Examining the "Repletion Reflex": The Association between Serum Potassium and Outcomes in Hospitalized Patients with Heart Failure.

Abstract

In patients hospitalized with heart failure (HF) exacerbations, physicians routinely supplement potassium to maintain levels ≥4.0 mEq/L. The evidence basis for this practice is relatively weak. We aimed to evaluate the association between serum potassium levels and outcomes in patients hospitalized with HF. We identified patients admitted with acute HF exacerbations to hospitals that contributed to an electronic health record-derived dataset. In a subset of patients with normal admission serum potassium (3.5-5.0 mEq/L), we averaged serum potassium values during a 72-hour exposure window and categorized as follows: <4.0 mEq/L (low normal), 4.0-4.5 mEq/L (medium normal), and >4.5 mEq/L (high normal). We created multivariable models examining associations between these categories and outcomes. We included 4,995 patients: 2,080 (41.6%), 2,326 (46.6%), and 589 (11.8%) in the <4.0, 4.0-4.5, and >4.5 mEq/L cohorts, respectively. After adjustment for demographics, comorbidities, and presenting severity, we observed no difference in outcomes between the low and medium normal groups. Compared to patients with levels <4.0 mEq/L, patients with a potassium level of >4.5 mEq/L had a longer length of stay (median of 0.6 days; 95% CI = 0.1 to 1.0) but did not have statistically significant increases in mortality (OR [odds ratio] = 1.51; 95% CI = 0.97 to 2.36) or transfers to the intensive care unit (OR = 1.78; 95% CI = 0.98 to 3.26). Inpatients with heart failure who had mean serum potassium levels of <4.0 showed similar outcomes to those with mean serum potassium values of 4.0-4.5. Compared with mean serum potassium level of <4.0, mean serum levels of >4.5 may be associated with increased risk of poor outcomes.