Prognostic Significance of Discharge Hyponatremia in Heart Failure Patients With Normal Admission Sodium (from the ESCAPE Trial)

Abstract

Hyponatremia in acute decompensated heart failure (HF) is indicative of a poor prognosis and predicts morbidity and mortality. We explored the predictive utility of hyponatremia at the time of hospital discharge among HF patients with normal admission sodium (Na). Characteristics and outcomes of HF patients enrolled in the Evaluation Study of Congestive Heart Failure and Pulmonary Artery Catheterization Effectiveness trial, who had normal Na on admission, were compared between those who were hyponatremic (Na <135 meq/L) or normonatremic on discharge. Three hundred six patients with normal admission Na had either hyponatremia (n = 86) or normal Na (n = 220) on discharge. Compared with patients with normal Na on discharge, hyponatremic patients were younger (p = 0.004), with lower discharge systolic (p <0.001) and diastolic (p = 0.004) blood pressure, higher discharge blood urea nitrogen (p = 0.011) despite similar creatinine (p = 0.566), lower ejection fraction (p = 0.007), and higher left ventricular end-diastolic (p = 0.028) and end-systolic (p = 0.007) dimensions. Despite comparable congestion on hospital admission, patients with discharge hyponatremia had a higher degree of decongestion throughout hospitalization evident in the significantly greater admission to discharge weight loss (p = 0.044) and admission to discharge reduction in inferior vena cava diameter (p = 0.014). Despite longer initial hospitalization (p = 0.004), total duration in hospital at 30 days (p = 0.004) and 6 months (p = 0.045), there were no significant differences between patients with discharge hyponatremia versus normal Na on discharge regarding rehospitalization (p = 0.386), all-cause mortality (p = 0.440), and composite of death, cardiac rehospitalization, and cardiac transplant (p = 0.799), all up to 6-month following randomization. Restricted cubic spline analysis also showed no significant relationships between discharge Na and the aforementioned 3 outcomes. Cox proportional hazards regressions showed that discharge hyponatremia did not significantly predict any of the 3 outcomes after adjustment for imbalances at baseline. Among patients with discharge hyponatremia, a poor outcome was more likely if they were also hyponatremic on admission: the composite end point occurred in 69.2% of those also hyponatremic on admission versus 51.2% of those with normal Na on admission but decreased Na on discharge (p = 0.045). Because the median discharge Na level in the discharge hyponatremia group was 132 meq/L, our findings suggest a benign nature of mild discharge hyponatremia in HF patients with normal admission Na.