Abstract 14859: Potassium Levels and Mortality in Patients With Newly Diagnosed Heart Failure: A Nationwide Study if US Veterans
Background: Patients with heart failure (HF) are at high risk for hypo- and hyperkalemia. However, their prognostic values in this clinical population have been only investigated in small samples, selected patients in trials, or in the context of specific drugs (e.g., hypokalemia under loop diuretics).
Methods: From a nationwide cohort of 3,504,732 U.S. veterans with preserved kidney function between 2004-06, we identified 224,858 patients with a hospitalization or two outpatient encounters with ICD code 428 for HF during follow-up through 2013. Of these, 46,206 cases with HF diagnosis in the first year from cohort entry were considered prevalent cases and were excluded. Of the remaining 178,652 patients, we could define baseline potassium levels (average over 6 months prior to HF diagnosis) and link to all-cause mortality in 142,087 patients.
Results: Hypokalemia (<4 mEq/L) was observed in 24.5% whereas hyperkalemia (≥5 mEq/L) was seen in 8.4%. For hypokalemia, lower cumulative survival compared to normal levels (4-<5 mEq/L) was found only below 3.5 mEq/L (prevalence of 4.0%), particularly in the first 5-6 years (Figure). In contrast, potassium of 5-<5.5 (prevalence 6.9%) and ≥5.5 mEq/L (1.5%) showed lower survival than the normal range. After accounting for demographic and clinical covariates (e.g., diabetes, kidney function, history of other cardiovascular diseases, and medications), potassium <3.5 and ≥5.5 mEq/L conferred 29% (95% CI 24%-35%) and 15% (7%-23%) greater risk of mortality, respectively, compared to 4-<5 mEq/L. Similar results were seen when we analyzed 3-month averages of potassium.
Conclusion: Historically, hypokalemia has received more attention compared to hyperkalemia among HF patients, but in our nationwide sample, both hypo- and hyperkalemia at HF diagnosis were associated with poor prognosis. Whether subsequent patterns of potassium levels and HF treatment modify these associations would require investigation.
Author Disclosures: K. Matsushita: None. Y. Sang: None. S.H. Ballew: None. M.E. Grams: None. J. Coresh: None. M.Z. Molnar: None.
- © 2016 by American Heart Association, Inc.