Nephrology

Rapid correction of hyperkalemia is associated with reduced mortality in ED patients

a b s t r a c t

Background: Hyperkalemia (HK) is common and associated with mortality. Our purpose was to determine if the rapid correction of elevated serum potassium level (K+) was associated with reduced mortality in emergency de- partment (ED) patients.

Methods: Design: We reviewed the electronic medical records (EMR) of ED patients with HK (K+ >= 5.5 mEq/L) from 10/2016-10/2017. Setting: Large, academic ED. Participants: Adult ED patients presenting with hyperkalemia. Main outcomes and measures: The main outcome was in-hospital mortality. We compared out- comes of patients whose K+ normalized (dropped below 5.5 mEq/L) with those whose K+ did not normalize using chi-square and multivariate analyses to determine the associations between predictor variables and out- comes.

Results: From 114,977 ED visits, 1033 patients (0.9%, 95%CI 0.85-0.95%) had HK. Their mean (SD) age was 60 (26) years and 58% were male. Of these, 884 had a second K+ measured within a median (IQR) of 5 (3-8) hours. Mor- tality and admission rates were higher in patients with HK vs. those with normal K+ (8.5% vs. 0.8%, P b 0.001 and 80% vs. 39%, P b 0.001, respectively). Mortality was lower in patients whose HK normalized compared with those whose K+ remained elevated (6.3% vs. 12.7%, P = 0.001). After adjusting for age, creatinine, comorbidities, and initial K+, normalization of K+ was associated with reduced mortality (OR 0.47, 95%CI 0.28 to 0.80).

Conclusions: Normalization of K+ during the ED stay in patients with HK is associated with a 50% mortality reduc- tion. Efforts to rapidly identify and treat HK in the ED are needed.

(C) 2019

Introduction

Hyperkalemia (HK), defined as a serum potassium (K+) N5.5 mEq/L, is responsible for an estimated 800,000 emergency department (ED) visits each year [1]. HK has the potential to cause fatal cardiac arrhyth- mias [2-4]. Several studies, both in [5] and out [6-10] of the ED, have confirmed that HK is associated with increased mortality, which in- creases proportional to the degree of K+ elevation. As a result, it is a common belief based on little confirmatory evidence, that a temporally sensitive urgent correction of HK is necessary to improve patient outcomes.

Unfortunately, there is little evidence to guide practitioners on how HK should best be managed. A recent study documented a wide varia- tion in how HK is managed throughout U.S. EDs [11]. It is unclear

* Corresponding author at: Department of Emergency Medicine, HSC-L4-050, Stony Brook University, Stony Brook, NY 11794, United States of America.

E-mail address: [email protected] (A.J. Singer).

whether the rate and extent of correction of raised K+ levels affect pa- tient outcomes such as mortality.

The goal of the current study was to describe the association be- tween the rate and extent of reducing elevated K+ levels and mortality in ED patients presenting with HK. We hypothesized that the mortality of patients whose K+ levels normalized during their ED stay would be lower than patients in whom K+ levels remained elevated.

Methods

Study design

We performed a structured, retrospective chart review, consistent with the recommended methodology of Gilbert and Lowenstein [12], as well as that of Kaji et al. [13], in all patients presenting to our ED with an elevated serum K+. Our study also followed the Strengthening of Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines for cross-sectional studies (http://www.equator-

https://doi.org/10.1016/j.ajem.2019.12.012

0735-6757/(C) 2019

2362 A.J. Singer et al. / American Journal of Emergency Medicine 38 (2020) 23612364

network.org/reporting-guidelines/strobe/). Because of the retrospective design, we received IRB approval with waiver of informed consent.

Patients and setting

Eligible adult patients (ages 18 and older) presented from October 2016 to October 2017 and had a serum K+ N5.5 mEq/L while in the ED. We excluded patients with evidence of blood sample hemolysis. Our ED is a large, tertiary care, suburban, academic medical center with over 100,000 annual ED visits. Our medical center is also a major referral center for the county of Suffolk, NY with over 1.5 million inhabitants.

Data source and collection

We performed a computerized search of our electronic medical re- cords to identify patients meeting all the inclusion criteria. For eligible patients, we extracted patient demographic information (age, sex), co- morbidities (e.g., chronic kidney disease), current medications (e.g., ACE inhibitors), laboratory values (e.g., K+ and Creatinine levels), and treatments received while in the ED for managing HK. We defined all study data and variables prior to initiating the study and trained our data abstractors using a library of definitions. We periodically mon- itored data collection and determined the inter-observer agreement on the primary outcome on a randomly selected sample of 20 study pa- tients. Interobserver agreement for mortality was 100%.

Study outcomes

The primary outcome was inpatient mortality. Secondary outcomes included admission to the hospital, admission to an intensive care unit (ICU) at any time, and hospital length of stay.

Data analysis

Data are summarized as numbers and frequencies for binary data and means with standard deviations (SD), or medians with inter- quartile ranges (IQR), for continuous data. Patients were divided into two groups based on whether their K+ was reduced to b5.5 mEq/L while still in the ED, or not. Comparisons between groups were per- formed using Chi-square or Fisher’s exact test for binary data and t– test or Mann Whitney U test for continuous data. Mortality rates were adjusted using age, sex, Charlson Comorbidity Index (CCI) [14], initial K+ and creatinine concentrations. Multivariate analysis was performed with forward stepwise logistic regression for binary data and linear re- gression for continuous data. Alpha was defined as a P value of 0.05 or less.

Results

Of 114,977 ED patient visits during the study period, 61,492 had at least one K+ measured. Among all patients 1033 (0.9%, 95%CI 0.85-0.95%) were hyperkalemic. Of hyperkalemic patients, 884 had a second K+ measured within a median (IQR) of 5 (3-8) hours, and rep- resent the analytic cohort. Mean (SD) age of the analytic cohort was 64

(22) years, and 42% were female. Inpatient mortality and admission rates were higher in patients with HK compared with those with a nor- mal presentation K+ (8.5% vs. 0.8%, P b 0.001 and 80% vs. 39%, P b 0.001, respectively).

Of the 884 patients who had a second K+ measured while still in the ED, K+ levels normalized in 576 patients (65%). In the remaining 308 patients, the second K+ remained above normal. The median [IQR] ED lengths of stay for patients whose K+ normalized and those whose K+ did not normalize were similar (664 [428-1243] vs. 670 [428-1133] mi- nutes respectively, P = 0.94 using the median test). The mean (SD) rate of decline in K+ among hyperkalemic patients was 0.33 (3.24) mEq/L

Table 1

Baseline characteristics of study patients

2nd K+ b 5.5 (n = 576)

2nd K+ >= 5.5 (n = 308)

P

value

Mean age (sd)

64 (22)

63 (20)

0.50

Male, %

59

57

0.62

Mean (SD) 1st K+ level, mEq/L

5.9 (0.6)

6.3 (0.8)

b0.001

Mean (SD) 1st creatinine, mg/dL

2.8 (3.0)

4.8 (4.3)

b0.001

% Creatinine >=1.5 mg/dL

57

81

b0.001

% Charlson index N5

33

36

0.41

per hour and increased with higher initial K+ levels. The absolute reduc- tion in K+ between the first and second measurements was greater in patients whose K+ normalized compared with those whose K+ remained elevated (1.2 [0.8] and 0.3 [0.7] respectively, P b 0.001). A comparison of baseline characteristics of patients whose K+ normalized vs whose K+ did not is presented in Table 1. Patient groups were similar in age, sex, and CCI. However, normalized patients had lower initial K+ and creatinine levels. Overall Hospital admission rates were similar (87% and 89% among patients whose K+ did and did not normalize respec- tively, P = 0.48).

Mortality was significantly lower in patients whose K+ normalized compared with those whose K+ remained elevated (6.3% vs. 12.7%, P = 0.001). The association of predictor variables and mortality is pre- sented in Table 2. After adjusting for age, creatinine, CCI, and initial K+, normalization of K+ was still associated with reduced mortality (OR 0.47, 95%CI 0.28 to 0.80). Hospital and intensive care unit admission rates were similar between normalizing and persistently elevated co- horts (87% vs. 89% [P = 0.48] and 31% vs. 35% [P = 0.27], respectively). There were 75 deaths of which 37 were classified as non- hyperkalemic; 38 were hyperkalemic since the patient had hyperkalemia at the time of death and died due to cardiac related arrest. There was no difference in sex but the patients with hyperkalemic re- lated deaths were younger than those who died from other causes (mean [SD] age 79 [11] vs. 67 [18], P = 0.001). Both initial and 2nd po- tassium measurements were higher in the potassium related death group (6.5+-0.9 vs. 6.0+-0.7 [p = 0.008] and 5.7+-0.9 vs. 5.2+-0.9 [P =

0.02], respectively). The rate of mortality directly related to hyperkalemia was higher in patients whose potassium did not normal- ize while in the ED (8% vs. 2%, P b 0.001, respectively).

Of the 844 patients in the sample, 141 (16%) were diagnosed or were identified as having ESRD. Of these 141 patients 2 (1%) died. Neither of these patients were identified as being on dialysis although one of them was identified as being noncompliant with dialysis. Of the 141 patients with ESRD 8 (6%) were identified as being on dialysis; none of them died in-hospital.

Medications aimed at reducing K+ levels were given to 525 of 884 (59.4%) patients while still in the ED (Table 3). Patients receiving at least one therapy were older than the untreated group (66+-18 vs. 61

+-26, respectively, P = 0.002). There was no sex difference between treated and untreated patients (59% females vs. 57% females, respec- tively, P = 0.48). In general, the K+ levels were higher in those receiving all therapies except for oral sodium polystyrene sulfonate, which was given to b10% of patients in both groups (Table 4). Overall use of any K+ reducing medication did not differ between normalized and persis- tently elevated cohorts (60.8% vs. 56.8%, respectively, P = 0.26).

Table 2

Multivariate predictors of mortality

OR

95% CI

Age

1.026

1.010-1.042

+

1.35

1.02-1.79

2nd K+ normalized (b 5.5)

0.47

0.28-0.80

Creatinine >=1.5

1.92

1.03-3.57

Diabetes

0.45

0.26-0.78

Renal disease

0.56

0.42-0.76

Initial K level

A.J. Singer et al. / American Journal of Emergency Medicine 38 (2020) 23612364 2363

Table 3

Administration of K+-reducing medications by K+ normalization status.

2nd K+ b 5.5 (n = 576)

2nd K+ >= 5 (n = 308)

P

value

% Sodium bicarbonate

8.5

13.0

0.04

% Kayexalate

8.2

6.5

0.37

% Insulin

46

46

0.91

5 mg

4

6

10 mg

32

33

% Glucose

18

12

0.01

% Calcium

13.2

23.2

b0.001

% Albuterol

25.5

32.8

0.02

2.5 mg

10

13

10 mg

13

18

% Any K+-reducing med

60.8

56.8

0.26

However, intravenous calcium (aimed at stabilizing the cardiac mem- brane) and nebulized albuterol were more commonly used in the group of patients whose K+ levels remained elevated, most likely due to their higher presentation K+ concentration.

After excluding deaths, mean total hospital length of stay was not significantly different between those whose K+ normalized and those whose levels remained elevated (173 vs. 159 h, P = .38). Normal- ization of K+ was not associated with LOS after adjusting for age, sex, initial K+ and creatinine levels, and whether a K+-reducing medication was administered.

Discussion

Our study is the first showing that ED patients presenting with HK have lower mortality when their K+ levels normalize while still in the ED, compared with those whose K+ levels remain elevated. If prospec- tively confirmed, our findings will have implications for ED operations as to more rapidly identifying hyperkalemic patients (e.g., the more standard use of rapid point of care testing as a mortality reduction inter- vention), as well as suggesting the necessity of the immediate imple- mentation of K+ lowering therapy.

Patients who had a second potassium level >= 5.5 had markedly ele- vated creatinine levels compared to those with a lower second potas- sium. Thus, they may have been more likely to have chronic kidney disease or be on renal replacement therapy, and by definition be ex- pected to do worse. However, even after controlling for creatinine levels, normalization of potassium was associated with lower mortality. We also described therapeutic variance in the application of HK ther- apies, based on the persistence of elevated K+. While the overall use of any K+ lowering therapy was similar between the groups, the use of cal- cium (which stabilizes the cardiac membrane and does not alter the K+ concentration) and albuterol was more common in those whose K+ remained elevated. The increased use of more therapies in the setting of persistent K+ elevation may suggest that these interventions failed to initially normalize K+. Conversely, the increased use of calcium and

Table 4

+

albuterol may have simply been due to the fact that, in patients with higher initial K+, multiple therapies are likely, as would be consistent with the practice patterns described in the recent prospective REVEAL hyperkalemia intervention study [15].

Finally, the absolute reduction in K+ concentration was greatest in patients whose K+ normalized. Why therapeutic resistance occurred in some patients is unclear. Unfortunately, our methodology does not allow determination of which therapies are most effective in lowering K+. Lastly, as reported by others, we confirm the association between increased absolute K+ levels and mortality.

A previous study including 175 ED patients with HK found that the median (IQR) time from triage to initiating treatment was 117 (59-196) minutes [16]. The median (IQR) time to treatment initiation in another multicenter study was 2.7 (1.9-3.5) hours [11]. Delays in rec- ognizing potentially fatal HK can be reduced with point-of-care labora- tory testing performed in the ED, especially in patients at high risk of HK (e.g., chronic kidney disease) or with ECG findings (e.g., widened QRS) suggesting HK. However, overreliance on ECG findings is problematic due to its low sensitivity in identifying HK. In the study by Peacock et al., peaked T-waves were seen in b25% of patients with K+ levels less than or equal to 7.0 mEq/L, and in only 35% of those with levels N7.0 mEq/L. [11] A widened QRS was seen in b13% of those with K+ <= 7.0 mEq/L and in only 45% of those with K+ N 7.0 mEq/L.

While urgent reduction of K+ levels is necessary, evidence or guide- lines regarding which, or how many, therapies are most safe and effec- tive are lacking. A prior observational study of ED management of HK (defined as a K+ level greater than or equal to 5.5 mEq/L) included 203 patients across 14 U.S. sites [11]. In this study, the median (IQR) K+ in patients not receiving immediate dialysis was 6.2 (5.7-6.8) mEq/L, and the median (IQR) reduction in K+ was 0.9 (0.4-1.5) mEq/L within 4 h of initiating treatment. Here too, treatments varied greatly with insulin/glucose being most common (64%) followed by intrave- nous calcium (55%), inhaled beta-2 agonists (33%), oral sodium polysty- rene sulfonate (31%), IV bicarbonate (29%), hemodialysis (24%) and IV diuretics (5%). In this study, seven different therapies were used in 144 different application strategies, thus emphasizing the need for bet- ter guidance on how best to manage HK. As expected, hemodialysis was found to be the most effective treatment for HK reducing the median (IQR) K+ levels from 6.6 (6.2-7.0) to 3.8 (3.7-4.2) mEq/L within 2 h of initiation.

Limitations

Our study has several major limitations. Most importantly, due to the retrospective and observational nature of the study, we cannot ex- clude residual confounding that may have affected the association be- tween the decline in K+ levels and outcomes. We also did not compare ECG findings in patients whose K+ normalized or remained el- evated. Thus, our outcomes should be considered hypothesis generating associations. The study results may also not be generalizable to other medical settings outside of our own. While we excluded patients with

obvious hemolysis, we cannot exclude the possibility of pseudo

Mean (SD) of 1st K

level by medication administration.

Mean (SD) of 1st K measurement P value Did not receive med Received med

hyperkalemia in some patients. Finally, the recent availability of newer K+ binding agents was not evaluated in this data set; thus, our findings must be limited to the seven therapies discussed herein.

Sodium bicarbonate 6.0 (0.7) 6.5 (0.7) b0.001

Kayexalate 6.0 (0.7) 6.1 (0.6) 0.32

Insulin 5.9 (0.6) 6.2 (0.8) b0.001

Conclusions

Our results demonstrate that, in ED patients presenting with hyperkalemia, rapid K+ lowering while still in the ED is associated with as much as a 50% reduction in inpatient mortality. Future studies should determine whether Rapid reduction of K is the cause of a reduced mortality. Until then, greater efforts to recognize and initiate therapy in ED patients with hyperkalemia are needed.

5 mg

6.2 (0.7)

10 mg

6.3 (0.8)

Glucose

6.1 (0.7)

5.9 (0.6)

b0.001

Calcium carbonate

5.9(0.7)

6.6 (0.7)

b0.001

Albuterol

5.9 (0.6)

6.3 (0.8)

b0.001

2.5 mg

6.2 (0.9)

10 mg

6.4 (0.7)

Any K+-reducing med

5.9 (0.6)

6.2 (0.7)

b0.0001

2364 A.J. Singer et al. / American Journal of Emergency Medicine 38 (2020) 23612364

Authors contribution

AJS designed the study and drafted the first version, HCT analyzed data and contributed to the final paper, WFP assisted in data analysis and writing of the final paper.

Declaration of competing interest

AJS and WFP have received consulting fees from AstraZeneca and AJS is on the speakers’ bureau of AstraZeneca.

References

  1. Aggarwal S, Topalogu H, Kumar S. Trends in emergency room visits due to hyperkalemia in the United States. Value Health 2015;18:A386.
  2. Fisch C. Relation of electrolyte disturbances to cardiac arrhythmias. Circulation 1973;

47:408-19.

  1. Surawicz B. Relationship between electrocardiogram and electrolytes. Am Heart J 1967;73:814-34.
  2. Durfey N, Lehnhof B, Bergeson A, et al. Severe hyperkalemia: can the electrocardio- gram risk stratify for short-term adverse events? West J Emerg Med 2017;18: 963-71.
  3. Singer AJ, Thode HC, Peacock WF. A retrospective study of ED K+ disturbances: se- verity, treatment and outcomes. Clin Exp Emerg Med 2017;4:73-9.
  4. Khanagavi J, Gupta T, Aronow WS, et al. Hyperkalemia among hospitalized patients and association between duration of hyperkalemia and outcomes. Arch Med Sci 2014;10:251-7.
  5. An JN, Lee JP, Jeon HJ, et al. Severe hyperkalemia requiring hospitalization: predic- tors of mortality. Crit Care 2012;16:R225.
  6. Jain N, Kotla S, Little BB, et al. Predictors of hyperkalemia and death in patients with cardiac and renal disease. Am J Cardiol 2012;109:1510-3.
  7. Goyal A, Spertus JA, Gosch K, Venkitachalam L, Jones PG, Van den Berghe G, et al. Serum K+ levels and mortality in acute myocardial infarction. JAMA 2012;307: 157-64.
  8. McMahon GM, Mendu ML, Gibbons FK, Christopher KB. Association between hyperkalemia at critical care initiation and mortality. Intensive Care Med 2012;38: 1834-42.
  9. Peacock WF, Rafique Z, Clark CL, Singer AJ, Turner S, Miller J, et al. Real world evi- dence for treatment of hyperkalemia in the emergency department (REVEAL-ED): a multicenter, prospective, observational study. J Emerg Med 2018;55:741-50.
  10. Gilbert EH, Lowenstein SR, Koziol-McLain J, Barta DC, Steiner J. Chart reviews in emergency medicine research: where are the methods? Ann Emerg Med 1996; 27:305-8.
  11. Kaji AH, Schriger D, Green S. Looking through the retrospectoscope: reducing bias in emergency medicine chart review studies. Ann Emerg Med 2014;64(3):292-8. https://doi.org/10.1016/j.annemergmed.2014.03.025.
  12. Charlson ME, Pompei P, Ales KL, MacKenzie CR. New method classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40:373-83.
  13. Peacock WF, Miller J, char CL, et al. Burden of hyperkalemia and treatment patterns in the emergency department setting: results from the REVEAL-ED study. Ann Emerg Med 2016;68:S1-2.
  14. Freeman K, Feldman JA, Mitchell P, Donovan J, Dyer S, Elisio L, et al. Effects of presen- tation and electrocardiogram on time to treatment of hyperkalemia. Acad Emerg Med 2008;15:239-49.

Leave a Reply

Your email address will not be published. Required fields are marked *