Elevated Admission Potassium Levels and 1-Year and 10-Year Mortality Among Patients With Heart Failure

Author’s Accepted Manuscript Elevated Admission Potassium Levels and 1-Year and 10-Year Mortality Among Patients with Heart Failure Anan Younis, Ilan Goldenberg, Ronen Goldkorn, Arwa Younis, Yael Peled, Boaz Tzur, Robert Klempfner www.elsevier.com

PII: DOI: Reference:

S0002-9629(17)30405-6 http://dx.doi.org/10.1016/j.amjms.2017.07.006 AMJMS508

To appear in: The American Journal of the Medical Sciences Received date: 7 April 2017 Revised date: 14 July 2017 Accepted date: 17 July 2017 Cite this article as: Anan Younis, Ilan Goldenberg, Ronen Goldkorn, Arwa Younis, Yael Peled, Boaz Tzur and Robert Klempfner, Elevated Admission Potassium Levels and 1-Year and 10-Year Mortality Among Patients with Heart F a i l u r e , The American Journal of the Medical Sciences, http://dx.doi.org/10.1016/j.amjms.2017.07.006 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Elevated admission potassium levels and 1-year and 10-year mortality among patients with heart failure Anan Younis MD¹, Ilan Goldenberg MD1,2,3, Ronen Goldkorn MD1,Arwa Younis MD1, Yael Peled MD1, Boaz Tzur MD1, Robert Klempfner MD1,2 1

The Leviev Heart Center, Sheba Medical Center, Tel Hashomer, Sheba Road 2, 52620 Ramat

Gan, Israel. 2

Sakler School of Medicine, Tel Aviv University, Ramat Gan, Israel.

3

Heart Research Follow-up Program, University of Rochester, Rochester, NY USA.

[email protected], [email protected], [email protected], [email protected] [email protected], [email protected] [email protected]

Corresponding author: Anan Younis, MD. Address: Heart Center, Sheba Medical Center, Tel Hashomer,Sheba Road 2, Ramat Gan, Israel [email protected], Tel: +972544731936, Fax: +97235305905

1

Short Title: potassium and mortality among heart failure patients. Funding: This research received no grant from any funding agency in the public, commercial, or not-for-profit sectors. Acknowledgments: The study was possible by the combined efforts of the HFSIS Study Group and the Israeli Association for Cardiovascular Trials. The Authors declare no conflicts of interest. Keywords: Hyperkalemia, Heart Failure, all-cause mortality, short term and long term outcomes.

Abstract: Background: Limited, contradictory data exists regarding the effect of hyperkalemia on both short and long-term all-cause mortality among hospitalized patients with heart failure (HF). Methods: We analyzed 4,031 patients who were enrolled in the Heart Failure Survey in Israel. The study patients were grouped into 3 different potassium (K) categories. Multivariate analysis was used to determine the association of potassium levels as well as 1 and 10 year all-cause mortality. Results: A total of 3,349 patients (83%) had K<5mEq/L, whereas 461 patients (11%) had serum K≥5mEq/L but≤ 5.5mEq/L, and 221 patients (6%) had K>5.5mEq/L. Survival analysis showed 2

that 1-year mortality rates were significantly higher among patients with K> 5.5mEq/L (40%) and those with serum K≥5mEq/L but ≤ 5.5mEq/L (34%) compared to those with K<5mEq/L (27%);(all log- rank p <0.01). Similarly, 10-year mortality rates among those with K>5.5mEq/L were 92% whereas among those with serum K≥5mEq/L but ≤ 5.5mEq/L rates were 88% and in those with K<5mEq/L rates were 82%; (all log-rank p <0.001). Consistently, multivariate analysis showed that compared to patients with K<5mEq/L, patients with K>5.5mEq/L had an independently 51% and 31% higher mortality risk at 1 year and 10 years, respectively (1 year Hazard ratio 1.51; 95%CI 1.04-2.2; 10 years Hazard ratio 1.31; 95%CI 1.035-1.66), whereas patients with serum K≥5mEq/L but≤ 5.5mEq/L had comparable adjusted mortality risk to patients with K<5mEq/L at 1 and 10 years. Conclusions: Among hospitalized patients with HF, admission K >5.5mEq/L was independently associated with increased short and long-term mortality, whereas serum K≥5mEq/L but≤ 5.5mEq/L was not independently associated with worse outcomes.

Abbreviations: ACEI = Angiotensin Converting Enzyme Inhibitor; ARBs = Angiotensin II Receptor Blockers; BMI = Body Mass Index; COPD = Chronic Obstructive Pulmonary Disease; CVA/TIA = Cerebrovascular Accident/ Transient Ischemic Attack; DM = Diabetes Mellitus; eGFR = Estimated Glomerular Filtration rate; HF = Heart Failure; HR= Hazard Ratio; HFSIS = Heart Failure Survey In Israel; LVEF = Left Ventricular Ejection Fraction; NYHA = New York Heart

3

Association; MRAs = Mineralocortecoid Receptor Antagonists; RAASi = Renin Angiotensin Aldosterone System Inhibitors.

1. Introduction Potassium is an important factor in determining myocardial function[1]. Low as well as high levels have been associated with sudden cardiac death and fatal arrhythmias [2-5]. Heart failure (HF) activates the Renin Angiotensin Aldosterone system (RAAS), sympathetic nervous system and induces hypokalemia. Non-potassium-sparing diuretics aggravate hypokalemia and heighten neurohormonal activation [6-8]. Many studies have shown an increased risk of cardiac and all-cause mortality rates among patients receiving non-potassium sparing diuretics, mainly due to the effect of hypokalemia [9-12]. Angiotensin Converting Enzyme Inhibitors (ACEI)/ Angiotensin II receptor blockers (ARBs) mineralocorticoid receptor antagonists (MRAs), and angiotensin receptor neprilysin inhibitor (ARNI) improve prognosis [1320]. Moreover, their mortality benefit may be partly attributed to the increased serum potassium and ventricular arrhythmia reduction [13-15, 18, 20]. Data regarding the effect of hyperkalemia on both short and long term all-cause mortality outcomes among hospitalized patients with HF is both limited and contradictory. One study showed that potassium levels of 5 - 5.5mEq/L may be a useful biomarker of poor prognosis in

4

HF.[21] Another recent study showed that potassium levels >5mEq/L were found to be associated with increased 12-week all-cause mortality among HF patients [22]. Moreover, a recent study showed that even serum potassium levels higher than 4.6 mEq/L are associated with increased risk of mortality [23]. It is well established that the addition of MRAs to beta blockers and either ACEI or ARBs has been shown to improve clinical outcomes, including all-cause mortality. However, with the widespread use of these agents, concern has been raised regarding an increased incidence of life –threatening hyperkalemia [24]. In fact, a rising potassium level represents a frequent cause for renin–angiotensin-aldosterone system inhibitors (RAASi) dose reduction or even discontinuation, actions that may deprive patients of therapy shown to improve clinical outcomes[25, 26]. Accordingly, the aims of our study are: (1) To evaluate the short and long-term effect of both slightly and more markedly elevated serum potassium levels at admission on the 1-year and 10year all-cause mortality outcome among patients hospitalized with HF. (2) To explore the interaction between slightly and markedly elevated serum potassium levels and major comorbidities on the short-and long-term all-cause mortality outcome.

5

2. Materials and Methods 2.1. Study population The present study population was comprised of 4,089 patients enrolled in the Heart Failure Survey in Israel (HFSIS). The prospective-cohort HFSIS was conducted in March and April 2003 in all 25 public hospitals in Israel. The study included 93 of the 98 internal medicine, and 24 of the 25 cardiology departments in Israel at that time; its design and methods have been previously published in detail [27-29]. Included in the study were patients ≥50 years of age who were hospitalized with acute, new- onset HF or exacerbation of chronic HF. Data abstracted from the HFSIS files or the present study included patients’ baseline characteristics such as medical history, New York Heart association (NYHA) functional classification, echocardiography data, laboratory indices, pre-admission and discharge medications. Following exclusion of those patients with missing serum potassium values at admission, our final analysis included a total of 4,031 patients. Informed consent was obtained from each patient before entry into the database, and the study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a priori approval by the institutions human research committee. The protocol was approved by the Ethics Committee at each of the participating hospitals. We excluded subjects with potassium levels below 3 or above 6.

6

2.1.1. Heart Failure diagnosis The criteria used for diagnosing HF were symptoms of HF (at rest or during exercise) and objective evidence of cardiac dysfunction at rest. Acute HF was defined as a rapid onset or change in the signs or symptoms of HF, resulting in the need for urgent therapy. Acute HF may be either acute de novo HF or acute exacerbation of chronic HF. NYHA functional class was determined according to patients’ functional status and symptoms before index hospitalization. Left ventricular ejection fraction (LVEF), collected by chart review, was determined by echocardiography and classified as follows: normal ≥50%, mildly impaired 40– 49%, moderately impaired 30–39%, and severely impaired <30%.

2.1.2. Serum potassium groups definition and classification Based on the current recommendations of the optimal serum potassium levels in HF we divided the study population into 3 different groups according to the measured admission potassium levels. Thus, those with admission serum potassium levels K <5mEq/L were classified as having low-normal serum potassium levels (group with either normal levels or hypokalemia), whereas those with admission potassium levels of serum K≥5mEq/L but≤ 5.5mEq/L were classified as having borderline-high serum potassium levels (i.e. mild hyperkalemia). Patients with admission potassium levels of K >5.5mEq/L were classified as having hyperkalemia.

7

2.2. Primary end points of the study The primary endpoint of the present study was all-cause mortality at 1 year and 10 years of follow-up. Mortality data were obtained by matching the patient’s identification numbers with their vital status in the National Population Registry of Israel. It should be emphasized that each matched record was checked for correct identification by matching the study recorded date of birth, during enrollment with the date of birth stored at the national registry. The secondary endpoint was 30-day mortality.

2.3. Statistical analysis Continuous variables are expressed as means ± standard deviation (SD), and categorical data are summarized as percentages. The clinical characteristics of the patients at baseline by the different admission serum potassium groups were compared with the use of the 1-way ANOVA for continuous variables with post hoc analysis and Tamhan’s test as needed, and the chisquare, with Z test and Bonferroni correction test, for the categorical variables. Kaplan–Meier survival analysis was used to graphically present survival estimates according to the different admission serum potassium groups and the subsequent 1-year and 10-year long-term survival probability. Cumulative event rates were compared using the log rank test.

8

Multivariate Cox proportional hazard regression modeling was used to assess the independent effect of potassium on the primary endpoint of all-cause mortality. The following covariates were introduced using the best subset method, following a univariate analysis of all relevant variables: age, gender, renal failure (eGFR< 60 ml/min/1.73m²), and diagnosis of diabetes mellitus (DM), severe HF (NYHA >2), history of ischemic heart disease (IHD), past cerebrovascular accident/ Transient ischemic attack (TIA/ CVA), chronic obstructive lung disease (COPD). Left ventricular ejection fraction (EF) <40% and Body Mass Index (BMI). Proportionality of hazard assumption was verified using Schoenfeld residuals and the log minus log (LML) method. In order to further explore the independent risk associated with hyperkalemia and normal-high serum potassium levels based on the different classifications of potassium levels in pre-specified patient subgroups (including age [≥ 65 years], sex, Diabetes Mellitus, renal dysfunction [eGFR<60ml/min/1.73m2], and New York Heart Association (NYHA) functional class >2, and discharged with prescription for digoxin or ACEI/ARBs. We performed interaction term analysis by the introduction of hyperkalemia-by-risk-subgroup interaction-term to the multivariate age adjusted Cox model. Interaction analysis is graphically presented in the form of a Forest plot. The results were further adjusted to age, gender, renal failure (eGFR<60ml/min/1.73m2), and diagnosis of diabetes mellitus (DM), severe heart failure (NYHA >2), history of ischemic heart disease (IHD), past cerebrovascular accident/ Transient ischemic attack (CVA/TIA), chronic obstructive lung disease (COPD), Left Ventricular ejection fraction (LVEF) <40% Body Mass Index (BMI), and discharge recommendations including: ,ACEI/ARBs, Digoxin and mineralocorticoid receptor antagonists. 9

Statistical significance was accepted for a 2-sided p < 0.05. The statistical analysis was performed with IBM SPSS version 20.0 (Chicago, IL, USA) and SAS version 9.2 (SAS institute Inc.).

3. Results Based on our classification of potassium levels, 3,349 patients (83%) had low-normal serum potassium levels, whereas 461patients (11%) had borderline-high serum potassium, and 221 patients (6%) had hyperkalemia. Baseline characteristics of the study patients according to the different potassium groups are presented in Table 1. Furthermore, preadmission and discharge medications are presented in Table 2.

3.1. Baseline clinical characteristics Most of the baseline demographic and past medical history characteristics were similar among the different potassium groups, except for the fact that patients with hyperkalemia and those with borderline-high serum potassium were more likely to have DM in comparison to those with low-normal serum potassium. In addition, those with hyperkalemia were more likely to have hypertension compared to those with low-normal serum potassium.

10

Patients with hyperkalemia had higher creatinine concentration compared to those with low to normal as well as those with borderline-high, serum potassium levels. Also, those with borderline-high potassium had higher creatinine concentration in comparison to those with low-normal potassium. However, the distribution of other laboratory indices was similar between the different potassium groups. Importantly, those with hyperkalemia as well as those with borderline-high potassium had higher rates of severe HF (NYHA >2) compared to those with low-normal potassium. Regarding HF, both those with hyperkalemia and those with borderline-high potassium were more likely to have diabetes compared to those with low-normal potassium. Furthermore, those with hyperkalemia were more likely to have hypertension in comparison with those with low-normal potassium.

3.2. Admission and discharge medications The rates of most medications recommended at discharge were similar among the different potassium groups, except for lower ACEI/ARBs rates among those with hyperkalemia compared to those with low-normal potassium, but similar rates to those with borderline-high potassium (Table 2).

11

Moreover, patients with hyperkalemia as well as those with borderline-high potassium had higher rates of spironolactone use at admission compared with those with low-normal potassium, yet discharge rates were similar. Importantly, patients with low-normal potassium were significantly more likely to receive ACEI/ARBs, as well as spironolactone at discharge, compared to the admission rates (Table 2).

3.3. Short term, 1-year all-cause mortality according to the different potassium level groups Unadjusted 30-day survival demonstrated increased mortality rate among the hyperkalemia group (13%) compared to those with low- normal potassium (7%); (P=0.002; Table 1), whereas following adjustment to age, gender and other comorbidities, hyperkalemia was not associated with statistically significant increased mortality risk (HR = 1.57; 95% CI 0.73-3.35). Kaplan-Meier survival analysis showed that at 1 year of follow-up the rate of all-cause mortality was significantly higher among those with hyperkalemia (40%) and those with borderline-high potassium (34%) compared to those with low-normal potassium (27%), (all log rank p<0.01; Figure 1A). Adjusted for age, gender and other major comorbidities, hyperkalemia was associated with an approximately 51% independently increased 1-year mortality risk (HR = 1.51; 95%CI 1.04- 2.2;

12

Table 3). Whereas borderline-high potassium was associated with a non-significant risk of mortality (HR = 0.79; 95%CI 0.57- 1.09; Table 3). Other independent predictors of 1 year mortality were: NYHA >2, eGFR < 60 ml/min/1.73 m², Age, COPD, and LVEF < 40% (Table 3).

3.3.1. Subgroup analysis for the 1-year outcome We further explored the independent association of hyperkalemia and 1-year all-cause mortality in predefined subgroups. Thus, patients with hyperkalemia had consistently similar increased risk of all-cause mortality among the subgroups of risk, except a pronounced increased risk among those discharged with digoxin (HR= 3.96; 95%CI 1.78-8.83; P for interaction= 0.005; Figure 2A). 3.4. Long term, 10-year all-cause mortality according to the different potassium level groups Kaplan-Meier survival analysis showed that at 10 years of follow-up the rate of all-cause mortality was significantly higher among those with hyperkalemia (92%) and those with borderline-high potassium (88%) compared to those with low-normal potassium (82%; log rank p <0.001; Figure 1B). Moreover, Mortality rates from all-cause were significantly higher among those with hyperkalemia compared to those with borderline-high potassium (log rank p =0.016).

13

Multivariate analysis adjusted for age, gender and comorbidities indicated hyperkalemia was independently associated with 31% increased risk of all-cause mortality (HR= 1.31; 95%CI 1.0351.66); P-value= 0.025; Table 3), whereas borderline-high potassium was associated with nonsignificant greater risk of mortality (HR= 0.985; 95%CI 0.83-1.16; P value= 0.86, Table 3). Other independent predictors of 10-year all-cause mortality were: NYHA Class >2, LVEF< 40%, COPD, male gender, age, renal failure (eGFR< 60 ml/min /1.73 m²) and DM. Consistent results were obtained when subjects with admission K<3.5mEq/L were excluded.

3.4.1. Subgroup analysis for the 10-year outcome We further explored the independent association of hyperkalemia and 10-year all-cause mortality. Thus, patients with hyperkalemia had consistently similar increased risk of all-cause mortality among the subgroups of risk, except there was a more pronounced increased risk among those discharged with digoxin (HR= 3.28; 95% CI 1.93- 7.21; P for interaction <0.001; Figure 2B).

14

4. Discussion The primary results of our study are: (1) among hospitalized patients with HF, admission hyperkalemia is independently associated with approximately 51% and 31% increased risk of 1year and 10-year all-cause mortality, respectively; (2) borderline- high potassium level was not significantly associated with risk of mortality risk at 1 or 10 years of follow-up; (3) hyperkalemia is consistently associated with increased risk of mortality which is similar among most of the subgroups at 1- and 10-year of follow-up, respectively, with a more pronounced effect among those discharged with digoxin at both 1 and 10 years of follow-up. Previous studies concerning the association between potassium levels and mortality among cardiovascular patients had shown an increased risk of mortality at the extremes of potassium levels [22, 23, 30-32]. Studies focusing on heart failure patients consistently demonstrated an association of serum potassium levels lower than 4mEq/L with increased mortality risk [23, 30, 33], however, 1 review had suggested the lower cut off to be 4.5mEq/L [1]. Regarding the upper cut off, most studies found that serum potassium levels more than 5mEq/L were associated with significant increased risk of mortality[22, 23, 33], whereas, Krogager et al[23], showed that even levels of 4.6mEq/L or higher were associated with increased 90 days mortality risk. In contrast to these studies, our study found that borderline-high potassium levels were not associated with all-cause mortality over 10 years of follow-up. Our findings extend previous 15

data published in 2010, when a study by M.I. Ahmed et al. showed that potassium levels of 55.5mEq/L appeared to be relatively safe (only when propensity matching was performed for the normal potassium group), yet the median follow-up was under 3 years [21]. Moreover, our study is the first to examine the association between short and long term all-cause mortality and potassium at levels higher than 5.5mEq/L among hospitalized patients with HF. Hyperkalemia can be life threatening because of the associated risk for fatal arrhythmias and conduction system abnormalities [34, 35]. Severe hyperkalemia is a medical emergency that can lead to significant morbidity and mortality[36]. The independent association of admission hyperkalemia with short as well as long term mortality may be related to several factors: firstly in our study patients with hyperkalemia at admission had higher degrees of DM as well as renal dysfunction compared to those with low-normal potassium levels thus progression of theses chronic diseases increases the risk of hyperkalemia and the associated fatal arrhythmias[37, 38]. Secondly, patients with HF at advanced stages are more likely to develop moderate or severe chronic kidney dysfunction, thus contributing to the increased risk of development of life threatening hyperkalemia especially among those with baseline hyperkalemia [39]. Thirdly hyperkalemia itself among HF patients increases suppression of the RAAS thus reducing renal potassium excretion , thus increasing serum potassium levels and eventually leading to more risk of rhythm disturbances [40]. Finally, admission hyperkalemia is also a marker of greater burden of comorbidities thus contributing to elevated risk not necessarily mediated through directs effects of potassium. 16

An, important finding of our study is the increased risk of mortality among patients with hyperkalemia discharged with digoxin prescription. This finding may be related to the fact that digoxin increases the serum potassium levels by inhibiting the NA+ - K+ ATPase enzyme [41].Thus increasing the risk of life threatening arrhythmias. Moreover, hyperkalemic HF patients who develop either acute or chronic digoxin toxicity are at significant increased risk of mortality as showed by several studies [42-44]. Furthermore, hyperkalemia may potentiate the the suppressive effect of digoxin on the atrio-ventricular node [45]. Antagonism of the neurohormonal systems have driven the greatest pharmacological improvement in the natural history of patients with HF, especially those with reduced LVEF [13, 15, 18, 19]. However, these actions come with some costs and limitations. Both beta adrenergic receptor antagonists and RAASi may cause hyperkalemia. Thus with the widespread use of these agents especially in higher doses and in combination, concern has been raised regarding an increased incidence of life threatening hyperkalemia [24, 46-48]. As a result, a rising potassium level has led to RAASi dose reduction or even discontinuation, actions that may deprive patients of optimal therapy and dosage proven to improve outcomes[25, 26]. Our study results show that admission hyperkalemia is associated with worse outcomes. Based on data from previous studies, discontinuation or dose reduction of the RAASi especially spironolactone should be considered among this group of HF patients [24, 25].

17

Moreover, based on previous studies it may be reasonable to reconsider digoxin use in patients with hyperkalemia, unless the cause of hyperkalemia is being corrected, because of the substantial increased mortality risk if these patients develop digoxin toxicity [43, 44]. In our analysis borderline-high potassium levels were not associated with all-cause mortality. Data from previous large studies suggested that discontinuation of RAASi and beta adrenergic receptor blockers is not obligatory in these patients [25, 26, 49, 50]. However, because mild hyperkalemia is a harbinger of more severe hyperkalemia, these patients may quickly develop severe hyperkalemia especially in the presence of chronic kidney disease and DM [51-53], thus increased vigilance is important. HF patients with serum potassium levels between 5 and 5.5mEq/L still require long term serial and careful monitoring of electrolytes for early identification of progression to more severe hyperkalemia in which case it may be prudent to reduce the dose of the offending drug. Accordingly, Eplerenone, a selective aldosterone receptor inhibitor, in 25 to 50 mg dosages, has been shown to reduce mortality in post-acute myocardial infarction patients with systolic HF treated with standard therapy without causing severe hyperkalemia when serum potassium was periodically monitored [53]. Another important finding of our study is the fact that there was a significant increase in ACEI/ARBs and spironolactone prescription at discharge compared to admission rates among patients with low-normal potassium, while among the groups of elevated potassium there were only minor changes between admission and discharge prescription rates, thus possibly strengthening the concept that low-normal potassium encourages and facilitate more 18

prescription (and up-titration, not explored in this study) of medications that form the cornerstone of HF treatment.

4.1. Limitations The present study is an observational study restricted to Israeli adults, and as such it is unclear whether these findings could be generalized to other populations. We have only data on allcause mortality and we do not know whether the increased mortality among patients with hyperkalemia was related to cardiovascular causes. In addition, we do not have information about the long-term treatment of patients after discharge from hospital, which could affect their survival and hospitalizations. Finally, we had no data on serum potassium during followup, and additionally underestimation of true associations due to regression dilution is possible [54]. 5. Conclusions In hospitalized patients with HF, admission hyperkalemia is associated with significantly increased 1 and 10-year mortality risk. These results are consistent in all subgroups, with the tendency to further increased risk among patients discharged with digoxin. Conversely, borderline-high potassium was not associated with increased all-cause mortality risk.

19

References: [1] Macdonald JE, Struthers AD. What is the optimal serum potassium level in cardiovascular patients? Journal of the American College of Cardiology. 2004;43:155-61. [2] Bielecka-Dabrowa A, Mikhailidis DP, Jones L, et al. The meaning of hypokalemia in heart failure. International journal of cardiology. 2012;158:12-7. [3] Podrid PJ. Potassium and ventricular arrhythmias. The American journal of cardiology. 1990;65:33E44E; discussion 52E. [4] Bieleka-Dabrowa A RJ, Mikhailidis DP, Banach M. What is the risk of hyperkalemia in heart failure? Expert Opin Pharmacotherapy 2011; 12; 2329-2338. 2011. [5] Nordrehaug JE, Johannessen KA, von der Lippe G. Serum potassium concentration as a risk factor of ventricular arrhythmias early in acute myocardial infarction. Circulation. 1985;71:645-9. [6] Steiness E, Olesen KH. Cardiac arrhythmias induced by hypokalaemia and potassium loss during maintenance digoxin therapy. British heart journal. 1976;38:167-72. [7] Fitzpatrick MA, Nicholls MG, Ikram H, et al. Stability and inter-relationships of hormone, haemodynamic and electrolyte levels in heart failure in man. Clinical and experimental pharmacology & physiology. 1985;12:145-54. [8] Knight RK, Miall PA, Hawkins LA, et al. Relation of plasma aldosterone concentration to diuretic treatment in patients with severe heart disease. British heart journal. 1979;42:316-25. [9] Cooper HA, Dries DL, Davis CE, et al. Diuretics and risk of arrhythmic death in patients with left ventricular dysfunction. Circulation. 1999;100:1311-5. [10] Whelton PK, Watson AJ. Diuretic-induced hypokalemia and cardiac arrhythmias. The American journal of cardiology. 1986;58:5A-10A. [11] Ahmed A, Husain A, Love TE, et al. Heart failure, chronic diuretic use, and increase in mortality and hospitalization: an observational study using propensity score methods. European heart journal. 2006;27:1431-9. [12] Cohen JD, Neaton JD, Prineas RJ, et al. Diuretics, serum potassium and ventricular arrhythmias in the Multiple Risk Factor Intervention Trial. The American journal of cardiology. 1987;60:548-54. [13] Effect of ramipril on mortality and morbidity of survivors of acute myocardial infarction with clinical evidence of heart failure. The Acute Infarction Ramipril Efficacy (AIRE) Study Investigators. Lancet. 1993;342:821-8. [14] Effects of enalapril on mortality in severe congestive heart failure. Results of the Cooperative North Scandinavian Enalapril Survival Study (CONSENSUS). The CONSENSUS Trial Study Group. The New England journal of medicine. 1987;316:1429-35. [15] Pfeffer MA, Braunwald E, Moye LA, et al. Effect of captopril on mortality and morbidity in patients with left ventricular dysfunction after myocardial infarction. Results of the survival and ventricular enlargement trial. The SAVE Investigators. The New England journal of medicine. 1992;327:669-77. [16] Studies of left ventricular dysfunction (SOLVD)--rationale, design and methods: two trials that evaluate the effect of enalapril in patients with reduced ejection fraction. The American journal of cardiology. 1990;66:315-22. [17] Effect of enalapril on mortality and the development of heart failure in asymptomatic patients with reduced left ventricular ejection fractions. The SOLVD Investigattors. The New England journal of medicine. 1992;327:685-91.

20

[18] Pitt B, Remme W, Zannad F, et al. Eplerenone, a selective aldosterone blocker, in patients with left ventricular dysfunction after myocardial infarction. The New England journal of medicine. 2003;348:1309-21. [19] Pitt B, Zannad F, Remme WJ, et al. The effect of spironolactone on morbidity and mortality in patients with severe heart failure. Randomized Aldactone Evaluation Study Investigators. The New England journal of medicine. 1999;341:709-17. [20] McMurray JJ, Packer M, Desai AS, et al. Angiotensin-neprilysin inhibition versus enalapril in heart failure. The New England journal of medicine. 2014;371:993-1004. [21] Ahmed MI, Ekundayo OJ, Mujib M, et al. Mild hyperkalemia and outcomes in chronic heart failure: a propensity matched study. International journal of cardiology. 2010;144:383-8. [22] Velavan P, Khan NK, Goode K, et al. Predictors of short term mortality in heart failure - insights from the Euro Heart Failure survey. International journal of cardiology. 2010;138:63-9. [23] Krogager ML, Eggers-Kaas L, Aasbjerg K, et al. Short-term mortality risk of serum potassium levels in acute heart failure following myocardial infarction. European heart journal Cardiovascular pharmacotherapy. 2015;1:245-51. [24] Juurlink DN, Mamdani MM, Lee DS, et al. Rates of hyperkalemia after publication of the Randomized Aldactone Evaluation Study. The New England journal of medicine. 2004;351:543-51. [25] Vardeny O, Claggett B, Anand I, et al. Incidence, predictors, and outcomes related to hypo- and hyperkalemia in patients with severe heart failure treated with a mineralocorticoid receptor antagonist. Circulation Heart failure. 2014;7:573-9. [26] Rossignol P, Dobre D, McMurray JJ, et al. Incidence, determinants, and prognostic significance of hyperkalemia and worsening renal function in patients with heart failure receiving the mineralocorticoid receptor antagonist eplerenone or placebo in addition to optimal medical therapy: results from the Eplerenone in Mild Patients Hospitalization and Survival Study in Heart Failure (EMPHASIS-HF). Circulation Heart failure. 2014;7:51-8. [27] Barsheshet A, Shotan A, Cohen E, et al. Predictors of long-term (4-year) mortality in elderly and young patients with acute heart failure. European journal of heart failure. 2010;12:833-40. [28] Garty M, Shotan A, Gottlieb S, et al. The management, early and one year outcome in hospitalized patients with heart failure: a national Heart Failure Survey in Israel--HFSIS 2003. The Israel Medical Association journal : IMAJ. 2007;9:227-33. [29] Lewis BS, Shotan A, Gottlieb S, et al. Late mortality and determinants in patients with heart failure and preserved systolic left ventricular function: the Israel Nationwide Heart Failure Survey. The Israel Medical Association journal : IMAJ. 2007;9:234-8. [30] Ahmed A, Zannad F, Love TE, et al. A propensity-matched study of the association of low serum potassium levels and mortality in chronic heart failure. European heart journal. 2007;28:1334-43. [31] Choi JS, Kim YA, Kim HY, et al. Relation of serum potassium level to long-term outcomes in patients with acute myocardial infarction. The American journal of cardiology. 2014;113:1285-90. [32] Patel RB, Tannenbaum S, Viana-Tejedor A, et al. Serum potassium levels, cardiac arrhythmias, and mortality following non-ST-elevation myocardial infarction or unstable angina: insights from MERLINTIMI 36. European heart journal Acute cardiovascular care. 2015. [33] Nolan J, Batin PD, Andrews R, et al. Prospective study of heart rate variability and mortality in chronic heart failure: results of the United Kingdom heart failure evaluation and assessment of risk trial (UK-heart). Circulation. 1998;98:1510-6. [34] An JN, Lee JP, Jeon HJ, et al. Severe hyperkalemia requiring hospitalization: predictors of mortality. Critical care. 2012;16:R225. [35] Spodick DH. Effects of severe hyperkalemia. The American heart hospital journal. 2008;6:68. 21

[36] Gennari FJ. Disorders of potassium homeostasis. Hypokalemia and hyperkalemia. Critical care clinics. 2002;18:273-88, vi. [37] Luo J, Brunelli SM, Jensen DE, et al. Association between Serum Potassium and Outcomes in Patients with Reduced Kidney Function. Clinical journal of the American Society of Nephrology : CJASN. 2016;11:90-100. [38] Jain N, Kotla S, Little BB, et al. Predictors of hyperkalemia and death in patients with cardiac and renal disease. The American journal of cardiology. 2012;109:1510-3. [39] Butler J, Givertz MM. Response to Sexton: Inhibiting the renin-angiotensin-aldosterone system in patients with heart failure and renal dysfunction: common sense or nonsense? Circulation Heart failure. 2014;7:537-40. [40] Dargie HJ. Interrelation of electrolytes and renin-angiotensin system in congestive heart failure. The American journal of cardiology. 1990;65:28E-32E; discussion 52E. [41] MG K. Cardiac Drug therapy. 6th ed. Philadelphia: W.B Saunders; 2003. [42] Bismuth C, Gaultier M, Conso F, et al. Hyperkalemia in acute digitalis poisoning: prognostic significance and therapeutic implications. Clinical toxicology. 1973;6:153-62. [43] Wenger TL, Butler VP, Jr., Haber E, et al. Treatment of 63 severely digitalis-toxic patients with digoxin-specific antibody fragments. Journal of the American College of Cardiology. 1985;5:118A-23A. [44] Manini AF, Nelson LS, Hoffman RS. Prognostic utility of serum potassium in chronic digoxin toxicity: a case-control study. American journal of cardiovascular drugs : drugs, devices, and other interventions. 2011;11:173-8. [45] Fisch C, Greenspan K, Edmands RE. Complete atrioventricular block due to potassium. Circulation research. 1966;19:373-7. [46] McMurray JJ, Ostergren J, Swedberg K, et al. Effects of candesartan in patients with chronic heart failure and reduced left-ventricular systolic function taking angiotensin-converting-enzyme inhibitors: the CHARM-Added trial. Lancet. 2003;362:767-71. [47] Nakao N, Yoshimura A, Morita H, et al. Combination treatment of angiotensin-II receptor blocker and angiotensin-converting-enzyme inhibitor in non-diabetic renal disease (COOPERATE): a randomised controlled trial. Lancet. 2003;361:117-24. [48] Cohn JN, Tognoni G, Valsartan Heart Failure Trial I. A randomized trial of the angiotensin-receptor blocker valsartan in chronic heart failure. The New England journal of medicine. 2001;345:1667-75. [49] Rossignol P, Menard J, Fay R, et al. Eplerenone survival benefits in heart failure patients postmyocardial infarction are independent from its diuretic and potassium-sparing effects. Insights from an EPHESUS (Eplerenone Post-Acute Myocardial Infarction Heart Failure Efficacy and Survival Study) substudy. Journal of the American College of Cardiology. 2011;58:1958-66. [50] Vardeny O, Wu DH, Desai A, et al. Influence of baseline and worsening renal function on efficacy of spironolactone in patients With severe heart failure: insights from RALES (Randomized Aldactone Evaluation Study). Journal of the American College of Cardiology. 2012;60:2082-9. [51] Desai AS, Swedberg K, McMurray JJ, et al. Incidence and predictors of hyperkalemia in patients with heart failure: an analysis of the CHARM Program. Journal of the American College of Cardiology. 2007;50:1959-66. [52] de Denus S, Tardif JC, White M, et al. Quantification of the risk and predictors of hyperkalemia in patients with left ventricular dysfunction: a retrospective analysis of the Studies of Left Ventricular Dysfunction (SOLVD) trials. American heart journal. 2006;152:705-12. [53] Pitt B, Bakris G, Ruilope LM, et al. Investigators E. Serum potassium and clinical outcomes in the Eplerenone Post-Acute Myocardial Infarction Heart Failure Efficacy and Survival Study (EPHESUS). Circulation. 2008;118:1643-50. 22

[54] Clarke R, Shipley M, Lewington S, et al. Underestimation of risk associations due to regression dilution in long-term follow-up of prospective studies. American journal of epidemiology. 1999;150:34153.

Figure 1A: Kaplan Meier 1year survival estimates according to the 3 potassium groups (Hyperkalemia = K>5.5mEq/L, borderline- high potassium= serum K≥5mEq/L but ≤ 5.5mEq/L, low-normal Potassium = k<5mEq/L). (Log rank P<0.01 comparing K>5.5mEq/L or serum K≥5mEq/L but ≤ 5.5mEq/L to k<5mEq/L, Log rank P=0.1 comparing K>5.5mEq/L to serum K≥5mEq/L but ≤ 5.5mEq/L).

Figure 1B: Kaplan Meier 10 year survival estimates according to the 3 potassium groups (Hyperkalemia = K>5.5mEq/L, borderline- high potassium= K≥5mEq/L but≤ 5.5mEq/L, low-normal Potassium = k<5mEq/L). (Log rank P-value<0.001, comparing K>5.5mEq/L or serum K≥5mEq/L but≤ 5.5mEq/L to k<5mEq/L, Log rank P=0.016 comparing K>5.5mEq/L to serum K≥5mEq/L but≤ 5.5mEq/L).

Figure 2A: Adjusted Hazard ratios of mortality for K>5.5mEq/L vs. k<5mEq/L in selected subgroups of patients for the 1year mortality outcome. Figure 2B: Adjusted Hazard ratios of mortality for K>5.5mEq/L vs. k<5mEq/L in selected subgroups of patients for the 10 year mortality outcome.

23

* model is further adjusted to: Age, Sex, ischemic heart disease (IHD), chronic obstructive pulmonary disease (COPD), New York Heart Association( NYHA)>2, Body Mass Index (BMI), renal failure (eGFR< 60 ml/min/1.73m²) , and diagnosis of diabetes mellitus (DM), severe heart failure (NYHA >2), past cerebrovascular accident/ Transient ischemic attack (TIA/ CVA), Left ventricular ejection fraction (EF) <40% .and discharge recommendations including: ,ACEI/ARBs, Digoxin and mineralocorticoid receptor antagonists.

Table 1: baseline characteristics of the patients based on the 3 different admission serum potassium groups.

K <5

K≥5mEq/L but≤ 5.5mEq/L

K >5.5

Demographics

N= 3349(83%)

N= 461(11%)

N= 221(6%)

Age^

73±12

73±12

74±11

0.764

Male

1900(57%)

272(59%)

122(55%)

0.569

BMI ≥30

468(24%)

71(28%)

30(27%)

0.471

HTN

2264(67%)

328(71%)

170(77%)*

0.007

DM

1334(40%)

236(51%)*

126(57%)*

<0.001

COPD

649(19%)

368(20%)

168(24%)

0.242

IHD

2578(78%)

358(80%)

173(80%)

0.696

Past MI

1266(38%)

195(42%)

88(40%)

0.161

Past CVA/TIA

404(12%)

66(14%)

36(16%)

0.089

p-value @

Past medical history

24

Past PCI

508(15%)

78(17%)

40(18%)

0.346

Past CABG

515(15%)

72(16%)

42(19%)

0.355

Atrial fibrillation

1116(33%)

155(34%)

72(33%)

0.964

Current or past smoker

972(29%)

143(31%)

155(30%)

0.665

Chronic CHF

1367(42%)

192(43%)

71(36%) #

0.022

Acute-on-chronic

1284(40%)

198(44%)

109(50%)*

0.002

HFrEF

1777(53%)

235(51%)

105(48%)

0.217

NYHA class >2

1290(40%)

208(46%)*

107(49%)*

0.001

LVEF <40%

1191(51%)

178(57%)

87(60%)

0.019

Coronary artery disease

2354(70%)

344(75%)

161(73%)

0.128

Diabetes mellitus(DM)

734(22%)

129(28%)*

75(34%)*

<0.001

Hypertensive Heart Disease

1384(41%)

214(46%)

113(51%)*

0.003

Any cardiomyopathy

472(14%)

67(15%)

17(8%)*#

0.025

Any Valvular Heart Disease

824(25%)

110(24%)

52(24%)

0.891

Serum Creatinine mg/dl

1.4±0.9

1.8±1.22*

2.4±1.7*#

<0.001

HDL mg/dl

43±27

43±15

41±17

0.613

LDL mg/dl

110±48

107±42

94±42*#

0.009

Triglycerides mg/dl

134±84

138±78

144±96

0.321

Plasma glucose mg/dl

157±81

177±103*

197±140*

<0.001

Sodium mEq/L

139±31

137±6

135±11

0.126

30-day mortality

235(7.1%)

42(9.1%)

29(13.2%)*

0.002

Heart Failure features

Congestive heart failure etiology

Laboratory values

25

1-year mortality

895(27%)

155(34%)*

87(40%)*

<0.001

10-year mortality

2748(82%)

407(88%)*

203(92%)*

<0.001

^ Continuous variables are reported as mean ±standard deviation if normally distributed; otherwise, as median with 25th – 75th range. Categorical variables are reported as numbers (%). @ denotes for P-value of overall comparison. * denotes for P-value <0.05 comparing hyperkalemic patients (K>5.5mEql/L), or patients with borderline-high potassium levels (serum K≥5mEq/L but≤ 5.5mEq/L) to patients with low-normal potassium levels (K<5mEq/L). # denotes for P-value <0.05 comparing hyperkalemic patients (K>5.5mEq/L) to patients with borderline-high potassium levels (serum K≥5mEq/L but≤ 5.5mEq/L). BMI = body mass index; COPD = chronic obstructive pulmonary disease; CHF- Congestive Heart failure; CABG = coronary artery bypass grafting; CVA/TIA = cerebral vascular accident/ transient ischemic attack; DM = diabetes mellitus; HTN = hypertension; HFrEF = heart failure with reduced ejection fraction; HDL = high density lipoprotein; IHD = ischemic heart disease; LDL = low density lipoprotein; MI = myocardial infarction; NYHA = New York Heart Association; PCI = percutaneous coronary intervention.

26

Table 2: Pre-admission and Discharge Medications of hospitalized patients according to the different potassium groups. Preadmission Medications K<5

K≥5mEq/L but≤ 5.5mEq/L

K>5.5

N=3349(83%)

N=461(11%)

N=221(6%)

Aspirin/ Clopidogrel

1991(60.0%)

283(61%)

144(65%)

ACEI/ARBs

1936(58%) ^

277(60%)

BB

1604(48%)

CCB

Discharge Medications Pvalue#

K<5

K≥5mEq/L but≤ 5.5mEq/L

K>5.5

Pvalue#

N=3349(83%)

N=461(11%)

N=221(6%)

0.198

2234(70%)

295(68%)

141(71%)

0.815

123(56%)

0.504

2212(69%) ^

279(65%)

115(58%)*

0.001

253(55%)*

120(54%)

0.005

1888(59%)

264(61%)

129(65%)

0.231

879(26%)

114(25%)

71(32%)

0.109

795(25%)

112(26%)

63(32%)

0.105

Loop Diuretics

2117(63%)

330(72%)*

152(69%)

0.001

2405(75%)

342(79%)

152(76%)

0.165

Spironolactone

422(13%) ^

109(24%)*

44(20%)*

<0.001

639(20%) ^

97(23%)

36(18%)

0.346

Digoxin

448(13%)

67(15%)

36(16%)

0.403

476(15%)

57(13%)

27(14%)

0.593

Statins

1137(34%)

179(39%)

83(38%)

0.078

1292(40%)

193(45%)

81(41%)

0.218

Insulin

304(9%)

61(13%)

32(15%)

0.001

285(9%)

55(13%)*

28(14%)*

0.004

Amiodarone

323(10%)

51(11%)

22(10%)

0.630

349(11%)

57(13%)

22(11%)

0.357

# denotes for P-value of overall comparison. *denotes for P-value <0.05 comparing hyperkalemic patients (K>5.5mEq/L), or patients with borderline-high potassium levels (serum K≥5mEq/L but≤ 5.5mEq/L) to patients with low-normal potassium levels (K<5mEq/L). All P-values comparing hyperkalemic patients (K>5.5mEq/L) to patients with borderline-high potassium levels (serum K≥5mEq/L but≤ 5.5mEq/L) are >0.05.

27

^ P-value<0.05 comparing pre-admission and discharge medication prescription rates among patients with low-normal potassium levels at admission (K< 5mEq/L). ACEI = angiotensin converting enzyme inhibitor; ARBS =angiotensin II receptor blockers; BB = beta blockers; CCB = calcium channel blockers.

Table 3 combined: independent predictors of 1 and 10 years all-cause mortality outcomes. 1 year

HR (95%CI)

10 years

P-value

HR (95%CI)

P-value

K > 5.5

1.513;(1.040-2.200)

0.030

K>5.5

1.312;(1.035-1.663)

0.025

K≥5 but ≤5.5

0.786;(0.567-1.090)

0.150

K≥5 but ≤5.5

0.985;(0.834-1.163)

0.86

Age^

1.026;(1.016-1.037)

<0.001

Age^

1.044;(1.038-1.050)

<0.001

NYHA class >2

2.372;(1.913-2.941)

<0.001

NYHA class >2

1.777;(1.584-1.993)

<0.001

eGFR< 60 ml/min/ 1.73m²

1.654;(1.308-2.092)

<0.001

eGFR< 60 ml/min/1.73m²

1.657;(1.470-1.867)

<0.001

LVEF< 40%

1.333;(1.062-1.673)

0.013

LVEF< 40%

1.243;(1.105-1.399)

<0.001

COPD

1.344;(1.057-1.710)

0.016

COPD

1.271;(1.112-1.453)

<0.001

Male gender

0.969;(0.776-1.210)

0.782

Male gender

1.181;(1.048-1.330)

0.006

DM

1.114;(0.902-1.376)

0.316

DM

1.466;(1.308-1.642)

<0.001

28

^ denotes covariates introduced as continuous variables into the Cox model. $ hazard ratios compared to K<5mEq/L. COPD = chronic obstructive pulmonary disease ; CI = Confidence Interval; DM (Diabetes Mellitus); eGFR (estimated glomerular filtration rate); HR = Hazard Ratio; LVEF = (Left ventricular ejection fraction); NYHA = New York Heart association. Hazard ratios are further adjusted to: Sex, Age, renal failure (eGFR< 60 ml/min/ 1.73m²), chronic obstructive pulmonary disease (COPD), cerebrovascular accident (CVA), history of ischemic heart disease(IHD), NYHA Class >2, LVEF < 40%, DM (Diabetes Mellitus), BMI (Body Mass Index).

29