ability of clinicians to address symptoms effectively is hindered by the paucity of published data on how symptoms change after patients are discharged for HF admissions, which is a barrier to clinical management in targeting critical time points for intervention. The limited longitudinal studies examining HF symptom trajectories in patients receiving medical therapy (not including mechanical circulatory support or other study interventions) have generally shown 2 groups of patients: those with higher symptoms and those with lower symptoms over time.^24,25 Psychological symptoms (depressive and anxiety) tended to decline over time, while physical symptoms (fatigue and dyspnea) were consistently stable over time.^24,25 These studies were often limited by relatively short periods of follow-up (5 days to 3 months), the use of different symptom measures, or a lack of a specific starting point from where symptoms might be expected to change.^24–26 Furthermore, there is a lack of data on how symptoms may differ by SDOH and some demographic characteristics, particularly after discharge from HFHs, which may result in less effective treatment and disparities in health outcomes. Thus, there is an urgent need to better understand patterns of patient-reported outcomes in the longer term after hospitalization, and how SDOH and demographic factors are associated with outcome trajectories.

The purpose of this study was to explore common HF symptoms and HF-specific HRQOL from hospitalization for HF exacerbation through 6 months postdischarge. In addition, SDOH and demographic characteristics were examined for associations with HF symptoms and HF-specific HRQOL at hospitalization and with changes from hospitalization through 6 months postdischarge. Insight into how symptoms and HRQOL change after discharge for HF exacerbation can provide important information about the evolution of symptoms during recovery from HFHs to guide clinical care. Examining SDOH and demographic characteristics that may be associated with symptoms and HRQOL can further elucidate differences in symptoms over time that may allow more effective symptom management and identification of disparities in the symptom experience.

Methods

Design and Setting

An observational prospective longitudinal design from HFH through 6 months postdischarge was used to conduct this study. The electronic health record was used to identify potential participants admitted for HF. Potential participants meeting the inclusion criteria were approached by the research staff for recruitment during their HFH and given information about the study. If interested in participating, individuals provided written informed consent. Surveys were collected in-person on a digital device using REDCap during the recruitment hospital admission and at 1, 3, and 6 months postdischarge from March 2020 through December 2022. Participants unable to meet in-person at 1, 3, or 6 months were sent an email to complete the survey electronically using REDCap or the survey was completed over the phone. If no response to the email was received within 7 days, participants were contacted to complete surveys via SMS texting and phone calls. Demographic and SDOH data were collected at hospitalization, and clinical characteristics were assessed from the electronic health record. Participants were recruited from 3 medical centers in an urban area in the Pacific Northwest. The study was reviewed and approved by the University of Washington Institutional Review Board. Given that this study was conducted during the COVID-19 pandemic, all hospital and clinic protocols to ensure staff and patient safety were followed.

Sample

A total of 49 patients with New York Heart Association (NYHA) class III HF participated in the study. Inclusion criteria included: (1) current hospital admission for HF, (2) NYHA class III functional status, (3) ability to read and comprehend English, (4) >18 years of age, and (5) willing and able to provide informed consent. Exclusion criteria included: (1) having received mechanical circulatory support or heart transplantation, (2) a major uncorrected hearing impairment, (3) diagnosis of severe cognitive impairment (ie, Alzheimer’s), and (4) engagement in end-of-life care or inability to complete the requirements of the study. Only those with NYHA class III HF were recruited to reduce the participant heterogeneity. NYHA class was adjudicated by an HF-trained physician.

Measures

Dyspnea was measured with the 6-item Heart Failure Somatic Perception Scale (HFSPS) dyspnea subscale. The HFSPS dyspnea asks how much the participant was bothered by difficulty breathing or shortness of breath during the last week and provides 6 response options ranging from 0 (not at all) to 5 (extremely bothersome). The HFSPS dyspnea has excellent internal consistency (α = 0.89) and concordant validity with the Kansas City Cardiomyopathy Questionnaire (KCCQ) functional limitations scale (r = 0.53) and independently predicts HF-specific clinical events.²⁷

Fatigue was measured with the 6-item PROMIS-Fatigue scale. The 6-item PROMIS-Fatigue measures the experience of fatigue during the past 7 days. The items are measured on a 5-point scale from 1 (not at all) to 5 (very much). A raw score is calculated by summing the 6 items with a range of 6–30. Raw scores correspond to T scores based on large nationally representative samples.²⁸ Higher scores indicate more fatigue, with a score of 50 being the average for the U.S. general population.²⁸

Sleep quality was measured using the Pittsburgh Sleep Quality Index (PSQI).²⁹ The PSQI assesses 7 patient-reported sleep domains over the last month, including subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbances, sleep medication, and daytime dysfunction. A global score for sleep quality was calculated, with a higher score indicating worse sleep quality. A global score > 5 is indicative of poor overall sleep quality. The PSQI has been shown to distinguish between poor and good sleep quality with good internal consistency (α = 0.83).²⁹

Pain interference was measured using the 6-item PROMIS-Pain Interference scale.³⁰ The 6-item PROMIS-Pain Interference scale measures the consequences of pain on social, cognitive, emotional, physical, and recreational activities. The items are measured on a 5-point scale from 1 (not at all) to 5 (very much). A raw score is calculated by summing the 6 items with a range of 6–30. Raw scores correspond to T scores based on large nationally representative samples.³⁰ Higher scores indicate more pain interference, with a score of 50 being the average for the U.S. general population.³⁰

Anxiety was measured with the 6-item PROMIS-Anxiety scale.³¹ The 6-item PROMIS-Anxiety scale asks respondents to rate their level of fear, anxiousness, and worry with 5 response options of never, rarely, sometimes, often, always (scale of 1–5). A raw score is calculated by summing the 6 items with a range of 6–30. Raw scores correspond to T scores based on large nationally representative samples.³¹ Higher scores indicate more anxiety, with a score of 50 being the average for the U.S. general population.³¹

Depressive symptoms were measured using the 8-item Patient Health Questionnaire (PHQ8). The PHQ8 provides 4 response options: 0 (not at all) to 3 (nearly every day) based on the Diagnostic and Statistical Manual, 4th Edition. Scores range from 0 to 24, with higher scores indicating more depression. The PHQ8 is sensitive and specific for detecting major depression with excellent internal consistency (α = 0.86–0.89).³² The severity of depression measured with PHQ8 includes scores 0–4 (none), 5–9 (mild), 10–14 (moderate), 15–19 (moderately severe), and >19 (severe).³²

Heart failure–specific HRQOL was measured with the KCCQ Short Version (KCCQ-12), a 12-item Likert scale that quantifies symptom frequency, physical and social function, and quality of life.³³ Scores range from 0 to 100, with higher scores reflecting better function. The KCCQ-12 had excellent concordant validity (0.93–0.99) with the original well-validated and reliable 23-item measure and excellent test–retest reliability (0.76–0.92).³³

Social determinants of health were collected with the baseline survey. The SDOH were selected based on the National Institute of Health’s conceptualization of SDOH, including social, economic (income, education, and employment), and legal structures (marital status).³⁴ The SDOH included in this study were income, marital or partner status, employment status, and educational attainment. Demographic characteristics included were age, sex, and self-identified race. The survey choices for sex included male, female, and an option for participants to state an identity other than male or female. Racial identity categories included American Indian, Asian, Black, Native Hawaiian or Pacific Islander, White, and an option to state a race that was not previously provided. Participants also had an option to select 2 or more racial identities. The survey asked participants to select 1 of 6 categories of annual income from <$19 999 to >$100 000. For marital status, participants were asked to select whether they were single, married or living with a domestic partner, divorced or separated, or widowed. Employment status included not being employed, employed part-time, and employed full-time. The educational attainment question asked participants to select their highest level of education, including less than high school, high school, some college, associate’s degree, bachelor’s degree, master’s degree, or doctoral degree.

Analysis

Standard descriptive statistics of frequency, central tendency, and dispersion were used to describe the sample. Piecewise or multiphase latent growth curve modeling (LGCM)³⁵ was used to characterize distinct trajectories of change in physical and psychological symptoms from hospitalization to 6 months postdischarge (R version 4.4.1 with lavaan package 0.6-19, Vienna, Austria).³⁶ Piecewise LGCM is a form of longitudinal analysis based on structural equation modeling that can accommodate nonlinear change. Since the change in symptoms from hospitalization to 6 months was not linear (more change early after discharge with less change later after discharge), piecewise LGCM was used to estimate different trajectories of change over time. The initial symptom value at hospitalization (intercept) and the rate of change (slope) over 2 time periods (slope 1 = hospitalization to 1 month postdischarge; slope 2 = 1 to 6 months post-discharge) were estimated using continuous symptom measures gathered at the 4 time points. The intercept P value indicates the extent to which the baseline value can be accurately estimated, whereas the slope P value indicates whether the change in the outcome is different than 0. In other words, a nonsignificant P value for a slope estimate indicates no significant change over time. The LGCM used full information maximum likelihood estimation to accommodate missing data. Full information maximum likelihood estimation does not rely on imputed data but maximizes the joint likelihood of missing data across all participants in the study. Full information maximum likelihood estimation handles missing data probabilistically without dropping participants who are missing data but have outcome data at a minimum of 1 time point. In smaller samples, full information maximum likelihood estimates have been shown to be more precise and less biased than multiple imputations.³⁷ Slope estimates are reported as the amount of change in the outcome per month. Standard structural equation modeling statistics were used to assess the LGCM model fit.³⁸ Furthermore, the symptom intercept and slope estimates were regressed on demographic characteristics and SDOH (age, sex, race, education, income, and marital status) to determine associations at hospitalization and changes in symptoms for slope 1 and slope 2. The regression models were adjusted for the number of days participants were hospitalized before completing the symptom survey to account for differences in the amount of time undergoing HF treatment during hospitalization.

Due to the sample size and 4 time points being used for piecewise LCGM, constraints were imposed on the LCGM models to facilitate model identification. For example, the slope 2 (1–6 months) variance was extremely small and nonsignificant in all models and was, therefore, constrained to 0. Additionally, outcome residual variances were constrained to equality. Post hoc power analysis using Monte Carlo simulations (Mplus, v11.2, Los Angeles, California), including missing data patterns, demonstrated the sample size (N = 49) had >80% power to identify moderate-to-large (>0.5 standard deviation; standard deviation = standard error × n) differences in symptoms for the piecewise LCGM models (shown in Table 3). The regression analyses examining differences in symptoms by SDOH and demographic characteristics (shown in Figures 1–3) had >80% power to detect large differences (>1 standard deviation).

TABLE 3.

Piecewise Growth Models of Heart Failure Symptoms and Heart Failure–Specific Health-Related Quality of Life Over 6 Months After Heart Failure Hospitalization Discharge

	Intercept (SE) Hospitalization	Intercept P Value	Slope 1 (SE) Hospitalization to 1 Month	Slope 1 P Value	Slope 2 (SE) 1–6 Months	Slope 2 P Value
Dyspnea	14.80 (1.18)	<.001	−7.93 (1.43)	<.001	−0.13 (0.16)	.431
Fatigue	62.91 (1.25)	<.001	−7.09 (1.97)	<.001	−0.02 (0.18)	.933
Sleep quality	10.67 (0.67)	<.001	−0.92 (0.75)	.222	−0.04 (0.14)	.785
Pain interference	54.69 (1.61)	<.001	−0.56 (1.67)	.737	0.35 (0.28)	.211
Anxiety	53.46 (1.44)	<.001	−2.93 (1.41)	.039	−0.71 (0.50)	.155
Depressive symptoms	6.40 (0.58)	<.001	0.12 (0.68)	.860	0.03 (0.12)	.823
HF-specific HRQOL	33.85 (2.82)	<.001	15.76 (3.51)	<.001	0.45 (0.67)	.502

Standard SEM statistics were used to assess model fit. Bolded text indicates P Value is <0.05.

All models resulted in χ² test > 0.05, CFI > 0.95, TLI > 0.95, RMSEA < 0.06, and SRMR ≤ 0.08.

Abbreviations: CFI, comparative fit index; HF, heart failure; HRQOL, health-related quality of life; RMSEA, root mean square error of approximation; SE, standard error; SEM, structural equation modeling;

SRMR, standardized root mean square residual; TLI,Tucker-Lewis Index.

FIGURE 1.

Differences in symptoms and HRQOL between female and male participants. iβ, s1β, and s2β are the regression coefficients showing the symptom intercepts (hospitalization, iβ) and slope (change over time, s1β and s2β) estimates for women compared with men. Coefficients and error bars display ±the standard error. A positive intercept coefficient indicates a worse symptom (better HRQOL) for women at hospitalization, with a negative intercept coefficient indicating a better symptom (worse HRQOL) for women at hospitalization. A positive slope coefficient indicates less symptom (more HRQOL) improvement for women compared with men. A negative slope coefficient indicates more symptom (less HRQOL) improvement than men. Significant associations are bolded. HF, heart failure; HRQOL, health-related quality of life.

FIGURE 3.

Differences in symptoms between participants with a college degree and no college degree. iβ, s1β, and s2β are the regression coefficients showing the symptom intercepts (hospitalization, iβ) and slope (change over time, s1β and s2β) estimates for those with a college degree compared with none. Coefficients and error bars display ±the standard error. A positive intercept indicates a worse symptom for those with a college degree at hospitalization, with a negative intercept indicating a better symptom for those with a college degree at hospitalization. A positive slope indicates less symptom improvement for a college degree compared with no degree. A negative slope indicates more symptom improvement for those with a college degree. Significant associations are bolded. HF, heart failure.

Dichotomized demographic and SDOH variables were created for sex, race, education, income, and marital status. Since no participants identified as other than male or female, the sex variable was categorized as men (male) or women (female). The self-identified race variable was grouped by those who identified as White or as Persons of Color, which included participants who described themselves as Black, American Indian, or Asian. The education variable was divided into those with a college degree (associate’s, bachelor’s, master’s, and PhD) and those without a degree. The income variable was separated into participants with less than $60 000 annual income and those with $60 000 or more annual income. The split at $60 000 was chosen as a reasonable separation between lower income and moderate-to-higher income while maintaining enough participants in each group for analysis. Marital/partner status was split by participants who were married or living with a partners or those who were not married or living with a partner. Since so few participants were employed (n = 5), employment status was not used in the SDOH analysis. A P value ≤.05 was considered significant with 2-tailed tests, and model estimates and coefficients are reported with standard error.

Results

The sample was comprised of 49 patients with NYHA class III HF. The average age was 68 ± 10 years, with over half of the participants being women (53%). The racial composition of the sample self-identified as White (70%), Black (22%), American Indian (6%), or Asian (2%). Most participants (57%) had an income of <$60 000/year, well below the median income ($115 000/year) for the urban area in which the study was conducted and 49% of the sample had college degrees. Participants who were married or living with a partner comprised 37% of the sample and only 10% of subjects were employed at enrollment (Table 1). The sample had a similar number of participants with reduced (42.9%) and preserved (44.9%) ejection fraction, and most subjects had a nonischemic etiology of HF (79.6%) (Table 2). The median length of the baseline hospitalization was 8 days (interquartile range: 6–14), with most baseline data collected 3 days (median) before discharge (interquartile range: 1–9).

TABLE 1.

Demographic Characteristics (N = 49)

	Mean ± SD/n (%)
Age	68.27 ± 10.14
Gender (female)	26 (53.1)
Race
American Indian/Asian	4 (8.1)
Black	11 (22.4)
White	34 (69.4)
Education
≤High school	8 (16.3)
Some college	22 (44.9)
Bachelor’s degree	10 (20.4)
Graduate degree	7 (14.3)
Marital status (married or living with partner)	18 (36.7)
Annual income
<$40 000	18 (36.7)
$40 000–59 999	10 (20.4)
$60 000–80 000	6 (12.2)
>$80 000	10 (20.4)
Employment (employed)	5 (10.2)

Abbreviation: SD, standard deviation.

TABLE 2.

Clinical Characteristics at Hospitalization (N = 49)

	Mean ± SD/n (%)
LVEF	43.98 ± 19.88
LVID	5.33 ± 1.02
Years with HF	3.56 ± 4.19
HF phenotype
EF < 40%	21 (42.9)
EF 40%–50%	5 (10.2)
EF > 50%	22 (44.9)
Nonischemic etiology	39 (79.6)
Diuretics	42 (87.5)
Beta blockers	31 (67.4)
ACEi/ARB/Sac-Nep	28 (58.3)
Aldosterone agonist	29 (59.2)
SGLTi	10 (20.4)
BMI	32.16 ± 10.01
Comorbidities	1.93 ± 1.07
ICD/pacemaker	16 (32.7)
Creatinine	1.56 ± 0.62
BUN	42.53 ± 21.75
HCT	36.78 ± 7.20
HGB	11.73 ± 2.47
Albumin	3.82 ± 0.43
Days hospitalized before data collection	4.73 ± 2.75

Abbreviations: ACEi, angiotensin converting enzyme inhibitor; ARB, angiotensin receptor blocker; BMI, body mass index; BUN, blood urea nitrogen; EF, ejection fraction; HCT, hematocrit; HF, heart failure; HGB, hemoglobin; ICD, implantable cardioverter defibrillator; LVEF, left ventricular ejection fraction; LVID, left ventricular internal diameter; Sac-Nep, sacubitril neprilysin inhibitor; SD, standard deviation; SGLTi, sodium glucose cotransporter two inhibitor.

The number of participants who completed the symptom survey included 48 (98%) at baseline, 32 (65%) at 1 month, 31 (63%) at 3 months, and 25 (51%) at 6 months. At baseline, 1 participant was discharged from their hospitalization before completing the survey, and 23 (47%) participants were not able to be contacted at 1 or more time points to complete the survey. Over the course of the study, 7 (14%) participants died, and 4 (8%) participants asked to stop participating in the study (3 after entering hospice and 1 after receiving a left ventricular assist device). Excluding those who died or withdrew from the study, no significant differences in age, sex, race, or other social factors were observed between those who completed all surveys and those who did not (P > .05).

Trajectories of Heart Failure Symptoms and Heart Failure–Specific Health-Related Quality of Life After Hospital Discharge

At hospitalization, the mean dyspnea score was 14.80 ± 1.18, and the mean fatigue score was rated at 62.91 ± 1.25. Both dyspnea and fatigue declined substantially by 1 month after discharge (Dyspnea slope 1 = −7.93 ± 1.43, P < .001; Fatigue slope 1 = −7.09 ± 1.97, P < .001), then did not change significantly to 6 months (Dyspnea slope 2 = −0.13 ± 0.16; P = .431; Fatigue slope 2 = −0.02 ± 0.18, P = .933) (Table 3). Mean anxiety at hospitalization was 53.46 ± 1.44 and declined from hospitalization to 1 month postdischarge (slope 1 = −2.93 ± 1.41, P = .039), then remained stable from 1 month to 6 months (slope 2 = −0.71 ± 0.50, P = .155). At hospitalization, mean sleep quality was 10.67 ± 0.67, pain interference was 54.69 ± 1.61, and depressive symptoms were 6.40 ± 0.58. None of these symptoms changed significantly at any point from hospitalization through 6 months (all slope P values > .200) (Table 3). At hospitalization, HF-specific HRQOL was 33.85 ± 2.82, which improved significantly from hospitalization to 1 month post-discharge (slope 1 = 15.76 ± 3.51, P < .001) and remained stable to 6 months (slope 2 = 0.45 ± 0.67, P = .502) (Table 3). In summary, all the gains in symptom reduction and improvements in HRQOL were between hospitalization and 1 month after discharge.

Post hoc power analysis using Monte Carlo simulations demonstrated that most models shown in Table 3 had adequate power (>80%) to detect that the intercept (hospitalization value) was different than 0 and that slope 1 (hospitalization to 1 month) was different than 0, indicating a change over time. Anxiety was the only model that was not adequately powered (power = 58%) to detect the differences for slope 1. For slope 2 (1–6 months), there were no differences in symptoms over time (Table 3).

Social Determinants of Health, Demographic Characteristics, and Symptom Trajectories

Several SDOH and demographic characteristics showed significant associations with HF symptoms at hospitalization and with changes in symptoms over time. Self-identified sex was associated with differences in symptoms and HF-specific HRQOL (Figure 1). Those identifying as women reported significantly worse dyspnea (β = 5.87 ± 1.97, P = .002), fatigue (β = 6.34 ± 2.23, P = .005), sleep quality (β = 2.92 ± 1.28, P = .023), anxiety (β = 5.68 ± 2.82, P = .044), and depressive symptoms (β = 2.51 ± 1.05, P = .017) at hospitalization (Figure 1). In addition, women reported significantly worse HRQOL at hospitalization than men (β = −14.25 ± 5.12, P = .005). Women participants reported more improvement in sleep quality (β = −2.46 ± 1.2, P = .042) than men from hospitalization to 1 month postdischarge (Figure 1). Changes in all symptoms and HRQOL were similar between women and men between 1 month and 6 months after discharge.

Identifying as a Person of Color was associated with differences in symptom trajectories compared with those who identified as White (Figure 2). At hospitalization, Persons of Color experienced similar levels of symptoms and HRQOL as White participants. After discharge, however, Persons of Color had significantly more improvement in dyspnea (β = −6.15 ± 2.60, P = .018), fatigue (β = −9.30 ± 3.90, P = .017), sleep quality (β = −2.74 ± 1.17, P = .019), and anxiety (β = −9.35 ± 2.69, P = .001) at 1 month postdischarge. Although not significantly different at baseline, depressive symptoms in Persons of Color improved significantly more over 6 months (β = −0.54 ± 0.26, P = .036) than those who identified as White (Figure 2).

FIGURE 2.

Differences in symptoms and HRQOL between People of Color and White participants. iβ, s1β, and s2β are the regression coefficients showing the symptom intercepts (hospitalization, iβ) and slope (change over time, s1β and s2β) estimates for People of Color compared with White participants. Coefficients and error bars display ±the standard error. A positive intercept coefficient indicates a worse symptom (better HRQOL) for People of Color at hospitalization, with a negative intercept coefficient indicating a better symptom (worse HRQOL) for People of Color at hospitalization. A positive slope coefficient indicates less symptom (more HRQOL) improvement for People of Color compared with White participants. A negative slope coefficient indicates more symptom (less HRQOL) improvement than White participants. Significant associations are bolded. HF, heart failure; HRQOL, health-related quality of life.

Participants with a college degree also experienced differences in symptoms compared with those without a college degree (Figure 3). Sleep quality (β = −2.63 ± 1.28, P = .039) and pain interference (β = −8.47 ± 3.00, P = .005) were better at hospitalization for those with college degrees. Although both sleep quality and pain interference worsened from 1 to 6 months in participants with college degrees, the differences over time were not statistically significant compared with those with no degree (Figure 3). Although similar at hospitalization, anxiety (β = −2.02 ± 1.00, P = .043) and depressive symptoms (β = −0.49 ± 0.25, P = .044) improved significantly more from 1 to 6 months for participants without a college degree (Figure 3).

Only sleep quality and anxiety showed significant associations with income. Participants with higher income experienced similar levels of anxiety at hospitalization as those with lower income, but anxiety for higher income participants declined significantly more from 1 to 6 months (β = −1.98 ± 0.97, P = .042). Having a higher income, however, was associated with worsening sleep quality from 1 to 6 months postdischarge compared with having a lower income (β = 0.46 ± 0.22, P = .040). Older age was associated with more improvement in sleep quality 1–6 months postdischarge (slope 2 β = −0.03 ± 0.01; P = .032) than being younger, and marital/partner status was not significantly associated with symptoms at hospitalization or with changes in symptoms after discharge (Supplemental Digital Content 1, Table S1, https://links.lww.com/JCN/A372 and Supplemental Digital Content 2, Table S2, https://links.lww.com/JCN/A373).

Post hoc power analysis showed that most models were not powered >80% to detect differences in symptoms at hospitalization or in the change over time between the SDOH and demographic characteristics. There were 2 exceptions: the difference in anxiety by race and the difference in dyspnea by sex were adequately powered (>80%).

Discussion

In this longitudinal observational study of 49 patients with NYHA class III HF, we characterized distinct trajectories of common HF symptoms up to 6 months after discharge from an HFH. Additionally, we identified novel associations between SDOH and demographic characteristics and HF symptoms. This study provides new insights into how diverse patients experience common HF symptoms while hospitalized and the evolution of symptoms after discharge from a HF exacerbation in people with NYHA class III HF.

Three symptoms declined significantly after discharge from a hospitalization to treat congestion, namely dyspnea, fatigue, and anxiety. The decline in these symptoms occurred concurrently with a dramatic increase in HF-specific HRQOL. Dyspnea, fatigue, and anxiety are symptoms closely linked to cardiopulmonary congestion, and worsening dyspnea is a main reason patients with HF are admitted for hospitalization.^39,40 Since decongestion to lower filling pressures and reducing symptoms during hospitalization is a primary clinical goal,⁴¹ it is notable that symptoms of cardiopulmonary congestion also decline in the short term (up to 1 month postdischarge) and remain lower in the longer term after HFH (up to 6 months postdischarge). The rapid decline and stability over time for dyspnea, fatigue, and anxiety suggest that ongoing clinical outpatient management and patient HF self-care were likely addressing symptoms of congestion for most participants. Importantly, improvements in dyspnea, fatigue, and anxiety after HFH reflected concurrent improvements in HF-specific HRQOL, highlighting the importance of these symptoms as drivers of HRQOL in HF.⁴²

Other HF symptoms assessed in this study, including sleep quality, pain interference, and depressive symptoms, did not improve from hospitalization for HF to 6 months after discharge. For the sample as a whole, lack of improvement in these symptoms may reflect the relatively mild levels of symptoms throughout the study. While those with higher levels of anxiety (PROMIS-Anxiety > 60, n = 13) did have significant reductions in anxiety over time (t = 3.34, P = .016), participants with symptoms of major depression (PHQ8 > 9, n = 10) did not experience a significant change in depressive symptoms (t = −1.42; P = .228) suggesting that those most in need of assistance with depression were not having their symptoms effectively addressed after hospitalization. These observations showing a lack of improvement in symptoms less commonly related to congestion, such as pain, depression, and poor sleep, suggest that these symptoms are infrequently assessed by clinicians.^43,44 Although such symptoms are common in people with HF, they may be the result of multimorbidity (eg, chronic obstructive pulmonary disease, atrial fibrillation, obesity, and coronary artery disease) with less clear etiologies, reducing the frequency of assessment.^45,46

Research on interventions to address distressing symptoms in people with cancer has found that integrating regular reports of patient-reported symptoms into clinical care with provider guidance to address symptoms reduced symptom distress and increased patient–provider communication.^47–49 Systematic and repeated symptom assessment and clinician guidance for worsening HF symptoms may allow for earlier and more tailored symptom management. For those with escalating HF symptoms, additional interventions may include referral to palliative care, cardiac rehab, or mental health services. Current studies indicate models of care focusing on more collaborative approaches to address patient needs can be successful in reducing symptom burden in people with HF,^50,51 while allowing for more engagement and shared decision-making to improve patient trust and satisfaction with their care.⁵²

Social determinants of health, demographic characteristics, and HF symptomology are not often described together. Some SDOH are known to significantly contribute to health outcomes for patients with HF, particularly outcomes involving hospital admission and mortality.^1,53 Yet, the associations of SDOH and demographic characteristics and HF symptoms have been infrequently studied and mainly include the influence of age and sex on symptoms.^19,54 This study identified associations of HF symptoms with additional SDOH and demographic characteristics, including race, education, and income during hospital admission and, importantly, in the evolution of symptoms after discharge.

Participants identifying as women consistently reported higher symptom burden (worse dyspnea, fatigue, sleep quality, anxiety, depression, and HRQOL) during hospitalization and demonstrated more improvement in dyspnea and sleep quality over 6 months (although not at P < .05) compared with men. Most symptoms in women started out worse and stayed worse than in men throughout the study. These results are consistent with prior studies showing women often rate symptoms more severely compared with men.^25,55 Additionally, a recent cross-sectional study by Stutsman et al⁵⁶ showed that more fatigue was associated with higher peripheral catecholamine levels in women with HF but not in men with HF, providing insight into the biological underpinnings of sex differences in fatigue.

Addressing a prominent gap in the literature regarding the evolution of HF symptoms, this study advances the understanding of how symptoms may change over time after hospitalization in women and men with NYHA class III HF. Women demonstrated much more improvement in dyspnea up to 1 month postdischarge (10-point improvement) than men (5-point improvement). More relief from dyspnea for women while hospitalized has been previously reported.²⁶ Importantly, in the study by Meyer et al,²⁶ being a woman was an independent predictor of dyspnea relief, even though women had less diuretics administered. In contrast, this study had no significant differences between women and men in the proportion of patients receiving diuretics or HF medications and demonstrated longer-term changes in dyspnea after hospitalization. In summary, women in our study experienced more severe dyspnea during hospitalization, with dyspnea improving more than men over 6 months after discharge. Additional research is needed to elucidate the role of sex in dyspnea relief to provide tailored and effective care for both women and men.

Differences in symptom trajectories were identified by race. Persons of Color rated symptoms at hospitalization similar to those identifying as White. Often, symptoms improved significantly more for Persons of Color compared with White people, such that Persons of Color rated lower levels of symptoms and better HRQOL at 6 months (Figure 2). Differences in HF symptoms by race have been infrequently studied in HF, particularly with changes in symptoms over time. The results from this study provide novel evidence that symptoms may change differently over time depending on the self-identified race of participants. One explanation from this study as to why symptoms for Persons of Color improved more after HFH is a difference in the proportion of Persons of Color and White people with reduced versus preserved ejection fraction HF (HFpEF). More Persons of Color were diagnosed with HF with reduced ejection fraction (<40%; HFrEF) compared with White people (48% vs. 19%, respectively; P = .031). The observed difference in HFrEF between Persons of Color and White people is consistent with larger studies showing that Black individuals are at an increased risk for HFrEF.¹ Since guideline-directed medical therapy has been shown to be more effective in those with HFrEF,^3,57–59 the differences in symptom trajectories by race may reflect challenges in medical management to reduce symptom burden for patients with HFpEF.³ Furthermore, other studies suggest that differences in social support, access to care, and the built environment in different racial groups may also contribute to differences in HF outcomes after HFH.^60–63 These results demonstrate that symptoms may change differently in different racial groups and may illustrate limitations in the treatment of patients with HFpEF to reduce symptom burden. Further study is needed to understand the drivers of these differences to ensure all patients have effective and equitable care.

This study also noted variations in symptoms based on educational attainment (no college degree vs. college degree). Consistent with the literature in HF,^64,65 symptoms of pain interference and sleep quality were significantly worse during hospitalization for those without a college degree. Those with no college degree, however, reported greater declines in anxiety over 6 months, particularly from 1 to 6 months postdischarge. This observation somewhat contrasts with one study that demonstrated people with HF with less education may have more emotional distress, impacting functional limitations.⁶⁴ Our study highlights the need to examine symptom trajectories over time to more fully understand how social factors such as education influence symptom progression after admission for HF exacerbation.

Participants with higher income reported more improvement in anxiety after discharge, particularly over 1–6 months after leaving the hospital. These results are consistent with research showing that lower socioeconomic status is associated with higher levels of stress and anxiety in people with HF.⁶⁶ Yet, sleep quality worsened for those with higher income over the same period (1–6 months), suggesting anxiety may not be a strong driver of sleep quality in HF. In addition, older age was associated with improvements in sleep quality from 1 to 6 months, which contrasts with research showing worse sleep quality in older adults.⁶⁷ Most studies examining age and sleep quality in people with HF, however, were cross-sectional, providing little information about how sleep quality changes differently over time in older and younger patients with HF.^67,68

Marital status was not significantly associated with any symptoms at hospitalization or change in symptoms over time. Social support from spouses or partners is generally reported as beneficial to people with HF, with more social support associated with less depression and anxiety, better self-care, and physical function.^69,70 Although marital or partner status in this study was not significantly associated with any symptom at hospitalization or with changes in symptoms over time, most symptoms and HF-specific HRQOL were consistently better at hospitalization (although not at P < .05) and at each time point in those with spouses or those who were partnered (data not shown). Thus, the lack of significant association between symptoms and marital or partner support in this study may reflect the relatively small sample size for the marital/partner group (n = 18) and symptom heterogeneity common with HF.

Strengths and Limitations

Study strengths include the variety of physical and affective HF symptoms assessed, a longitudinal design, and data analysis using piecewise LCGM to examine multiple symptom trajectories after hospitalization. Full information maximum likelihood estimation was used to address missing data, and the sample included roughly an equal number of women and men. Limitations include an exploratory design with a relatively small sample size, limiting the ability to detect significant differences and changes over time. The results examining differences in symptoms by SDOH and demographic characteristics were mostly underpowered (Figures 1–3). Although this indicates a lower likelihood of significant differences in SDOH and demographic characteristics being present in the population, this study provides useful information about the magnitude and direction of those relationships that may generate new hypotheses and further research. Furthermore, the limited data in the literature regarding how HF symptoms change over time and the clinically meaningful differences in symptoms noted in this study provide a strong rationale for reporting underpowered results. Additionally, this exploratory study with a relatively small sample tested a large number of hypotheses without statistical adjustment, which increased the likelihood that some results were discovered by chance. Further larger studies should take care to adjust their analysis to reduce the risk of false discovery. The sample was primarily older white participants, limiting our understanding of symptoms in younger and Black, American Indian, and Asian populations. Finally, data collection during the COVID-19 pandemic may have influenced survey responses and difficulty with participant follow-up, contributing to missing data. For example, many participants transitioned to remote clinic visits during the pandemic, drastically reducing in-person data collection and responses to requests to complete surveys.

Conclusions

This prospective longitudinal study examining common HF symptoms from hospitalization through 6 months postdischarge in patients with NYHA class III described changes in several common HF symptoms and identified novel associations between symptoms, SDOH, and demographic characteristics. Dramatic reductions in symptoms related to cardiopulmonary congestion and concurrent improvements in HF-specific HRQOL were observed. Importantly, depressive symptoms for those with moderate depression did not improve over 6 months after hospital discharge, indicating a need for additional assessment and collaboration with mental health providers. The associations of SDOH and demographic characteristics with symptom trajectories add to our understanding of the complexity and heterogeneity of HF symptoms. There are many factors influencing symptom burden in HF beyond objective clinical markers of HF. These data, demonstrating that changes in symptoms over time may differ by social and demographic characteristics, suggest that collecting reliable and repeated symptom assessments after HFH may help providers better address symptom burden earlier and more equitably.

Supplementary Material

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s Web site (www.jcnjournal.com).

What’s New and Important.

• Symptoms of cardiopulmonary congestion (dyspnea, fatigue, and anxiety) improve by 1 month after hospitalization for heart failure.

• Sleep quality, pain, and depression did not change from hospitalization through 6 months after discharge.

• Novel associations between social determinants of health, demographic characteristics, and symptom trajectories were identified that may inform clinical heart failure management.