Abstract
Background
After coronary artery bypass grafting (CABG), maintaining serum potassium at the upper normal limit lacks clear evidence for its impact on clinical outcomes. Also, potassium variability might signal poor prognosis. This retrospective study aimed to assess how postoperative serum potassium levels and variability affect 1-year mortality rate in CABG patients.
Methods
This retrospective study was based on data from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database. We analyzed the relationships between the average value, coefficient of variation (CV), and change trajectory of postoperative serum potassium levels and 1-year mortality rate, using logistic regression as the primary analytical method.
Results
A total of 50,903 potassium measurements, all collected during the postoperative ICU stay, from 4,606 patients were analyzed in this study. The patients whose average serum potassium level was maintained at 3.5-4 mmol/L had the lowest mortality rate of 1.33% (p = 0.004) and pointed a significant inflection point at 3.85mmo/L. The CV in deceased patients was greater than that in surviving patients (p < 0.001). Patients with CV < 5% exhibited a mortality rate of 1.36%. After adjustment for age, sex and comorbidities, those with CV > 20% had significantly higher mortality (11.63%; adjusted OR = 4.72, 95% CI: 1.32–15.17, p = 0.011). In addition, the patients with highly fluctuating potassium change trajectories had the highest 1-year mortality rate of 19.72%. Moreover, the shorter the time to the stability of the serum potassium level, the lower the mortality rate.
Conclusions
In this retrospective cohort of patients who underwent CABG, those with a mean potassium level between 3.5 and 4.0 mmol/l or minimal potassium variability correlated with the lowest observed 1-year mortality risk. Higher levels above 3.85 mmol/L correlated with increased mortality; mild hypokalemia risk remained inconclusive.
Supplementary Information
The online version contains supplementary material available at 10.1186/s13019-026-03956-7.
Keywords: CABG, Potassium, Potassium variability, Potassium trajectory, Mortality
Introduction
Coronary artery disease (CAD) remains one of the major causes of morbidity and mortality worldwide [1]. Approximately 1.5 million cardiac surgical procedures are performed annually worldwide, with coronary artery bypass grafting (CABG) being the most common [2]. Compared with medical therapy and percutaneous coronary intervention (PCI), CABG can improve survival rates in stable patients with multivessel or left main coronary artery disease [3]. However, the incidence of atrial fibrillation after CABG is as high as 25% [4], which is associated with the occurrence of stroke in the medium and long term [5], sternum and respiratory tract infections, gastrointestinal dysfunction, renal dysfunction [6], increased early and late mortality rates, prolonged stays in the intensive care unit and in the hospital, increased medical costs, etc [7, 8].
During CABG with cardiopulmonary bypass (CPB), electrolyte depletion is very common, especially a decrease in serum potassium [9, 10], several factors influence serum potassium levels. Acute hemodilution due to CPB priming fluid can lead to hypokalemia [11], the administration of cardioplegic solution is associated with increased fluctuations in serum potassium levels [12]. Additionally, ischemia-reperfusion injury and mechanical stress from the extracorporeal circuit release intracellular potassium [13]. CPB-related acidosis and iatrogenic factors like diuretics or insulin further modulate potassium levels, while renal hypoperfusion [14] may impair potassium excretion, contributing to hyperkalemia, especially in patients with preexisting kidney disease [9, 15, 16]. As one of the three major ions in the human body, potassium plays a crucial role in cellular metabolism, osmotic pressure regulation, maintenance of the acid‒base balance, and regulation of nerve and muscle activities [17]. Hypokalemia or hyperkalemia often leads to arrhythmia and may cause cardiac arrest in severe cases [18]. Other cardiovascular diseases caused by potassium disorders, such as hypertension and heart failure, further increase the risk of death [19–21]. Potassium plays a vital role in cardiac action potential, and the human body strictly controls the potassium ion concentration to maintain it within the currently recognized normal range, that is, 3.5–5.5 mmol/L [22]. For critically ill patients, maintaining the potassium level within the range of 3.5–4.5 mmol/L can significantly reduce the risk of mortality [23]. According to the 2004 American College of Cardiology/American Heart Association (ACC/AHA) guidelines, the potassium concentration should be maintained above 4.0 mmol/L to reduce the risk of ventricular fibrillation [24]. Patients with atrial fibrillation tend to have lower serum potassium levels [25]. To prevent atrial fibrillation after cardiac surgery, the serum potassium level after cardiac surgery is currently maintained at the upper limit of the normal range (4.5–5.5 mmol/L) in many medical centers worldwide [6, 7]. However, this practice is not uniformly advocated across different regions, reflecting the current lack of evidence proving the effectiveness of this strategy [8, 26].
In addition, most previous studies were based on serum potassium levels, and research on the influence of potassium fluctuations is limited. Emerging evidence suggests that greater potassium variability may independently predict adverse outcomes in critically ill populations [27, 28]. This highlights a potentially significant yet poorly characterized link between dynamic potassium instability and prognosis in high-risk surgical patients [29]. Nevertheless, research on the relationship between potassium variability and long-term survival in patients after CABG is still scarce. This study was designed to examine, within a large-scale clinical database, how postoperative mean potassium levels, their variability, and dynamic trajectories are associated with 1-year mortality in the CABG patient population.
Methods
Study design
We conducted a retrospective cohort study based on a large US-based database called the Medical Information Mart for Intensive Care IV (MIMIC-IV, version 3.1). The MIMIC-IV is a large and freely available database that comprises anonymous medical records of patients who were admitted to the ICUs of Beth Israel Deaconess Medical Center between 2008 and 2019 [30]. The copies of the datasets used in this study are available from the MIMIC database. Consequently, this research trial did not require ethical approval or the acquisition of informed consent. The data used are de-identified, and the requirement for individual patient informed consent was waived. After the completion of the Collaborative Institutional Training Initiative program, one author (Xue Z) obtained access to the database and was responsible for data extraction (certification number 64760929). Our study complied with the Reporting of Studies Conducted via the Observational Routinely Collected Health Data (RECORD) statement [31].
Selection of participants
We retrospectively analyzed the clinical data of patients who underwent CABG and were admitted to the intensive care unit (ICU) after surgery from the MIMIC-IV database on the basis of the International Classification of Diseases (ICD)−9 code 36.1 and ICD-10 code I36.1. For patients with multiple ICU admissions, only their first hospitalization records were considered in our analysis. We excluded patients who met the following criteria: under 18 years of age (n = 0), length of ICU stay < 24 h (n = 378), malignant tumors (n = 926), chronic kidney disease (n = 992, ICD-9 code 585, V451 or ICD-10 code N18, Z99.2.), potassium level data is less than three (n = 1), and missing 1-year survival status data (n = 0). Finally, a total of 4,606 patients were included in the study cohort. The flowchart of the patient inclusion process is shown in Fig. 1.
Fig. 1.
Flow diagram of patient recruitment according to the cohort selection and exclusion criteria. MIMIC-IV: Medical Information Mart for Intensive Care IV; CABG: coronary artery bypass grafting; ICU: intensive care unit
Variable extraction
Data were extracted from the MIMIC-IV database (version 3.1) through PostgreSQL programming in Navicat Premium software, and each potassium measurement obtained during the ICU stay after CABG was recorded, with frequency determined by clinical needs. The monitoring starting point is the first serum potassium value captured after ICU admission. The baseline characteristics of the patients, including age, sex, height, and weight, were also recorded. The vital signs measured within the first 24 h after admission to the ICU included systolic blood pressure, diastolic blood pressure, mean arterial pressure, heart rate, temperature, and respiratory rate; the laboratory indicators measured within the first 24 h included the white blood cell, hemoglobin, platelet, serum creatinine, lactate, and blood glucose. The Sequential Organ Failure Assessment (SOFA) score, duration of mechanical ventilation, and survival information were also recorded. Using the ICD-9 codes recorded in the MIMIC-IV database, information on comorbidities, including heart failure, peripheral vascular disease, hypertension, cerebrovascular disease, diabetes mellitus, and hyperlipidemia, was collected for analysis. We recorded the most recent preoperative serum creatinine measurement as the baseline value and defined acute kidney injury (AKI) according to the Kidney Disease Improving Global Outcomes (KDIGO) definition [32].
Outcomes
The main outcome measure of this study was the 1-year mortality rate, and the secondary outcome measures included the length of hospital stay and the mechanical ventilation time.
Statistical analysis
For normally distributed quantitative data, t tests and analysis of variance (ANOVA) were employed to evaluate the significance between patient outcomes (alive vs. deceased), whereas nonparametric tests were applied for nonnormally distributed data. Descriptive statistics were utilized to delineate baseline clinical and demographic characteristics. Patients were initially stratified into six groups on the basis of their average serum potassium levels. Continuous variables are presented as means ± standard deviations, and categorical variables are expressed as percentages. The threshold effect model was used to analyze the non - linear association between the average serum potassium level and the mortality rate, identify the important inflection points of serum potassium, and determine the significance threshold through the likelihood ratio test to evaluate the overall significance of the model and the importance of variable relationships.
Serum potassium variability is defined by the coefficient of variation (CV). This is calculated as the ratio of the standard deviation to the average serum potassium concentration, expressed as a percentage.(CV= standard deviation/mean× 100%) [33]. A higher CV indicates an increase in the fluctuation of serum potassium ion levels, that is, greater variability. Patients were divided into five groups according to the CV of their serum potassium. To assess the robustness of the association between potassium variability and 1-year mortality, we constructed a series of progressively adjusted models. We first performed an unadjusted analysis, followed by a model adjusting for basic demographic factors. Subsequent models incorporated additional clinical confounders, markers of illness severity, and indicators of treatment intensity. This stepwise adjustment strategy allowed us to evaluate whether the observed association between CV and mortality remained consistent after accounting for increasing levels of clinical complexity and potential confounding.
For the trajectory analysis of potassium levels, we collected serial potassium measurements following CABG surgery. Trajectory analysis was performed using Python (version 3.10), applying the Gaussian Mixture Model (GMM) implemented in the scikit-learn library to identify distinct potassium trajectory patterns. Ultimately selecting a three-class model (Trajectory 1: initial stability; Trajectory 2: mid-term stability; Trajectory 3: continuous fluctuations). Pearson correlation tests were used to assess relationships between time to potassium stability and mortality, as well as between the average potassium level at stability and mortality.
A multiple logistic regression model was used to analyze the relationships between serum potassium levels, CV, trajectories, and the 1-year mortality rate. Multiple models were constructed to control the influence of confounding factors and more accurately reveal the independent associations between variables. Finally, odds ratio (OR) and 95% confidence interval (CI) were reported. Statistical significance was defined as a p value < 0.05.
Results
Baseline characteristics
After screening, a total of 4,606 patients were included in this study (Fig. 1). The serum potassium level was divided into six groups on the basis of the average serum potassium level measured after surgery. Q1 was < 3.5 mmol/L, Q2 was 3.5–4 mmol/L, Q3 was 4–4.5 mmol/L, Q4 was 4.5–5 mmol/L, Q5 was 5–5.5 mmol/L, and Q6 was > 5.5 mmol/L. The baseline characteristics of the patients are shown in Table 1. The mean age of the included patients was 67.08 ± 10.09 years, 76.62% were male, and the proportion of males increased significantly in the low- and high-potassium groups (82.28% in Q1 and 80.49% in Q2, p = 0.002). The overall 1-year mortality rate was 3.10%, and the difference between the groups was significant (p = 0.004). As the serum potassium value increased, the SOFA score (5.84 ± 3.21 in group Q6, p < 0.001) and white blood cell count (18.89 ± 4.31 in group Q6, p < 0.001) gradually increased, indicating that a high serum potassium concentration may be related to disease severity and the severity of infection. The proportion of patients with diabetes in Q4 was the highest (45.28%, p < 0.001), and the proportion of patients with acute renal failure in Q6 increased significantly (29.27%, p < 0.001); the mean mechanical ventilation time in Q1 was significantly prolonged to 55 ± 28 h, p < 0.001, and there was no significant difference in the length of hospital stay between the groups (p = 0.058). Postoperative ICU stay duration differed significantly among potassium level groups (p < 0.001). Q2 had the shortest ICU stays (1.99 ± 1.67), whereas Q6 required significantly longer ICU care (3.63 ± 3.89).
Table 1.
Baseline characteristics according to the mean potassium range
| Variables | Missing(%) | Overall n = 4606 |
Q1(< 3.5) n = 79 |
Q2(3.5–4.5) n = 679 |
Q3(4–4.5.5) n = 3125 |
Q4(4.5–5.5) n = 625 |
Q5(5–5.5.5) n = 57 |
Q6(> 5.5) n = 41 |
p value |
|---|---|---|---|---|---|---|---|---|---|
| Baseline characteristics, mean ± SD | |||||||||
| Age, years | 0.0 | 67.08 ± 10.09 | 66.38 ± 9.65 | 67.60 ± 10.70 | 67.16 ± 9.93 | 66.07 ± 10.39 | 68.14 ± 8.34 | 67.39 ± 9.80 | 0.108 |
| BMI, kg/m2 | 4.3 | 30.17 ± 5.81 | 30.04 ± 4.81 | 30.16 ± 6.30 | 30.18 ± 5.76 | 30.17 ± 5.58 | 29.88 ± 6.82 | 29.80 ± 5.26 | 0.957 |
| Male, n (%) | 0.0 | 3529 (76.62) | 65 (82.28) | 478 (70.40) | 2424 (77.57) | 486 (77.76) | 43(75.44) | 33(80.49) | 0.002 |
| Laboratory parameters, mean ± SD | |||||||||
| Hemoglobin, g/dL | 0.0 | 9.89 ± 1.86 | 9.91 ± 1.43 | 9.92 ± 1.97 | 9.97 ± 1.84 | 9.55 ± 1.83 | 8.92 ± 1.48 | 9.35 ± 2.37 | < 0.001 |
| Platelet count, ×109/L | 0.3 | 158.47 ± 56.60 | 165.23 ± 57.46 | 155.01 ± 56.75 | 157.59 ± 55.05 | 164.09 ± 60.91 | 174.54 ± 76.16 | 163.05 ± 62.85 | 0.023 |
| White blood cell, ×109/L | 0.1 | 13.12 ± 6.51 | 12.74 ± 5.18 | 12.92 ± 5.36 | 13.11 ± 5.55 | 13.12 ± 5.52 | 17.17 ± 32.00 | 18.89 ± 4.31 | < 0.001 |
| Lactate, mmol/L | 1.6 | 2.17 ± 1.02 | 2.21 ± 0.90 | 2.22 ± 1.12 | 2.14 ± 0.98 | 2.20 ± 1.09 | 2.27 ± 1.22 | 2.34 ± 1.25 | 0.775 |
| Blood glucose, mg/dL | 0.0 | 123.85 ± 35.68 | 128.92 ± 59.32 | 126.03 ± 37.63 | 122.94 ± 34.19 | 125.40 ± 36.16 | 118.02 ± 30.01 | 132.46 ± 48.68 | 0.073 |
| Vital Signs, mean ± SD | |||||||||
| Heart rate, bpm | 0.0 | 80.28 ± 10.88 | 79.63 ± 10.67 | 80.66 ± 11.07 | 79.90 ± 10.76 | 81.59 ± 10.87 | 80.49 ± 10.53 | 83.66 ± 15.19 | 0.010 |
| Systolic blood pressure, mmHg | 2.5 | 112.66 ± 17.25 | 111.09 ± 17.13 | 112.43 ± 18.17 | 112.94 ± 17.06 | 111.94 ± 17.05 | 111.91 ± 18.75 | 110.13 ± 18.17 | 0.640 |
| Diastolic blood pressure, mmHg | 2.5 | 58.83 ± 10.29 | 58.05 ± 10.81 | 59.69 ± 10.87 | 58.98 ± 10.18 | 57.58 ± 9.94 | 57.00 ± 11.49 | 56.33 ± 9.22 | 0.001 |
| Mean arterial pressure, mmHg | 2.5 | 77.35 ± 15.46 | 76.20 ± 12.90 | 78.72 ± 19.79 | 77.45 ± 14.65 | 75.61 ± 12.94 | 78.42 ± 26.42 | 74.20 ± 10.40 | 0.019 |
| Respiratory rate, bpm | 0.1 | 14.77 ± 5.31 | 14.53 ± 4.71 | 15.04 ± 4.73 | 14.70 ± 4.63 | 14.75 ± 8.34 | 14.33 ± 4.06 | 16.61 ± 6.24 | 0.038 |
| Temperature, ℉ | 3.0 | 97.77 ± 3.33 | 97.86 ± 1.01 | 97.99 ± 0.99 | 97.69 ± 3.82 | 97.86 ± 2.67 | 97.91 ± 0.87 | 97.94 ± 0.84 | 0.250 |
| Comorbidities, n (%) | |||||||||
| Peripheral vascular disease | 0.0 | 117 (2.54) | 3(3.80) | 18(2.65) | 75(2.40) | 20(3.20) | 0(0.00) | 1(2.44) | 0.643 |
| Cerebrovascular disease | 0.0 | 296 (6.43) | 7(8.86) | 46(6.77) | 199(6.37) | 35(5.60) | 4(7.02) | 5(12.20) | 0.557 |
| Diabetes mellitus | 0.0 | 1708 (37.08) | 28(35.44) | 229(33.73) | 1130(36.16) | 283(45.28) | 25(43.86) | 13(31.71) | < 0.001 |
| Hypertension | 0.0 | 3324(72.17) | 54(68.35) | 509(74.96) | 2,250(72.00) | 440(70.40) | 38(66.67) | 33(80.49) | 0.260 |
| Acute kidney injury | 0.0 | 493 (10.70) | 7(8.86) | 67(9.87) | 299(9.57) | 96(15.36) | 14(24.56) | 12(29.27) | < 0.001 |
| Hyperlipidemia | 0.0 | 3384 (73.47) | 60(75.95) | 480(70.69) | 2,312(73.98) | 461(73.76) | 40(70.18) | 31(75.61) | 0.578 |
| Heart failure | 0.0 | 933(20.26) | 14(17.72) | 133(19.59) | 611(19.55) | 149(23.84) | 16(28.07) | 10(24.39) | 0.108 |
| Myocardial infarction | 0.0 | 826 (17.93) | 10(12.66) | 107(15.76) | 558(17.86) | 128(20.48) | 15(26.32) | 8(19.51) | 0.100 |
| Others, mean ± SD | |||||||||
| Ventilation hour, h | 1.7 | 55.00 ± 48.08 | 51.80 ± 63.78 | 41.52 ± 33.83 | 34.26 ± 22.16 | 37.76 ± 34.28 | 45.99 ± 33.64 | 47.34 ± 48.93 | < 0.001 |
| SOFA | 0.0 | 5.27 ± 2.52 | 4.58 ± 2.37 | 4.96 ± 2.58 | 4.83 ± 2.48 | 5.17 ± 2.58 | 5.42 ± 2.60 | 5.84 ± 3.21 | < 0.001 |
| Death within 1- year, n (%) | 0.0 | 143(3.10) | 4(5.06) | 9(1.33) | 94(3.01) | 30(4.80) | 3(5.26) | 3(7.32) | 0.004 |
| Hospital day | 0.0 | 8.65 ± 7.15 | 9.14 ± 8.21 | 7.75 ± 4.68 | 8.52 ± 5.88 | 8.86 ± 7.76 | 8.83 ± 5.96 | 8.53 ± 4.28 | 0.058 |
| ICU day | 0.5 | 2.77 ± 3.80 | 2.67.23 ± 2.89 | 1.99 ± 1.67 | 2.77 ± 3.84 | 2.44 ± 3.94 | 2.36 ± 3.29 | 3.63 ± 3.89 | < 0.001 |
Patient characteristics stratified by the average potassium concentration. The data are presented as means ± SDs or as the number of patients and percentages. (n) denotes the total number of patients present in each group. The p value was calculated via the chi-square test for categorical variables and the Kruskal‒Wallis test for nonparametric variables. BMI: body mass index; SOFA: Sequential Organ Failure Assessment
Influence of the average serum potassium level after CABG on the 1-year mortality rate
As shown in Table 1, after the average postoperative serum potassium level was stratified, the 1-year mortality rate was the lowest at 1.33% in patients whose serum potassium level was maintained at 3.5–4.0 mmol/L, and the mortality rate was the highest at 7.32% when the serum potassium level was greater than 5.5 mmol/L (p = 0.004). As evident from Figures. 2A, the CI for each group indicates that as the serum potassium level deviates from the normal range, the degree of uncertainty increases. Notably, the standard error (SE) in group Q6 is relatively high. In addition, a threshold effect was detected and illustrate in Table 2.. Among patients with serum potassium levels < 3.85 mmol/L, the OR for 1-year mortality was 0.18 (95% CI 0.03–1.29; p = 0.089), indicating that there was no statistically significant association between hypokalemia and 1- year mortality. While beyond the inflection point of 3.85 mmol/L, a subtle positive correlation emerged, linking serum potassium levels to 1-year mortality (OR = 2.73, 95% CI 1.68–4.37, p < 0.001), and an increase in serum potassium may increase the 1-year mortality rate of patients. The p value of the likelihood ratio test was 0.031, indicating a significant difference between the models of different serum potassium level groups.
Fig. 2.
1-year mortality rates stratified by average serum potassium level (A), serum potassium coefficient of variation (B), and serum potassium trajectory (C). Error bars denote standard error (SE), the shaded areas denote the 95% confidence intervals (CI). Numbers below the x-axis indicate the number of patients in each category
Table 2.
Threshold effect analysis of the relationship between the average serum potassium level and the 1-year mortality rate
| Adjusted OR_95CI | p value | |
|---|---|---|
| Two model | ||
| Average serum potassium<3.85mmol/L | 0.18 (0.03-1.29) | 0.089 |
| Average serum potassium>3.85mmol/L | 2.72 (1.69-4.37) | <0.001 |
| Likelihood ratio test | - | 0.031 |
Influence of serum potassium variability after CABG on the 1-year mortality rate
We collected all the measured serum potassium values of the patients and calculated the potassium coefficient of variation for each patient and confirmed via Table S1 that the overall cohort had a median of 7 postoperative potassium measurements (IQR: 4). All potassium CV categories had a minimum of 3 measurements per patient (median range: 6–12). Over time, the serum potassium level in the alive group was stable and gradually decreased, while that in the deceased group fluctuated greatly. The results shown in Fig. 3 indicate that the overall distribution range of potassium coefficient of variation values in the deceased group was wider, and the overall level was higher than that in the alive group. We performed a t test to analyze the potassium coefficient of variation of patients in the deceased group and the alive group. The potassium CV of the deceased group was significantly greater than that of the alive group (T=−6.57, p < 0.001). Maintaining serum potassium levels between 3.5 and 4.0 mmol/L and minimizing variability are associated with the lowest mortality rates.
Fig. 3.
Relationship between the coefficient of variation (CV) of the serum potassium concentration and the 1-year mortality rate. The potassium CV of the death group was significantly greater than that of the survival group
We divided the CVs into five groups (CV1 < 5%, CV2 5%–10%, CV3 10%–15%, CV4 15%–20%, and CV5 > 20%). As the CV gradually increased, the mortality rate also correspondingly rose from 1.36% to 11.63%. (p < 0.001). Figures 2B demonstrate that subgroups with high variability (e.g., CV5) exhibit significantly wider CI and elevated SE, collectively indicating pronounced data dispersion and reduced precision in mortality estimates for these cohorts. As shown in Table S2, higher potassium CV was associated with progressively increased rates of death within 1 year, acute kidney injury (AKI), and renal replacement therapy (RRT). Similarly, the incidence of AKI and RRT rose markedly across increasing CV, while baseline serum creatinine showed a modest upward trend.
Unadjusted and adjusted logistic regression revealed the association between the coefficient of variation of serum potassium and 1-year mortality (Table 3). In the unadjusted analysis, the OR for 1-year mortality significantly escalated with higher CV categories. Compared with the reference group, CV > 20% exhibited the highest OR (OR = 9.52, 95% CI: 2.81–28.99 p < 0.001). After adjusting for age and sex in Model 1, the association remained statistically significant, albeit with attenuated effect sizes (e.g.CV5: OR = 8.14, 95% CI 2.38–25.08, p < 0.001). Further adjustment for comorbidities in Model 2 revealed a consistent trend, CV > 20% still demonstrating a significantly elevated mortality risk (OR = 4.72, 95% CI: 1.32–15.17, p = 0.011). This association remained robust in sensitivity analyses that additionally adjusted for markers of illness severity (SOFA score, lactate, serum creatinine, white blood cells, hemoglobin) and treatment intensity (mechanical ventilation duration.
Table 3.
Association between serum potassium variability and the risk of 1-year mortality
| Unadjusted | Model 1 | Model 2 | Model 3 | Model 4 | ||||||
|---|---|---|---|---|---|---|---|---|---|---|
| OR (95% CI) | p value | OR (95% CI) | p value | OR (95% CI) | p value | OR (95% CI) | p value | OR (95% CI) | p value | |
| CV1 (< 5%) | 1(Reference) | 1(Reference) | 1(Reference) | 1(Reference) | 1(Reference) | |||||
| CV2 (5%−10%) | 1.85 (0.97–3.99) | 0.086 | 1.63 (0.85–3.54) | 0.172 | 1.42 (0.74–3.09) | 0.331 | 2.25 (0.89–5.67) | 0.085 | 2.15 (0.85–5.42) | 0.106 |
| CV3 (10%−15%) | 3.47 (1.77–7.62) | 0.001 | 2.79 (1.41–6.18) | 0.006 | 1.93 (0.96–4.32) | 0.082 | 3.32 (1.29–8.54) | 0.013 | 2.92 (1.13–7.58) | 0.027 |
| CV4 (15%−20%) | 7.23 (3.16–17.52) | < 0.001 | 5.73 (2.48–13.99) | < 0.001 | 3.25 (1.36–8.15) | 0.009 | 4.81 (1.65–13.98) | 0.004 | 4.54 (1.54–13.36) | 0.006 |
| CV5 (> 20%) | 9.52 (2.81–28.99) | < 0.001 | 8.14 (2.38–25.08) | < 0.001 | 4.72 (1.32–15.17) | 0.011 | 8.44 (2.26–31.50) | 0.002 | 8.82 (2.39–32.62) | 0.001 |
Model 1 was adjusted for sex and age; Model 2 was adjusted for sex and age plus acute myocardial infarction, heart failure, acute kidney failure, and diabetes mellitus. Model 3 was further adjusted for markers of illness severity (SOFA score, lactate, serum creatinine, white blood cell count, and hemoglobin). Model 4 was additionally adjusted for mechanical ventilation duration
Serum potassium dynamic trajectory and the 1-year mortality rate
We applied the mixed - effects models to identify serum potassium level trajectories, selecting a three-class model (Fig. 4). Trajectory 1: Stability in the initial stage (3978 (86.4%) patients). Trajectory 2: Stability in the middle stage (557 (12.1%) patients). Trajectory 3: Continuous fluctuations (71 (1.5%) patients).
Fig. 4.
Postoperative serum potassium trajectories identified by Gaussian mixture modeling. The x-axis represents the Measurement Index (sequential order of potassium measurements after ICU admission). Trajectory 1: The serum potassium level stabilized in the initial stage. Trajectory 2: The serum potassium level stabilized in the middle stage. Trajectory 3: The serum potassium level fluctuated continuously
There were significant differences in the baseline characteristics of patients with these trajectories (Table 4). The patients in Trajectory 3 had the longest hospital stay (23.87 ± 14.65, p < 0.001), the longest ICU stay (12.86 ± 12.74, p < 0.001), the highest Sequential Organ Failure Assessment (SOFA) score indicating severe disease (6.38 ± 3.71, p < 0.001), the longest mechanical ventilation time (181.51 ± 162.88 h, p < 0.001), and the highest 1-year mortality rate (19.72%, p < 0.001). Figures 2C demonstrate narrow CI across trajectory groups, reflecting high precision in mortality estimates. Notably, group Trajectory 3 (comprising only 71 patients) exhibited significantly elevated SE compared to Trajectory 1, suggesting pronounced within-group heterogeneity in this cohort and wide uncertainty in its mortality estimates. The Pearson correlation test revealed a weak but significant positive correlation between the time to potassium level normalization and mortality, and the longer the time to stability, the slightly higher the risk of death (correlation coefficient: 0.04, p = 0.005); there was a more obvious positive correlation between the average value at the time of stability and mortality. Although the correlation was still weak but statistically significant (correlation coefficient: 0.08, p < 0.001).
Table 4.
Baseline characteristic comparisons among the three serum potassium level trajectories
| Variable | Trajectory 1 (n = 3978) |
Trajectory 2 (n = 557) |
Trajectory 3 (n = 71) |
p value |
|---|---|---|---|---|
| Baseline characteristics | ||||
| Age, years | 66.72 ± 9.98 | 69.35 ± 10.36 | 69.17 ± 11.53 | < 0.001 |
| BMI, kg/m2 | 30.15 ± 5.74 | 30.37 ± 6.41 | 29.66 ± 4.81 | 0.981 |
| Male, n (%) | 3111(78.21) | 372(66.80) | 46(65.00) | < 0.001 |
| Laboratory parameters, mean ± SD | ||||
| Hemoglobin, g/dL | 9.85 ± 1.77 | 10.06 ± 2.36 | 10.57 ± 2.39 | 0.061 |
| Platelet count, ×109/L | 155.27 ± 51.64 | 176.88 ± 75.77 | 193.66 ± 95.49 | < 0.001 |
| White blood cell, ×109/L | 13.07 ± 6.56 | 13.34 ± 6.12 | 14.09 ± 6.15 | 0.282 |
| Lactate, mmol/L | 2.16 ± 0.93 | 2.24 ± 1.44 | 2.03 ± 1.52 | < 0.001 |
| Blood glucose, mg/dL | 120.78 ± 29.16 | 141.30 ± 58.32 | 151.74 ± 69.30 | < 0.001 |
| Vital Signs, mean ± SD | ||||
| Heart rate, bpm | 79.70 ± 10.23 | 83.27 ± 12.84 | 89 ± 18.81 | < 0.001 |
| Systolic blood pressure, mmHg | 112.82 ± 16.82 | 111.56 ± 20.01 | 111.36 ± 20.70 | 0.031 |
| Diastolic blood pressure, mmHg | 59.05 ± 10.05 | 57.22 ± 11.54 | 57.41 ± 13.47 | < 0.001 |
| Mean arterial pressure, mmHg | 77.44 ± 14.03 | 76.54 ± 22.35 | 78.11 ± 30.55 | 0.033 |
| Respiratory rate, bpm | 14.49 ± 5.26 | 16.40 ± 5.24 | 17.46 ± 5.64 | < 0.001 |
| Temperature, ℉ | 97.74 ± 3.11 | 97.97 ± 3.84 | 97.56 ± 7.78 | < 0.001 |
| Others, mean ± SD/n(%) | ||||
| Death within 1- year | 85(2.14) | 44(7.90) | 14(19.72) | < 0.001 |
| Acute kidney injury | 266(6.69) | 176(31.60) | 51(71.83) | < 0.001 |
| ICU day | 2.23 ± 2.75 | 5.39 ± 4.80 | 12.86 ± 12.74 | < 0.001 |
| Hospital day | 7.65 ± 6.24 | 13.79 ± 7.43 | 23.87 ± 14.65 | < 0.001 |
| Ventilation hour, h | 33.31 ± 36.16 | 88.75 ± 79.61 | 181.51 ± 162.88 | < 0.001 |
| SOFA | 4.79 ± 2.34 | 5.51 ± 3.33 | 6.38 ± 3.71 | < 0.001 |
The patients were divided into three groups according to the trajectory of changes in the serum potassium concentration. Trajectory 1: stability in the initial stage, Trajectory 2: stability in the middle stage, Trajectory 3: Continuous fluctuations
Discussion
Among CABG patients in this large retrospective cohort, a postoperative potassium range of 3.5–4.0 mmol/L and reduced potassium variability were linked to the lowest 1-year mortality.; moreover, the shorter the time to stability, the more favorable the prognosis, but also more efficient use of critical care resources.
There is an association between abnormal potassium concentrations and an increased risk of death, which is as expected. A retrospective study in Denmark revealed that potassium disorders in patients before and after CABG are not common, and patients whose postoperative serum potassium concentration is maintained at 4.0–4.6 mmol/L have the lowest risk of death [34]. However, Patel et al. reported a U-shaped relationship between the serum potassium level of patients with acute myocardial infarction and the risk of early and late cardiovascular death. Although mild hypokalemia (< 3.85 mmol/L) was not associated with increased mortality in our cohort, careful monitoring and appropriate correction of marked hypokalemia (< 3.5 mmol/L) remains clinically prudent given its known arrhythmogenic potential. When the serum potassium level is between 3.5 and 4.5 mmol, the cardiovascular mortality rate is the lowest, and when the serum potassium level is ≥ 4.5 mmol/L, the 1-year cardiovascular death risk tends to increase [35].For patients in the ICU, maintaining a serum potassium target of 3.5–4.0 mmol/L and minimizing variability are beneficial for improving patient prognosis.
Moreover, for patients receiving potassium supplementation during hospitalization, both their average serum potassium concentration and serum potassium variability increase, and the mortality rate also increases [36]. The results of a recent real-world study revealed that when potassium supplementation was administered for a serum potassium level was less than 3.6 mmol/L, compared with the conventional practice of potassium supplementation when the level was less than 4.5 mmol/L, the incidence of atrial fibrillation after cardiac surgery did not increase [37], which is consistent with the results of our study. Maintaining the serum potassium level at 3.5–4 mmol/L after CABG is beneficial for reducing the 1-year mortality rate, and active potassium supplementation to maintain a slightly elevated normal serum potassium level is instead unfavorable for the patient’s prognosis.
Increased postoperative serum potassium variability is associated with an elevated mortality rate. This may be because fluctuations in serum potassium levels impair the stability of cardiac electrophysiological activities, thereby increasing the risk of arrhythmia. This study acknowledges potential confounding by ICU length of stay and monitoring intensity, as sicker patients at higher mortality risk may have longer ICU stays and experience greater potassium variability due to more frequent interventions and more intensive monitoring, which could introduce bias in the observed associations; residual confounding may still exist given the retrospective nature of this study. Research by Zhang et al. revealed that the serum potassium variability of ICU patients is related to the 28-day mortality rate [38]. For patients with acute myocardial infarction, the in-hospital mortality rate is the lowest when the serum potassium variability is minimal [27]. In a hospital-based study, Lombardi et al. confirmed that even when the potassium ion concentration is within the normal range, patients with greater potassium variability are still at high risk of death [39]. Hessels et al. reported that there is an independent association between potassium variability and mortality [40]. Our findings suggest that a CV within 5 is associated with the lowest 1-year mortality, likely reflecting better overall clinical stability.
The specific mechanism of the fluctuation of patients’ serum potassium has not been clearly defined at present and may be the result of a combination of various factors. After CABG, stress hyperglycemia, insulin secretion, elevated catecholamine levels, activation of Na+-K+-ATPase, myocardial ischemia-reperfusion injury, potassium loss through the gastrointestinal tract, secondary to acid-base imbalance, and iatrogenic causes such as the use of diuretics and RAS inhibitors may affect the serum potassium concentration [41]. Another important factor is the occurrence of postoperative AKI. Studies revealed that patients with high potassium variability often have increased creatinine levels [27]. The kidney is the main organ involved in regulating water and electrolyte balance, and concurrent acute kidney injury indicates insufficient blood volume and decreased renal regulatory capacity [42]. Kidney function may partly explain the observed association between potassium variability and mortality. Higher CV categories were associated with higher baseline serum creatinine levels and markedly increased rates of postoperative AKI and RRT, suggesting that potassium variability may reflect underlying renal dysfunction and overall clinical instability rather than a direct causal effect. We need for enhanced renal monitoring in patients with unstable potassium levels, as these parameters may interact in a vicious cycle of mutual deterioration. The use of computer assistance can effectively ameliorate potassium disorders and variability problems [40].
To prevent the occurrence of adverse events such as cardiovascular disease and death in patients after CABG, maintaining a certain serum potassium concentration is reasonable, and excessive fluctuations in potassium concentration caused by potassium supplementation should be avoided. Although the dose price of potassium agents is low, in many cardiac departments, the cumulative annual expenditure on intravenous potassium supplementation is greater than that of most other drugs [43]. The time spent by medical staff on potassium supplementation interventions further increases the monetary cost and opportunity cost. Potassium supplementation also has negative effects on patients. Adverse gastrointestinal reactions caused by oral potassium supplementation are relatively common, and patients have poor tolerance [44]. If intravenous access is maintained for a long time exclusively for potassium supplementation, the risk of infection increases, and intravenous potassium supplementation may lead to fluid overload [44]. Rapid infusion may be fatal [45].
The novelty of the present study lies in its comprehensive and dynamic characterization of postoperative potassium homeostasis in patients undergoing CABG. In contrast to prior studies that predominantly focused on isolated potassium measurements or short-term clinical outcomes, we simultaneously evaluated mean potassium levels, potassium variability, and postoperative potassium trajectories, and examined their associations with 1-year mortality in a large retrospective cohort. Our results suggest that potassium instability and delayed stabilization are strongly associated with poorer long-term outcomes, independent of average potassium concentrations.
At present, it is not possible to obtain information from this database regarding whether CABG surgery was performed using an off-pump method. CABG surgery is usually performed with the assistance of cardiopulmonary bypass [46]. In 2016, approximately 25% of CABG surgeries were performed using the off-pump method [47]. Since the incidence of postoperative atrial fibrillation for both off-pump and on-pump CABC surgeries is similar [48] and the early and late survival rates are similar [47], we believe that the lack of information on whether the surgery was performed using cardiopulmonary bypass or the off-pump method will not have any impact on the results of this study [46]. Whereas, the database further lacks critical details: emergency surgery status, New York Heart Association Functional Classification (NYHA class), complexity of coronary artery lesions, the number of bypass long-term outpatient potassium data, post-discharge medication adherence are unavailable, which may confound the association between potassium variability and mortality. Moreover, intraoperative serum potassium levels remain inaccessible—fluctuations in this intraoperative period are likely to be more pronounced, representing another notable limitation, while the postoperative stabilization patterns we identified have independent prognostic value and more direct implications for clinical practice.
This study is limited by its retrospective design, which is known for selection bias and missing data. we could not control for intraoperative potassium fluctuations, timing of supplementation, or institutional protocols—factors that may affect potassium trajectories. Selection bias may also exist due to exclusion criteria (e.g., chronic kidney disease, short ICU stays), limiting generalizability to these subgroups. Despite adjusting for age, sex, and comorbidities, variables like nutritional status, muscle mass, or concurrent medications (e.g., insulin) could modify potassium-outcome relationships.
Despite analyzing a large volume of data, there were discrepancies in data stability across different groups. Notably, the serum potassium > 5.5 mmol/L subgroup exhibited poor stability, potentially compromising the accuracy of its mortality estimates. Similarly, groups with high potassium variability (e.g., CV > 20%) showed wide CI, reflecting substantial uncertainty and suggesting potential bias in risk assessment. We found that patients with continuous fluctuations (Trajectory 3) had the highest 1-year mortality. However, this small sample size introduced substantial uncertainty into the mortality estimate and limited the generalizability of subgroup-specific findings. This study focused on 1-year mortality and did not assess detailed short-term complications. Future studies incorporating both short- and long-term outcomes are needed. Furthermore, the database lacks formal surgical risk scores, limiting precise characterization of baseline operative risk and comparison with established risk prediction models. Thus, further research is warranted to elucidate the precise mechanisms underlying the association between serum potassium parameters and mortality, and to evaluate their incremental prognostic value when integrated into comprehensive risk stratification frameworks, thereby providing a more solid foundation for developing precise potassium management strategies in clinical practice.
Conclusion
In this large retrospective cohort of CABG patients, postoperative potassium levels between 3.5 and 4.0 mmol/L and lower potassium variability were associated with the lowest 1-year mortality. Prospective studies are warranted to determine whether targeting this range and minimizing variability can improve outcomes.
Electronic Supplementary Material
Below is the link to the electronic supplementary material.
Acknowledgements
The authors would like to thank the team at the Laboratory of Computational Physiology at the Massachusetts Institute of Technology for developing and maintaining the MIMIC-IV database.
Abbreviations
- CABG
Coronary artery bypass grafting
- MIMIC-IV
The medical information mart for intensive care IV
- CV
Coefficient of variation
- CAD
Coronary artery disease
- PCI
Percutaneous coronary intervention
- ACC/AHA
American college of cardiology/American heart association
- ICU
Intensive care unit
- ICD
International classification of diseases
- SOFA
Sequential organ failure assessment
- BMI
Body mass index
Author contributions
XZ: Conceptualization, Writing - original draft, Writing - review & editing, Formal analysis, Data curation. JF, YJ: Formal analysis, Visualization, Validation. JY, PZ, ZY: Writing - review & editing, Data curation. HF, CM: Supervision, Methodology. XS, YX: Conceptualization, Methodology. WC: Conceptualization, Funding acquisition, Writing - review & editing. All authors read and approved the final manuscript.
Funding
The research was supported by the National Natural Science Foundation of China (82272192, 82572465), Shanghai Eastern Talent Plan (BJWS2025009), Shanghai “Yiyuan New Star” Outstanding Youth Medical Talent Program (20244Z0012).
Data availability
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
Declarations
Ethics approval and consent to participate
The MIMIC-IV database received approval from the Massachusetts Institute of Technology and the Beth Israel Deaconess Medical Center. The copies of the datasets used in this study are available from the MIMIC database. Consequently, this research trial did not require ethical approval or the acquisition of informed consent.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Xue Zhang, Juncheng Fang and Yucheng Jin these authors are co-first authors and contributed equally to this work.
Xia Sun, Yun Xiong and Wankun Chen these corresponding authors contributed equally to this work..
Contributor Information
Xia Sun, Email: 18017312998@163.com.
Yun Xiong, Email: yunx@fudan.edu.cn.
Wankun Chen, Email: chenwank@163.com.
References
- 1.Tsao CW, Aday AW, Almarzooq ZI, et al. Heart disease and stroke statistics-2023 update: a report from the American Heart Association. Circulation. 2023;147(8):e93–621. 10.1161/cir.0000000000001123. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Vervoort DLG, Ghandour H, et al. Global cardiac surgical volume and gaps: trends, targets,and way forward. Annals of Thoracic Surgery Short Reports. 2024;2(2):320–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Lawton JS, Tamis-Holland JE, Bangalore S, et al. 2021 ACC/AHA/SCAI Guideline for Coronary Artery Revascularization: A Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. Circulation Jan. 2022;18(3):e18–114. 10.1161/cir.0000000000001038. [DOI] [PubMed] [Google Scholar]
- 4.D’Agostino RS, Jacobs JP, Badhwar V, et al. The Society of Thoracic Surgeons Adult Cardiac Surgery Database: 2018 Update on Outcomes and Quality. The Annals of thoracic surgery. 2018;105(1):15–23. 10.1016/j.athoracsur.2017.10.035. [DOI] [PubMed] [Google Scholar]
- 5.Benedetto U, Gaudino MF, Dimagli A, et al. Postoperative atrial fibrillation and long-term risk of stroke after isolated coronary artery bypass graft surgery. Circulation. 2020;142(14):1320–9. 10.1161/circulationaha.120.046940. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Burrage PS, Low YH, Campbell NG, O’Brien B. New-onset atrial fibrillation in adult patients after cardiac surgery. Current anesthesiology reports. 2019;9(2):174–93. 10.1007/s40140-019-00321-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.O’Brien B, Burrage PS, Ngai JY, et al. Society of Cardiovascular Anesthesiologists/European Association of Cardiothoracic Anaesthetists Practice Advisory for the Management of Perioperative Atrial Fibrillation in Patients Undergoing Cardiac Surgery. J Cardiothorac Vasc Anesth. 2019;33(1):12–26. 10.1053/j.jvca.2018.09.039. [DOI] [PubMed] [Google Scholar]
- 8.Caldonazo T, Kirov H, Rahouma M, et al. Atrial fibrillation after cardiac surgery: a systematic review and meta-analysis. J Thorac Cardiovasc Surg. 2023;165(1):94-103.e24. 10.1016/j.jtcvs.2021.03.077. [DOI] [PubMed] [Google Scholar]
- 9.Polderman KH, Girbes ARJ. Severe electrolyte disorders following cardiac surgery: a prospective controlled observational study. Critical Care. 2004. 10.1186/cc2973. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Ghorbaninezhad K, Bakhsha F, Yousefi Z, Halakou S, Mehrbakhsh Z. Comparison effect of tranexamic acid (TA) and tranexamic acid combined with vitamin C (TXC) on drainage volume and atrial fibrillation arrhythmia in patients undergoing cardiac bypass surgery: randomized clinical trial. Anesthesiology and pain medicine. 2019;9(5):e96096. 10.5812/aapm.96096. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Hessel EA 2nd. What’s new in cardiopulmonary bypass. J Cardiothorac Vasc Anesth. 2019;33(8):2296–326. 10.1053/j.jvca.2019.01.039. [DOI] [PubMed]
- 12.Buggeskov KB, Maltesen RG, Rasmussen BS, et al. Lung Protection Strategies during Cardiopulmonary Bypass Affect the Composition of Blood Electrolytes and Metabolites-A Randomized Controlled Trial. Journal of clinical medicine. 2018. 10.3390/jcm7110462. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Dekker NAM, van Leeuwen ALI, van de Ven PM, et al. Pharmacological interventions to reduce edema following cardiopulmonary bypass: a systematic review and meta-analysis. J Crit Care. 2020;56:63–72. 10.1016/j.jcrc.2019.12.006. [DOI] [PubMed] [Google Scholar]
- 14.Pontes JC, Silva GV, Benfatti RA, Machado NP, Pontelli R, Pontes ER. Risk factors for the development of acute renal failure following on-pump coronary artery bypass grafting. Rev Bras Cir Cardiovasc. 2007;22(4):484–90. 10.1590/s0102-76382007000400016. [DOI] [PubMed] [Google Scholar]
- 15.Polderman KH, Peerdeman SM, Girbes AR. Hypophosphatemia and hypomagnesemia induced by cooling in patients with severe head injury. J Neurosurg. 2001;94(5):697–705. 10.3171/jns.2001.94.5.0697. [DOI] [PubMed] [Google Scholar]
- 16.Young R. Perioperative fluid and electrolyte management in cardiac surgery: a review. J Extra-Corp Tech. 2012;44(1):P20-6. [PMC free article] [PubMed] [Google Scholar]
- 17.Gumz ML, Rabinowitz L, Wingo CS. An integrated view of potassium homeostasis. N Engl J Med. 2015;373(1):60–72. 10.1056/NEJMra1313341. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Cheung CM, Ponnusamy A, Anderton JG. Management of acute renal failure in the elderly patient: a clinician’s guide. Drugs Aging. 2008;25(6):455–76. 10.2165/00002512-200825060-00002. [DOI] [PubMed] [Google Scholar]
- 19.Sica DA. Antihypertensive therapy and its effects on potassium homeostasis. The Journal of Clinical Hypertension. 2006;8(1):67–73. 10.1111/j.1524-6175.2006.05139.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Urso C, Brucculeri S, Caimi G. Acid-base and electrolyte abnormalities in heart failure: pathophysiology and implications. Heart Fail Rev. 2015;20(4):493–503. 10.1007/s10741-015-9482-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Krogager ML, Torp-Pedersen C, Mortensen RN, et al. Short-term mortality risk of serum potassium levels in hypertension: a retrospective analysis of nationwide registry data. Eur Heart J. 2017;38(2):104–12. 10.1093/eurheartj/ehw129. [DOI] [PubMed] [Google Scholar]
- 22.Erratum Regarding. Physiology and Pathophysiology of Potassium Homeostasis: Core Curriculum 2019 (Am J Kidney Dis. 2019;74[5]:682–695). American journal of kidney diseases: the official journal of the National Kidney Foundation. Nov 2022;80(5):690. 10.1053/j.ajkd.2022.09.002 [DOI] [PubMed]
- 23.Ookuma T, Miyasho K, Kashitani N, et al. The clinical relevance of plasma potassium abnormalities on admission in trauma patients: a retrospective observational study. J Intensive Care. 2015;3(1):37. 10.1186/s40560-015-0103-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Antman EM, Anbe DT, Armstrong PW, et al. ACC/AHA Guidelines for the Management of Patients With ST-Elevation Myocardial Infarction—Executive Summary. Circulation. 2004;110(5):588–636. 10.1161/01.Cir.0000134791.68010.Fa. [DOI] [PubMed] [Google Scholar]
- 25.Krijthe BP, Heeringa J, Kors JA, et al. Serum potassium levels and the risk of atrial fibrillation: the Rotterdam Study. International J cardiology Oct. 2013;15(6):5411–5. 10.1016/j.ijcard.2013.08.048. [DOI] [PubMed] [Google Scholar]
- 26.Lai FY, Abbasciano RG, Tabberer B, Kumar T, Murphy GJ. Identifying research priorities in cardiac surgery: a report from the James Lind Alliance Priority Setting Partnership in adult heart surgery. BMJ Open. 2020;10(9):e038001. 10.1136/bmjopen-2020-038001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Zhang X, Wang M, Zhu Z, et al. Serum potassium level, variability and in-hospital mortality in acute myocardial infarction. Eur J Clin Invest. 2022;52(7):e13772. 10.1111/eci.13772. [DOI] [PubMed] [Google Scholar]
- 28.Zhang XL, Cai HX, Wang SJ, et al. Potassium variability during hospitalization and outcomes after discharge in patients with acute myocardial infarction. Journal geriatric cardiology: JGC Jan. 2021;28(1):10–9. 10.11909/j.issn.1671-5411.2021.01.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Lin Z, Zheng J, Liu X, Hu X, Fuxian R, Gao D. Assessing potassium levels in critically ill patients with heart failure: application of a group-based trajectory model. ESC Heart Fail. 2023;10(1):57–65. 10.1002/ehf2.14161. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Johnson AEW, Bulgarelli L, Shen L et al. MIMIC-IV, a freely accessible electronic health record dataset. Scientific data. Jan 3. 2023;10(1):1. 10.1038/s41597-022-01899-x [DOI] [PMC free article] [PubMed]
- 31.Benchimol EI, Smeeth L, Guttmann A, et al. The REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) statement. PLoS Med. 2015;12(10):e1001885. 10.1371/journal.pmed.1001885. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Birkelo BC, Pannu N, Siew ED. Overview of diagnostic criteria and epidemiology of acute kidney injury and acute kidney disease in the critically ill patient. Clin J Am Soc Nephrol. 2022;17(5):717–35. 10.2215/cjn.14181021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Botta-Dukát Z. Quartile coefficient of variation is more robust than CV for traits calculated as a ratio. Sci Rep. 2023;13(1):4671. 10.1038/s41598-023-31711-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Kjeldgaard M, Mæng MO, Torp-Pedersen C, et al. 90-day mortality risk related to postoperative potassium levels in patients undergoing coronary bypass surgery. Journal of Molecular and Cellular Cardiology Plus. 2023. 10.1016/j.jmccpl.2023.100035. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Patel RB, Tannenbaum S, Viana-Tejedor A, et al. Serum potassium levels, cardiac arrhythmias, and mortality following non-ST-elevation myocardial infarction or unstable angina: insights from MERLIN-TIMI 36. Eur Heart J Acute Cardiovasc Care. 2017;6(1):18–25. 10.1177/2048872615624241. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Engelhardt LJ, Balzer F, Müller MC, et al. Association between potassium concentrations, variability and supplementation, and in-hospital mortality in ICU patients: a retrospective analysis. Ann Intensive Care. 2019;9(1):100. 10.1186/s13613-019-0573-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.O’Brien B, Campbell NG, Allen E, et al. Potassium Supplementation and Prevention of Atrial Fibrillation After Cardiac Surgery. Jama. 2024. 10.1001/jama.2024.17888. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Zhang Q, Zhao R, Wu Y, Zhang L, Feng Y. Etomidate combined with propofol versus remimazolam for sedation in elderly patients during gastrointestinal endoscopy: a randomized prospective clinical trial. Drug Des Devel Ther. 2024;18:2681–92. 10.2147/dddt.S454314. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Lombardi G, Gambaro G, Ferraro PM. Serum potassium variability is associated with increased mortality in a large cohort of hospitalized patients. Nephrol Dial Transplant. 2022;37(8):1482–7. 10.1093/ndt/gfab211. [DOI] [PubMed] [Google Scholar]
- 40.Hessels L, Hoekstra M, Mijzen LJ, et al. The relationship between serum potassium, potassium variability and in-hospital mortality in critically ill patients and a before-after analysis on the impact of computer-assisted potassium control. Critical Care. 2015;19(1):4. 10.1186/s13054-014-0720-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Eliacik E, Yildirim T, Sahin U, et al. Potassium abnormalities in current clinical practice: frequency, causes, severity and management. Med Princ Pract. 2015;24(3):271–5. 10.1159/000376580. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Dixon DL, Abbate A. Potassium levels in acute myocardial infarction: definitely worth paying attention to. Eur Heart J Cardiovasc Pharmacother. 2015;1(4):252–3. 10.1093/ehjcvp/pvv029. [DOI] [PubMed] [Google Scholar]
- 43.Campbell NG, Allen E, Sanders J, et al. The impact of maintaining serum potassium ≥ 3.6 mEq/L vs ≥ 4.5 mEq/L on the incidence of new-onset atrial fibrillation in the first 120 hours after isolated elective coronary artery bypass grafting – study protocol for a randomised feasibility trial for the proposed Tight K randomized non-inferiority trial. Trials. 2017. 10.1186/s13063-017-2349-x. [DOI] [PMC free article] [PubMed]
- 44.Cohn JN, Kowey PR, Whelton PK, Prisant LM. New guidelines for potassium replacement in clinical practice: a contemporary review by the National Council on Potassium in Clinical Practice. Arch Intern Med. 2000;160(16):2429–36. 10.1001/archinte.160.16.2429. [DOI] [PubMed] [Google Scholar]
- 45.Weiner ID, Wingo CS. Hypokalemia–consequences, causes, and correction. J Am Soc Nephrol. 1997;8(7):1179–88. 10.1681/asn.V871179. [DOI] [PubMed] [Google Scholar]
- 46.Bakaeen FG, Shroyer AL, Gammie JS, et al. Trends in use of off-pump coronary artery bypass grafting: Results from the Society of Thoracic Surgeons Adult Cardiac Surgery Database. The Journal of thoracic and cardiovascular surgery. 2014;148(3):856–3. 10.1016/j.jtcvs.2013.12.047. [DOI] [PubMed]
- 47.Fudulu D, Benedetto U, Pecchinenda GG, et al. Current outcomes of off-pump versus on-pump coronary artery bypass grafting: evidence from randomized controlled trials. J Thorac Dis. 2016;8(Suppl 10):S758-s771. 10.21037/jtd.2016.10.80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Böning A, Diegeler A, Hilker M, et al. Preoperative atrial fibrillation and outcome in patients undergoing on-pump or off-pump coronary bypass surgery: lessons learned from the GOPCABE trial. Interactive cardiovascular and thoracic surgery. 2015;20(1):74–8. 10.1093/icvts/ivu331. [DOI] [PubMed]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.