Quality of life post heart failure diagnosis: population-level trends in the U.S.

Abstract

Background

Individuals with heart failure (HF) experience various symptoms making both diagnosis and disease burden estimates challenging. While HF-specific patient-reported outcome measures (PROMs) are widely used, their focus on clinical cohorts limits their generalizability. Preference-based measures like the EQ-5D enable standardized health-related quality of life (HRQoL) comparisons across conditions, supporting resource allocation decisions. The CDC’s Healthy Days (HD) Survey—a simple two-question tool that can be mapped to the EQ-5D—offers a broader approach to tracking HRQoL but remains underutilized in HF populations.

Methods

Using a nationally representative U.S. sample, we mapped HD Survey responses to EQ-5D utility scores to compare HRQoL between individuals with and without HF and examined changes in HRQoL over time. We assessed whether HD-derived scores align with HF-specific utility measures to support population-level health monitoring.

Results

Individuals with HF report significantly more physically unhealthy days (8.46 vs. 3.42) and mentally unhealthy days (5.42 vs. 3.86) compared to those without HF. HF respondents are, on average, 20 years older than those without HF, consistent with HF’s prevalence in older adults. The likelihood of an HF diagnosis is similar for men and women but higher among non-Hispanic whites and blacks than Hispanics and other races. Those with HF are more likely to have health insurance. Adjusting for age, sex, race, and insurance, mean EQ-5D utility scores for individuals with and without HF are 0.785 (95% CI: 0.714–0.825) and 0.840 (95% CI: 0.827–0.851), respectively. Utility scores for HF patients remain significantly lower than those without HF up to 10 years post-diagnosis.

Conclusion

HF reduces HRQoL by 6.55%, surpassing the clinically significant threshold of a 1–2% decrement. These findings highlight the potential of the HD Survey to inform public health monitoring and underscore the need for tailored interventions to address HRQoL deficits in HF populations.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12955-025-02372-0.

Background

Heart failure (HF) has diverse and variable symptoms such as dyspnea, fatigue, edema, fluid accumulation, irregular heartbeats, cognitive impairment, and chest pain [1]. These symptoms complicate diagnosis and make assessing the disease burden challenging. Research on health-related quality of life (HRQoL) in HF patients and how this differs from that of the general population is limited and often outdated. Most studies focus on HF-specific patient-reported outcome measures (PROMs) collected from clinical cohorts which often do not fully generalize to the broader population with the disease (i.e., they skew younger, healthier, have more specialized care and have higher socio-economics statuses), restricting comparisons with other populations [2]. Furthermore, the research available is predominantly international, with minimal representation of U.S.-based populations [3–8].

The two most common assessments for heart failure are The Kansas City Cardiomyopathy Questionnaire (KCCQ) [9] and the Minnesota Living with Heart Failure Questionnaire (MLHFQ) [2, 10]. Both assessment tools have been designed to measure the effects of heart failure on patients’ daily lives and have been validated, showing strong sensitivity to changes in health status over time [9, 11] and predicting clinical outcomes such as hospitalizations and mortality [12–14]. While both instruments are valuable tools for assessing the impact of HF, both the KCCQ and the MLHFQ need algorithms that map the questionnaire responses to utility scores obtained from preference-based measures such as the EuroQol five-dimensional (EQ-5D) to produce utility scores suitable for broader comparisons and cost-effectiveness analyses [15, 16].

Preference-based measures are preferable to HF-specific PROMs because they provide a standardized way of assessing health-related quality of life across different health conditions, allowing for comparability between studies and facilitating decision-making in healthcare resource allocation. Mapping to the EQ-5D or other preference-based algorithms that generate utility scores allow disease-specific questionnaires that focus narrowly on aspects relevant to a specific condition to be eventually transformed into single utility values [17]. Preference-based measures linked to more general HRQoL questionnaires, like the EQ-5D, use information representative of the broader population to elicit preferences for different health states through valuation exercises, such as time trade-offs or standard gambling methods [18]. This approach produces values that reflect societal preferences regarding the desirability of different health outcomes and can be used to calculate quality-adjusted life years (QALYs), which are needed in cost-effectiveness analyses.

The Healthy Days (HD) Survey is an instrument developed by the Centers for Disease Control and Prevention (CDC) to assess the general population’s self-reported health-related quality of life, rather than HF-specific HRQoL. To our knowledge, the Healthy Days survey has not yet been used to assess HRQOL in patients with HF. The HD survey consists of two questions and can be found in CDC-sponsored surveys, such as the Behavioral Risk Factor Surveillance System (BRFSS) [19] and the National Health and Nutrition Examination Survey (NHANES) [20], which are openly available to the public. Participants are asked to report the number of days in the past month when they experienced poor physical and mental health (referred to as “unhealthy days”). Jia & Lubetkin (2008) proposed a mapping for this survey to obtain preference-based values for the EQ-5D utility score [21]. As such, the HD survey could be more widely used in public health research and surveillance to monitor population health and track trends over time. We use the HD instrument to compare HRQoL between people with and without heart failure and for tracking the change in HRQoL over time for people diagnosed with the condition in a US representative population.

Because the Healthy Days survey is considerably shorter and, therefore, easier to implement than disease specific questionnaires, we ask whether, at a population level, it can produce similar HRQoL values for people with and without HF by comparing our results to those in the literature using a common instrument (in our case the Eq. 5d mapped from the KCCQ) to elicit population-based preferences.

Data and methods

The HD questionnaire in the publicly available NHANES is available for the years 2001 through 2012, covering six two-year survey cohorts. The complete set of core HD questions is available in Appendix A. From a total of 61,951 individuals interviewed across these survey years, 32,766 (52.89%) responded to both the physically and mentally unhealthy days questions. The HD questionnaire was administered to respondents aged 12 years and older [22]. We restrict our sample to individuals older than 18 years to align with the focus on adult health outcomes. Detailed sample sizes by cohort and variables used in the analysis are provided in Additional File 1, Appendices B and C. According to U.S. Department of Health and Human Services regulations (45 CFR 46.102) and institutional guidelines, studies using publicly available datasets do not constitute ‘human subjects research’ and are therefore exempt from Institutional Review Board (IRB) review. Since this dataset is explicitly designed and released for public use, ethics approval was not required for this study.

NHANES employs a stratified multi-stage, unequal probability cluster sampling design to ensure that the sample is representative of the total civilian non-institutionalized population of the United States. Each participant is assigned a sample weight reflecting the number of people represented by that individual, accounting for survey non-response, over-sampling, post-stratification, and sampling error. These weights were incorporated into all analyses using the svyset and svy commands in Stata/SE 18 Stata (College Station, TX, USA: StataCorp LP).

Population preference-based HRQoL is a weighted measure on a scale from 0 to 1, where 0 represents death and 1 represents perfect health. We computed HRQoL using NHANES survey data and Jia and Lubetkin’s (2008) mapping to obtain preference-based values for the EQ-5D questionnaire index, based on respondents’ HD questions. This allow us to estimate average patient utility levels for persons with HF and without HF. Additional Files 1, Appendix D shows Jia and Lubetkin’s (2008) algorithm producing a single utility score that ranges from 0 to 1 where 1 represents perfect health and 0 represents death.

To create a baseline HRQOL, we define:

where physically unhealthy days (PUD) and mentally unhealthy days (MUD) were combined, capped at a maximum of 30 days. The resulting measure was then transformed using the Jia and Lubetkin algorithm to produce a single utility score ranging from 0 to 1.

Analytical approach

To examine the association between HF and HRQoL, we conducted two main analyses:

1. Longitudinal Impact of HF: We assessed HRQoL for individuals with HF at different time points (years since HF diagnosis) to evaluate disease progression and its impact on health status. Years since diagnosis were calculated by subtracting the reported age of HF diagnosis from the age at the time of the survey. We estimated mean EQ-5D utility scores for individuals with and without HF after controlling for the mean age at diagnosis and estimated annual average marginal EQ-5D utility score decrement following the year of diagnosis. By design, pre-HF utility scores are not available. To address this limitation, we compared the utility scores of individuals with HF from age 48 onward to those without HF at the same age.

2. Comparison of HF and Non-HF Groups: We compared HRQoL between individuals with and without HF, controlling for confounders, including age, sex, race, and insurance status.

We employed the post-double lasso (PDL) method to select covariates and estimate causal effects for robustness. In the first stage, separate lasso regressions were performed: one to identify predictors of HRQoL and another to identify confounders associated with HF. Variables selected in these lasso regressions were included in a second-stage ordinary least squares (OLS) regression to estimate the effect of HF on HRQoL while minimizing bias [23, 24].

We used t-tests to compare utility scores between HF and non-HF groups. Since six cohorts were combined, weighted averages were calculated to account for differences in sample size and population representation across cycles [25].

All statistical analyses were conducted using Stata/SE 18. The NHANES survey design and weights were incorporated into all analyses using the svy commands to ensure valid population estimates and variance calculations.

Our findings were compared against previously published results using the EQ-5D instrument. Literature comparisons were based on criteria such as methodology, time since diagnosis, and confounding adjustments to contextualize and validate our results within the broader body of evidence on HF and HRQoL.

Results

Table 1 presents the demographic and health-related characteristics of individuals with and without HF. Respondents with HF reported significantly more physically and mentally unhealthy days compared to those without HF. Specifically, the mean number of physically unhealthy days was 8.46 among those with HF, compared to 3.42 among those without HF. Similarly, the mean number of mentally unhealthy days was 5.42 for individuals with HF, compared to 3.86 for those without HF. Missing responses to the HD questionnaire were comparable between the two groups, with 14% missing in the HF group and 13% in the non-HF group.

Table 1.

Demographic and health characteristics of individuals with and without heart failure

Characteristics	HF (N = 1,142)	No HF (N = 31,879)
Unhealthy days*—Physical (mean, (SD))	8.46 (11.47)	3.42 (8.07)
Unhealthy days*—Mental (mean, (SD))	5.42 (9.45)	3.86 (7.88)
Missing**	159 (14%)	4,219 (13%)
Age (mean, (SD))	66.72 (12.86)	46.06 (18.45)
Age at diagnosis (mean, (SD))	57.71 (17.05)	N/A
Male (%)	50.80	47.95
Race (%)
White	73.87	69.64
Black	14.52	11.22
Hispanic	6.47	13.01
Other/non-available	5.12	6.12
Insurance (any, yes = 1) (%)	92.36	80.29
Private	44.39	64.03
Medicare	66.42	15.65
Medicaid	14.61	5.37

*coded as a number between 0 and 30. ** Not all survey participants respond to the HD questionnaire

Individuals with HF were significantly older than those without HF, with a mean age of 66.72 years compared to 46.06 years, reflecting the higher prevalence of HF among older adults. The average age at HF diagnosis was approximately 57.71 years.

The distribution of HF diagnoses was similar between men and women, with 50.80% of individuals with HF identifying as male compared to 47.95% in the non-HF group. However, racial and ethnic disparities were evident. Non-Hispanic White (73.87%) and Black (14.52%) individuals were more likely to report a diagnosis of HF compared to Hispanic (6.47%) and other racial/ethnic groups (5.12%).

Insurance coverage was significantly higher among individuals with HF, with 92.36% reporting any form of insurance compared to 80.29% among those without HF. Medicare and Medicaid coverage were notably more prevalent in the HF group (66.42% and 14.61%, respectively) compared to the non-HF group (15.65% and 5.37%, respectively). Conversely, private insurance was more common among individuals without HF (64.03%) than those with HF (44.39%).

These results highlight the greater physical and mental health burden, older age profile, and distinct demographic and insurance patterns among individuals with HF compared to those without.

Figure 1 shows the quality of life (HRQoL) scores and their annual rate of change for individuals with and without HF, starting at a baseline age of 48 years. At age 48, individuals with HF have a significantly lower baseline HRQoL score (0.74; 95% CI: 0.67–0.80) compared to those without HF (0.853; 95% CI: 0.849–0.856). For individuals with HF, HRQoL declines annually at an estimated rate of -0.0056 (95% CI: -0.0071 to 0.0001). This indicates a slightly faster decline in HRQoL over time, though the confidence interval includes zero, suggesting large variations across patients. For individuals without HF, HRQoL is more consistent, and confidence intervals are narrower. The results show that quality of life is significantly and substantially lower for individuals with HF compared to their non-HF counterparts, and these differences persist over time following diagnosis.

Fig. 1.

QoL and Annual Rate of Change for individuals with and without HF starting at age = 48

Table 2 presents the coefficients and 95% confidence intervals (CIs) from two regression models comparing utility scores between individuals with and without heart failure (HF). Model 1 excludes covariates, while Model 2 includes adjustments for sex, age, race, and insurance status. Both models indicate statistically significant differences in HRQoL between individuals with and without HF. Model 1 estimates a HRQoL reduction of -0.107 (95% CI: -0.123, -0.091) for individuals with HF compared to those without. After adjusting for covariates, Model 2 shows that approximately half of this decrement (-0.055; 95% CI: -0.071, -0.040) can be attributed to differences in demographic and socioeconomic factors. Among covariates, being male or privately insured are associated with higher utility, while Medicaid coverage and increasing age are associated with lower utility. Race/ethnicity shows only modest associations.

Table 2.

Regression analysis (HF vs. No HF) with (model 2) and without covariates (model 1)

Dependent variable: utility scores	Coefficient (model 1)	95% CI Lower; Upper	Coefficient (model 2)	95% CI Lower; Upper
HF(= 1)	-0.107	-0.123; -0.091	-0.055	-0.071; -0.040
Current Age-48			-0.002	-0.002; -0.001
Male(= 1)			0.032	0.028; 0.037
Hispanic			0.000	-0.011; 0.012
NH White			-0.012	-0.023; -0.001
NH Black			-0.007	-0.018; 0.005
Private insurance			0.025	0.019; 0.031
Medicare			-0.017	-0.026; -0.007
Medicaid			-0.053	-0.064; -0.042
constant	0.853	0.849; 0.856	0.840	0.827; 0.851

Comparison group for insurance status: other insurance or no insurance. Comparison group for race: Other race. Comparison group for male: female

We put into perspective our results by comparing the decrement found here with that of other studies. Based on the most recent meta-analyses published at the time of writing [26] only two studies are US based, all other studies represent samples from Germany, Japan, Scandinavia and Canada. In the US studies the decrement in utility from having HF compared to not having HF is -0.064 and − 0.042 [27, 28], respectively. Our estimate of -0.055 aligns closely with these findings. The decrement in quality of life over time among patients diagnosed with HF appears to be considerably larger than those found in this study, albeit the follow up in these studies is less than one year [3, 29].

Discussion

The Healthy Days survey provides a unique opportunity to derive preference-based HRQoL measures from just two simple questions, offering a practical tool for evaluating population health. To our knowledge, this survey has not been previously used to compare HRQoL between individuals with and without (HF or to track changes in HRQoL among HF populations over time. By mapping these data to preference-based indicators, we gain insights into utilities that, when combined with life expectancy, can generate QALYs. QALYs are essential for translating health gains into standardized metrics, facilitating comparisons across conditions and enabling the calculation of incremental cost-effectiveness ratios (ICERs) at optimal thresholds.

Nationally representative surveys like NHANES are particularly valuable for tracking HRQoL in diverse populations, as they provide robust and generalizable data. Unfortunately, the Healthy Days survey is no longer administered, leaving a critical gap in the ability to track HRQoL at a national level with a tool that can be easily mapped to utility scores. Given its simplicity and utility, our findings indicate the need for the reintroduction of the HD survey in national health datasets to monitor and address the burden of chronic diseases like HF.

It is important to note that statistical significance does not necessarily equate to clinical significance. On the 0–1 utility scale, decrements exceeding 10% are often associated with substantial changes in patient-reported outcomes, potentially necessitating adjustments to treatment protocols [30]. The literature on minimum clinically important differences in HRQoL in patients with chronic disease indicates that these differences typically correspond to half a standard deviation [31], equivalent to a 1–2% deviation from the mean in most samples.

One limitation of this analysis is the inability to stratify HRQoL by HF severity, as individuals with varying degrees of disease progression are pooled together. Notably, we are unable to account for decrements during acute events that characterize HF (i.e. hospitalizations and the refractory period of HRQoL recovery after an acute decompensation). This aggregation could mask nuanced differences in HRQoL among subgroups. Additionally, response bias might skew the results if healthier individuals are more likely to participate in the survey. However, our findings indicate that non-response rates for the HF questionnaire are nearly identical between those with HF (14%) and those without (13%), suggesting that response bias may not be a significant concern (Table 1).

Conclusion

Our findings underscore the profound and ongoing impact of HF on health-related quality of life. People with HF not only exhibit significantly lower HRQoL compared to those without HF but also experience a faster rate of decline in HRQoL over time compared to those without HF. These results highlight the importance of addressing the long-term burden of HF and prioritizing interventions that improve HRQoL in affected populations. While the heterogeneity of HF patients, ranging from mildly symptomatic to critically ill, presents a limitation to direct application to specific subgroups, our study provides valuable insights at a population level. The analysis remains useful for understanding the steady state of HF HRQoL. The methods employed here, including the use of the Healthy Days survey, offer a robust framework for assessing HRQoL in potentially any population where such measures are collected. Reintroducing these straightforward but impactful survey questions into national datasets would be a critical step in capturing the societal burden of chronic conditions like HF. Such data would enhance our ability to monitor trends, inform targeted interventions, and ultimately improve the lived experience of individuals coping with HF at the population level.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

HF

Heart Failure

PROMs

Patient reported outcome measures

HRQoL

Health related quality of life

HD

Healthy Days

KCCQ

Kansas City Cardiomyopathy Questionnaire

MLHFQ

Minnesota Living with Heart Failure Questionnaire

EQ-5D

EuroQol five-dimensional

CDC

Centers for Disease Control and Prevention

BRFSS

Behavioral Risk Factor Surveillance System

NHANES

National Health and Nutrition Examination Survey

PUD

Physically unhealthy days

MUD

Mentally unhealthy days

PDL

Post-double lasso

OLS

Ordinary least squares

CIs

Confidence Intervals

QALYs

Quality-adjusted life years

ICERs

Incremental cost-effectiveness ratios

Author contributions

Full access to all of the study’s data and is responsible for its integrity and accuracy: M.A.Concept and designs: M.A., S.D., S.S., P.K., K.A., Z.Z., W.WAcquisition and analysis of data: M.A.Drafting of the manuscript: M.A.Read, edited and approved the final manuscript: M.A., S.D., S.S., P.K., K.A., Z.Z., W.W.

Funding

This work was supported in part by Lexicon Pharmaceuticals.

Data availability

The datasets used and/or analyzed in the study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.