Study design and population

This cohort study utilized data from the NHANES, an ongoing, biannual survey designed to assess the health and nutritional status of the civilian, non-institutionalized U.S. population. NHANES employs a complex, stratified, multistage probability sampling design, selecting participants at the county, census tract, household, and individual levels across all 50 states and the District of Columbia. Sample weights are applied to account for oversampling, survey non-response, and post-stratification, ensuring nationally representative estimates. Following a structured home interview, participants underwent standardized physical examinations and laboratory assessments at mobile examination centers. All study protocols were approved by the National Center for Health Statistics Research Ethics Review Board, with written informed consent obtained from all participants. For this analysis, we pooled data from eight continuous NHANES cycles (2003–2018), focusing on adults aged 18 years or older with available data on CHF, SDoH, and mortality outcomes. We excluded: (1) individuals under 18 years (n = 32,549); (2) those with missing CHF information (n = 3,108); (3) participants with incomplete SDoH data (n = 4,210); and (4) individuals with missing covariate or mortality information (n = 1,508). The final analytical sample consisted of 38,937 participants (Fig. 1).

Fig. 1 Flowchart of study design. Full size image

Assessment of SDoH

SDoH were assessed using eight self-reported indicators across five domains, based on definitions from the Healthy People 2030 initiative and established literature13,14. These indicators included employment status, family poverty-income ratio, food security, education level, health insurance coverage, home ownership, and marital status (detailed definitions provided in Supplementary Table S1).Each SDoH indicator was categorized as favorable (0) or unfavorable (1) according to established thresholds. A cumulative unfavorable SDoH score was calculated by summing these dichotomized indicators (range: 0–8). Participants were classified into two groups based on their cumulative score: low burden (SDoH ≤ 2) or high burden (SDoH > 2), consistent with prior research15. To address potential selection bias from excluded participants, we performed supplementary analyses exploring the relationship between the cumulative SDoH burden and mortality.

Assessment of CHF

One NHANES 2003–2018 question assessed CHF manifesting in jugular venous distention, dyspnea, orthopnea, bendopnea, leg edema and elevated B-type natriuretic peptide (BNP) or N-terminal prohormone of B-type natriuretic peptide (NT-proBNP)1. CHF were assessed by ‘Has a doctor or other health professional ever told you that you had congestive heart failure?’ with ‘yes’, ‘no’, ‘refused’ and ‘don’t know’ responses, excluding ‘refused’ or ‘don’t know’.

Mortality outcomes

The primary outcome was all-cause mortality, while secondary outcome included cause-specific mortality from cardiovascular disease (CVD). Mortality data were obtained through linkage with the National Death Index, with follow-up extending until December 31, 2019. Causes of death were classified using the International Classification of Diseases, Tenth Revision (ICD-10) code: CVD mortality (I00–I09, I11, I13, I20–I51, I60–I69)16. Follow-up time was measured from the baseline examination date to either the recorded date of death or the last known survival date, whichever came first.

Covariates

Covariates included demographic characteristics (age, sex, race/ethnicity), lifestyle factors (smoking status, drinking status, physical activity [PA]), and clinical parameters (body mass index [BMI], hypertension, diabetes mellitus [DM], and hyperlipidemia). Race/ethnicity was categorized as non-Hispanic White, non-Hispanic Black, Hispanic, or other. Smoking status was defined as current (≥ 100 cigarettes in lifetime and currently smoking), former (≥ 100 cigarettes but quit), or never (< 100 cigarettes). Drinking status was classified as: never ( 1 to ≤ 2 drinks/day for women or > 2 to ≤ 3 for men), and heavy (> 2 drinks/day for women or > 3 for men). PA was quantified using metabolic equivalent task (MET)-minutes/week, calculated by multiplying activity duration by standard MET values based on reported frequency and duration. BMI was computed as weight (kg) divided by height squared (m²), measured during physical exams. Hypertension was defined by self-reported diagnosis, antihypertensive medication use, or blood pressure ≥ 140/90 mmHg. DM was identified by self-report, antidiabetic treatment, or meeting any diagnostic thresholds: fasting plasma glucose ≥ 7.0 mmol/L, 2-hour glucose ≥ 11.1 mmol/L, or HbA1c ≥ 6.5%. Hyperlipidemia was defined by self-report, lipid-lowering therapy, or abnormal lipid levels: total cholesterol ≥ 200 mg/dL, triglycerides ≥ 150 mg/dL, LDL ≥ 130 mg/dL, or HDL < 40 mg/dL (men) or < 50 mg/dL (women).

Statistical analyses

All statistical analyses adhered to NHANES analytic guidelines, accounting for sample weights, clustering, and stratification to ensure nationally representative estimates. The masked variance of the primary sampling unit (sdmvpsu) and pseudo-stratum masked variance (sdmvstra) were incorporated to adjust for the survey’s complex sampling design. Sampling weights for individual participants were calculated by dividing the 2-year mobile examination center weight by eight, reflecting the integration of data from eight NHANES cycles.

Continuous variables were summarized as weighted means with standard errors (SE), while categorical variables were described as counts with weighted percentages. Group differences were assessed using analysis of variance for continuous variables and the chi-square test for categorical variables. Hazard ratios (HRs) and 95% confidence intervals (CIs) for the individual and joint associations of CHF and SDoH with all-cause and cardiovascular mortality were estimated using Cox proportional hazards regression models. Model 1 was adjusted for age, sex, and race/ethnicity. Model 2 further adjusted for smoking, drinking, hypertension, diabetes, hyperlipidemia, physical activity, and BMI. Kaplan-Meier (KM) curves were used to visualize survival probabilities by SDoH burden and CHF status. Restricted cubic spline (RCS) analyses were performed to assess potential non-linear dose-response relationships between SDoH scores and mortality risk, with knots placed at the 10th, 50th, and 90th percentiles of the SDoH distribution. Participants were categorized into four groups according to CHF status (yes/no) and unfavorable SDoH burden (low/high), with the group without CHF and low SDoH burden serving as the reference. Stratified analyses were conducted across key demographic and clinical subgroups.

To maximize statistical power and minimize potential bias from missing covariate data, we performed multiple imputation using chained equations with five iterations, implemented via the mice package in R. For sensitivity analyses, we conducted a complete-case analysis by excluding individuals with any missing covariate data. Additionally, we performed a second sensitivity analysis excluding participants who died within the first two years of follow-up to reduce the risk of reverse causation.

All analyses were performed using R version 4.2.2, with statistical significance defined as P < 0.05.

Ethics statement

The authors take full responsibility for all aspects of the work, ensuring that any questions concerning the accuracy or integrity of any part of the work are properly investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013).