Metra, Rossignol, Bayés-Genís, Lupón, Gorriz, Bodí, Núñez, Sanchis, Navarro, Santas, Chorro, Zannad, Núñez, Carratalá, Husser, Mollar, Miñana
Background -The prognostic value of long-term potassium monitoring and dynamics in heart failure (HF) has not been characterized completely. We sought to determine the association between serum potassium values collected at follow-up with all-cause mortality in a prospective and consecutive cohort of patients discharged from a previous acute HF admission. Methods -Serum potassium was measured at every physician-patient encounter, including hospital admissions and ambulatory settings. The multivariable-adjusted association of serum potassium with mortality was assessed using comprehensive state-of-the-art regression methods that can accommodate time-dependent exposure modeling. Results -The study sample included 2164 patients with a total of 16,116 potassium observations. Mean potassium at discharge was 4.3±0.48 mEq/L. Hypokalemia (<3.5 mEq/L), normokalemia (3.5 to 5.0 mEq/L), and hyperkalemia (>5 mEq/L) were observed at the index admission in 77 (3.6%), 1965 (90.8%), and 122 (5.6%) patients, respectively. At a median follow-up of 2.8 years (range=0.03-12.8 years), 1090 patients died (50.4%). On a continuous scale, the multivariable-adjusted association of potassium values and mortality revealed a non-linear association (U-shaped) with higher risk at both ends of its distribution (omnibus p-value=0.001). Likewise, the adjusted hazard ratios (HRs) for hypokalemia and hyperkalemia - normokalemia as reference - were 2.35 (95% confidence interval [CI]:1.40-3.93; p=0.001) and 1.55 (95% CI:1.11-2.16; p=0.011), respectively (omnibus p-value=0.0003). Furthermore, dynamic changes in potassium were independently associated with substantial differences in mortality risk. Potassium normalization was independently associated with lower mortality risk (p=0.001). Conclusions -Either modeled continuously or categorically, serum potassium levels during long-term monitoring were independently associated with mortality in patients with HF. Likewise, persistence of abnormal potassium levels was linked to higher risk of death compared with patients who maintained or returned to normal values.
