Hyperkalemia is a common electrolyte imbalance in adults with a potentially peri-arrest risk, but of reversible possibility when diagnosed and managed in time. This study highlights the challenges associated with the diagnosis and management of hyperkalemia. Hyperkalemic patients in our study tend to be older adults and elderly and suffer from several comorbidities, such as HTN, DM, and chronic kidney disease. Besides, drugs inducing hyperkalemia were common, including analgesics, beta-blockers, CCBs, ACEIs, and ARBs. Given the non-specific clinical presentations of hyperkalemia, and about one-quarter of our patients were asymptomatic, we evaluated ECG as an attainable test to raise the possibility of hyperkalemia. The results indicated a poor sensitivity of initial ECG and presenting symptoms in detecting hyperkalemia as it ranged between 0.28 and 0.36 and improved minimally when potassium ≥6.5 mmol/L with peaked T wave was significantly more observed than in mild and moderate hyperkalemia. Thus, the absence of ECG alterations suggestive of hyperkalemia should not lower the physician’s suspicion of the presence of hyperkalemia in high-risk patients. Also, a mean delay of 1 h from triage to initial hyperkalemia treatment was observed, which is an alarming finding.
Hyperkalemia is usually multifactorial in etiology, and we did not attempt to investigate the specific causes for each case of hyperkalemia. However, in this study, our patients with elevated potassium levels had one or more conditions causing hyperkalemia such as comorbid HTN, DM, or CKD, and high rates of certain medications use, including non-steroidal anti-inflammatory drugs (NSAIDs) beta-blockers, CCBs, ACEIs, and ARBs. These findings are concordant with previous reports on the risk factors of hyperkalemia [6, 14, 15, 47,48,49,50,51,52,53]. Our study as well reported high rates of statins and PPIs to use within hyperkalemic patients. Previous reports indicated that statin use as a cytochrome P450s 3A4 inhibitor, especially in combination with antihypertensive medications, could contribute to acute kidney injury and hyperkalemia [54, 55]. Previous investigators also observed elevated serum potassium levels among PPI users [56, 57]. However, as this is a cross-sectional study, we could not establish causality associations between statins and PPIs with hyperkalemia. Our population’s high rates of insulin use represent a marker for more advanced contributing comorbidities such as DM, which may induce hyperkalemia through developing type IV renal tubular acidosis. Thus, we could conclude that the patients with multiple comorbidities such as HTN, DM, or CKD and those on hyperkalemia-inducing medications were eligible for regular monitoring of electrolytes disturbances.
ECGs had abnormalities consistent with hyperkalemia among around two third of the studied patients with hyperkalemia, and the most common alterations were elevations of T wave amplitude and QRS duration. As well, increased PR interval and QRS duration and presence of peaked T wave were correlated with serum potassium levels. These findings align with previous reports of a higher frequency of ECG alterations suggestive of hyperkalemia with elevated serum potassium levels [33, 58]. Trail has found that ECG disturbances, including peaked T waves and an increase in the duration of the QRS complex, were associated with hyperkalemia and more evident with a serum potassium level of ≥7.8 mEq/L [34]. T wave in ECG occurs due to repolarization of ventricles, whereas QRS duration represents the time for ventricular depolarization, and PR interval represents the time between atrial depolarization and ventricular depolarization. Hyperkalemia causes an increase in the velocity of phase 3 of the action potential, which is associated with the peaking of the T wave. Also, hyperkalemia causes a decrease in the resting membrane potential of myocardial cells with less negativity which causes conduction defects and prolongation of the PR intervals and QRS complexes [59,60,61].
Varga et al. reported the QRS widening, peaked T waves, first-degree heart block, and bradycardia as the most frequent ECG alterations suggestive of hyperkalemia in severely hyperkalemic patients with serum potassium levels of >7 mmol/L (31.6%, 18.4%, 18.4%, 18.4%, respectively) [58]. These ECG alterations were significantly more common among severely hyperkalemic patients than in normokalemia patients (8.2, 4.7, 7.1, and 6.5%, respectively) [58]. Hicks, in his case report, assessed hyperkalemia-associated ECG findings in a 34-year-old female with DM, abnormal cardiac rhythm, and no known history of renal failure presented to the ED [62]. The most significant findings were peaked T waves and widening QRS complex with a potassium level of 7.6 mmol/L [62]. Peaked T waves could be considered one of the typical and earliest ECG signs of elevated serum potassium levels [2, 22, 32,33,34, 62, 63].
Other ECG alterations suggestive of hyperkalemia in our study included flattening and disappearance of P waves, RBBB, and ST elevations were observed. Similarly, in a clinical review, a woman presented to the ED with respiratory distress and altered mental status and had an elevated serum potassium level of 9.6 mmol/L; her ECG recorded ST elevations, RBBB, and loss of P wave amplitude [64]. ST segment is the state of the ventricles between repolarization and depolarization. ST elevation had been previously linked to hyperkalemia and called “pseudo-infarction”; therefore, hyperkalemia is a potential differential diagnosis for the cause of elevations in the ST segments [65]. However, the mechanism of ST segment elevation due to hyperkalemia is not already known [64]. Also, ST depression and shortening of the QTc interval had been reported in several investigations as manifestations of hyperkalemia [23, 66, 67].
Most studies suggested an association between lower potassium levels and a higher risk of atrial fibrillation [68,69,70,71]. However, our results reported atrial fibrillation as the most common arrhythmia observed among hyperkalemic patients. This finding is concordant with Varga et al.’s findings that aThe trail fibrillation was more prevalent in severely hyperkalemic patients than normokalemia patients [58]. We attribute these results to the synergistic effect of CHF and CKD, which often present in patients with high serum potassium levels. Hyperkalemia and CHF are common in chronic kidney disease, and CHF could cause atrial fibrillation. Thus, atrial fibrillation occurs not as the result of hyperkalemia but rather as the consequence of illnesses often associated with hyperkalemia.
ECG is an inexpensive, broadly available, and easily attainable test. There have been conflicting reports about its sensitivity and specificity to capture elevated serum potassium levels [22, 29,30,31,32, 37, 63]. Our study indicated poor sensitivity of initial ECG and clinical presentation in detecting hyperkalemia as ranged between 0.28 and 0.36. These results are concordant with previous studies showing that physicians’ ability to predict hyperkalemia from the ECG was low with sensitivities between 0.43 and 0.34, and experienced readers’ ability to predict the severity of hyperkalemia was likewise poor [22, 32]. Similarly, Rafique et al. reported a mean sensitivity of 0.19 (± 0.16) for the emergency physicians detecting hyperkalemia based on the ECG, and this sensitivity improved to 0.29 (± 0.20) in severe hyperkalemia [39]. Varga et al. captured ECG alterations suggestive of hyperkalemia among 46% of the hyperkalemic patients, and surprisingly 24% of normokalemia patients exhibited such ECG alterations [58]. Thus, based on ECG analysis and with or without presenting symptoms knowledge, the physician could not confirm or exclude hyperkalemia, and serum laboratory tests should be conducted for accurate hyperkalemia diagnosis. Montague et al. had conducted a study on ninety patients diagnosed with hyperkalemia as serum potassium of ≥6 mmol/L [33]. The authors reported that the probability of ECG changes increased with increasing potassium levels, but the sensitivity and specificity of ECG changes in diagnosing hyperkalemia were poor [33]. It could be concluded that the management of hyperkalemia should be guided by the clinical scenario and serial laboratory potassium measurements, and the absence of ECG alterations suggestive of hyperkalemia should not lower the physician’s concern for the presence of hyperkalemia in high-risk patients.
Although the lack of sensitivity in detecting hyperkalemia based on ECGs ultimately depends on physicians’ interpretations, other confounding factors could not be excluded. First, the possible effects of other electrolytes, such as calcium and magnesium, in mitigating the ECG changes suggestive of hyperkalemia as proposed by prior investigators [31, 72,73,74]. Second, 64% of our patients suffered from CKD, and about one third of participants were on regular dialysis, which could cause the non-specificity of ECG abnormalities. It was reported that hemodialysis patients with hyperkalemia were less likely to show ECG changes despite the risk of suddenly developing arrhythmias as the myocytes were less sensitive to electrolyte changes in these patients; therefore, hyperkalemia did not manifest in them its typical forms [25, 37, 75]. Third, the rate of increase in serum potassium levels could affect the development of ECG changes [1, 21, 31]. As the velocity of serum potassium concentrations risen was unknown to the readers, their insensitivity could be attributed to the slowly rising potassium levels, especially in the setting of CKD. Fourth, patients’ medications such as digitalis could have interacted with the effects of electrolytes on myocytes and masked the effects of hyperkalemia. Lastly, metabolic acidosis and ischemia could be associated with arrhythmias and ST and T wave alterations in the patterns suggestive of hyperkalemia [76]. Although including serum markers of cardiac ischemic injury and ABGs for acidosis detection, the absence of abnormalities in these serum markers does not exclude them as confounding factors. There is also the potential that elevated serum potassium levels may potentiate arrhythmias that could be attributable to other causes. However, these possibilities could not be ruled out as they are part of clinical practice.
One of our most striking findings was that the meantime from triage to initial hyperkalemia treatment of more than one hour. Freeman et al. investigated the possible effects of presentations and ECGs on triage time to the initial hyperkalemia management [22]. The authors found that most hyperkalemic patients waited for a median of 2 h from triage to initial treatment, even though ECG was performed before the laboratory serum potassium measurement [22]. Also, the delay in hyperkalemia treatment was reported among hospitalized patients, with approximately 2 h delays from laboratory notification of potassium to initiation of treatment [2].
In our study, we observed some behaviors in hyperkalemia treatment, which would explain the delays. In some cases, the initial physician response to an unexpectedly elevated potassium level was to repeat the serum potassium test and obtain intravenous access. In other cases, difficulties in performing intravenous access were documented, which is expected for a population including 64% suffered from CKD and 24% of patients on regular dialysis. However, other treatments should be considered in patients with difficult intravenous access, including high-dose inhaled beta-agonist therapy or direct intravenous injection of hyperkalemia therapies in a life-threatening situation.
Strengths, implications, and limitations of this study
The strengths of this study included the prospective nature of data collection and processing, uniform data collection, consistent definitions applied, and detailed periodic review of the abstracted data, which support the integrity and validity of the collected data. Also, the ECGs were conducted simultaneously within 1 h of serum potassium level measurement and before initiation of therapy, which assured tight data pairing. In addition, the initial ECGs and patients’ symptoms at the time of ED presentation were interpreted by an emergency physician and a cardiologist to predict hyperkalemia while blinding to all laboratory values, study design, patients’ diagnoses and comorbidities, and each other’s readings to reduce bias. Since the readers considered the possibility of non-hyperkalemia diagnoses when interpreting the initial ECGs, our study was more realistic and similar to the clinical emergency practice and reduced the readers’ reported sensitivities. Also, this study highlighted the delay in hyperkalemia treatment registration that might be considered a failure due to a missed diagnosis based on clinical presentations and initial ECG alterations.
All these factors contributed to a more robust study, which supports the conclusions of previous reports that the clinical presentations and ECGs are not reliable tools in the diagnosis of hyperkalemia [22, 32, 39, 58]. Although our study indicated that the recognition of hyperkalemia is challenging and the initiation of appropriate therapy is frequently delayed, it had been suggested by Riccardi et al. that initiation of intravenous calcium gluconate as a life-saving treatment to stabilize the cardiac membrane in suspected hyperkalemia before laboratory confirmation would be advisable [77]. However, previous investigations reported that the empiric hyperkalemia therapy based on ECG solely was associated with the mistreatment of approximately 15% of patients [32]. Thus, given the non-specific nature of the patients’ clinical presentations and the variability of ECG presentations of hyperkalemia, it is prudent to delay hyperkalemia treatment in relatively stable patients until laboratory confirmation of hyperkalemia. In unstable patients, intravenous calcium gluconate as an empirical treatment for hyperkalemia could be initiated based on ECG alterations solely. Also, the absence of ECG alterations suggestive of hyperkalemia should not affect the physician’s suspicions of hyperkalemia in high-risk patients. In light of these facts and the reported delays for the initiation of hyperkalemia therapy, technological advancements should be considered in high-risk patients, such as using finger-stick testing [78] and incorporating artificial intelligence into the ECGs [78].
Our study has several limitations. Firstly, the inherent limitations associated with the cross-sectional design could not establish causality inference. This study did not include a control group with normal serum potassium levels to compare, limiting our results’ internal validity; therefore, we were unable to calculate the specificity and predictive values of ECG in diagnosing hyperkalemia. Also, it was conducted on a relatively small sample size of patients admitted to emergency care and a small number of ECG readers. Moreover, a sample size calculation was not done. This study was conducted at a single center and on a narrow ethnicity of patients, which limits the generalizability of our results and conclusions beyond our patients. Hence, a larger number of hyperkalemic patients with different ethnicities and appropriate disease prevalence rates calculation with involving more evaluators would help improve the robustness of the study. Also, there might have been human error in conducting and interpreting ECGs, which was not being considered. However, we uploaded a color copy of the ECG with the highest quality to reduce misinterpretations, and this study included a relatively small number of ECGs to reduce the possibility of readers’ fatigue.
Another limitation is the interpretation of the initial ECGs solely as isolated tracings in this study without comparing them with the previous ECG tracings. Also, the notations of dynamic ECG changes during the emergency course were not recorded and were unavailable to the readers to interpret the initial ECGs. However, in the actual practice, a comparison of the initial ECGs with prior ECGs might contribute to the delay in hyperkalemia therapy administration. Also, there is potential for confounding factors in the interpretation of ECGs that might modify the ability of physicians to predict hyperkalemia since several ECG alterations could be attributed to other causes than hyperkalemia or ECG changes due to other conditions might mask the ECG signs of hyperkalemia. These confounding factors included other electrolyte abnormalities, such as calcium and magnesium, CKD, hemodialysis, and drugs such as digitalis, metabolic acidosis, and myocardial ischemia that might mask the ECG changes suggestive of hyperkalemia. However, this study collected and included the patients’ comorbidities, used medications, ABG measurements, and cardiac ischemic injury serum markers. Finally, the time to hyperkalemia treatment was calculated based on the triage time and the first hyperkalemia therapy administration by the nurse. This methodology could mask other potential causes for delay in hyperkalemia treatment, such as the time spent before triage and the time between physician order and nursing administration of treatment. Further prospective larger-scale studies examining the ECG alterations suggestive of hyperkalemia in hyperkalemic patients compared with normokalemia individuals with attention to other electrolytes levels would be needed to confirm our findings.