- Email: [email protected]
Incidence and determinants of hyperkalemia and hypokalemia in a large healthcare system
IJCA-25256; No of Pages 8 International Journal of Cardiology xxx (2017) xxx–xxx
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Incidence and determinants of hyperkalemia and hypokalemia in a large healthcare system Erik Nilsson a,b,h,1, Alessandro Gasparini c,1, Johan Ärnlöv d,1, Hairong Xu e,1, Karin M. Henriksson d,1, Josef Coresh f,1, Morgan E. Grams f,1, Juan Jesus Carrero g,⁎,1 a
Division of Renal Medicine, Department of Clinical Science, Intervention and Technology, Karolinska Institutet, Stockholm, Sweden Department of Internal Medicine, School of Medical Sciences, Örebro University, Örebro, Sweden Department of Health Sciences, University of Leicester, United Kingdom d Department of Medical Sciences, Uppsala University Hospital, Uppsala, Sweden e AstraZeneca, Gaithersburg, MD, USA f Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA g Department of Medical Epidemiology and Biostatistics (MEB), Karolinska Institutet, Sweden h Division of Baxter Novum, Department of Clinical Science, Intervention and Technology, Karolinska Institutet, Stockholm, Sweden b c
a r t i c l e
i n f o
Article history: Received 1 May 2017 Received in revised form 4 July 2017 Accepted 12 July 2017 Available online xxxx Keywords: Epidemiology Hyperkalemia Hypokalemia Renin-angiotensin-aldosterone system inhibitors
a b s t r a c t Background: Hypo- and hyperkalemia in clinical settings are insufficiently characterized and large-scale data from Europe lacking. We studied incidence and determinants of these abnormalities in a large Swedish healthcare system. Methods: Observational study from the Stockholm CREAtinine Measurements project, including adult individuals from Stockholm accessing healthcare in 2009 (n = 364,955). Over 3-years, we estimated the incidence of hypokalemia, defined as potassium b 3.5 mmol/L, hyperkalemia, defined as potassium N 5 mmol/L, and moderate/severe hyperkalemia, defined as potassium N 5.5 mmol/L. Kidney function was assessed by estimated glomerular filtration rate (eGFR). Results: Of 364,955 participants, 69.4% had 1–2 potassium tests, 16.7% had 3–4 tests and the remaining 13.9% had N 4 potassium tests/year. Hypokalemia occurred in 49,662 (13.6%) individuals, with 33% recurrence. Hyperkalemia occurred in 25,461 (7%) individuals, with 35.7% recurrence. Moderate/severe hyperkalemia occurred in 9059 (2.5%) with 28% recurrence. The frequency of potassium testing was an important determinant of dyskalemia risk. The incidence proportion of hyperkalemia was higher in the presence of diabetes, lower eGFR, myocardial infarction, heart failure (HF), or use of renin angiotensin-aldosterone system inhibitors (RAASi). In adjusted analyses, women and use of loop/thiazide diuretics were associated with lower hyperkalemia risk. Older age, lower eGFR, diabetes, HF and use of RAASi were associated with higher hyperkalemia risk. On the other hand, women, younger age, higher eGFR and baseline use of diuretics were associated with higher hypokalemia risk. Conclusion: Hypo- and hyperkalemia are common in healthcare. Optimal RAASi and diuretics use and careful potassium monitoring in the presence of certain comorbidities, especially lower eGFR, is advocated. © 2017 The Authors. Published by Elsevier Ireland Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
1. Introduction Maintenance of K+ homeostasis is important for many physiologic processes, such as cardiac electrical conduction and inotropy, smooth muscle tone, neuronal signaling and acid-base balance [1,2]. Potassium is regulated mainly in the tubuli and collecting ducts of the kidney
⁎ Corresponding author at: Department of Medical Epidemiology and Biostatistics (MEB), Karolinska Institutet, Nobels väg 12A, Box 281, 171 77 Stockholm, Sweden. E-mail address: [email protected] (J.J. Carrero). 1 This author takes responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation.
where aldosterone stimulates its excretion. Situations that affect this pathway are thought to increase the risk of dyskalemia, including comorbidities such as chronic kidney disease (CKD), heart failure (HF), cardiovascular disease (CVD) and diabetes mellitus (DM) [3,4], as well as various common medication classes such as inhibitors of the renin-angiotensin aldosterone system (RAASi), beta blockers, calcium channel blockers and diuretics. These comorbidities are often interconnected, and medications affecting potassium homeostasis are commonly used in their treatment. The incidence of dyskalemia in real-world clinical settings is insufficiently characterized and likely not reflected by the controlled environment of randomized controlled trials. The RALES trial, which studied
http://dx.doi.org/10.1016/j.ijcard.2017.07.035 0167-5273/© 2017 The Authors. Published by Elsevier Ireland Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Please cite this article as: E. Nilsson, et al., Incidence and determinants of hyperkalemia and hypokalemia in a large healthcare system, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.07.035
2
E. Nilsson et al. / International Journal of Cardiology xxx (2017) xxx–xxx
effect of spironolactone in heart failure patients on ACEi, reported a 2% rate of severe hyperkalemia in the treatment group [5], but observational reports in the U.S. revealed that 11% of outpatients prescribed with ACEi developed hyperkalemia over a period of two years [6]. Differences in reported rates of dyskalemia may be explained by the inclusion of more heterogeneous populations [7] and diverse clinical practices, such as variations in the frequency of potassium monitoring, which is reported as suboptimal in real-world settings [8–10]. Studies with broader coverage, such as those derived from healthcare records may be relevant to inform clinical practice. Available reports in this regard are scarce and derived from North American materials [6,11,12]. Because of differences in practice patterns, drug usage and healthcare access, comparison with other countries may be useful to detect population segments where more intensive K-monitoring and/ or preventive strategies are needed. In this study, we investigated the frequency, severity and recurrence of hypo- and hyperkalemia in a large Swedish healthcare system as well as comorbidities and medications associated. 2. Methods 2.1. Data sources This project is based on the Stockholm CREAtinine Measurements (SCREAM) project [13], a healthcare-utilization cohort including all residents in the region of Stockholm, Sweden, undertaking at least one measurement of serum creatinine in inpatient or outpatient care during 2006–11. Additional laboratory data, including all serum potassium measurements in the region were extracted. Data was thereafter linked with regional and national administrative databases for information on healthcare utilization (International Classification of Diseases, Tenth Revision [ICD-10] codes and therapeutic procedures [14]), complete information of drugs dispensed at Swedish pharmacies [15], validated renal replacement therapy endpoints [16] and vital status, with no loss to follow up. 2.2. Study population For this study, we included all individuals ≥ 18 years of age who accessed outpatient care and were registered in SCREAM between January 1st, 2009, and December 31st, 2009. Index date was set at January 1st 2009, the point at which the study covariates were calculated. An additional inclusion criterion was the existence of at least one ambulatory measurement of serum creatinine within the preceding year, in order to estimate kidney function. After applying study eligibility criteria, all available potassium measurements during the following three years were extracted until death, migration from the region or end of follow-up (December 31st, 2011). The study was approved by the Regional Ethics Review Board, Stockholm, Sweden. 2.3. Study covariates Inter- as well as intra-laboratory variation in laboratory measurements was considered minimal among the three laboratories providing services to the region, being frequently audited for quality and harmonization by the national organization EQUALIS (www.equalis.se). The serum creatinine measured closest to index date was used to estimate glomerular filtration rate (eGFR). To this end, creatinine measurements in connection with a hospital stay were excluded, as they may represent acute illness rather than stable kidney function. Implausible serum creatinine (b25 and N 1500 μmol/l) values were also excluded. All serum creatinine tests were measured with either enzymatic or corrected Jaffe methods. The 2009 CKD-EPI creatinine-based equation was used to calculate eGFR [17]. In the absence of information on albuminuria we defined eGFR categories rather than CKD stages. eGFR categories were considered as follows: G1–2 = eGFR
≥ 60 ml/min/1.73 m 2 ; G3 = eGFR ≥ 30 and b60 ml/min/ 1.73 m2; G4+ = eGFR b 30 ml/min/1.73 m2, also including individuals undergoing renal replacement therapy (dialysis or transplantation), which was ascertained via linkage with the Swedish Renal Registry (http://www.snronline.se). Other covariates included age, sex, and selected comorbidities based on ICD codes (chronic kidney disease, hypertension, diabetes mellitus [DM], myocardial infarction [MI], heart failure [HF], peripheral vascular disease [PVD] and cerebrovascular disease [CeVD], see Supplemental Table S1) [12,13]. Cardiovascular disease history was defined as any event of MI, HF, PVD and CeVD. To better define diabetes and hypertension, ICD-10 codes were enriched with information on current intake of related medication (purchase of oral antidiabetics, Anatomical Therapeutic Classification [ATC] code A10; purchase of anti-hypertensives, ATC codes C03, C07, C08, C09) up to 6 months before index date. We also extracted, at index date, information on concurrent use of medication that may affect potassium homeostasis. These included non-steroidal anti-inflammatory drugs (NSAID), renin-angiotensinaldosterone system inhibitors (RAASi), angiotensin converting enzyme inhibitors (ACEi), angiotensin receptor inhibitors (ARB), mineralocorticoid receptor antagonists (MRA), beta-blockers, potassium sparing diuretics, thiazide/loop diuretics, and other blood pressure medications (which included calcium channel blockers, central agonists, direct vasodilators, and α-blockers) (see Supplemental Table S2). Direct renin inhibitors were not available in Sweden during the study period. 2.4. Study outcome The study outcomes, determined by potassium laboratory tests and defined from commonly used clinical thresholds, were: a) hyperkalemia defined as potassium N 5 mmol/L; b) moderate/severe hyperkalemia defined as potassium N 5.5 mmol/L and c) hypokalemia, defined as potassium b 3.5 mmol/L. We included all available plasma (reference range 3.5–4.4 mmol/L) and serum (reference range 3.6–4.6 mmol/L) potassium measurements in our healthcare system [18], assessed by potentiometric titration. Potassium values above 10 mmol/L were considered implausible and discarded. When more than two measurements of potassium were available in the same day, we selected their median value. Repeated tests denoting hypo-/hyperkalemia within 7 days from each other were considered as part of the same event and the date of the first abnormal potassium was used. For all outcomes, we also examined the pattern of dyskalemia over the 3-year observation period, defining patterns as never-, transient- (only 1 occurrence), or recurrent (N1 event during observation). 2.5. Statistical methods This study utilizes actual health care data, with an indication behind each available test. Because the incidence and patterns of dyskalemia are dependent on the regularity of potassium measurements, we categorized patients by the following frequencies of testing: never, 1– 2, 3–4 and N 4 measurements of potassium/year, as a proxy for contact with the medical system. We used this categorization as a stratification factor in the comparisons and as a covariate in the models. As a first step, we estimated the incidence proportion and recurrence of dyskalemias over the three-year observation period overall and in suspected populations by age strata, eGFR strata, HF, MI, DM and RAASi use. Secondly, we estimated incidence rates (IR). Adjusted IR were obtained from a zero-inflated negative binomial model and presented fixing the value of the adjustment factors to their mean value. Logistic regression was used to examine baseline patient characteristics and medications associated with dyskalemia incidence; multinomial logistic regression was used to study patterns of dyskalemia. Covariates were selected a priori based on known associations with circulating potassium. These included age, sex, eGFR strata, comorbidities (CKD, hypertension, DM, MI, HF, PVD and CeVD) and medication
Please cite this article as: E. Nilsson, et al., Incidence and determinants of hyperkalemia and hypokalemia in a large healthcare system, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.07.035
E. Nilsson et al. / International Journal of Cardiology xxx (2017) xxx–xxx
use (NSAIDs, ACEi, ARBs, MRAs, beta-blockers, potassium sparing diuretics, thiazide/loop diuretics and other blood pressure medications). The frequency of potassium measurements per year was included as a covariate with the reference category of 1–2 potassium tests. All analyses were performed using R (www.r-project.org) and Stata version 14 (www.stata.com). 3. Results 3.1. Population characteristics There were 498,716 eligible individuals accessing healthcare and enrolled in SCREAM during 2009 [9]. Of those, 24.5% had no potassium test taken during observation (Table 1). These were younger, had lower proportion of comorbidities and more seldom used medications that affect potassium balance. As many as 8.9% of diabetics, 4.5% of HF patients and between 4.5 and 6.0% of patients with CKD did not undergo annual K testing (Supplemental Table S3). The remaining 364,955 participants (75.5%) underwent at least one potassium test and were therefore included in our study, and 91.1% of all potassium tests recorded during this period were performed in plasma. They underwent a median number of 2 measurements/ year (interquartile range 1–5), and their total time at risk was 1,050,392 person years. Among participants with potassium test, 69.4% had 1–2 potassium tests/year, 16.7% had 3–4 tests/year and the remaining 13.9% had N4 potassium tests/year. Across more frequent potassium testing categories, participants were older, more often men, and more often had DM, hypertension, cardiovascular disease and low eGFR. More frequent potassium testing was also associated with increased use of all studied medications. 3.2. Proportion and incidence rates of hyperkalemia Of the 364,955 participants with available potassium measurements, 334,628 (91.7%) were followed for the whole period, 29,292 (8.0%) died before end of follow up and 1035 (0.3%) migrated to other regions of Sweden.
3
As many as 25,461 (6.98%) participants had at least one episode of hyperkalemia (N 5.0 mmol/l) over the 3-year period. Of these, 64.3% experienced only one event (transient hyperkalemia), and 35.7% had recurrent hyperkalemia (N1 event). Analogously, 9059 (2.48%) individuals had at least one episode of moderate/severe hyperkalemia (N5.5 mmol/L). Of these, 72.0% had transient hyperkalemia and 28.0% had recurrence of hyperkalemia. In a sensitivity analysis we calculated the number of events using only plasma potassium tests, observing a similar incidence: 6.67% for K N 5.0 mmol/L and 2.47% for K N 5.5 mmol/L. The incidence proportion of hyperkalemia, was higher among patients with more frequent potassium measurements; 1.6%, 8.4%, and 32.3% experienced potassium N 5 mmol/L among those with 1–2, 3–4 and N4 potassium checks/year, respectively. Likewise, the 3-year risk of moderate/severe hyperkalemia was 0.2%, 1.9%, and 14.3% for those with 1–2, 3–4 and N4 potassium checks/year, respectively. Recurrent hyperkalemia was more common among individuals with N 4 potassium tests/year; For potassium N 5.0 mmol/L, 45.8% of participants with N 4 potassium checks/year had recurrent hyperkalemia, as compared to 25.0% and 12.5% in individuals with 1–2 or 2–4 K measures/year, respectively. The incidence proportion of hyperkalemia was higher among suspected comorbid populations (older age, CKD, HF, MI, DM) and among users of RAASi (Fig. 1), with the same pattern seen across categories of potassium testing frequency (Supplemental Table S4). For instance, 24.0% (range 4.33% to 43.3% depending on the frequency of K testing) of patients with heart failure had at least one potassium N 5.0 mmol/L within 3 years. Individuals with lower kidney function had the highest hyperkalemia incidence proportions: 55.0% and 33.0% of patients with eGFR G4 + had a K N 5 or N 5.5 mmol/L, respectively, during observation. Likewise, 20.3% and 7.64% of patients with eGFR G3 had a K N 5 or N 5.5 mmol/L, respectively. Supplemental Table S5 shows the incidence rates for hyperkalemia for the abovementioned high-risk comorbidities according to RAASi use. In all cases, hyperkalemia incidence was higher when consuming RAASi. Table 2 shows crude and adjusted incidence rates (IR) of hyperkalemia. Hyperkalemia incidence became increasingly more common when potassium was measured more often. The crude overall IR
Table 1 Descriptive characteristics of excluded (without K+ tests) and included participants (with K+ tests). Included participants are further stratified by the annual frequency of potassium testing. Category
Without K+ tests
With K+ tests
1–2 K tests/year
3–4 K tests/year
N4 K test/year
Total N Age, years Gender, male eGFR, ml/min/1.73 m2 eGFR Categories G1–2 (eGFR ≥ 60 ml/min/1.73 m2) G3 (eGFR60–30 ml/min/1.73 m2) G4+ (eGFR b 30 ml/min/1.73 m2) Diabetes mellitus Hypertension Cardiovascular disease Congestive heart failure Myocardial infarction Peripheral vascular disease Cerebrovascular disease RAASi ACEi ARBs MRAs Beta-blockers K-sparing diuretics Loop/thiazide diuretics Other blood pressure medications NSAIDs
118,644 (24.53%) 47.00 (35.17–60.09) 54,207 (45.69%) 99.18 (87.29–111.27) – 115,660 (97.48%) 2755 (2.32%) 229 (0.19%) 5543 (4.67%) 22,540 (19.00%) 4730 (3.99%) 1384 (1.17%) 1257 (1.06%) 911 (0.77%) 2324 (1.96%) 8476 (7.14%) 6111 (5.15%) 5417 (4.57%) 541 (0.46%) 10,110 (8.52%) 638 (0.54%) 3475 (2.93%) 5318 (4.48%) 16,990 (14.32%)
364,955 (75.47%) 63.34 (50.25–74.51) 161,125 (44.15%) 87.12 (71.72–99.61) – 316,847 (86.82%) 42,709 (11.70%) 5399 (1.48%) 57,343 (15.71%) 198,798 (54.47%) 70,282 (19.26%) 29,684 (8.13%) 22,159 (6.07%) 14,716 (4.03%) 30,377 (8.32%) 85,320 (23.38%) 61,776 (16.93%) 52,285 (14.33%) 10,659 (2.92%) 99,911 (27.38%) 12,264 (3.36%) 52,903 (14.50%) 55,879 (15.31%) 65,204 (17.87%)
252,211 (52.15%) 60.17 (46.51–69.92) 108,884 (43.17%) 90.75 (77.55–102.36) – 234,447 (92.96%) 17,256 (6.84%) 508 (0.20%) 31,133 (12.34%) 118,057 (46.81%) 30,464 (12.08%) 9299 (3.69%) 9367 (3.71%) 5537 (2.20%) 13,872 (5.50%) 49,458 (19.61%) 36,203 (14.35%) 31,218 (12.38%) 3705 (1.47%) 55,715 (22.09%) 4339 (1.72%) 22,378 (8.87%) 32,208 (12.77%) 44,592 (17.68%)
61,464 (12.71%) 69.26 (58.17–79.67) 27,143 (44.16%) 80.86 (64.53–93.83) – 49,355 (80.30%) 11,207 (18.23%) 902 (1.47%) 13,167 (21.42%) 41,887 (68.15%) 17,309 (28.16%) 7360 (11.97%) 5362 (8.72%) 3516 (5.72%) 7403 (12.04%) 18,138 (29.51%) 12,953 (21.07%) 11,257 (18.31%) 2652 (4.31%) 22,106 (35.97%) 3123 (5.08%) 12,781 (20.79%) 12,176 (19.81%) 11,908 (19.37%)
51,280 (10.60%) 74.67 (63.51–83.51) 25,098 (48.94%) 71.13 (51.47–87.94) – 33,045 (64.44%) 14,246 (27.78%) 3989 (7.78%) 13,043 (25.43%) 38,854 (75.77%) 22,509 (43.89%) 13,025 (25.40%) 7430 (14.49%) 5663 (11.04%) 9102 (17.75%) 17,724 (34.56%) 12,620 (24.61%) 9810 (19.13%) 4302 (8.39%) 22,090 (43.08%) 4802 (9.36%) 17,744 (34.60%) 11,495 (22.42%) 8704 (16.97%)
Values are presented as median and inter-quartile interval [IQI], or number and percentage. Abbreviations: eGFR, estimated glomerular filtration rate; RAASi, renin-angiotensinaldosterone system inhibitors; ACEi, angiotensin converting enzyme inhibitors; ARB, angiotensin receptor inhibitors; MRA, mineralocorticoid receptor antagonists; NSAID, non-steroidal anti-inflammatory drugs.
Please cite this article as: E. Nilsson, et al., Incidence and determinants of hyperkalemia and hypokalemia in a large healthcare system, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.07.035
4
E. Nilsson et al. / International Journal of Cardiology xxx (2017) xxx–xxx
Fig. 1. Incidence proportion of hyperkalemia over three years for selected comorbidities and use of RAASi at index date. Abbreviations: eGFR, estimated glomerular filtration rate; RAASi, renin-angiotensin-aldosterone-system inhibitors.
for mild hyperkalemia (K N 5 mmol/L) was 50.0 (95% CI 49.5–50.4) per 1000 person years. Crude and adjusted IR increased in the presence of older age, male sex, worse kidney function categories, comorbidities
and RAASi use. Participants within the lowest eGFR stratum (G4 +) experienced the highest hyperkalemia incidence (adjusted IR 133.5, 95%CI 123.0–143.8 events per 1000 person years).
Table 2 Crude and adjusted incidence rates of hyperkalemia (and 95% confidence intervals) by risk factors, per 1000 person years. Category Overall All Age category (years) 18–44 45–64 65–74 ≥75 eGFR category G1–2 G3 G4+ Heart failure No Yes Myocardial infarction No Yes Diabetes mellitus No Yes RAASi No Yes Frequency of potassium tests per year 1–2 3–4 N4
Crude IR K N 5.0 mmol/L
Adjusted IR K N 5.0 mmol/L
Crude IR K N 5.5 mmol/L
Adjusted IR K N 5.5 mmol/L
49.9 (49.5–50.3)
−
14.6 (14.4–14.9)
−
15.6 (15.1–16.2) 33.1 (32.5–33.7) 57.2 (56.2–58.2) 99.8 (98.5–101.1)
8.4 (7.8–9.0) 12.4 (11.9–12.8) 14.6 (14.0–15.2) 15.4 (14.8–16.1)
5.7 (5.4–6.1) 10.4 (10.1–10.8) 16.4 (15.9–17.0) 27.3 (26.7–28.0)
2.0 (1.7–2.2) 2.2 (2.0–2.3) 2.2 (2.0–2.4) 1.9 (1.8–2.1)
22.6 (22.3–23.0) 156.3 (154.0–158.6) 1070.4 (1052.7–1088.5)
10.8 (10.5–11.1) 30.5 (29.1–31.9) 133.5 (123.3–143.8)
5.5 (5.4–5.7) 42.9 (41.7–44.1) 418.8 (407.7–430.2)
1.8 (1.6–1.9) 5.1 (4.6–5.5) 27.5 (24.1–30.9)
37.5 (37.1–37.9) 211.9 (208.6–215.2)
12.3 (12.0–12.6) 18.7 (17.6–19.8)
10.6 (10.4–10.8) 66.9 (65.0–68.7)
2.0 (1.9–2.1) 3.0 (2.7–3.3)
43.1 (42.7–43.5) 164.1 (160.9–167.4)
12.6 (12.3–13.0) 14.6 (13.6–15.6)
12.5 (12.3–12.8) 49.5 (47.7–51.3)
2.1 (1.9–2.2) 2.2 (1.9–2.4)
36.6 (36.2–37.0) 123.1 (121.4–124.8)
11.3 (11.0–11.6) 23.3 (22.3–24.2)
10.6 (10.4–10.8) 37.0 (36.0–37.9)
1.9 (1.7–2.0) 3.7 (3.4–4.0)
34.2 (33.8–34.7) 102.2 (101.0–103.5)
11.75 (11.42–12.07) 16.80 (16.13–17.47)
9.8 (9.6–10.0) 30.7 (30.1–31.5)
1.9 (1.8–2.0) 2.8 (2.6–3.1)
5.6 (5.52–5.8) 41.3 (40.3–42.2) 334.4 (331.2–337.7)
− − −
0.78 (0.72–0.84) 7.5 (7.1–7.9) 110.5 (108.6–112.4)
− − −
Analysis was done in patients with at least one potassium measurement. Adjusted incident rates are obtained from a zero-inflated negative binomial model, and are adjusted (when applicable) by age categories, sex, renal function categories, history of heart failure, myocardial infarction, hypertension, diabetes mellitus, use of RAASi, and frequency of K+ tests per year. Adjusted incident rates are presented fixing the value of the adjustment factors to their mean value. Abbreviations: eGFR, estimated glomerular filtration rate.
Please cite this article as: E. Nilsson, et al., Incidence and determinants of hyperkalemia and hypokalemia in a large healthcare system, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.07.035
E. Nilsson et al. / International Journal of Cardiology xxx (2017) xxx–xxx
The crude overall IR for moderate/severe hyperkalemia (K N 5.5 mmol/L) was 14.7 (95% CI 14.4–14.9) per 1000 person years. Again, IR increased across older age or worsening kidney function strata and in the presence of other comorbidities or RAASi use. Again, participants within the lowest kidney function stratum experienced the highest IR. Adjusted IR showed the same increasing trend across worsening kidney function strata or diabetes, but did not dramatically differ across age strata or in the presence of other comorbidities and RAASi. 3.3. Proportion and incidence rates of hypokalemia Hypokalemia occurred in 49,662 (13.6%) individuals, and 33% of those had recurrent hypokalemia. Hypokalemia incidence proportion over 3 years is shown in Supplemental Table S6. Again, the incidence proportion of hypokalemia was higher among patients with more frequent potassium measurements and among suspected comorbid populations (older age, CKD, HF), but did not importantly differ among individuals with MI, DM, or users of RAASi. The crude incidence rate of hypokalemia was 61.3 (60.9–61.7) per 1000 person years. Hypokalemia incidence became increasingly more common when potassium was measured more often; crude IR increased for instance in the presence of older age (from 68.9 (67.7–70.1) per 1000 person years in individuals aged 18–44 years to 110.9 (109.6–112.3) per 1000 person years in individuals aged N 75 years), male sex, worse kidney function categories (from 74.8 (74.3–75.4) per 1000 person years in individuals with eGFR G1–2 to 269.8 (260.9–278.9) per 1000 person years in individuals with eGFR G4+) or heart failure (76.6 (76.1–77.2) per 1000 person years in non-HF individuals versus 145.6 (142.9–148.4) per 1000 person years in HF individuals). 3.4. Baseline characteristics associated with dyskalemias In adjusted analysis (Table 3), older age categories, male sex, worse kidney function, diabetes, hypertension, heart failure and peripheral
5
vascular disease were all associated with higher hyperkalemia risk. Use of ACE, ARBs or MRAs were also associated with increased risk. These predictors were consistent risk factors for hyperkalemia of all patterns (single/recurrent, Supplemental Tables S7 and S8) and severity (K N 5 or N 5.5 mmol). Whereas use of beta-blockers associated with higher risk of K N 5.0 mmol/L, and use of Loop/Thiazide diuretics or other blood pressure medication associated with a reduced risk, these medications did not significantly associate with K N 5.5 mmol/L (Table 3). Further, people with CeVD were less likely to experience K N 5.5 mmol/L, but this was not the case for K N 5 mmol/L. Predictors of single and recurrent K N 5.5 mmol/L differed slightly (Table S8): whereas older age associated with the risk of single K N 5.5 mmol/L, recurrence was associated with younger age categories. Use of loop/thiazide diuretics was associated with a lower risk of single K N 5.5 mmol/L, but with a higher risk of recurrence (Table S5). Predictors of hypokalemia are also depicted in Table 3. Decreased risk of hypokalemia was seen with baseline older age, lower eGFR, presence of comorbidities (DM, HF, MI, PVD) as well as with use of ACEi, ARB, MRA, beta blocker and NSAID. Hypokalemia risk was higher among women, patients with hypertension or CVD, and especially among those treated with loop/thiazide diuretics or other blood pressure medication. The predictors of single/recurrent hypokalemia did not vary (data not shown). 3.5. Clinical case: heart failure patients To test the consistency of our findings, hyperkalemia incidence proportion and determinants was studied separately in the 29,684 adult patients with HF and at least one K measurement during the study period (Supplemental Tables S9 and S10). Similar to the main overall analysis, HF patients with more frequent K testing tended to be older, more often men and with a higher number of comorbidities and medications affecting K balance. The multivariable predictors of hyperkalemia and hypokalemia were also similar to the main analysis (Supplemental Table S10), but in some cases with broader confidence
Table 3 Baseline risk factors for hyperkalemia and hypokalemia occurrence over 3 years. Baseline risk factors
Age: 18–44 years Age: 45–64 years Age: 65–74 years Age: ≥75 years Gender: female eGFR: G1–2 eGFR: G3 eGFR: G4+ Diabetes mellitus Hypertension Heart failure Myocardial infarction Peripheral vascular disease Cerebrovascular disease ACEi ARBs MRAs Beta-blockers NSAIDs Loop/thiazide diuretics K-sparing diuretics Other blood pressure medications Frequency of K measurements: 1–2 per year Frequency of K measurements: 3–4 per year Frequency of K measurements: N4 per year
Odds ratio (95% confidence interval) K N 5.0 mmol/L
K N 5.5 mmol/L
K b 3.5 mmol/L
Ref. 1.52 (1.42–1.63) 1.82 (1.69–1.95) 1.87 (1.74–2.01) 0.79 (0.77–0.81) Ref. 2.14 (2.07–2.22) 5.60 (5.24–5.98) 1.73 (1.67–1.79) 1.05 (1.00–1.10) 1.14 (1.09–1.19) 1.03 (0.98–1.08) 1.22 (1.15–1.28) 0.97 (0.93–1.01) 1.51 (1.46–1.57) 1.21 (1.17–1.26) 1.66 (1.41–1.95) 1.06 (1.03–1.10) 1.00 (0.96–1.04) 0.94 (0.90–0.97) 1.08 (0.92–1.26) 0.88 (0.85–0.91) Ref. 4.17 (3.99–4.36) 17.26 (16.58–17.97)
Ref. 1.17 (1.04–1.31) 1.22 (1.09–1.37) 1.07 (0.95–1.19) 0.79 (0.76–0.83) Ref. 2.25 (2.12–2.38) 6.39 (5.93–6.89) 1.62 (1.54–1.71) 1.10 (1.02–1.19) 1.14 (1.07–1.22) 0.98 (0.91–1.05) 1.15 (1.07–1.23) 0.92 (0.87–0.98) 1.57 (1.49–1.66) 1.22 (1.15–1.29) 1.44 (1.13–1.82) 1.00 (0.94–1.05) 0.94 (0.89–1.01) 1.01 (0.95–1.07) 1.07 (0.85–1.35) 0.96 (0.91–1.01) Ref. 6.31 (5.70–7.00) 40.13 (36.65–43.94)
Ref. 0.59 (0.57–0.61) 0.49 (0.48–0.51) 0.43 (0.42–0.45) 1.72 (1.68–1.75) Ref. 0.68 (0.66–0.71) 0.54 (0.51–0.58) 0.68 (0.66–0.70) 1.80 (1.74–1.86) 0.84 (0.80–0.87) 0.82 (0.79–0.86) 0.87 (0.83–0.91) 1.06 (1.02–1.09) 0.71 (0.69–0.73) 0.94 (0.91–0.97) 0.70 (0.61–0.80) 0.89 (0.86–0.91) 0.99 (0.96–1.02) 1.76 (1.71–1.81) 1.07 (0.95–1.21) 1.37 (1.33–1.40) Ref. 3.74 (3.64–3.84) 8.63 (8.39–8.87)
Shown are adjusted multivariable odds ratios derived from separate logistic regression models. Statistically significant associations (P b 0.05) are marked in bold. Adjustment was for age category, female, eGFR category, diabetes mellitus, hypertension, chronic heart failure, myocardial infarction, peripheral vascular disease, cerebrovascular disease, ace inhibitor, angiotensin receptor inhibitor, mineralocorticoid receptor blocker, beta blocker, non-steroidal anti-inflammatory drug use, loop- or thiazide diuretic, potassium sparing diuretic, other blood pressure medication, potassium samplings per year. Abbreviations: eGFR, estimated glomerular filtration rate; ACEi, angiotensin converting enzyme inhibitors; ARB, angiotensin receptor inhibitors; MRA, mineralocorticoid receptor antagonists; NSAID, non-steroidal anti-inflammatory drugs.
Please cite this article as: E. Nilsson, et al., Incidence and determinants of hyperkalemia and hypokalemia in a large healthcare system, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.07.035
6
E. Nilsson et al. / International Journal of Cardiology xxx (2017) xxx–xxx
intervals. Age was not a risk factor for moderate/severe HK among HF patients, perhaps explained by the older age of this population (median age 79.84 (interquartile range 70.59–86.34) years). 4. Discussion In this study we comprehensively evaluate the frequency of potassium testing in the large healthcare system of Stockholm, Sweden, and analyze with precision the rates and likelihood of recurrence of dyskalemias by clinical screening over a 3-year period. We demonstrate that hypo- and hyperkalemia is common in Swedish healthcare and that incidence figures to a large extent are influenced by the frequency of testing. Hyperkalemia was particularly common among those with comorbidities and, notably, among those with reduced kidney function. We also demonstrate that even after adjusting for these comorbidities, the use of RAASi was strongly associated with hyperkalemia occurrence. Hypokalemia was also common, especially in diuretic users. As expected, associations between risk factors and hypokalemia were generally in the opposite direction of the associations between risk factors and hyperkalemia. Strengths of our analysis include the complete regional capture of healthcare use in a country with universal healthcare access. The ascertainment of kidney function by eGFR in our analysis is a strength, as kidney dysfunction is one of the strongest risk factors for elevated potassium levels [19–23], but this condition is generally affected by poor awareness and underutilization of ICD diagnoses in healthcare [24,25]. Likewise, the use of potassium tests as the study exposure allows us to quantify the burden of mild and moderate hyperkalemia in the outpatient setting. These are clinical events otherwise not captured by ICD diagnoses [26,27], but that have been repeatedly associated with increased mortality risk in various settings, including acute MI [28,29], hypertension [30], HF [31] and CKD [32]. Evaluating the frequency of dyskalemia events in healthcare is important, as the reported incidence in controlled RCTs clearly does not match that of the real-world clinical setting. In our study, approximately 7.0% and 2.5% of observed individuals experienced mild and moderate/severe hyperkalemia, respectively, over a three-year period. These incidence proportions are comparable to those in US healthcare users undergoing blood pressure testing (10.8% and 2.3% three-year incidence proportion for K N 5.0 and K N 5.5, respectively [12]). In another large study of US healthcare users with at least one inpatient hospital visit, 13.7% had hyperkalemia ≥ 5.5 mmol/L within one year [11]. Characterizing hyperkalemia risk in different high-risk groups highlights the need for hyperkalemia-preventive strategies in these conditions. Abnormalities in potassium balance are attributable to multiple conditions affecting potassium intake, distribution and excretion, including demographic characteristics, comorbidities and medications [3,4]. Beyond the importance of underlying kidney function discussed above, our multivariate risk analyses confirm the diversity of the risk factors involved in this condition, which include older age, male sex and important comorbidities such as HF, diabetes, hypertension or CVD [11,21,22,33,34]. The use of RAASi emerged as another notable risk predictor per se (with an overall risk increase of 43.0%) and in addition to each studied comorbidity (Supplemental Table S5). Disentangling the risk attributed to medications from that of comorbidities is neither possible in our observational analysis nor in real life, as these comorbidities and medications are intricately connected. Interestingly, we found in our multivariate analysis that the hyperkalemia risk associated with RAASi use varies by medication class. Specifically, and alike a previous report [12], we found that the hyperkalemia risk associated to ACEi or MRA were similar, and higher than that of the risk associated to ARB or potassium-sparing diuretics. Our estimates are therefore in line with some [35,36], but not all [23] trials suggesting that ACEi have stronger effect in raising potassium than ARBs. Recent RCTs suggest that the use of potassium binder medications may allow the use of RAASi in patients prone to hyperkalemia [37–40], but the
long-term benefit of such strategies on clinical outcomes is yet to be demonstrated. In the meantime, we speculate that switching medications (for instance, from ACEi to ARBs) may be a possible hyperkalemia management strategy to consider. We found an even higher incidence proportion of hypokalemia (13.6%) than for hyperkalemia. Associations between risk factors and hypokalemia were generally in the opposite direction of the associations between risk factors and hyperkalemia. In addition, diuretics appeared as the strongest baseline risk factor, in agreement with the current understanding of this iatrogenically induced electrolyte abnormality [41–43]. According to previous studies, up to 56% of individuals on diuretics are likely to develop hypokalemia, although it is clinically relevant only in 4% to 5% of these cases [44–46]. Occurrence of dyskalemias was, not surprisingly, dependent on the frequency of potassium testing, and individuals with frequent potassium monitoring were found to have more comorbid conditions and were at higher dyskalemia risk. Such findings are in line with a recent U.S. analysis [12], and overall justify our decision to include frequency of testing as a covariate in our adjustments. Potassium monitoring is, after all, performed in clinical practice by judgement of the clinician, which is affected by the patient's condition, the perceived dyskalemia risk and variations in clinical practice. To our knowledge, no recommendations exist on how often to monitor for potassium in the ambulatory setting, or after the occurrence of dyskalemia. We observe that approximately 9% of patients with diabetes, 5% of patients with HF or CKD, all at high-risk of hyperkalemia did not undergo annual K screening. Current guidelines recommend to closely monitoring potassium during the initiation of RAASi therapy [47–49], and this observation is in line with real-world data suggesting that adequate monitoring is not taking place [8–10]. An additional interesting aspect of our analysis pertains to the recurrence of hyperkalemia: one in three patients experiencing mild or moderate/severe hyperkalemia developed a second episode. While some studies have addressed this issue in different populations and/or with different definitions for recurrence, they find comparable proportions [11,12]. Our multivariate analysis identifies similar baseline risk factors (old age, male sex, low eGFR, DM, HT, HF, CVD, RAASi and frequency of measurement) for both hyperkalemia occurrence and recurrence, with the exception of higher odds for hyperkalemia recurrence among younger age categories. We speculate that this may be attributed to higher risk of mortality among the elderly with hyperkalemia or more frequent discontinuation of potassium-sparing medications after a single hyperkalemia event across older age categories. Limitations in the interpretation of our study are its retrospective nature and the lack of information on other suggested risk factors for hyperkalemia, such as blood pressure or body mass index. Despite our attempts to identify erroneous values, false dyskalemias may have occurred and result in overestimation of true incidences. Data reflects routine care in the region of Stockholm, and findings may not necessarily extrapolate to other areas. Finally, as any observational study, we are impacted by residual confounding due to unmeasured/undetected factors and by confounding by indication bias, making impossible to separate, for instance, the risk of dyskalemia associated with medication use from risk due to the underlying condition that prompted such prescription. 5. Conclusion Both hypo- and hyperkalemia are frequent events among Swedish healthcare users. We found substantial variability in the frequency of potassium monitoring, but patients with older age and greater number of comorbidities (especially CKD), were consistently associated with higher dyskalemia occurrence, recurrence and event severity. These event rates increased in the presence of medication use that affects potassium balance, namely RAASi and diuretics. Further studies
Please cite this article as: E. Nilsson, et al., Incidence and determinants of hyperkalemia and hypokalemia in a large healthcare system, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.07.035
E. Nilsson et al. / International Journal of Cardiology xxx (2017) xxx–xxx
are needed to optimize potassium monitoring strategies to detect dyskalemias and to allow the implementation of preventive/corrective strategies. Conflict of interest Dr. Xu is an employee of AstraZeneca. The rest of authors report no relationships that could be construed as a conflict of interest. Acknowledgement This study was supported by an institutional grant from AstraZeneca to Karolinska Institutet. In addition, we acknowledge other grant support from the Swedish Heart and Lung Foundation, the Stockholm County Council, Vifor Fresenius Medical Care Renal Pharma, Martin Rind's and Westman's Foundations. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.ijcard.2017.07.035. References [1] C.P. Kovesdy, Management of hyperkalemia: an update for the internist, Am. J. Med. 128 (12) (2015) 1281–1287, http://dx.doi.org/10.1016/j.amjmed.2015.05.040. [2] C.P. Kovesdy, Updates in hyperkalemia: outcomes and therapeutic strategies, Rev. Endocr. Metab. Disord. (September 2016) 1–7, http://dx.doi.org/10.1007/s11154016-9384-x. [3] I. Lazich, G.L. Bakris, Prediction and management of hyperkalemia across the spectrum of chronic kidney disease, Semin. Nephrol. 34 (3) (2014) 333–339, http://dx.doi.org/10.1016/j.semnephrol.2014.04.008. [4] C.B. Salem, A. Badreddine, N. Fathallah, R. Slim, H. Hmouda, Drug-induced hyperkalemia, Drug Saf. 37 (9) (2014) 677–692, http://dx.doi.org/10.1007/ s40264-014-0196-1. [5] B. Pitt, F. Zannad, W.J. Remme, et al., The effect of spironolactone on morbidity and mortality in patients with severe heart failure. Randomized aldactone evaluation study investigators, N. Engl. J. Med. 341 (10) (1999) 709–717, http://dx.doi.org/ 10.1056/NEJM199909023411001. [6] L.C. Reardon, D.S. Macpherson, Hyperkalemia in outpatients using angiotensinconverting enzyme inhibitors. How much should we worry? Arch. Intern. Med. 158 (1) (1998) 26–32. [7] M. Epstein, Hyperkalemia constitutes a constraint for implementing reninangiotensin-aldosterone inhibition: the widening gap between mandated treatment guidelines and the real-world clinical arena, Kidney Int. Suppl. 6 (1) (2016) 20–28, http://dx.doi.org/10.1016/j.kisu.2016.01.004. [8] L.B. Cooper, B.G. Hammill, E.D. Peterson, et al., Characterization of mineralocorticoid receptor antagonist therapy initiation in high-risk patients with heart failure, Circ. Cardiovasc. Qual. Outcomes 10 (1) (2017)http://dx.doi.org/10.1161/CIRCOUTCOMES. 116.002946. [9] L.B. Cooper, B.G. Hammill, E.D. Peterson, et al., Consistency of laboratory monitoring during initiation of mineralocorticoid receptor antagonist therapy in patients with heart failure, JAMA 314 (18) (2015) 1973–1975, http://dx.doi.org/10.1001/jama. 2015.11904. [10] L.A. Allen, S.M. Shetterly, P.N. Peterson, et al., Guideline concordance of testing for hyperkalemia and kidney dysfunction during initiation of mineralocorticoid receptor antagonist therapy in patients with heart failure, Circ. Heart Fail. 7 (1) (2014) 43–50, http://dx.doi.org/10.1161/CIRCHEARTFAILURE.113.000709. [11] L.M. Einhorn, M. Zhan, V.D. Hsu, et al., The frequency of hyperkalemia and its significance in chronic kidney disease, Arch. Intern. Med. 169 (12) (2009) 1156, http://dx.doi.org/10.1001/archinternmed.2009.132. [12] A.R. Chang, Y. Sang, J. Leddy, et al., Antihypertensive medications and the prevalence of hyperkalemia in a large health system, Hypertension 67 (6) (2016) 1181–1188, http://hyper.ahajournals.org/content/67/6/1181.short. [13] B. Runesson, A. Gasparini, A.R. Qureshi, et al., The Stockholm CREAtinine measurements (SCREAM) project: protocol overview and regional representativeness, Clin. Kidney J. 9 (1) (2016) 119–127, http://dx.doi.org/10.1093/ckj/sfv117. [14] M. Rosén, National health data registers: a Nordic heritage to public health, Scand. J. Public Health 30 (2) (2002) 81–85, http://dx.doi.org/10.1080/140349401753683444. [15] B. Wettermark, N. Hammar, C.M. Fored, et al., The new Swedish prescribed drug register–opportunities for pharmacoepidemiological research and experience from the first six months, Pharmacoepidemiol. Drug Saf. 16 (7) (2007) 726–735, http://dx.doi.org/10.1002/pds.1294. [16] M. Evans, M.M. Suttorp, R. Bellocco, et al., Trends in haemoglobin, erythropoietinstimulating agents and iron use in Swedish chronic kidney disease patients between 2008 and 2013, Nephrol. Dial. Transplant. 31 (4) (2016) 628–635, http://dx.doi.org/ 10.1093/ndt/gfv298.
7
[17] A.S. Levey, L.A. Stevens, C.H. Schmid, et al., A new equation to estimate glomerular filtration rate, Ann. Intern. Med. 150 (9) (2009) 604–612http://www.ncbi.nlm.nih. gov/pmc/articles/PMC2763564/. [18] P. Rustad, P. Felding, L. Franzson, et al., The Nordic reference interval project 2000: recommended reference intervals for 25 common biochemical properties, Scand. J. Clin. Lab. Invest. 64 (4) (2004) 271–284, http://dx.doi.org/10.1080/ 00365510410006324. [19] P.A. Sarafidis, R. Blacklock, E. Wood, et al., Prevalence and factors associated with hyperkalemia in predialysis patients followed in a low-clearance clinic, Clin. J. Am. Soc. Nephrol. 7 (8) (2012) 1234–1241, http://dx.doi.org/10.2215/CJN.01150112. [20] J. Hayes, K. Kalantar-Zadeh, J.L. Lu, S. Turban, J.E. Anderson, C.P. Kovesdy, Association of hypo- and hyperkalemia with disease progression and mortality in males with chronic kidney disease: the role of race, Nephron Clin. Pract. 120 (1) (2012) c8–16, http://dx.doi.org/10.1159/000329511. [21] N. Jain, S. Kotla, B.B. Little, et al., Predictors of hyperkalemia and death in patients with cardiac and renal disease, Am. J. Cardiol. 109 (10) (2012) 1510–1513, http://dx.doi.org/10.1016/j.amjcard.2012.01.367. [22] O. Moranne, M. Froissart, J. Rossert, et al., Timing of onset of CKD-related metabolic complications, J. Am. Soc. Nephrol. 20 (1) (2009) 164–171, http://dx.doi.org/10. 1681/ASN.2008020159. [23] J.M. Weinberg, L.J. Appel, G. Bakris, et al., Risk of hyperkalemia in nondiabetic patients with chronic kidney disease receiving antihypertensive therapy, Arch. Intern. Med. 169 (17) (2009) 1587–1594, http://dx.doi.org/10.1001/archinternmed. 2009.284. [24] A. Gasparini, M. Evans, J. Coresh, et al., Prevalence and recognition of chronic kidney disease in stockholm healthcare, Nephrol. Dial. Transplant. 31 (12) (2016) 2086–2094, http://dx.doi.org/10.1093/ndt/gfw354. [25] L.C. Plantinga, L.E. Boulware, J. Coresh, et al., Patient awareness in chronic kidney disease: trends and predictors, Arch. Intern. Med. 168 (20) (2008) 2268–2275, http://dx.doi.org/10.1001/archinte.168.20.2268. [26] M.A. Raebel, M.L. Smith, G. Saylor, et al., The positive predictive value of a hyperkalemia diagnosis in automated health care data, Pharmacoepidemiol. Drug Saf. 19 (11) (2010) 1204–1208, http://dx.doi.org/10.1002/pds.2030. [27] J.L. Fleet, S.Z. Shariff, S. Gandhi, M.A. Weir, A.K. Jain, A.X. Garg, Validity of the international classification of diseases 10th revision code for hyperkalaemia in elderly patients at presentation to an emergency department and at hospital admission, BMJ Open 2 (6) (2012) e002011, http://dx.doi.org/10.1136/bmjopen-2012-002011. [28] J.S. Choi, Y.A. Kim, H.Y. Kim, et al., Relation of serum potassium level to long-term outcomes in patients with acute myocardial infarction, Am. J. Cardiol. 113 (8) (2014) 1285–1290, http://dx.doi.org/10.1016/j.amjcard.2014.01.402. [29] A. Goyal, J.A. Spertus, K. Gosch, et al., Serum potassium levels and mortality in acute myocardial infarction, JAMA 307 (2) (2012) 157–164, http://dx.doi.org/10.1001/ jama.2011.1967. [30] M.L. Krogager, C. Torp-Pedersen, R.N. Mortensen, et al., Short-term mortality risk of serum potassium levels in hypertension: a retrospective analysis of nationwide registry data, Eur. Heart J. 38 (2) (2017) 104–112, http://dx.doi.org/10.1093/ eurheartj/ehw129. [31] J. Tromp, J.M. ter Maaten, K. Damman, et al., Serum potassium levels and outcome in acute heart failure (data from the PROTECT and COACH trials), Am. J. Cardiol. 119 (2) (2017) 290–296, http://dx.doi.org/10.1016/j.amjcard.2016.09.038. [32] J. Luo, S.M. Brunelli, D.E. Jensen, A. Yang, Association between serum potassium and outcomes in patients with reduced kidney function, Clin. J. Am. Soc. Nephrol. 11 (1) (2016) 90–100, http://dx.doi.org/10.2215/CJN.01730215. [33] A.S. Desai, K. Swedberg, J.J. McMurray, et al., Incidence and predictors of hyperkalemia in patients with heart failure, J. Am. Coll. Cardiol. 50 (20) (2007) 1959–1966, http://dx.doi.org/10.1016/j.jacc.2007.07.067. [34] K. Takaichi, F. Takemoto, Y. Ubara, Y. Mori, Analysis of factors causing hyperkalemia, Intern. Med. 46 (12) (2007) 823–829, http://dx.doi.org/10.2169/internalmedicine. 46.6415. [35] E. Espinel, J. Joven, I. Gil, et al., Risk of hyperkalemia in patients with moderate chronic kidney disease initiating angiotensin converting enzyme inhibitors or angiotensin receptor blockers: a randomized study, BMC Res. Notes 6 (2013) 306, http://dx.doi.org/10.1186/1756-0500-6-306. [36] R.S. McKelvie, S. Yusuf, D. Pericak, et al., Comparison of candesartan, enalapril, and their combination in congestive heart failure, Circulation 100 (10) (1999) 1056–1064, http://dx.doi.org/10.1161/01.CIR.100.10.1056. [37] B. Pitt, S.D. Anker, D.A. Bushinsky, et al., Evaluation of the efficacy and safety of RLY5016, a polymeric potassium binder, in a double-blind, placebo-controlled study in patients with chronic heart failure (the PEARL-HF) trial, Eur. Heart J. 32 (7) (2011) 820–828, http://dx.doi.org/10.1093/eurheartj/ehq502. [38] J.M. Buysse, I.-Z. Huang, B. Pitt, PEARL-HF: prevention of hyperkalemia in patients with heart failure using a novel polymeric potassium binder, RLY5016, Futur. Cardiol. 8 (1) (2012) 17–28, http://dx.doi.org/10.2217/fca.11.71. [39] M.R. Weir, G.L. Bakris, D.A. Bushinsky, et al., Patiromer in patients with kidney disease and hyperkalemia receiving RAAS inhibitors, N. Engl. J. Med. 372 (3) (2015) 211–221, http://dx.doi.org/10.1056/NEJMoa1410853. [40] M. Kosiborod, H.S. Rasmussen, P. Lavin, et al., Effect of sodium zirconium cyclosilicate on potassium lowering for 28 days among outpatients with hyperkalemia: the HARMONIZE randomized clinical trial, JAMA 312 (21) (2014) 2223–2233, http://dx.doi.org/10.1001/jama.2014.15688. [41] R.L. Bloomfield, D.J. Wilson, V.M. Buckalew, The incidence of diuretic-induced hypokalemia in two distinct clinic settings, J. Clin. Hypertens. 2 (4) (1986) 331–338. [42] F.J. Gennari, Hypokalemia, N. Engl. J. Med. 339 (7) (1998) 451–458, http://dx.doi. org/10.1056/NEJM199808133390707. [43] I.D. Weiner, C.S. Wingo, Hypokalemia–consequences, causes, and correction, J. Am. Soc. Nephrol. 8 (7) (1997) 1179–1188.
Please cite this article as: E. Nilsson, et al., Incidence and determinants of hyperkalemia and hypokalemia in a large healthcare system, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.07.035
8
E. Nilsson et al. / International Journal of Cardiology xxx (2017) xxx–xxx
[44] O. Paltiel, E. Salakhov, I. Ronen, D. Berg, A. Israeli, Management of severe hypokalemia in hospitalized patients: a study of quality of care based on computerized databases, Arch. Intern. Med. 161 (8) (2001) 1089–1095. [45] J. Pepin, C. Shields, Advances in diagnosis and management of hypokalemic and hyperkalemic emergencies, Emerg. Med. Pract. 14 (2) (2012) 1–17 (quiz 17–18). [46] H.-J. Kim, S.-W. Han, Therapeutic approach to hyperkalemia, Nephron 92 (Suppl. 1) (2002) 33–40 (doi:65375). [47] M. Jessup, W.T. Abraham, D.E. Casey, et al., 2009 focused update: ACCF/AHA guidelines for the diagnosis and management of heart failure in adults, Circulation 119 (14) (2009) 1977–2016, http://dx.doi.org/10.1161/CIRCULATIONAHA.109.192064.
[48] J. McMurray, A. Cohen-Solal, R. Dietz, et al., Practical recommendations for the use of ACE inhibitors, beta-blockers, aldosterone antagonists and angiotensin receptor blockers in heart failure: putting guidelines into practice, Eur. J. Heart Fail. 7 (5) (2005) 710–721, http://dx.doi.org/10.1016/j.ejheart.2005.07.002. [49] A. Levin, P.E. Stevens, R.W. Bilous, et al., Kidney disease: improving global outcomes (KDIGO) CKD work group. KDIGO 2012 clinical practice guideline for the evaluation and management of chronic kidney disease, Kidney Int. Suppl. 3 (1) (2013) 1–150, http://dx.doi.org/10.1038/kisup.2012.73.
Please cite this article as: E. Nilsson, et al., Incidence and determinants of hyperkalemia and hypokalemia in a large healthcare system, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.07.035
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Incidence and determinants of hyperkalemia and hypokalemia in a large healthcare system Erik Nilsson a,b,h,1, Alessandro Gasparini c,1, Johan Ärnlöv d,1, Hairong Xu e,1, Karin M. Henriksson d,1, Josef Coresh f,1, Morgan E. Grams f,1, Juan Jesus Carrero g,⁎,1 a
Division of Renal Medicine, Department of Clinical Science, Intervention and Technology, Karolinska Institutet, Stockholm, Sweden Department of Internal Medicine, School of Medical Sciences, Örebro University, Örebro, Sweden Department of Health Sciences, University of Leicester, United Kingdom d Department of Medical Sciences, Uppsala University Hospital, Uppsala, Sweden e AstraZeneca, Gaithersburg, MD, USA f Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA g Department of Medical Epidemiology and Biostatistics (MEB), Karolinska Institutet, Sweden h Division of Baxter Novum, Department of Clinical Science, Intervention and Technology, Karolinska Institutet, Stockholm, Sweden b c
a r t i c l e
i n f o
Article history: Received 1 May 2017 Received in revised form 4 July 2017 Accepted 12 July 2017 Available online xxxx Keywords: Epidemiology Hyperkalemia Hypokalemia Renin-angiotensin-aldosterone system inhibitors
a b s t r a c t Background: Hypo- and hyperkalemia in clinical settings are insufficiently characterized and large-scale data from Europe lacking. We studied incidence and determinants of these abnormalities in a large Swedish healthcare system. Methods: Observational study from the Stockholm CREAtinine Measurements project, including adult individuals from Stockholm accessing healthcare in 2009 (n = 364,955). Over 3-years, we estimated the incidence of hypokalemia, defined as potassium b 3.5 mmol/L, hyperkalemia, defined as potassium N 5 mmol/L, and moderate/severe hyperkalemia, defined as potassium N 5.5 mmol/L. Kidney function was assessed by estimated glomerular filtration rate (eGFR). Results: Of 364,955 participants, 69.4% had 1–2 potassium tests, 16.7% had 3–4 tests and the remaining 13.9% had N 4 potassium tests/year. Hypokalemia occurred in 49,662 (13.6%) individuals, with 33% recurrence. Hyperkalemia occurred in 25,461 (7%) individuals, with 35.7% recurrence. Moderate/severe hyperkalemia occurred in 9059 (2.5%) with 28% recurrence. The frequency of potassium testing was an important determinant of dyskalemia risk. The incidence proportion of hyperkalemia was higher in the presence of diabetes, lower eGFR, myocardial infarction, heart failure (HF), or use of renin angiotensin-aldosterone system inhibitors (RAASi). In adjusted analyses, women and use of loop/thiazide diuretics were associated with lower hyperkalemia risk. Older age, lower eGFR, diabetes, HF and use of RAASi were associated with higher hyperkalemia risk. On the other hand, women, younger age, higher eGFR and baseline use of diuretics were associated with higher hypokalemia risk. Conclusion: Hypo- and hyperkalemia are common in healthcare. Optimal RAASi and diuretics use and careful potassium monitoring in the presence of certain comorbidities, especially lower eGFR, is advocated. © 2017 The Authors. Published by Elsevier Ireland Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
1. Introduction Maintenance of K+ homeostasis is important for many physiologic processes, such as cardiac electrical conduction and inotropy, smooth muscle tone, neuronal signaling and acid-base balance [1,2]. Potassium is regulated mainly in the tubuli and collecting ducts of the kidney
⁎ Corresponding author at: Department of Medical Epidemiology and Biostatistics (MEB), Karolinska Institutet, Nobels väg 12A, Box 281, 171 77 Stockholm, Sweden. E-mail address: [email protected] (J.J. Carrero). 1 This author takes responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation.
where aldosterone stimulates its excretion. Situations that affect this pathway are thought to increase the risk of dyskalemia, including comorbidities such as chronic kidney disease (CKD), heart failure (HF), cardiovascular disease (CVD) and diabetes mellitus (DM) [3,4], as well as various common medication classes such as inhibitors of the renin-angiotensin aldosterone system (RAASi), beta blockers, calcium channel blockers and diuretics. These comorbidities are often interconnected, and medications affecting potassium homeostasis are commonly used in their treatment. The incidence of dyskalemia in real-world clinical settings is insufficiently characterized and likely not reflected by the controlled environment of randomized controlled trials. The RALES trial, which studied
http://dx.doi.org/10.1016/j.ijcard.2017.07.035 0167-5273/© 2017 The Authors. Published by Elsevier Ireland Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Please cite this article as: E. Nilsson, et al., Incidence and determinants of hyperkalemia and hypokalemia in a large healthcare system, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.07.035
2
E. Nilsson et al. / International Journal of Cardiology xxx (2017) xxx–xxx
effect of spironolactone in heart failure patients on ACEi, reported a 2% rate of severe hyperkalemia in the treatment group [5], but observational reports in the U.S. revealed that 11% of outpatients prescribed with ACEi developed hyperkalemia over a period of two years [6]. Differences in reported rates of dyskalemia may be explained by the inclusion of more heterogeneous populations [7] and diverse clinical practices, such as variations in the frequency of potassium monitoring, which is reported as suboptimal in real-world settings [8–10]. Studies with broader coverage, such as those derived from healthcare records may be relevant to inform clinical practice. Available reports in this regard are scarce and derived from North American materials [6,11,12]. Because of differences in practice patterns, drug usage and healthcare access, comparison with other countries may be useful to detect population segments where more intensive K-monitoring and/ or preventive strategies are needed. In this study, we investigated the frequency, severity and recurrence of hypo- and hyperkalemia in a large Swedish healthcare system as well as comorbidities and medications associated. 2. Methods 2.1. Data sources This project is based on the Stockholm CREAtinine Measurements (SCREAM) project [13], a healthcare-utilization cohort including all residents in the region of Stockholm, Sweden, undertaking at least one measurement of serum creatinine in inpatient or outpatient care during 2006–11. Additional laboratory data, including all serum potassium measurements in the region were extracted. Data was thereafter linked with regional and national administrative databases for information on healthcare utilization (International Classification of Diseases, Tenth Revision [ICD-10] codes and therapeutic procedures [14]), complete information of drugs dispensed at Swedish pharmacies [15], validated renal replacement therapy endpoints [16] and vital status, with no loss to follow up. 2.2. Study population For this study, we included all individuals ≥ 18 years of age who accessed outpatient care and were registered in SCREAM between January 1st, 2009, and December 31st, 2009. Index date was set at January 1st 2009, the point at which the study covariates were calculated. An additional inclusion criterion was the existence of at least one ambulatory measurement of serum creatinine within the preceding year, in order to estimate kidney function. After applying study eligibility criteria, all available potassium measurements during the following three years were extracted until death, migration from the region or end of follow-up (December 31st, 2011). The study was approved by the Regional Ethics Review Board, Stockholm, Sweden. 2.3. Study covariates Inter- as well as intra-laboratory variation in laboratory measurements was considered minimal among the three laboratories providing services to the region, being frequently audited for quality and harmonization by the national organization EQUALIS (www.equalis.se). The serum creatinine measured closest to index date was used to estimate glomerular filtration rate (eGFR). To this end, creatinine measurements in connection with a hospital stay were excluded, as they may represent acute illness rather than stable kidney function. Implausible serum creatinine (b25 and N 1500 μmol/l) values were also excluded. All serum creatinine tests were measured with either enzymatic or corrected Jaffe methods. The 2009 CKD-EPI creatinine-based equation was used to calculate eGFR [17]. In the absence of information on albuminuria we defined eGFR categories rather than CKD stages. eGFR categories were considered as follows: G1–2 = eGFR
≥ 60 ml/min/1.73 m 2 ; G3 = eGFR ≥ 30 and b60 ml/min/ 1.73 m2; G4+ = eGFR b 30 ml/min/1.73 m2, also including individuals undergoing renal replacement therapy (dialysis or transplantation), which was ascertained via linkage with the Swedish Renal Registry (http://www.snronline.se). Other covariates included age, sex, and selected comorbidities based on ICD codes (chronic kidney disease, hypertension, diabetes mellitus [DM], myocardial infarction [MI], heart failure [HF], peripheral vascular disease [PVD] and cerebrovascular disease [CeVD], see Supplemental Table S1) [12,13]. Cardiovascular disease history was defined as any event of MI, HF, PVD and CeVD. To better define diabetes and hypertension, ICD-10 codes were enriched with information on current intake of related medication (purchase of oral antidiabetics, Anatomical Therapeutic Classification [ATC] code A10; purchase of anti-hypertensives, ATC codes C03, C07, C08, C09) up to 6 months before index date. We also extracted, at index date, information on concurrent use of medication that may affect potassium homeostasis. These included non-steroidal anti-inflammatory drugs (NSAID), renin-angiotensinaldosterone system inhibitors (RAASi), angiotensin converting enzyme inhibitors (ACEi), angiotensin receptor inhibitors (ARB), mineralocorticoid receptor antagonists (MRA), beta-blockers, potassium sparing diuretics, thiazide/loop diuretics, and other blood pressure medications (which included calcium channel blockers, central agonists, direct vasodilators, and α-blockers) (see Supplemental Table S2). Direct renin inhibitors were not available in Sweden during the study period. 2.4. Study outcome The study outcomes, determined by potassium laboratory tests and defined from commonly used clinical thresholds, were: a) hyperkalemia defined as potassium N 5 mmol/L; b) moderate/severe hyperkalemia defined as potassium N 5.5 mmol/L and c) hypokalemia, defined as potassium b 3.5 mmol/L. We included all available plasma (reference range 3.5–4.4 mmol/L) and serum (reference range 3.6–4.6 mmol/L) potassium measurements in our healthcare system [18], assessed by potentiometric titration. Potassium values above 10 mmol/L were considered implausible and discarded. When more than two measurements of potassium were available in the same day, we selected their median value. Repeated tests denoting hypo-/hyperkalemia within 7 days from each other were considered as part of the same event and the date of the first abnormal potassium was used. For all outcomes, we also examined the pattern of dyskalemia over the 3-year observation period, defining patterns as never-, transient- (only 1 occurrence), or recurrent (N1 event during observation). 2.5. Statistical methods This study utilizes actual health care data, with an indication behind each available test. Because the incidence and patterns of dyskalemia are dependent on the regularity of potassium measurements, we categorized patients by the following frequencies of testing: never, 1– 2, 3–4 and N 4 measurements of potassium/year, as a proxy for contact with the medical system. We used this categorization as a stratification factor in the comparisons and as a covariate in the models. As a first step, we estimated the incidence proportion and recurrence of dyskalemias over the three-year observation period overall and in suspected populations by age strata, eGFR strata, HF, MI, DM and RAASi use. Secondly, we estimated incidence rates (IR). Adjusted IR were obtained from a zero-inflated negative binomial model and presented fixing the value of the adjustment factors to their mean value. Logistic regression was used to examine baseline patient characteristics and medications associated with dyskalemia incidence; multinomial logistic regression was used to study patterns of dyskalemia. Covariates were selected a priori based on known associations with circulating potassium. These included age, sex, eGFR strata, comorbidities (CKD, hypertension, DM, MI, HF, PVD and CeVD) and medication
Please cite this article as: E. Nilsson, et al., Incidence and determinants of hyperkalemia and hypokalemia in a large healthcare system, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.07.035
E. Nilsson et al. / International Journal of Cardiology xxx (2017) xxx–xxx
use (NSAIDs, ACEi, ARBs, MRAs, beta-blockers, potassium sparing diuretics, thiazide/loop diuretics and other blood pressure medications). The frequency of potassium measurements per year was included as a covariate with the reference category of 1–2 potassium tests. All analyses were performed using R (www.r-project.org) and Stata version 14 (www.stata.com). 3. Results 3.1. Population characteristics There were 498,716 eligible individuals accessing healthcare and enrolled in SCREAM during 2009 [9]. Of those, 24.5% had no potassium test taken during observation (Table 1). These were younger, had lower proportion of comorbidities and more seldom used medications that affect potassium balance. As many as 8.9% of diabetics, 4.5% of HF patients and between 4.5 and 6.0% of patients with CKD did not undergo annual K testing (Supplemental Table S3). The remaining 364,955 participants (75.5%) underwent at least one potassium test and were therefore included in our study, and 91.1% of all potassium tests recorded during this period were performed in plasma. They underwent a median number of 2 measurements/ year (interquartile range 1–5), and their total time at risk was 1,050,392 person years. Among participants with potassium test, 69.4% had 1–2 potassium tests/year, 16.7% had 3–4 tests/year and the remaining 13.9% had N4 potassium tests/year. Across more frequent potassium testing categories, participants were older, more often men, and more often had DM, hypertension, cardiovascular disease and low eGFR. More frequent potassium testing was also associated with increased use of all studied medications. 3.2. Proportion and incidence rates of hyperkalemia Of the 364,955 participants with available potassium measurements, 334,628 (91.7%) were followed for the whole period, 29,292 (8.0%) died before end of follow up and 1035 (0.3%) migrated to other regions of Sweden.
3
As many as 25,461 (6.98%) participants had at least one episode of hyperkalemia (N 5.0 mmol/l) over the 3-year period. Of these, 64.3% experienced only one event (transient hyperkalemia), and 35.7% had recurrent hyperkalemia (N1 event). Analogously, 9059 (2.48%) individuals had at least one episode of moderate/severe hyperkalemia (N5.5 mmol/L). Of these, 72.0% had transient hyperkalemia and 28.0% had recurrence of hyperkalemia. In a sensitivity analysis we calculated the number of events using only plasma potassium tests, observing a similar incidence: 6.67% for K N 5.0 mmol/L and 2.47% for K N 5.5 mmol/L. The incidence proportion of hyperkalemia, was higher among patients with more frequent potassium measurements; 1.6%, 8.4%, and 32.3% experienced potassium N 5 mmol/L among those with 1–2, 3–4 and N4 potassium checks/year, respectively. Likewise, the 3-year risk of moderate/severe hyperkalemia was 0.2%, 1.9%, and 14.3% for those with 1–2, 3–4 and N4 potassium checks/year, respectively. Recurrent hyperkalemia was more common among individuals with N 4 potassium tests/year; For potassium N 5.0 mmol/L, 45.8% of participants with N 4 potassium checks/year had recurrent hyperkalemia, as compared to 25.0% and 12.5% in individuals with 1–2 or 2–4 K measures/year, respectively. The incidence proportion of hyperkalemia was higher among suspected comorbid populations (older age, CKD, HF, MI, DM) and among users of RAASi (Fig. 1), with the same pattern seen across categories of potassium testing frequency (Supplemental Table S4). For instance, 24.0% (range 4.33% to 43.3% depending on the frequency of K testing) of patients with heart failure had at least one potassium N 5.0 mmol/L within 3 years. Individuals with lower kidney function had the highest hyperkalemia incidence proportions: 55.0% and 33.0% of patients with eGFR G4 + had a K N 5 or N 5.5 mmol/L, respectively, during observation. Likewise, 20.3% and 7.64% of patients with eGFR G3 had a K N 5 or N 5.5 mmol/L, respectively. Supplemental Table S5 shows the incidence rates for hyperkalemia for the abovementioned high-risk comorbidities according to RAASi use. In all cases, hyperkalemia incidence was higher when consuming RAASi. Table 2 shows crude and adjusted incidence rates (IR) of hyperkalemia. Hyperkalemia incidence became increasingly more common when potassium was measured more often. The crude overall IR
Table 1 Descriptive characteristics of excluded (without K+ tests) and included participants (with K+ tests). Included participants are further stratified by the annual frequency of potassium testing. Category
Without K+ tests
With K+ tests
1–2 K tests/year
3–4 K tests/year
N4 K test/year
Total N Age, years Gender, male eGFR, ml/min/1.73 m2 eGFR Categories G1–2 (eGFR ≥ 60 ml/min/1.73 m2) G3 (eGFR60–30 ml/min/1.73 m2) G4+ (eGFR b 30 ml/min/1.73 m2) Diabetes mellitus Hypertension Cardiovascular disease Congestive heart failure Myocardial infarction Peripheral vascular disease Cerebrovascular disease RAASi ACEi ARBs MRAs Beta-blockers K-sparing diuretics Loop/thiazide diuretics Other blood pressure medications NSAIDs
118,644 (24.53%) 47.00 (35.17–60.09) 54,207 (45.69%) 99.18 (87.29–111.27) – 115,660 (97.48%) 2755 (2.32%) 229 (0.19%) 5543 (4.67%) 22,540 (19.00%) 4730 (3.99%) 1384 (1.17%) 1257 (1.06%) 911 (0.77%) 2324 (1.96%) 8476 (7.14%) 6111 (5.15%) 5417 (4.57%) 541 (0.46%) 10,110 (8.52%) 638 (0.54%) 3475 (2.93%) 5318 (4.48%) 16,990 (14.32%)
364,955 (75.47%) 63.34 (50.25–74.51) 161,125 (44.15%) 87.12 (71.72–99.61) – 316,847 (86.82%) 42,709 (11.70%) 5399 (1.48%) 57,343 (15.71%) 198,798 (54.47%) 70,282 (19.26%) 29,684 (8.13%) 22,159 (6.07%) 14,716 (4.03%) 30,377 (8.32%) 85,320 (23.38%) 61,776 (16.93%) 52,285 (14.33%) 10,659 (2.92%) 99,911 (27.38%) 12,264 (3.36%) 52,903 (14.50%) 55,879 (15.31%) 65,204 (17.87%)
252,211 (52.15%) 60.17 (46.51–69.92) 108,884 (43.17%) 90.75 (77.55–102.36) – 234,447 (92.96%) 17,256 (6.84%) 508 (0.20%) 31,133 (12.34%) 118,057 (46.81%) 30,464 (12.08%) 9299 (3.69%) 9367 (3.71%) 5537 (2.20%) 13,872 (5.50%) 49,458 (19.61%) 36,203 (14.35%) 31,218 (12.38%) 3705 (1.47%) 55,715 (22.09%) 4339 (1.72%) 22,378 (8.87%) 32,208 (12.77%) 44,592 (17.68%)
61,464 (12.71%) 69.26 (58.17–79.67) 27,143 (44.16%) 80.86 (64.53–93.83) – 49,355 (80.30%) 11,207 (18.23%) 902 (1.47%) 13,167 (21.42%) 41,887 (68.15%) 17,309 (28.16%) 7360 (11.97%) 5362 (8.72%) 3516 (5.72%) 7403 (12.04%) 18,138 (29.51%) 12,953 (21.07%) 11,257 (18.31%) 2652 (4.31%) 22,106 (35.97%) 3123 (5.08%) 12,781 (20.79%) 12,176 (19.81%) 11,908 (19.37%)
51,280 (10.60%) 74.67 (63.51–83.51) 25,098 (48.94%) 71.13 (51.47–87.94) – 33,045 (64.44%) 14,246 (27.78%) 3989 (7.78%) 13,043 (25.43%) 38,854 (75.77%) 22,509 (43.89%) 13,025 (25.40%) 7430 (14.49%) 5663 (11.04%) 9102 (17.75%) 17,724 (34.56%) 12,620 (24.61%) 9810 (19.13%) 4302 (8.39%) 22,090 (43.08%) 4802 (9.36%) 17,744 (34.60%) 11,495 (22.42%) 8704 (16.97%)
Values are presented as median and inter-quartile interval [IQI], or number and percentage. Abbreviations: eGFR, estimated glomerular filtration rate; RAASi, renin-angiotensinaldosterone system inhibitors; ACEi, angiotensin converting enzyme inhibitors; ARB, angiotensin receptor inhibitors; MRA, mineralocorticoid receptor antagonists; NSAID, non-steroidal anti-inflammatory drugs.
Please cite this article as: E. Nilsson, et al., Incidence and determinants of hyperkalemia and hypokalemia in a large healthcare system, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.07.035
4
E. Nilsson et al. / International Journal of Cardiology xxx (2017) xxx–xxx
Fig. 1. Incidence proportion of hyperkalemia over three years for selected comorbidities and use of RAASi at index date. Abbreviations: eGFR, estimated glomerular filtration rate; RAASi, renin-angiotensin-aldosterone-system inhibitors.
for mild hyperkalemia (K N 5 mmol/L) was 50.0 (95% CI 49.5–50.4) per 1000 person years. Crude and adjusted IR increased in the presence of older age, male sex, worse kidney function categories, comorbidities
and RAASi use. Participants within the lowest eGFR stratum (G4 +) experienced the highest hyperkalemia incidence (adjusted IR 133.5, 95%CI 123.0–143.8 events per 1000 person years).
Table 2 Crude and adjusted incidence rates of hyperkalemia (and 95% confidence intervals) by risk factors, per 1000 person years. Category Overall All Age category (years) 18–44 45–64 65–74 ≥75 eGFR category G1–2 G3 G4+ Heart failure No Yes Myocardial infarction No Yes Diabetes mellitus No Yes RAASi No Yes Frequency of potassium tests per year 1–2 3–4 N4
Crude IR K N 5.0 mmol/L
Adjusted IR K N 5.0 mmol/L
Crude IR K N 5.5 mmol/L
Adjusted IR K N 5.5 mmol/L
49.9 (49.5–50.3)
−
14.6 (14.4–14.9)
−
15.6 (15.1–16.2) 33.1 (32.5–33.7) 57.2 (56.2–58.2) 99.8 (98.5–101.1)
8.4 (7.8–9.0) 12.4 (11.9–12.8) 14.6 (14.0–15.2) 15.4 (14.8–16.1)
5.7 (5.4–6.1) 10.4 (10.1–10.8) 16.4 (15.9–17.0) 27.3 (26.7–28.0)
2.0 (1.7–2.2) 2.2 (2.0–2.3) 2.2 (2.0–2.4) 1.9 (1.8–2.1)
22.6 (22.3–23.0) 156.3 (154.0–158.6) 1070.4 (1052.7–1088.5)
10.8 (10.5–11.1) 30.5 (29.1–31.9) 133.5 (123.3–143.8)
5.5 (5.4–5.7) 42.9 (41.7–44.1) 418.8 (407.7–430.2)
1.8 (1.6–1.9) 5.1 (4.6–5.5) 27.5 (24.1–30.9)
37.5 (37.1–37.9) 211.9 (208.6–215.2)
12.3 (12.0–12.6) 18.7 (17.6–19.8)
10.6 (10.4–10.8) 66.9 (65.0–68.7)
2.0 (1.9–2.1) 3.0 (2.7–3.3)
43.1 (42.7–43.5) 164.1 (160.9–167.4)
12.6 (12.3–13.0) 14.6 (13.6–15.6)
12.5 (12.3–12.8) 49.5 (47.7–51.3)
2.1 (1.9–2.2) 2.2 (1.9–2.4)
36.6 (36.2–37.0) 123.1 (121.4–124.8)
11.3 (11.0–11.6) 23.3 (22.3–24.2)
10.6 (10.4–10.8) 37.0 (36.0–37.9)
1.9 (1.7–2.0) 3.7 (3.4–4.0)
34.2 (33.8–34.7) 102.2 (101.0–103.5)
11.75 (11.42–12.07) 16.80 (16.13–17.47)
9.8 (9.6–10.0) 30.7 (30.1–31.5)
1.9 (1.8–2.0) 2.8 (2.6–3.1)
5.6 (5.52–5.8) 41.3 (40.3–42.2) 334.4 (331.2–337.7)
− − −
0.78 (0.72–0.84) 7.5 (7.1–7.9) 110.5 (108.6–112.4)
− − −
Analysis was done in patients with at least one potassium measurement. Adjusted incident rates are obtained from a zero-inflated negative binomial model, and are adjusted (when applicable) by age categories, sex, renal function categories, history of heart failure, myocardial infarction, hypertension, diabetes mellitus, use of RAASi, and frequency of K+ tests per year. Adjusted incident rates are presented fixing the value of the adjustment factors to their mean value. Abbreviations: eGFR, estimated glomerular filtration rate.
Please cite this article as: E. Nilsson, et al., Incidence and determinants of hyperkalemia and hypokalemia in a large healthcare system, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.07.035
E. Nilsson et al. / International Journal of Cardiology xxx (2017) xxx–xxx
The crude overall IR for moderate/severe hyperkalemia (K N 5.5 mmol/L) was 14.7 (95% CI 14.4–14.9) per 1000 person years. Again, IR increased across older age or worsening kidney function strata and in the presence of other comorbidities or RAASi use. Again, participants within the lowest kidney function stratum experienced the highest IR. Adjusted IR showed the same increasing trend across worsening kidney function strata or diabetes, but did not dramatically differ across age strata or in the presence of other comorbidities and RAASi. 3.3. Proportion and incidence rates of hypokalemia Hypokalemia occurred in 49,662 (13.6%) individuals, and 33% of those had recurrent hypokalemia. Hypokalemia incidence proportion over 3 years is shown in Supplemental Table S6. Again, the incidence proportion of hypokalemia was higher among patients with more frequent potassium measurements and among suspected comorbid populations (older age, CKD, HF), but did not importantly differ among individuals with MI, DM, or users of RAASi. The crude incidence rate of hypokalemia was 61.3 (60.9–61.7) per 1000 person years. Hypokalemia incidence became increasingly more common when potassium was measured more often; crude IR increased for instance in the presence of older age (from 68.9 (67.7–70.1) per 1000 person years in individuals aged 18–44 years to 110.9 (109.6–112.3) per 1000 person years in individuals aged N 75 years), male sex, worse kidney function categories (from 74.8 (74.3–75.4) per 1000 person years in individuals with eGFR G1–2 to 269.8 (260.9–278.9) per 1000 person years in individuals with eGFR G4+) or heart failure (76.6 (76.1–77.2) per 1000 person years in non-HF individuals versus 145.6 (142.9–148.4) per 1000 person years in HF individuals). 3.4. Baseline characteristics associated with dyskalemias In adjusted analysis (Table 3), older age categories, male sex, worse kidney function, diabetes, hypertension, heart failure and peripheral
5
vascular disease were all associated with higher hyperkalemia risk. Use of ACE, ARBs or MRAs were also associated with increased risk. These predictors were consistent risk factors for hyperkalemia of all patterns (single/recurrent, Supplemental Tables S7 and S8) and severity (K N 5 or N 5.5 mmol). Whereas use of beta-blockers associated with higher risk of K N 5.0 mmol/L, and use of Loop/Thiazide diuretics or other blood pressure medication associated with a reduced risk, these medications did not significantly associate with K N 5.5 mmol/L (Table 3). Further, people with CeVD were less likely to experience K N 5.5 mmol/L, but this was not the case for K N 5 mmol/L. Predictors of single and recurrent K N 5.5 mmol/L differed slightly (Table S8): whereas older age associated with the risk of single K N 5.5 mmol/L, recurrence was associated with younger age categories. Use of loop/thiazide diuretics was associated with a lower risk of single K N 5.5 mmol/L, but with a higher risk of recurrence (Table S5). Predictors of hypokalemia are also depicted in Table 3. Decreased risk of hypokalemia was seen with baseline older age, lower eGFR, presence of comorbidities (DM, HF, MI, PVD) as well as with use of ACEi, ARB, MRA, beta blocker and NSAID. Hypokalemia risk was higher among women, patients with hypertension or CVD, and especially among those treated with loop/thiazide diuretics or other blood pressure medication. The predictors of single/recurrent hypokalemia did not vary (data not shown). 3.5. Clinical case: heart failure patients To test the consistency of our findings, hyperkalemia incidence proportion and determinants was studied separately in the 29,684 adult patients with HF and at least one K measurement during the study period (Supplemental Tables S9 and S10). Similar to the main overall analysis, HF patients with more frequent K testing tended to be older, more often men and with a higher number of comorbidities and medications affecting K balance. The multivariable predictors of hyperkalemia and hypokalemia were also similar to the main analysis (Supplemental Table S10), but in some cases with broader confidence
Table 3 Baseline risk factors for hyperkalemia and hypokalemia occurrence over 3 years. Baseline risk factors
Age: 18–44 years Age: 45–64 years Age: 65–74 years Age: ≥75 years Gender: female eGFR: G1–2 eGFR: G3 eGFR: G4+ Diabetes mellitus Hypertension Heart failure Myocardial infarction Peripheral vascular disease Cerebrovascular disease ACEi ARBs MRAs Beta-blockers NSAIDs Loop/thiazide diuretics K-sparing diuretics Other blood pressure medications Frequency of K measurements: 1–2 per year Frequency of K measurements: 3–4 per year Frequency of K measurements: N4 per year
Odds ratio (95% confidence interval) K N 5.0 mmol/L
K N 5.5 mmol/L
K b 3.5 mmol/L
Ref. 1.52 (1.42–1.63) 1.82 (1.69–1.95) 1.87 (1.74–2.01) 0.79 (0.77–0.81) Ref. 2.14 (2.07–2.22) 5.60 (5.24–5.98) 1.73 (1.67–1.79) 1.05 (1.00–1.10) 1.14 (1.09–1.19) 1.03 (0.98–1.08) 1.22 (1.15–1.28) 0.97 (0.93–1.01) 1.51 (1.46–1.57) 1.21 (1.17–1.26) 1.66 (1.41–1.95) 1.06 (1.03–1.10) 1.00 (0.96–1.04) 0.94 (0.90–0.97) 1.08 (0.92–1.26) 0.88 (0.85–0.91) Ref. 4.17 (3.99–4.36) 17.26 (16.58–17.97)
Ref. 1.17 (1.04–1.31) 1.22 (1.09–1.37) 1.07 (0.95–1.19) 0.79 (0.76–0.83) Ref. 2.25 (2.12–2.38) 6.39 (5.93–6.89) 1.62 (1.54–1.71) 1.10 (1.02–1.19) 1.14 (1.07–1.22) 0.98 (0.91–1.05) 1.15 (1.07–1.23) 0.92 (0.87–0.98) 1.57 (1.49–1.66) 1.22 (1.15–1.29) 1.44 (1.13–1.82) 1.00 (0.94–1.05) 0.94 (0.89–1.01) 1.01 (0.95–1.07) 1.07 (0.85–1.35) 0.96 (0.91–1.01) Ref. 6.31 (5.70–7.00) 40.13 (36.65–43.94)
Ref. 0.59 (0.57–0.61) 0.49 (0.48–0.51) 0.43 (0.42–0.45) 1.72 (1.68–1.75) Ref. 0.68 (0.66–0.71) 0.54 (0.51–0.58) 0.68 (0.66–0.70) 1.80 (1.74–1.86) 0.84 (0.80–0.87) 0.82 (0.79–0.86) 0.87 (0.83–0.91) 1.06 (1.02–1.09) 0.71 (0.69–0.73) 0.94 (0.91–0.97) 0.70 (0.61–0.80) 0.89 (0.86–0.91) 0.99 (0.96–1.02) 1.76 (1.71–1.81) 1.07 (0.95–1.21) 1.37 (1.33–1.40) Ref. 3.74 (3.64–3.84) 8.63 (8.39–8.87)
Shown are adjusted multivariable odds ratios derived from separate logistic regression models. Statistically significant associations (P b 0.05) are marked in bold. Adjustment was for age category, female, eGFR category, diabetes mellitus, hypertension, chronic heart failure, myocardial infarction, peripheral vascular disease, cerebrovascular disease, ace inhibitor, angiotensin receptor inhibitor, mineralocorticoid receptor blocker, beta blocker, non-steroidal anti-inflammatory drug use, loop- or thiazide diuretic, potassium sparing diuretic, other blood pressure medication, potassium samplings per year. Abbreviations: eGFR, estimated glomerular filtration rate; ACEi, angiotensin converting enzyme inhibitors; ARB, angiotensin receptor inhibitors; MRA, mineralocorticoid receptor antagonists; NSAID, non-steroidal anti-inflammatory drugs.
Please cite this article as: E. Nilsson, et al., Incidence and determinants of hyperkalemia and hypokalemia in a large healthcare system, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.07.035
6
E. Nilsson et al. / International Journal of Cardiology xxx (2017) xxx–xxx
intervals. Age was not a risk factor for moderate/severe HK among HF patients, perhaps explained by the older age of this population (median age 79.84 (interquartile range 70.59–86.34) years). 4. Discussion In this study we comprehensively evaluate the frequency of potassium testing in the large healthcare system of Stockholm, Sweden, and analyze with precision the rates and likelihood of recurrence of dyskalemias by clinical screening over a 3-year period. We demonstrate that hypo- and hyperkalemia is common in Swedish healthcare and that incidence figures to a large extent are influenced by the frequency of testing. Hyperkalemia was particularly common among those with comorbidities and, notably, among those with reduced kidney function. We also demonstrate that even after adjusting for these comorbidities, the use of RAASi was strongly associated with hyperkalemia occurrence. Hypokalemia was also common, especially in diuretic users. As expected, associations between risk factors and hypokalemia were generally in the opposite direction of the associations between risk factors and hyperkalemia. Strengths of our analysis include the complete regional capture of healthcare use in a country with universal healthcare access. The ascertainment of kidney function by eGFR in our analysis is a strength, as kidney dysfunction is one of the strongest risk factors for elevated potassium levels [19–23], but this condition is generally affected by poor awareness and underutilization of ICD diagnoses in healthcare [24,25]. Likewise, the use of potassium tests as the study exposure allows us to quantify the burden of mild and moderate hyperkalemia in the outpatient setting. These are clinical events otherwise not captured by ICD diagnoses [26,27], but that have been repeatedly associated with increased mortality risk in various settings, including acute MI [28,29], hypertension [30], HF [31] and CKD [32]. Evaluating the frequency of dyskalemia events in healthcare is important, as the reported incidence in controlled RCTs clearly does not match that of the real-world clinical setting. In our study, approximately 7.0% and 2.5% of observed individuals experienced mild and moderate/severe hyperkalemia, respectively, over a three-year period. These incidence proportions are comparable to those in US healthcare users undergoing blood pressure testing (10.8% and 2.3% three-year incidence proportion for K N 5.0 and K N 5.5, respectively [12]). In another large study of US healthcare users with at least one inpatient hospital visit, 13.7% had hyperkalemia ≥ 5.5 mmol/L within one year [11]. Characterizing hyperkalemia risk in different high-risk groups highlights the need for hyperkalemia-preventive strategies in these conditions. Abnormalities in potassium balance are attributable to multiple conditions affecting potassium intake, distribution and excretion, including demographic characteristics, comorbidities and medications [3,4]. Beyond the importance of underlying kidney function discussed above, our multivariate risk analyses confirm the diversity of the risk factors involved in this condition, which include older age, male sex and important comorbidities such as HF, diabetes, hypertension or CVD [11,21,22,33,34]. The use of RAASi emerged as another notable risk predictor per se (with an overall risk increase of 43.0%) and in addition to each studied comorbidity (Supplemental Table S5). Disentangling the risk attributed to medications from that of comorbidities is neither possible in our observational analysis nor in real life, as these comorbidities and medications are intricately connected. Interestingly, we found in our multivariate analysis that the hyperkalemia risk associated with RAASi use varies by medication class. Specifically, and alike a previous report [12], we found that the hyperkalemia risk associated to ACEi or MRA were similar, and higher than that of the risk associated to ARB or potassium-sparing diuretics. Our estimates are therefore in line with some [35,36], but not all [23] trials suggesting that ACEi have stronger effect in raising potassium than ARBs. Recent RCTs suggest that the use of potassium binder medications may allow the use of RAASi in patients prone to hyperkalemia [37–40], but the
long-term benefit of such strategies on clinical outcomes is yet to be demonstrated. In the meantime, we speculate that switching medications (for instance, from ACEi to ARBs) may be a possible hyperkalemia management strategy to consider. We found an even higher incidence proportion of hypokalemia (13.6%) than for hyperkalemia. Associations between risk factors and hypokalemia were generally in the opposite direction of the associations between risk factors and hyperkalemia. In addition, diuretics appeared as the strongest baseline risk factor, in agreement with the current understanding of this iatrogenically induced electrolyte abnormality [41–43]. According to previous studies, up to 56% of individuals on diuretics are likely to develop hypokalemia, although it is clinically relevant only in 4% to 5% of these cases [44–46]. Occurrence of dyskalemias was, not surprisingly, dependent on the frequency of potassium testing, and individuals with frequent potassium monitoring were found to have more comorbid conditions and were at higher dyskalemia risk. Such findings are in line with a recent U.S. analysis [12], and overall justify our decision to include frequency of testing as a covariate in our adjustments. Potassium monitoring is, after all, performed in clinical practice by judgement of the clinician, which is affected by the patient's condition, the perceived dyskalemia risk and variations in clinical practice. To our knowledge, no recommendations exist on how often to monitor for potassium in the ambulatory setting, or after the occurrence of dyskalemia. We observe that approximately 9% of patients with diabetes, 5% of patients with HF or CKD, all at high-risk of hyperkalemia did not undergo annual K screening. Current guidelines recommend to closely monitoring potassium during the initiation of RAASi therapy [47–49], and this observation is in line with real-world data suggesting that adequate monitoring is not taking place [8–10]. An additional interesting aspect of our analysis pertains to the recurrence of hyperkalemia: one in three patients experiencing mild or moderate/severe hyperkalemia developed a second episode. While some studies have addressed this issue in different populations and/or with different definitions for recurrence, they find comparable proportions [11,12]. Our multivariate analysis identifies similar baseline risk factors (old age, male sex, low eGFR, DM, HT, HF, CVD, RAASi and frequency of measurement) for both hyperkalemia occurrence and recurrence, with the exception of higher odds for hyperkalemia recurrence among younger age categories. We speculate that this may be attributed to higher risk of mortality among the elderly with hyperkalemia or more frequent discontinuation of potassium-sparing medications after a single hyperkalemia event across older age categories. Limitations in the interpretation of our study are its retrospective nature and the lack of information on other suggested risk factors for hyperkalemia, such as blood pressure or body mass index. Despite our attempts to identify erroneous values, false dyskalemias may have occurred and result in overestimation of true incidences. Data reflects routine care in the region of Stockholm, and findings may not necessarily extrapolate to other areas. Finally, as any observational study, we are impacted by residual confounding due to unmeasured/undetected factors and by confounding by indication bias, making impossible to separate, for instance, the risk of dyskalemia associated with medication use from risk due to the underlying condition that prompted such prescription. 5. Conclusion Both hypo- and hyperkalemia are frequent events among Swedish healthcare users. We found substantial variability in the frequency of potassium monitoring, but patients with older age and greater number of comorbidities (especially CKD), were consistently associated with higher dyskalemia occurrence, recurrence and event severity. These event rates increased in the presence of medication use that affects potassium balance, namely RAASi and diuretics. Further studies
Please cite this article as: E. Nilsson, et al., Incidence and determinants of hyperkalemia and hypokalemia in a large healthcare system, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.07.035
E. Nilsson et al. / International Journal of Cardiology xxx (2017) xxx–xxx
are needed to optimize potassium monitoring strategies to detect dyskalemias and to allow the implementation of preventive/corrective strategies. Conflict of interest Dr. Xu is an employee of AstraZeneca. The rest of authors report no relationships that could be construed as a conflict of interest. Acknowledgement This study was supported by an institutional grant from AstraZeneca to Karolinska Institutet. In addition, we acknowledge other grant support from the Swedish Heart and Lung Foundation, the Stockholm County Council, Vifor Fresenius Medical Care Renal Pharma, Martin Rind's and Westman's Foundations. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.ijcard.2017.07.035. References [1] C.P. Kovesdy, Management of hyperkalemia: an update for the internist, Am. J. Med. 128 (12) (2015) 1281–1287, http://dx.doi.org/10.1016/j.amjmed.2015.05.040. [2] C.P. Kovesdy, Updates in hyperkalemia: outcomes and therapeutic strategies, Rev. Endocr. Metab. Disord. (September 2016) 1–7, http://dx.doi.org/10.1007/s11154016-9384-x. [3] I. Lazich, G.L. Bakris, Prediction and management of hyperkalemia across the spectrum of chronic kidney disease, Semin. Nephrol. 34 (3) (2014) 333–339, http://dx.doi.org/10.1016/j.semnephrol.2014.04.008. [4] C.B. Salem, A. Badreddine, N. Fathallah, R. Slim, H. Hmouda, Drug-induced hyperkalemia, Drug Saf. 37 (9) (2014) 677–692, http://dx.doi.org/10.1007/ s40264-014-0196-1. [5] B. Pitt, F. Zannad, W.J. Remme, et al., The effect of spironolactone on morbidity and mortality in patients with severe heart failure. Randomized aldactone evaluation study investigators, N. Engl. J. Med. 341 (10) (1999) 709–717, http://dx.doi.org/ 10.1056/NEJM199909023411001. [6] L.C. Reardon, D.S. Macpherson, Hyperkalemia in outpatients using angiotensinconverting enzyme inhibitors. How much should we worry? Arch. Intern. Med. 158 (1) (1998) 26–32. [7] M. Epstein, Hyperkalemia constitutes a constraint for implementing reninangiotensin-aldosterone inhibition: the widening gap between mandated treatment guidelines and the real-world clinical arena, Kidney Int. Suppl. 6 (1) (2016) 20–28, http://dx.doi.org/10.1016/j.kisu.2016.01.004. [8] L.B. Cooper, B.G. Hammill, E.D. Peterson, et al., Characterization of mineralocorticoid receptor antagonist therapy initiation in high-risk patients with heart failure, Circ. Cardiovasc. Qual. Outcomes 10 (1) (2017)http://dx.doi.org/10.1161/CIRCOUTCOMES. 116.002946. [9] L.B. Cooper, B.G. Hammill, E.D. Peterson, et al., Consistency of laboratory monitoring during initiation of mineralocorticoid receptor antagonist therapy in patients with heart failure, JAMA 314 (18) (2015) 1973–1975, http://dx.doi.org/10.1001/jama. 2015.11904. [10] L.A. Allen, S.M. Shetterly, P.N. Peterson, et al., Guideline concordance of testing for hyperkalemia and kidney dysfunction during initiation of mineralocorticoid receptor antagonist therapy in patients with heart failure, Circ. Heart Fail. 7 (1) (2014) 43–50, http://dx.doi.org/10.1161/CIRCHEARTFAILURE.113.000709. [11] L.M. Einhorn, M. Zhan, V.D. Hsu, et al., The frequency of hyperkalemia and its significance in chronic kidney disease, Arch. Intern. Med. 169 (12) (2009) 1156, http://dx.doi.org/10.1001/archinternmed.2009.132. [12] A.R. Chang, Y. Sang, J. Leddy, et al., Antihypertensive medications and the prevalence of hyperkalemia in a large health system, Hypertension 67 (6) (2016) 1181–1188, http://hyper.ahajournals.org/content/67/6/1181.short. [13] B. Runesson, A. Gasparini, A.R. Qureshi, et al., The Stockholm CREAtinine measurements (SCREAM) project: protocol overview and regional representativeness, Clin. Kidney J. 9 (1) (2016) 119–127, http://dx.doi.org/10.1093/ckj/sfv117. [14] M. Rosén, National health data registers: a Nordic heritage to public health, Scand. J. Public Health 30 (2) (2002) 81–85, http://dx.doi.org/10.1080/140349401753683444. [15] B. Wettermark, N. Hammar, C.M. Fored, et al., The new Swedish prescribed drug register–opportunities for pharmacoepidemiological research and experience from the first six months, Pharmacoepidemiol. Drug Saf. 16 (7) (2007) 726–735, http://dx.doi.org/10.1002/pds.1294. [16] M. Evans, M.M. Suttorp, R. Bellocco, et al., Trends in haemoglobin, erythropoietinstimulating agents and iron use in Swedish chronic kidney disease patients between 2008 and 2013, Nephrol. Dial. Transplant. 31 (4) (2016) 628–635, http://dx.doi.org/ 10.1093/ndt/gfv298.
7
[17] A.S. Levey, L.A. Stevens, C.H. Schmid, et al., A new equation to estimate glomerular filtration rate, Ann. Intern. Med. 150 (9) (2009) 604–612http://www.ncbi.nlm.nih. gov/pmc/articles/PMC2763564/. [18] P. Rustad, P. Felding, L. Franzson, et al., The Nordic reference interval project 2000: recommended reference intervals for 25 common biochemical properties, Scand. J. Clin. Lab. Invest. 64 (4) (2004) 271–284, http://dx.doi.org/10.1080/ 00365510410006324. [19] P.A. Sarafidis, R. Blacklock, E. Wood, et al., Prevalence and factors associated with hyperkalemia in predialysis patients followed in a low-clearance clinic, Clin. J. Am. Soc. Nephrol. 7 (8) (2012) 1234–1241, http://dx.doi.org/10.2215/CJN.01150112. [20] J. Hayes, K. Kalantar-Zadeh, J.L. Lu, S. Turban, J.E. Anderson, C.P. Kovesdy, Association of hypo- and hyperkalemia with disease progression and mortality in males with chronic kidney disease: the role of race, Nephron Clin. Pract. 120 (1) (2012) c8–16, http://dx.doi.org/10.1159/000329511. [21] N. Jain, S. Kotla, B.B. Little, et al., Predictors of hyperkalemia and death in patients with cardiac and renal disease, Am. J. Cardiol. 109 (10) (2012) 1510–1513, http://dx.doi.org/10.1016/j.amjcard.2012.01.367. [22] O. Moranne, M. Froissart, J. Rossert, et al., Timing of onset of CKD-related metabolic complications, J. Am. Soc. Nephrol. 20 (1) (2009) 164–171, http://dx.doi.org/10. 1681/ASN.2008020159. [23] J.M. Weinberg, L.J. Appel, G. Bakris, et al., Risk of hyperkalemia in nondiabetic patients with chronic kidney disease receiving antihypertensive therapy, Arch. Intern. Med. 169 (17) (2009) 1587–1594, http://dx.doi.org/10.1001/archinternmed. 2009.284. [24] A. Gasparini, M. Evans, J. Coresh, et al., Prevalence and recognition of chronic kidney disease in stockholm healthcare, Nephrol. Dial. Transplant. 31 (12) (2016) 2086–2094, http://dx.doi.org/10.1093/ndt/gfw354. [25] L.C. Plantinga, L.E. Boulware, J. Coresh, et al., Patient awareness in chronic kidney disease: trends and predictors, Arch. Intern. Med. 168 (20) (2008) 2268–2275, http://dx.doi.org/10.1001/archinte.168.20.2268. [26] M.A. Raebel, M.L. Smith, G. Saylor, et al., The positive predictive value of a hyperkalemia diagnosis in automated health care data, Pharmacoepidemiol. Drug Saf. 19 (11) (2010) 1204–1208, http://dx.doi.org/10.1002/pds.2030. [27] J.L. Fleet, S.Z. Shariff, S. Gandhi, M.A. Weir, A.K. Jain, A.X. Garg, Validity of the international classification of diseases 10th revision code for hyperkalaemia in elderly patients at presentation to an emergency department and at hospital admission, BMJ Open 2 (6) (2012) e002011, http://dx.doi.org/10.1136/bmjopen-2012-002011. [28] J.S. Choi, Y.A. Kim, H.Y. Kim, et al., Relation of serum potassium level to long-term outcomes in patients with acute myocardial infarction, Am. J. Cardiol. 113 (8) (2014) 1285–1290, http://dx.doi.org/10.1016/j.amjcard.2014.01.402. [29] A. Goyal, J.A. Spertus, K. Gosch, et al., Serum potassium levels and mortality in acute myocardial infarction, JAMA 307 (2) (2012) 157–164, http://dx.doi.org/10.1001/ jama.2011.1967. [30] M.L. Krogager, C. Torp-Pedersen, R.N. Mortensen, et al., Short-term mortality risk of serum potassium levels in hypertension: a retrospective analysis of nationwide registry data, Eur. Heart J. 38 (2) (2017) 104–112, http://dx.doi.org/10.1093/ eurheartj/ehw129. [31] J. Tromp, J.M. ter Maaten, K. Damman, et al., Serum potassium levels and outcome in acute heart failure (data from the PROTECT and COACH trials), Am. J. Cardiol. 119 (2) (2017) 290–296, http://dx.doi.org/10.1016/j.amjcard.2016.09.038. [32] J. Luo, S.M. Brunelli, D.E. Jensen, A. Yang, Association between serum potassium and outcomes in patients with reduced kidney function, Clin. J. Am. Soc. Nephrol. 11 (1) (2016) 90–100, http://dx.doi.org/10.2215/CJN.01730215. [33] A.S. Desai, K. Swedberg, J.J. McMurray, et al., Incidence and predictors of hyperkalemia in patients with heart failure, J. Am. Coll. Cardiol. 50 (20) (2007) 1959–1966, http://dx.doi.org/10.1016/j.jacc.2007.07.067. [34] K. Takaichi, F. Takemoto, Y. Ubara, Y. Mori, Analysis of factors causing hyperkalemia, Intern. Med. 46 (12) (2007) 823–829, http://dx.doi.org/10.2169/internalmedicine. 46.6415. [35] E. Espinel, J. Joven, I. Gil, et al., Risk of hyperkalemia in patients with moderate chronic kidney disease initiating angiotensin converting enzyme inhibitors or angiotensin receptor blockers: a randomized study, BMC Res. Notes 6 (2013) 306, http://dx.doi.org/10.1186/1756-0500-6-306. [36] R.S. McKelvie, S. Yusuf, D. Pericak, et al., Comparison of candesartan, enalapril, and their combination in congestive heart failure, Circulation 100 (10) (1999) 1056–1064, http://dx.doi.org/10.1161/01.CIR.100.10.1056. [37] B. Pitt, S.D. Anker, D.A. Bushinsky, et al., Evaluation of the efficacy and safety of RLY5016, a polymeric potassium binder, in a double-blind, placebo-controlled study in patients with chronic heart failure (the PEARL-HF) trial, Eur. Heart J. 32 (7) (2011) 820–828, http://dx.doi.org/10.1093/eurheartj/ehq502. [38] J.M. Buysse, I.-Z. Huang, B. Pitt, PEARL-HF: prevention of hyperkalemia in patients with heart failure using a novel polymeric potassium binder, RLY5016, Futur. Cardiol. 8 (1) (2012) 17–28, http://dx.doi.org/10.2217/fca.11.71. [39] M.R. Weir, G.L. Bakris, D.A. Bushinsky, et al., Patiromer in patients with kidney disease and hyperkalemia receiving RAAS inhibitors, N. Engl. J. Med. 372 (3) (2015) 211–221, http://dx.doi.org/10.1056/NEJMoa1410853. [40] M. Kosiborod, H.S. Rasmussen, P. Lavin, et al., Effect of sodium zirconium cyclosilicate on potassium lowering for 28 days among outpatients with hyperkalemia: the HARMONIZE randomized clinical trial, JAMA 312 (21) (2014) 2223–2233, http://dx.doi.org/10.1001/jama.2014.15688. [41] R.L. Bloomfield, D.J. Wilson, V.M. Buckalew, The incidence of diuretic-induced hypokalemia in two distinct clinic settings, J. Clin. Hypertens. 2 (4) (1986) 331–338. [42] F.J. Gennari, Hypokalemia, N. Engl. J. Med. 339 (7) (1998) 451–458, http://dx.doi. org/10.1056/NEJM199808133390707. [43] I.D. Weiner, C.S. Wingo, Hypokalemia–consequences, causes, and correction, J. Am. Soc. Nephrol. 8 (7) (1997) 1179–1188.
Please cite this article as: E. Nilsson, et al., Incidence and determinants of hyperkalemia and hypokalemia in a large healthcare system, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.07.035
8
E. Nilsson et al. / International Journal of Cardiology xxx (2017) xxx–xxx
[44] O. Paltiel, E. Salakhov, I. Ronen, D. Berg, A. Israeli, Management of severe hypokalemia in hospitalized patients: a study of quality of care based on computerized databases, Arch. Intern. Med. 161 (8) (2001) 1089–1095. [45] J. Pepin, C. Shields, Advances in diagnosis and management of hypokalemic and hyperkalemic emergencies, Emerg. Med. Pract. 14 (2) (2012) 1–17 (quiz 17–18). [46] H.-J. Kim, S.-W. Han, Therapeutic approach to hyperkalemia, Nephron 92 (Suppl. 1) (2002) 33–40 (doi:65375). [47] M. Jessup, W.T. Abraham, D.E. Casey, et al., 2009 focused update: ACCF/AHA guidelines for the diagnosis and management of heart failure in adults, Circulation 119 (14) (2009) 1977–2016, http://dx.doi.org/10.1161/CIRCULATIONAHA.109.192064.
[48] J. McMurray, A. Cohen-Solal, R. Dietz, et al., Practical recommendations for the use of ACE inhibitors, beta-blockers, aldosterone antagonists and angiotensin receptor blockers in heart failure: putting guidelines into practice, Eur. J. Heart Fail. 7 (5) (2005) 710–721, http://dx.doi.org/10.1016/j.ejheart.2005.07.002. [49] A. Levin, P.E. Stevens, R.W. Bilous, et al., Kidney disease: improving global outcomes (KDIGO) CKD work group. KDIGO 2012 clinical practice guideline for the evaluation and management of chronic kidney disease, Kidney Int. Suppl. 3 (1) (2013) 1–150, http://dx.doi.org/10.1038/kisup.2012.73.
Please cite this article as: E. Nilsson, et al., Incidence and determinants of hyperkalemia and hypokalemia in a large healthcare system, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.07.035