Social determinants of health outperform classical clinical association for long term mortality after decompensated heart failure
G Dutra, B Ferraz de Oliveira Gomes, D de Vasconcellos Dias, F Cardoso de Oliveira, J L Fernandes Petriz, P Resende do Carmo Junior, E Matos Nascimento, B de Braganca Pereira, G M Moraes de OliveiraAbstract
Background
Decompensated heart failure (DHF) carries high short-term lethality, but the long-term prognostic impact of social determinants of health (SDH) in patients treated with standardized protocols in private tertiary hospitals remains poorly characterized, particularly in highly unequal Brazilian urban settings.
Purpose
To quantify the association of SDH with in-hospital and long-term all-cause mortality after DHF hospitalization, and compare their prognostic weight against conventional clinical variables using advanced survival modeling.
Methods
Retrospective analysis of a prospective cohort of 1.023 consecutive patients admitted with DHF (2011–2021) at two private tertiary centers serving contrasting socioeconomic profiles. Patients received standardized HF protocols. Primary outcome was all-cause mortality (in-hospital + post-discharge) over median 6.5-year follow-up (IQR 3.8–9.1). We used Kaplan–Meier curves stratified by hospital unit (Figure 1), survival trees identifying high-risk phenotypes (Figure 2), and multivariable Cox regression with Elastic Net variable selection to compare SDH (hospital unit, ethnicity) versus clinical predictors.
Results
Mean age was 76.3±13.4 years; 53.9% men. In-hospital mortality: 10.8%; post-discharge: 64.1%. Kaplan–Meier curves showed markedly inferior survival at the hospital serving lower socioeconomic populations (Figure 1). In the final Cox model, admission to this hospital (HR 2.43, 95% CI 2.06–2.86), Brown/Black ethnicity (HR 1.44, 95% CI 1.12–1.86), older age (HR 1.02/year, 95% CI 1.02–1.03), prior diabetes (HR 1.18, 95% CI 1.01–1.38), and prior HF (HR 1.32, 95% CI 1.10–1.59) independently predicted mortality; higher LVEF and systolic dysfunction (vs. preserved) were protective. Survival trees identified high-risk phenotypes combining hospital unit, age >79 years, and prior arrhythmias (Figure 2), with SDH hazards exceeding classical predictors.
Conclusions
In this Brazilian urban DHF cohort under standardized care, SDH—particularly hospital unit reflecting intra-urban inequality and race/skin color—were the strongest long-term mortality predictors, outperforming traditional clinical variables. Integrating SDH into risk models could refine post-discharge stratification and promote equitable cardiovascular careFigure 1Figure 2