Association between in-hospital changes in estimated plasma volume status and worsening renal function in patients with acute heart failure

Worsening renal function (WRF) during treatment for acute heart failure (AHF) has context-dependent clinical implications. Estimated plasma volume status (ePVS), calculated from hematocrit and body weight, may complement decongestion assessment. We evaluated the association between in-hospital change in ePVS (ΔePVS) and WRF. This single-center retrospective study included 309 adults hospitalized with AHF from 2021 to 2025. ΔePVS was defined as the last in-hospital ePVS minus admission ePVS; WRF was defined as an increase in serum creatinine of ≥ 26.5 µmol/L or ≥ 50%. Analyses included multivariable regression, bootstrap correction, and a 48-hour landmark analysis. Readmission analyses treated death as a competing event. WRF occurred in 92 patients (29.8%). ΔePVS was higher in patients with WRF than in those without WRF (1.0% vs. − 4.5%; P < 0.001). The adjusted odds ratio per 5% increase in ΔePVS was 1.54 (95% confidence interval [CI], 1.19–1.99). In sequential Firth models of 295 patients, the odds ratio was 1.39 (95% CI, 1.05–1.84) after adjustment for treatment intensity and diuretic response; the corresponding 48-hour landmark estimate was 1.36 (95% CI, 1.04–1.78). ΔePVS correlated modestly with changes in N-terminal pro-B-type natriuretic peptide (NT-proBNP) and clinical congestion signs and was driven mainly by hematocrit change. Adding ΔePVS increased the apparent area under the receiver operating characteristic curve (AUC) by 0.016 ( P = 0.098). The WRF/ΔePVS ≥ 0 phenotype had a greater 90-day readmission risk than the no-WRF/ΔePVS < 0 phenotype (subdistribution hazard ratio [sHR], 3.84; 95% CI, 1.91–7.72). A smaller decrease or an increase in ePVS was associated with WRF, less hemoconcentration, and a poorer early diuretic response. ΔePVS may complement multimodal decongestion assessment; however, its incremental discriminative and prognostic value requires external validation.