|Home | About | Journals | Submit | Contact Us | Français|
Hyperkalemia is common and potentially dangerous in hospitalized patients; its contemporary prevalence and prognostic importance following acute myocardial infarction are not well described.
In 38,689 consecutive acute myocardial infarction patients from the Cerner Health Facts database, we evaluated the association between maximum in-hospital potassium levels (max K) and in-hospital mortality. Patients were stratified by dialysis status, and grouped by max K as follows: <5 mEq/L, 5–<5.5 mEq/L, 5.5–<6.0 mEq/L, 6.0–<6.5 mEq/L, and ≥ 6.5 mEq/L. Multivariable logistic regression was used to adjust for multiple patient and site characteristics. The relationship between number of hyperkalemic values and in-hospital mortality was also evaluated.
Of 38,689 acute myocardial infarction patients, 886 were on dialysis. The rate of hyperkalemia (max K ≥ 5.0 mEq/L) was 22.6% in non-dialysis and 66.8% in dialysis patients. Moderate-severe hyperkalemia (max K ≥ 5.5 mEq/L) occurred in 9.8% of patients. There was a steep increase in mortality with higher max K levels. In-hospital mortality exceeded 15% once max K ≥5.5 mEq/L regardless of dialysis status. The relationship between higher max K and increased mortality risk persisted after multivariable adjustment. In addition, patients with greater number of hyperkalemic values (vs. a single value) experienced higher in-hospital mortality.
Hyperkalemia is common in patients hospitalized with acute myocardial infarction. Higher max K levels and number of hyperkalemic events are associated with a steep mortality increase; with higher risks for adverse outcomes observed even at mild levels of hyperkalemia. Whether more intensive management of hyperkalemia may improve outcomes in acute myocardial infarction patients merits further study.
Hyperkalemia is a common electrolyte abnormality that can lead to serious and potentially fatal cardiac dysrhythmias.1 Prior work has consistently supported the link between hyperkalemia and adverse cardiovascular outcomes.1–3 In the modern era, the prevalence of chronic kidney disease and diabetes, both associated with hyperkalemia, is rising.4,5 Additionally, there has been widespread adoption of practices that increase the risk of hyperkalemia in acute myocardial infarction survivors with or without incident heart failure. Examples of such therapies include guideline-directed use of beta-blockers5, mineralocorticoid receptor antagonists6, and renin angiotensin aldosterone system antagonists.7–9 The increase in the use of these medications has been associated with significant increases in hyperkalemia-related hospitalization and deaths.7,8,10 Further, procedures such as percutaneous intervention and coronary artery bypass grafting, are common in the contemporary management of acute myocardial infarction, and may indirectly contribute to the risk of hyperkalemia via the associated incidence of contrast-induced nephropathy11 and acute kidney injury12, respectively.
While prior studies have examined the U-shaped relationship between serum potassium levels and mortality following acute myocardial infarction1, neither the contemporary prevalence nor prognostic importance of various degrees of hyperkalemia have been described. In addition, it is unclear whether or not resolution of hyperkalemia results in improved outcomes. Further defining the risk associated with hyperkalemia in general, as well as with varying degrees of hyperkalemia, could help identify patients who may derive the greatest benefit from more aggressive hyperkalemia management. Accordingly, we analyzed data from Cerner Health Facts, a database of patients hospitalized with acute myocardial infarction in the United States between 2000–2008, to (1) examine the prevalence and prognosis associated with hyperkalemia in patients hospitalized for acute myocardial infarction, according to dialysis status; (2) describe mortality outcomes based on persistence versus resolution of hyperkalemia; and (3) evaluate mortality outcomes based on the number of recorded hyperkalemia events.
Details regarding the Health Facts® registry have been previously described.1,13 Briefly, Cerner Corporation’s Health Facts acute myocardial infarction database is a 67-center registry of 39,759 consecutive patients with acute myocardial infarction hospitalized between January 1, 2000 and December 31, 2008. Data collected included patients' demographic characteristics, medical history, comorbidities (using the International Classification of Diseases-Ninth Revision-Clinical Modification [ICD-9-CM], codes), laboratory studies, medications, procedures, and complications.
We queried the Health Facts database using ICD-9 codes. Documentation of acute myocardial infarction required a primary discharge diagnosis of acute myocardial infarction, using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnostic codes 410.xx, along with positive cardiac biomarkers. For this analysis, we included 38,689 patients with biomarker-confirmed acute myocardial infarction who also had at least 1 recorded potassium value (Figure 1). We included patients who had only whole blood potassium measurements recorded as we used serum potassium values for this analysis. Patients excluded from analysis were those admitted from hospice or transferred from a critical access hospital and those with a length of stay over 31 days. Data in the Health Facts database were obtained from patients' electronic medical records and included demographic (age, sex, and race) and clinical information (medical history, in-hospital procedures documented by ICD-9-CM codes), comprehensive laboratory data (including all in-hospital potassium measurements), pharmacy data, in-hospital mortality and hospital characteristics. All data were de-identified before being provided to the investigators; thus this analysis was considered exempt from human subjects research review by the Saint Luke's Hospital Institutional Review Board.
Hyperkalemia was defined as at least one maximum in-hospital potassium level measurement equaling 5 mEq/L or greater. Moderate-severe hyperkalemia was defined as a maximum potassium level equal to or greater than 5.5 mEq/L.
The Health Facts database included all acute myocardial infarction patients' serum potassium levels and their time of measurement relative to hospital admission. The maximum serum potassium level was defined as the highest potassium level at any point during hospitalization. Our primary focus was the relationship between maximum in-hospital potassium levels and outcomes. All serum potassium values were measured and reported in mEq/L (1 mEq/L = 1 mmol/L).
The primary outcome for this analysis was in-hospital mortality stratified by dialysis status, as documented in the Health Facts database. In secondary analyses, we examined in-hospital mortality according to number of hyperkalemia values (1 vs. 2 vs. 3 or greater). We subsequently evaluated mortality based on whether or not potassium normalized following the highest measurement. We defined normalization as a mean potassium level of less than 5.0 mEq/L following the maximum in-hospital potassium measurement, while non-normalization was defined as a mean potassium level greater than or equal to 5 mEq/L following the maximum in-hospital potassium measurement.
Baseline demographics and clinical characteristics were compared among patients categorized by the maximum in-hospital serum potassium levels: less than 5.0, 5.0 to less than 5.5, 5.5 to less than 6.0, 6.0 to less than 6.5, 6.5 or greater mEq/L. Continuous characteristics were compared using a linear trend test while categorical variables were compared using the Mantel-Haenszel trend test. Hierarchical logistic regression was then used (with hospital site as a random effect to account for clustering across centers) to assess the independent association between maximum serum potassium levels and mortality, after adjustment for potential patient- and hospital-level confounders.
Patients were stratified by dialysis status, and grouped into categories of max K (<5 mEq/L [reference group], 5–<5.5 mEq/L, 5.5–<6.0 mEq/L, 6.0–<6.5 mEq/L, and ≥ 6.5 mEq/L). For the multivariable models, predictor variables were chosen a priori based on factors previously shown to be associated with in hospital mortality. Covariates included in our main model assessing the association of mortality with hyperkalemia in non-dialysis dependent patients included age, sex, and race; baseline comorbidities captured by ICD-9-CM codes (diabetes, heart failure, hypertension, cerebrovascular disease, peripheral vascular disease, lung disease, dementia); other laboratory values on admission (glucose, white blood cell count, hematocrit, glomerular filtration rate); peak cardiac troponin level (a marker of infarct size); number of potassium checks per patient; cardiogenic shock and acute respiratory failure on admission (determined by ICD-9-CM codes); in-hospital procedures captured by ICD-9-CM codes (cardiac catheterization, percutaneous coronary intervention, and coronary artery bypass graft surgery); in-hospital complications (acute kidney injury defined by the Acute Kidney Injury Network as an increase in serum creatinine from laboratory (not ICD-9-M codes) by ≥0.3 mg/dL from baseline, or a relative increase in serum creatinine of ≥50%, during hospitalization); length of hospital stay; and medications during hospitalization (fibrinolytic therapy, aspirin, clopidogrel, ticlopidine, β-blockers, angiotensin-converting enzyme [ACE] inhibitors or angiotensin II receptor blockers, calcium channel blockers, nitrates, diuretics, bronchodilators, statins, insulin treatment, and oral antihyperglycemic agents).
For the model evaluating relationship of dialysis dependence and in-hospital mortality stratified by potassium level, the covariates included were nearly identical to those used in the main model, except we did not include acute kidney injury or initial glomerular filtration rate.
In secondary analyses, we examined crude mortality rate according to the number of hyperkalemic values (1 vs. 2 vs. 3 or greater). Next, we examined crude in-hospital mortality rates according to whether or not the potassium levels normalized following the highest potassium measurement using Chi square test. Additionally, a multivariable logistic regression model was constructed to examine the relationship between post-maximum potassium level normalization in patients with hyperkalemia and in-hospital mortality. For this model, we used the same covariates as those included in the logistic model evaluating the relationship of categories of hyperkalemia and in-hospital mortality. Lastly, we used a sensitivity analysis to evaluate the association between varying degrees of hyperkalemia and in-hospital mortality after excluding patients who died within 24 hours of admission.
Missing baseline data (mean number of missing items per patient= 0.34) were imputed using IVEware (Imputation and Variance Estimation Software; University of Michigan’s Survey Research Center, Institute for Social Research, Ann Arbor, MI). All remaining analyses were conducted using SAS v9.3 (SAS Institute, Inc., Cary, NC), and statistical significance was determined by a 2-sided p-value of <0.05.
Of 38,689 acute myocardial infarction patients in our analytic cohort, 886 (2.3%) were on prior maintenance dialysis. The median number of acute myocardial infarction patients from each hospital was 219 (interquartile range [IQR]: 48 to 1030). The 67 hospitals that contributed data to this analysis were comparable in their characteristics to those reported in other national registries: they were most commonly urban centers (88.5%), were less frequently teaching hospitals (35.9%), and represented all geographic regions of the United States (Northeast 38.5%, Midwest 25.6%, South 26.9%, and West 9%).
The prevalence of any hyperkalemia (max K >5 mEq/L) in the overall cohort was 23.6%, while moderate-severe hyperkalemia (max K ≥ 5.5 mEq/L) occurred in 9.8% of patients. The median number of potassium measurements per patient with hyperkalemia was 8.0 (IQR, 4.0–14.0) vs. 3.0 (IQR, 2.0–6.0) in patients without hyperkalemia. The median length of stay for patients with hyperkalemia was 7.3 days (IQR, 4.2–11.6) vs. 3.8 days (IQR, 2.5–6.0) in patients without hyperkalemia. A maximum potassium level of 5 mEq/L or greater occurred in 66.8% of patients who were dialysis-dependent and in 22.6% of patients who were not dialysis-dependent. Moderate-severe hyperkalemia occurred in 41.0% and 9.1% who were and were not dialysis-dependent, respectively. Patients with higher maximum potassium levels were older and had a greater burden of comorbidities including higher burden of heart failure, lung disease and acute kidney injury and any renal disease. They also had lower median estimated glomerular filtration rate, lower hemoglobin, and higher glucose and peak troponin levels (Table 1).
There was a steep increase in mortality with higher maximum potassium levels; this relationship was linear in non-dialysis patients, while a plateau was observed in dialysis patients (Figure 2). In the overall cohort, patients with higher max potassium levels experienced higher in-hospital mortality rates, as follows: 4.2%, 11.1%, 16.6%, 26.6%, 31.7% (p value for trend <0.001) for maximum potassium categories <5, 5 – <5.5, 5.5 – <6.0 mEq/L, 6 – <6.5 mEq/L, ≥6.5 mEq/L, respectively. The relationship between maximum potassium and mortality persisted within each dialysis subgroup after multivariable adjustment. Adjusted odds ratios (OR) for in-hospital mortality based on maximum potassium of 5 – <5.5, 5.5 – <6.0 mEq/L, 6 – <6.5 mEq/L, ≥6.5 mEq/L versus <5 mEq/L were 2.03 (95% CI 1.79–2.31), 2.79 (2.35–3.31), 4.62 (3.74–5.71), 4.97 (3.96–6.25) respectively for patients who were dialysis-dependent. The odds for in-hospital mortality among non-dialysis patients, in comparison to a reference potassium value of <5 mEq/L were 1.62 (1.41–1.87), 2.02 (1.68–2.43), 3.18 (2.51–4.03), 3.37 (2.60–4.36). In a sensitivity analysis excluding patients that died within 24 hours of admission from the primary model, we found that in the odds of in-hospital mortality rates were similar to those of the primary model (data not shown).
Of all patients with any hyperkalemia and at least one potassium measurement following the maximum potassium level (n=7549), 6788 experienced potassium normalization to <5 mEq/L, and 761 patients’ levels did not. Following multivariable adjustment, patients whose potassium level did not normalize were found to have greater than twice the odds of in-hospital mortality (OR 2.22, 95% CI 1.74–2.78) as compared with those whose potassium levels did normalize.
Among patients with in-hospital hyperkalemia, the mortality rate was 13.4% for patients with a single measurement of potassium ≥5 mEq/L, while it was 16.2% in patients who had 2 potassium measurements of a level of ≥5 mEq/L, and 19.8% in patients with 3 or more potassium measurements ≥5 mEq/L.
In this large multicenter, contemporary cohort with acute myocardial infarction, we found that hyperkalemia occurred in approximately twenty percent of non-dialysis patients and sixty five percent of dialysis patients. There was a graded mortality increase with higher maximum potassium levels, particularly in non-dialysis patients; and with greater number of hyperkalemia values. Furthermore, in-hospital mortality risk was more than double among patients with hyperkalemia whose potassium levels remained persistently hyperkalemic versus those whose levels normalized during hospitalization. To our knowledge, our study is among the first to describe the prevalence of hyperkalemia, and the steep increase in-hospital mortality associated with hyperkalemia in patients hospitalized with acute myocardial infarction who are and are not dependent on renal replacement therapy.
Reports of the prevalence of hyperkalemia in cardiovascular diseases have thus far been variable. In a large cohort of patients hospitalized with acute myocardial infarction, Goyal et al examined the association of potassium levels with in-hospital ventricular arrhythmias, cardiac arrest, finding an increase in the risk of in-hospital death among those with potassium levels of 5.5 mEq/L.1 Serum potassium levels in the high-normal or mildly elevated range (4.5 – 5.5 mEq/L) were also associated with significantly increased mortality.1,14 An older prior study included 1074 patients with acute myocardial infarction and demonstrated a U-shaped relationship between potassium levels and early postinfarction ventricular fibrillation events, but lacked power to show an association between potassium level and mortality, and was performed in the era preceding current advancement in acute myocardial infarction management.15
Several studies have reported increased hyperkalemia trends associated with widespread adoption of medications including angiotensin converting enzyme inhibitors16, aldosterone receptor blockers17, spironolactone7,18. Our study extends prior work by describing mortality according to varying degrees of hyperkalemia, as well as examining outcomes stratified by the number of hyperkalemia values and normalization (versus persistence) of hyperkalemia.
Recognition of hyperkalemia is especially important in patients with cardiovascular disease, such as those with acute myocardial infarction. Our findings highlight the frequency of hyperkalemia following acute myocardial infarction. Though treatment effect was not evaluated in this analysis, it is noteworthy that resolution of hyperkalemia does portend a better prognosis than persistent hyperkalemia. We feel that the poor prognosis associated with hyperkalemia should prompt clinicians to evaluate their patients for potentially modifiable factors that are associated with hyperkalemia following acute myocardial infarction.. In terms of hyperkalemia management, the few more traditional approaches to pharmacologic management of hyperkalemia, including intravenous insulin and sodium bicarbonate, have conflicting data supporting their efficacy. Several new products are in development, including patiromer calcium and sodium zirconium cyclosilicate (ZS-9).19–21 Phase 3 clinical trials have been recently completed in outpatients with chronic kidney disease and hyperkalemia. Although these novel treatment modalities may ultimately change the way hyperkalemia is managed in the hospital, they are still undergoing investigation in the acute and outpatient settings. Our study is particularly timely in light of development of new medications for the management of hyperkalemia, and highlights an opportunity to investigate the heterogeneity of treatment benefit of these agents. Furthermore, whether more aggressive treatment of hyperkalemia will improve patient outcomes is still unknown, and requires further study.
Our study findings should be considered in the context of several potential limitations. We recognize that hemolysis can contribute to hyperkalemia and that rates of hemolysis are higher in initial emergency room phlebotomy specimens compared to those taken later when in hospital. However, should a significant number of pseudohyperkalemic events have occurred due to hemolysis, we feel that this would have biased our results toward the null. In terms of the analysis, first, because of the observational nature of our analyses, we cannot rule out the possibility of residual confounding despite adjustment for numerous potential confounders. Therefore, our data cannot be used to determine whether hyperkalemia is a marker or mediator of adverse outcomes, particularly if the peak K occurred around the time of a patient’s death. Second, we did not evaluate hyperkalemia treatment interventions in this study, so no inference could be made regarding the cause of normalization (or lack thereof) of hyperkalemia. Last, our study focused on in-hospital mortality; and did not examine the relationship between hyperkalemia and other outcomes.
Our findings reveal that hyperkalemia is very common in patients hospitalized with acute myocardial infraction, and associated with marked increases in mortality, even with mild potassium elevations – both in patients with and without end-stage renal disease. Furthermore, patients with more frequent and persistent hyperkalemia experience higher in-hospital mortality as compared to those with transient potassium elevations. These results should prompt further investigation to determine whether more aggressive management of hyperkalemia can improve outcomes in patients hospitalized with acute myocardial infarction.
Dr. Grodzinsky is supported by a T32 training grant from the National Heart Lung and Blood Institute (T32HL110837). The funding agencies had no role in data collection, analysis, interpretation or the decision to submit the results. The content is solely the responsibility of the authors and does not necessarily represent the official views of the funding agencies.
Consulting work for United Healthcare, Amgen, Novartis and Janssen.
AM received fee as member of advisory board and/or Steering Committee from Cardiorentis, The Medicine Company, Adrenomed, MyCartis, ZS Pharma and Critical Diagnostics.
Dr. Kosiborod has received speaker fees from the American Association of Clinical Endocrinology, The American Diabetes Association, The Curators of the University of Missouri, Washington Med Center, Health Sciences Media, Inc, Heartland Mid America Chapter of AACE and R & R Healthcare Communications, Inc.
Publisher's Disclaimer: 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 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.
All authors had access to the data and a role in writing and revising this manuscript.
Dr. Spertus discloses the following potential conflicts of interest: Research Grants from Lilly, Abbott Vascular and Genentech.
Dr. Mebazaa discloses the following relationships: speaker's honoraria from The Medicines Company, Novartis, Orion, Roche, Servier, Vifor Pharma.
Dr. Kosiborod discloses the following relationships: consulting fee for Abbvie, AstraZeneca, Edwards Life Sciences, Gilead Sciences, Roche, St. Jude Medical, Genentech, Regeneron, Lilly, ZS Pharma.
All other authors report no relevant conflicts of interest or disclosures.