Among patients admitted to the intensive care unit (ICU), more than 70% are aged > 65 years. Owing to their underlying comorbidities and concomitant medications, the incidence of adverse drug reactions can be markedly higher among the elderly. As corrected QT interval (QTc) prolongation occurs temporarily and is often asymptomatic and unpredictable, it is almost impossible to detect QTc prolongation in the ambulatory population. Thus, although fluoroquinolones (FQs) available in the market are less likely to prolong QT interval than antiarrhythmic agents, their cardiac toxicity may be underestimated in real clinical field. Pharmaceutical companies have reported that the incidence of QTc prolongation caused by FQ administration ranges between 0.1% and 1%, but there is lack of FQ-induced long QT incidence and associated risk factors in critically ill patients who have any life-threatening condition that requires various pharmacological and/or mechanical support of vital organ functions. Medical ICU (MICU) is a specialized unit for providing the critical care and life support for critically ill patients under 24-hour hemodynamic monitoring including cardiac rhythm. This study carried out to determine the incidence of QT interval prolongation in patients receiving FQs and identify the associated risk factors in critically ill patients under 24-hour monitorable MICU.

Study design and population

This longitudinal, retrospective study included patients aged ≥ 18 years admitted to the MICU at a single tertiary hospital between January 1, 2018 and December 31, 2018 who received systemic levofloxacin, moxifloxacin, ciprofloxacin, or gemifloxacin antibacterial therapy. Patients were primarily excluded in case of congenital long QT syndrome, active chemotherapy, antiretroviral therapy, or no electrocardiogram (ECG) data.

Data collection

Patient data from electronic medical records, including demographic characteristics, laboratory data such as serum creatinine (Cr), estimated glomerular filtration rate (eGFR), and electrolytes, use of loop diuretics and number and classification of concomitant medications based on drug lists from CredibleMeds,¹⁾ were collected. Comorbidities were also reviewed, particularly present sepsis because it was one of the most common causes of MICU admission and associated with the administration of various drugs. In addition to the aforementioned list of QT-prolonging drugs, treatment with mannitol, which could provoke electrolyte imbalances, and chlorpheniramine, a first-generation histamine H1 receptor antagonist (H1A) that causes rare but severe cardiac arrhythmia, 2) was examined. Patients who received two types of FQs, irrespective of the occurrence of QTc prolongation, were classified as separate cases. eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKDEPI) equation based on sex, age, and creatinine level. Laboratory data were recorded at the time of QT prolongation, except for initial parameters. Serum magnesium concentrations were collected for each patient during the ICU stay, but not measured daily. Therefore, if there were no measured levels at the time of QT prolongation, the closest recorded values were used. Laboratory data of the non-QTc prolongation group were recorded the lowest levels during FQ administration. Hypokalemia, hypocalcemia, and hypomagnesemia were defined as ≤ 3.5 mEq/L, < 8.6 mg/dL, and < 1.5 mg/dL, respectively. Patients’ medical histories were classified using three categories: neurologic, cardiac, or endocrine disorders.

A 12-lead ECG (MAC 5500 HD, GE Healthcare, USA) was performed regularly daily and when QTc prolongation occurred by continuous bedside cardiac rhythm, lead II monitoring (CarescapeTM B850, GE Healthcare, USA). A initial QT interval was acquired within 4 h of ICU admission. The QT intervals for the detection of QT prolongation in this study were measured from lead Ⅱ of 12-lead ECG. The QT interval was defined as the period from the beginning of the QRS complex to the end of the T-wave and is influenced by various physiologic and pathologic factors, such as heart rate, diurnal variations, autonomic activities, and electrolyte disorders. 3) To account for the variability in QT interval due to differences in the heart rate, QT interval was corrected using the Bazett formula (QTc). For QTc prolongation occurring several times or consecutively, the QTc was recorded based on the time of first occurrence. In patients who did not develop QTc prolongation, the highest QTc was documented during FQ administration. The QT prolongation in this study was defined as QT prolongation that occurred during FQ use without other identifiable triggers such as electrolyte abnormalities, not based on the initial ECG. The criteria were defined as QTc ≥ 450 ms for men and ≥ 470 ms for women. 4),⁵⁾ Every patient in the ICU underwent 24-hour cardiac monitoring. This information was sent to central monitoring and one medical team member observed changes of hemodynamic, ECG, respiratory supports, etc., through central monitoring. In case of abnormality, the observer reported it to the attending physician and the intensivists. The attending physician immediately measured the 12-lead ECG. MAC 5500 HD is well known as a validated ECG analysis system, but it was evaluated again by two intensivists, and it was judged as QT prolongation when both of them agreed.

Statistical analysis

Data were compiled in Microsoft Excel (Microsoft Corporation, Redmond, WA, USA) and analyzed using IBM SPSS Statistics, version 21.0 (IBM Corporation, Armonk, NY, USA). Categorical variables are presented as counts and percentages. Continuous variables with normal distributions are presented as the mean and standard deviation (SD), and continuous variables with non-normal distributions are presented as the median and interquartile range (IQR). To determine normality of continuous parameters, Kolmogorov– Smirnov (n ≥ 50) and Shapiro–Wilk (n < 50) were used. Both groups were tested for normality, and if one or both did not follow normality, nonparametric methods were performed. Baseline characteristics were compared using chi-square tests or Fisher’s exact tests for categorical variables and Student’s t-tests or Mann–Whitney U tests for continuous variables.

Dummy variable regression analysis was performed to investigate the effect of the type of FQ on the QTc. In order to exclude other factors and determine whether QT prolongation is due to FQ administration, the difference in QTc between the day when long QT was observed and the day before was analyzed. The degree of long QT occurrence was compared with each FQ based on the most introduced levofloxacin. To identify the risk factors associated with QTc prolongation under systemic FQ treatment, multivariate logistic regression analysis was performed in patients who did or did not develop QTc prolongation. All variables with p < 0.1 in univariate analysis were included in the multivariate logistic regression analysis. Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated for each variable. Statistical significance was set at p < 0.05. Before performing logistic regression analysis, Pearson correlation analysis was used to determine the relative influence between variables. The variance inflation factor was used to detect multicollinearity in multivariate statistical analysis.

Ethics approval and consent to participate

This study was approved by the Institutional Review Board of the Jeonbuk National University Hospital (approval number 2020-09-023-001). This study was a retrospective analysis of electronic medical records, and all patient data were anonymized; therefore, the requirement for informed consent was waived by the institutional review board.

Baseline characteristics of the patients

A total of 455 patients were admitted to the MICU during the study period. A total of 126 patients were administered at least one of the four types of FQs. Of these, 13 patients who were considered to require FQ infusion despite QTc prolongation switched from one FQ to another FQ. Thus, we analyzed 139 patients treated with four types of FQs. Patients were counted as two cases when two types of FQs were administered (Fig. 1). The median age of the patients was 77 years (IQR = 63–81), and 50 patients (36.0%) were female. The median body mass index (BMI) was 22.5 kg/m² (IQR=19.5–24.8), and only four patients had a BMI ≥ 30 kg/m². The baseline characteristics of the patients are shown in Table 1.

Surprisingly, FQ administration in critically ill patients resulted in very high rate of QTc prolongation (119/139, 85.6%). More patients in the QTc prolongation group had a history of smoking (41.2% vs. 15.0%, p=0.025), sepsis (34.5% vs. 15.0%, p=0.08), and use of loop diuretics (88.2% vs. 55.0%, p=0.001) than those in the non-QTc prolongation group. A higher number of patients in the non-QTc prolongation group exhibited hypertension and renal impairment, with eGFR < 30 ml/min/1.73 m² (60.5% vs. 70.0%, p=0.42 and 23.5% vs. 45.0%, p=0.04, respectively; Table 1). Patients in the QTc prolongation group had lower serum potassium levels but were less likely to have hypomagnesemia and hypocalcemia than those in the non-QTc prolongation group (Table 2).

FQs and concomitant medications

This study analyzed patients treated with levofloxacin (n=43), moxifloxacin (n =35), gemifloxacin (n=15), and ciprofloxacin (n=46). In 130 patients, an FQ was administered within 14 days; in eight patients, it was administered for approximately 3 weeks depending on patient response and site of infection. There was only one patient who received levofloxacin for more than 50 days for the treatment of pulmonary tuberculosis. There was no difference among the groups in the duration of FQ administration (Table 3). The initial QTc median in the gemifloxacin group was 506.0 ms (IQR=479.0–524.0), which was the highest among all FQ groups. However, of the 15 gemifloxacin cases, eight were cases of second-line FQ use after the discontinuation of another FQ (Table 3).

Dummy variable regression was performed in 133 cases with regularly measured ECG records; 3 levofloxacin cases, 2 moxifloxacin cases, and 1 ciprofloxacin case were excluded from this analysis. We found that the difference in QTc between the day when QTc prolongation developed and the day prior differed depending on the type of FQ. Levofloxacin cases showed the largest difference in QTc among the other FQs, whereas gemifloxacin cases showed the smallest difference (Table 4). Concomitant medications with potential for QT prolongation were classified according to the risk stratification proposed by CredibleMeds, an online database of drugs that cause QT prolongation and TdP. In 106 (76.3%) patients who received FQs, at least one QTc-prolonging drug was administered. The most commonly used medications were nicardipine (34.5%), amiodarone (25.2%), and dexmedetomidine (22.7%; Table 5).

Risk factors for QTc prolongation

Baseline characteristics, laboratory data, and QTc on ECG were compared between patients with and without QTc prolongation using CredibleMeds. Univariate logistic regression showed that the risk factors for QTc prolongation were female sex, smoking history, liver failure, serum potassium level, eGFR ≤ 30 mL/min/1.73 m2, serum magnesium level ≤ 1.5 mg/dL, initial QTc ≥ 450 ms, loop diuretics use, total dose of diuretics within 48 h, sepsis, and number of medications with a known risk of TdP (Tables 1 and 2). Multivariate logistic regression analysis identified use of loop diuretics (p=0.038, OR=7.66, 95% CI=1.12–52.47), number of medications with known risk of TdP (p=0.028, OR =4.83, 95% CI=1.18–19.79), and sepsis (p=0.034, OR=8.81, 95% CI=1.18–65.96) as independent risk factors for the development of QTc prolongation (Table 6). Sepsis is the highest independent risk factor of long QT after FQ administration.

In this study, we evaluated the incidence of QTc prolongation following the administration of four types of FQs in the ICU through real-time ECG monitoring. The prevalence of FQ-induced QT prolongation was 85.6% in the ICU. Also, the frequency of TdP according to QT prolongation was investigated, but this was not observed. This is because FQ-induced long QT was confirmed through 24-hour ECG monitoring and drug administration was immediately stopped before a fatal arrhythmic course appeared. Levofloxacin was associated with longer QTc than other FQs. FQ-induced QT prolongation in critically ill patients was associated with several risk factors, including use of loop diuretics, sepsis, and use of medications with known risk of TdP according to Credible-Meds.

Previous studies on cardiac adverse effects associated with FQs found that moxifloxacin prolongs the QT interval with a clinically meaningful IC50 value, whereas levofloxacin and ciprofloxacin were not associated with QT prolongation.^6),7) This study showed that levofloxacin produced the largest QT interval prolongation, and that ciprofloxacin and gemifloxacin, but not moxifloxacin, had statistically significant effects. Unlike previous studies on intravenous levofloxacin (500 mg once a day), in this study, our patients received a standard dose of levofloxacin according to the appropriate indications.^8),9) Specifically, 750 mg levofloxacin is strongly recommended for the treatment of pneumonia, including community-acquired pneumonia. We adjusted the dose of levofloxacin based on eGFR by CKD-EPI and used a median daily dose of 750 mg (IQR=500–750 mg). Given that only a few studies have compared 750 mg levofloxacin with 400 mg moxifloxacin, our results could aid in the selection of more appropriate and safer FQs for the treatment of infections in elderly patients in clinical settings. Although moxifloxacin is classified as an intermediate inhibitor of the human Ethera- go-go (hERG) channel and levofloxacin as a weak inhibitor of the hERG channel based on the potency of each agent, there are insufficient data to determine the incidence of QT prolongation for individual FQs.¹⁰⁾ Our findings indicate that additional clinical studies accounting for specific risk factors are needed.

The female sex is a strong risk factor for the development of drug-induced arrhythmia. 11),¹²⁾ In addition, based on the known risk factors for TdP, women were more likely to exhibit prolonged cardiac repolarization than men, thus implicating the female sex as a risk factor for QT prolongation. 5),¹³⁾ However, we found no association between sex and delay in cardiac repolarization. It is noteworthy that, the average age of our patients was 77 years. Furthermore, the mechanism underlying the higher incidence of TdP and QT prolongation among women remains unclear. 14) This is speculated to be due to changes in sex hormones with age and not sex-related disparities. Estrogen reduces the repolarization reserve, which might increase the susceptibility to drug-induced QT prolongation among women.¹⁵⁾ In contrast, testosterone modulates cardiac repolarization by promoting hERG channel currents and reducing QT interval.¹⁶⁾ Whether estrogen or testosterone affects the QT interval remains unclear. Sex hormone deficiencies in both men and women may explain why we did not find any association between sex and QT prolongation in this study.

The administration of loop diuretics, especially at baseline, can increase the risk of QT prolongation and lead to TdP.¹⁷⁾ We identified the use of loop diuretics as an independent risk factor for QTc prolongation. Loop diuretics are mainly used for the treatment of pulmonary edema and acute kidney injury because they promote the excretion of water and sodium into the urine. Previous studies have shown that hypokalemia caused by loop diuretics lengthens the QT interval.^18),19) We recorded the concentrations of the electrolytes including the potassium at the time of QTc prolongation. Multivariate logistic regression showed no relationship between serum potassium levels and QTc prolongation. A previous study reported that diuretics, such as indapamide,²⁰⁾ block potassium currents directly, resulting in QT interval prolongation. The mechanism by which loop diuretics serve as a risk factor for QTc prolongation and TdP has not yet been clarified. However, despite our results, it may be related to potassium levels, particularly total body potassium levels. 17),²¹⁾ Most of the total body potassium is present in cells; only 2% is in the extracellular fluid, including the blood. Since the maintenance of potassium homeostasis between the intracellular and extracellular spaces is crucial for normal cell function, even small changes that cause potassium imbalance could lead to lifethreatening symptoms. Our simple logistic regression analysis results showed that the average potassium level was lower in the long QT group than in the nonlong QT group, although the average potassium level in both groups was within the normal range. Therefore, it may be helpful to maintain a potassium level in the upper normal range through sufficient administration of potassium when using loop diuretics.²¹⁾

TdP is an uncommon but potentially lifethreatening ventricular arrhythmia associated with congenital long QT syndrome or QT prolonging drugs.¹⁴⁾ Prolongation of the QT interval may increase the risk of TdP, which can lead to palpitations, syncope or sudden cardiac death. The exact prevalence of drug-induced TdP is not known. Though FQs-asociated QT prolongation in critically ill patients were more than 85% in this study, no TdP was observed with QT prolongation. Because the probable related drugs were discontinued and proper management of offending causes was performed immediately if QT prolongation identified and the possibility of FQs-associated QT prolongation could not be rule out transient finding, the frequency of death-related arrhythmia and TdP were not noted in the study. Among the drugs with known risk of causing TdP, the most used medication in combination with FQs in our patients was amiodarone. Many drugs, including amiodarone, prolong QT to inhibit the delayed rectifier K+ current (IKr). Antihistamines cause long QT in a similar manner²²⁾; however, we found that chlorpheniramine, an H1A, was not associated with QT prolongation. This may be due to the degree of IKr inhibition or the duration and dose of H1A therapy.²²⁾ We used a single dose of amiodarone, if necessary, to reduce the risk of QT prolongation when chlorpheniramine was co-administered and to prevent an overdose. Drug-induced QT prolongation increases the risk of TdP, which is associated with an increase in drug concentration in the body or the number of QT-prolonging drugs.²³⁾ Drug–drug interactions, renal or hepatic dysfunction, and the accumulation of concomitant drugs can increase the serum concentration of a drug. However, our multidisciplinary ICU team recognized the seriousness of drug interactions caused by pharmacokinetic and pharmacodynamic characteristics and conducted appropriate prescription reviews and medical interventions. Therefore, elevated drug concentrations cannot explain QT prolongation, especially in ICU patients. Rather, the concurrent use of several QT-prolonging drugs that block potassium currents may reduce repolarization reserve and lessen its protection against QT prolongation.²⁴⁾

Interestingly, sepsis is the highest independent risk factor of long QT after FQ administration. Sepsis is a condition in which a systemic inflammatory response occurs due to microbial infection. It is accompanied by organ failure, mainly involving the lungs, kidneys, and the cardiovascular system. In particular, sepsis-induced myocardial dysfunction (SIMD) is characterized by ECG changes similar to myocardial infarction (MI), including left bundle branch block, narrow QRS interval, and lengthening of the QT interval.²⁵⁾ However, a previous study showed that reduced coronary blood flow is not associated with myocardial dysfunction in sepsis.²⁶⁾ This means that SIMD is not triggered by acute coronary syndrome and particularly MI. Although the mechanism of SIMD has not yet been clarified, chemical mediators involved in sepsis may play an important role in myocardial depression.^26),27) In particular, sepsis is characterized by the release of proinflammatory cytokines, such as tumor necrosis factor alpha, interleukin (IL)-1β, and IL-6.25) These proinflammatory cytokines induce the release of nitric oxide (NO) by activating endothelial cells. Although NO acts as a regulator of coronary vascular tone and myocardial function, enhanced NO contributes to depressed cardiac function, leading to QT prolongation.^26),28)

Since this study was conducted under 24-hour monitoring, it had the advantage of being able to identify even hidden abnormal ECG changes without symptoms, but had some limitations in that critically ill patients were very heterogeneous including sepsis itself. First, it was retrospective in nature and included a relatively small number of patients from a single tertiary medical care institution. This made it difficult to adjust for the diverse and complex clinical variables occurring in the ICU. We attempted to compare the changes in electrolyte levels between baseline and the time of QT prolongation at the beginning of this study. However, it was a retrospective design and electrolyte levels were measured daily, but the electrolyte protocols were implemented accordingly. This made it difficult to identify the association between changes in parameters and QT prolongation. Second, the small size of the control group may have limited the evaluation of risk factors associated with prolonged QT interval. Third, some patients were duplicated in this analysis, raising concerns about independent events. Using logistic regression analysis, it was confirmed that a causal relationship was not established between variables.

Among critically ill patients, the use of QT-prolonging agents, such as FQs and antiarrhythmic agents, must be carefully considered particularly in sepsis. Identification of risk factors that predict the development of QT prolongation in individual patients is needed. If modifiable factors exist, pre-emptive correction should be implemented. A multicenter study to validate risk factors and establish a simple and systematic system for screening highrisk patients should be continued in the future.

We would like to thank Erin Haase PhD (www.BioScienceWriters.com) and Editage (www.editage.co.kr) for editing and proofreading this manuscript for English language. This study was supported by the Fund of Biomedical Research Institute, Jeonbuk National University Hospital.

Conflict of interests: No potential conflict of interest relevant to this article was reported.