QTc Interval Prolongation and Its Association With Electrolyte Abnormalities and Psychotropic Drug Use Among Patients With Eating Disorders

Abstract

Background

Eating disorders (EDs) often develop during adolescence with high mortality rates. Sudden cardiac death in these patients has been associated with corrected QT (QTc) interval prolongation. The significance of extrinsic factors on QTc prolongation in populations with EDs remains controversial. This study assessed the relationship between QTc prolongation in paediatric patients with EDs and extrinsic factors, such as QTc-prolonging medications and electrolyte abnormalities to investigate whether an ED alone is associated with an increased prevalence of QTc prolongation.

Methods

Electrocardiograms, electrolytes, and psychopharmaceutical usage were retrospectively analysed from the charts of 264 paediatric patients with EDs. Descriptive statistics were used to assess QTc prolongation and its relationship with electrolyte abnormalities and psychopharmaceuticals.

Results

Of 264 patients, 227 had normal QTc intervals (<440 ms), whereas 37 had borderline prolonged (440-460 ms) or prolonged (>460 ms) intervals. The prevalence of QTc intervals exceeding 440 ms in patients with normal electrolytes and not using QTc-prolonging psychotropics mirrored that of the general population (P = 0.59). Of the 23 patients taking psychotropics, 8 had abnormal QTc intervals. The average QTc was greater for patients using QTc-prolonging psychotropics (P = 0.05) with a correlation between interval length and psychotropic usage (P < 0.01). Average potassium (P = 0.08), calcium (P = 0.18), and magnesium (P = 0.08) levels did not significantly differ between those with normal and abnormal QTc intervals.

Conclusions

This study suggests that EDs alone may not prolong QTc intervals in paediatric patients with EDs, but psychotropics appear to be a salient external factor in QTc prolongation.

The increasing incidence and prevalence of eating disorders, including anorexia nervosa (AN) and bulimia nervosa (BN), pose a deleterious threat to the social, emotional, physical, and cognitive development of children and adolescents.^1,2

AN is defined by a severe reduction in caloric intake that results in low body weight and is driven by an individual’s disturbances in body perception and fear of gaining weight. The disorder may be accompanied by binge eating and purging behaviours, and it is often comorbid with other psychiatric conditions, including major depressive and obsessive-compulsive disorders.^3,4 Cardiovascular complications of AN include sinus bradycardia, reduced left ventricular myocardial mass, mitral valve prolapse, and corrected QT (QTc) interval prolongation.⁵ The risk of sudden cardiac death in this young population is particularly concerning, although the mechanism by which this occurs and the importance of external factors remain debated.⁶ Approximately one-third of deaths among patients with AN are due to cardiac causes, primarily sudden cardiac death. Frequent cardiovascular complications have been documented, with reported instances observed in up to 80% of cases.⁷

BN is characterized by frequent binge eating episodes during which an individual feels a lack of control over their caloric intake. Individuals with BN engage in inappropriate compensatory behaviours, including excessive exercise and self-induced vomiting, to avoid gaining weight. In addition, they disproportionately evaluate themselves based on their weight, but unlike AN, they are often at a normal weight or above average weight.³ BN is also highly comorbid with anxiety and mood disorders.⁸ Although it is typically less fatal than AN, cardiovascular complications are considered a key factor in the sudden death of patients with BN.⁹ Electrolyte abnormalities, such as hypokalaemia, are common in BN and have been associated with an increased risk of arrhythmias and myocardial degeneration.^9,10 Furthermore, QTc interval prolongation has also been observed in this population.^11,12

The QT interval is measured from the start of a QRS complex to the end of the subsequent T wave on an electrocardiogram (ECG) and represents the duration of ventricular depolarization to repolarization. As this interval varies with heart rate, QT must be adjusted using the RR interval to calculate the QTc interval.¹³ Although the Bazett formula (QTc = QT/√RR) has the potential to overestimate QTc length for patients with abnormal heart rates, this formula remains the most commonly used for correction in both the clinical and research settings.^14,15 In clinical practice for a general paediatric population, a normal QTc interval may be classified as less than 440 ms.¹⁶ QTc prolongation can be further categorized into borderline prolonged and prolonged.¹³ The definition of a prolonged QTc interval is still debated, with thresholds varying significantly based on the organization creating guidelines, research settings, and clinical practices. Depending on the definition used, the threshold for considering QTc as abnormal has ranged from 440 ms to 460 ms in paediatric populations.¹⁷ Moreover, before puberty, males and females exhibit relatively similar QTc intervals, whereas after puberty, females are more likely to have longer QT intervals.^18,19

Prolonged QTc intervals are associated with an increased risk of developing torsades de pointes. This polymorphic ventricular tachycardia often terminates spontaneously but can progress to pulseless ventricular tachycardia and result in sudden cardiac death.²⁰

QTc interval prolongation has been reported in both adult and paediatric populations with eating disorders.^11,21 Initially, this abnormal ECG finding was proposed to be inherent to AN and was associated with sudden death.²² Hypothesized mechanisms for QTc interval prolongation in eating disorders include reduced myocardial mass and altered myocardial excitability, which affect cardiac repolarization.^9,23 More recent studies suggest that extrinsic, rather than intrinsic, factors may influence QTc interval prolongation in populations with eating disorders.^24,25 Although the etiology of QTc prolongation remains debatable, both electrolyte abnormalities and psychotropic medications have been identified as potential risk factors of clinical relevance.^10,25, 26, 27, 28 QTc duration is associated with hypokalaemia, hypocalcaemia, and hypomagnesaemia, which are known complications of both AN and BN.^20,29 Similarly, psychotropic medications, which may be prescribed during eating disorder treatment, including antipsychotics and antidepressants, have also been implicated in QTc prolongation and torsades de pointes generation.^20,25

Previous investigations of QTc prolongation in paediatric populations with eating disorders have primarily focused on females with AN and have reported inconsistent findings of QTc prolongation within their small sample sizes.^{1,12,21,23,26,}30, 31, 32 Recently, a large cohort study assessing QTc duration in adult populations with eating disorders concluded that QTc prolongation is likely not inherent to eating disorders but, instead, may be related to confounding variables.²⁵

This study aimed to assess the relationship between QTc prolongation among paediatric patients with eating disorders and extrinsic factors, such as known QTc-prolonging medications and electrolyte abnormalities, and investigate whether the presence of an eating disorder alone is associated with a greater incidence of QTc prolongation than the general population.

Methods

Patients

Retrospective chart analysis was conducted for 264 patients between the ages of 7 and 17. All patients were diagnosed with an eating disorder by a physician and were assessed at the McMaster Children’s Hospital Eating Disorder Program (MEDP), during the period of January 13, 2013, to November 23, 2015, in Hamilton, Ontario, Canada. Because of the retrospective nature of this study, only patients with accessible medical charts and ECG reports were included.

Data collection

Electronic medical records were accessed using the MEDITECH system to obtain demographic information at the time of initial assessment at the MEDP including sex assigned at birth, age, and visit date. ECGs from the patients’ initial visit to the programme were collected to record parameters, such as heart rate, QT interval, RR interval, and QTc interval. QTc intervals were calculated using the Bazett formula to correct for heart rate.

Laboratory blood work results collected were specifically selected from a blood draw date that was closest to the patients’ initial visit date at the MEDP and were used to obtain serum electrolyte levels for potassium, calcium, and magnesium. In addition, any psychotropic medications known to prolong the QTc interval were recorded if the patient was taking the medication at the time of initial assessment.³³ ECGs used for data analysis were also performed during this initial assessment.

Statistical analysis

This study classified the QTc interval as normal if <440 ms, borderline prolonged if between 440 and 460 ms, and prolonged if >460 ms.^13,16 Prior cohort studies have reported that 8.7% of the adult general population have QTc intervals ≥440 ms.³⁴ Although this cited study involved adults, research on QTc prolongation with such a large sample size (n = 36,301) is seldom carried out in broader paediatric populations, and when attempted, these studies have often either not used QT correction or employ differing thresholds for QTc prolongation.34, 35, 36 This study adopted a QTc prolongation threshold of 440 ms, aligning with the clinical standard of overseeing paediatric cardiologists for this study’s patient population at the MEDP. This choice also enabled data comparison with Montanez et al.’s³⁴ research, which used the same threshold. Sex, prepubertal, and postpubertal differences in QTc prolongation definitions were not factored in due to varying definitions in the literature, lack of use among overseeing paediatric cardiologists at the MEDP, and to stay consistent with Montanez et al.’s³⁴ study that also did not account for these factors in their analysis.¹⁷

Normal electrolyte ranges were defined as potassium (K⁺) 3.5-5.0 mmol/L, calcium (Ca²⁺) 2.30-2.62 mmol/L, and magnesium (Mg²⁺) 0.70-0.91 mmol/L by the McMaster University Medical Centre laboratory. Unless otherwise specified, the criterion for normal electrolytes refers to normality of all K⁺, Ca²⁺, and Mg²⁺ serum electrolyte values. For the purposes of statistical analysis, patient usage of at least 1 QTc-prolonging medication was defined as a categorical variable. Relationships between QTc interval length, serum electrolyte levels, and psychotropic medication usage were assessed using the 1-sample t test, F test, Welch 2-sample approximated t test, Fisher exact test, Pearson product-moment correlation, Pearson χ², and cumulative logistic regression modelling as appropriate. Statistical analysis of patient data was performed by statisticians using the R software.³⁷ For all analyses, a statistical P value of ≤0.05 was considered to be statistically significant.

Results

Study patients

As summarized in Table 1, ECG parameters and serum electrolyte values were analysed from the initial visits of 264 paediatric patients with eating disorders aged 7-17 years. The majority of patients were of female sex (90.5%) with an average age of 14.3 (±1.9) years. At the time of intake, most (91.3%) patients were not taking psychotropics known to prolong the QTc interval (QTc-prolonging psychotropics). As corrected by the Bazett formula, the average QTc interval for all 264 patients was classified as a normal length (411.0 ± 25.9 ms) with interval lengths ranging from a minimum of 347 ms to a maximum duration of 483 ms. Patients with QTc intervals classified as either borderline prolonged or prolonged (≥440 ms) collectively represented 14.0% of the total patients (Table 2). A comparison between males and females revealed a lack of statistically significant differences between average ages and mean QTc intervals (respectively 13.8 and 14.4 years, P = 0.19, and 405.2 and 411.6 ms, P = 0.24). Consequently, patient sex was not considered in the subsequent analyses of electrolyte abnormality and usage of QTc-prolonging psychotropics.

Table 1.

Baseline characteristics of this study’s 264 paediatric patients with eating disorders

Characteristic	Value
Demographics
Age (y)	14.3 ± 1.9
Female	239 (90.5)
QTc-prolonging psychotropic use	23 (8.7)
ECG parameters
Heart rate (beats per minute)	65.6 ± 15.8
R-R interval (s)	1.0 ± 0.2
QT interval (ms)	400.8 ± 38.3
QTc interval (ms)	411.0 ± 25.9
Serum electrolyte levels (mmol/L)
Potassium∗	3.9 ± 0.4
Calcium†	2.4 ± 0.1
Magnesium‡	0.8 ± 0.1

Values are presented as mean ± standard deviation or number (%). Normal electrolyte ranges were defined as potassium 3.5-5.0 mmol/L, calcium 2.30-2.62 mmol/L, and magnesium 0.70-0.91 mmol/L.

ECG, electrocardiogram; QTc, corrected QT.

^∗Potassium serum levels unavailable for 18 patients.

^†Calcium serum levels unavailable for 32 patients.

^‡Magnesium serum levels unavailable for 29 patients.

Table 2.

QTc interval length distribution for patient subgroups

QTc interval range (ms)	Total patients	Serum electrolyte values		QTc-prolonging psychotropics†
		Normal (n = 152)∗	≥1 abnormal (n = 79)∗	No usage (n = 241)	Usage (n = 23)
<440	227 (86.0)	134 (88.2)	65 (82.3)	212 (88.0)	15 (65.2)
440-460	26 (9.8)	12 (7.9)	11 (13.9)	23 (9.5)	3 (13.0)
>460	11 (4.2)	6 (3.9)	3 (3.8)	6 (2.5)	5 (21.7)

Values are presented as number (%). The criterion for normal electrolytes refers to the normality of all potassium, calcium, and magnesium serum electrolyte values.

QTc, corrected QT.

^∗Excludes 33 patients with incomplete electrolyte data that were missing at least one of potassium, calcium, or magnesium serum values.

^†The distribution of QTc intervals across the categories of QTc length was significantly different between those who used and did not use QTc-prolonging psychotropics (P < 0.01).

The proportion of patients with QTc intervals ≥440 ms who did not use QTc-prolonging psychotropics and had normal serum electrolyte levels was not found to be statistically different from the prevalence of prolonged QTc (8.7%) among the general population (10.1%, P = 0.59).³⁰ However, the proportion of patients with QTc intervals ≥440 ms who did not use QTc-prolonging psychotropics but had at least 1 abnormal serum electrolyte level exceeded 8.7% (18.4%, P < 0.01).

Serum electrolyte abnormalities

A comparison of the average serum electrolyte values between patients with normal QTc intervals and those with intervals ≥440 ms revealed no significant differences for K⁺ (P = 0.08), Ca²⁺ (P = 0.18), or Mg²⁺ (P = 0.08). The average number of days between the date of electrolyte results and date of ECGs used for analysis when QTc was prolonged was ±9.2 days. The relationship between patient QTc interval length and serum electrolyte levels demonstrated weak negative linear correlations for the 3 electrolytes of interest (K⁺ r = −0.1, Ca²⁺ r = −0.2, and Mg²⁺ r = −0.2), with only Ca²⁺ and Mg²⁺ having correlations of statistical significance (K⁺ P = 0.19, Ca²⁺ P = 0.01, and Mg²⁺ P < 0.01). As outlined in Table 2, individuals with normal electrolytes and those who had 1 or more abnormal electrolytes were not found to have statistically different distributions across the QTc interval lengths (P = 0.36). Analysing the independence between normality or abnormality of serum electrolytes and classification of patient QTc interval further supported the lack of association between these variables (χ² = 1.6, P = 0.44).

Usage of known QTc-prolonging psychotropics

Antidepressants and antipsychotics known to prolong the QTc interval were used by 23 patients. The most frequently used psychotropic was fluoxetine (n = 7), followed by sertraline (n = 6), quetiapine (n = 5), escitalopram (n = 4), venlafaxine (n = 3), olanzapine (n = 2), and trazodone (n = 2). Although each medication was only used by 1 patient, imipramine (n = 1), mirtazapine (n = 1), aripiprazole (n = 1), and risperidone (n = 1) were also reported within the sample. While the average serum electrolyte values for K⁺ and Ca²⁺ were not statistically different between these patients and those who were not taking any QTc-prolonging psychotropics (K⁺ P = 0.82 and Ca²⁺ P = 0.10), the average serum Mg²⁺ values were found to be statistically different between these 2 groups (P = 0.02) (Table 3).

Table 3.

Serum electrolyte values and the use of known QTc-prolonging psychotropics

QTc-prolonging psychotropics	Potassium (mmol/L)	Calcium (mmol/L)	Magnesium (mmol/L)
No (n = 241)∗	3.9 ± 0.4	2.4 ± 0.1	0.9 ± 0.1
Yes (n = 23)†	3.9 ± 0.4	2.4 ± 0.2	0.80 ± 0.1
P value	0.82	0.10	0.02

Values are presented as mean ± standard deviation.

QTc, corrected QT.

^∗Potassium serum levels unavailable for 18 patients, calcium serum levels unavailable for 29 patients, and magnesium serum levels unavailable for 27 patients.

^†Calcium serum levels unavailable for 3 patients and magnesium serum levels unavailable for 2 patients.

The average QTc interval was significantly greater for patients taking QTc-prolonging psychotropics than those who were not taking these medications (P = 0.05) (Fig. 1). In addition, the distribution of QTc intervals across the categories of QTc length was significantly different between these 2 groups (P < 0.01) (Table 2). Among patients on psychotropic medications, 21.7% of them had a prolonged QTc interval, whereas 2.5% of patients not taking psychotropic medications had QTc prolongation. Furthermore, the use of these antidepressants or antipsychotics was related to the classification of patients’ QTc intervals as normal, borderline prolonged, or prolonged (χ² = 20.2, P < 0.01).

Figure 1.

QTc interval lengths for patients who did and did not use known QTc-prolonging psychotropic medications. X represents the mean values of 409.7 ± 24.8 ms for patients without psychotropic use and 424.6 ± 33.4 ms for patients using psychotropics (P = 0.05). The horizontal lines indicate the inclusive medians of 409 ms and 421 ms for their respective groups. A circle is used to represent an outlier value of 483 ms for the group of patients not using psychotropics. QTc, corrected QT.

Logistic regression was used to model the probability of the QTc interval categorization as normal, borderline prolonged, or prolonged, depending on whether QTc-prolonging psychotropics were used by the patient (Fig. 2). The fitted probabilities suggested that individuals using QTc-prolonging psychotropics were more likely to be classified into the borderline prolonged or prolonged groups than the normal QTc group. This cumulative logistic regression model revealed that QTc tends to be prolonged with QTc-prolonging psychotropic usage (P < 0.01).

Figure 2.

Logistic regression denoting the fitted probability of QTc interval categorization (normal, borderline prolonged, or prolonged) depending on whether QTc-prolonging psychotropics were used by the patient. The QTc tends to be prolonged with QTc-prolonging psychotropic medication usage (P < 0.01). QTc, corrected QT.

Discussion

Interpretation

This study examined the relationship between the observed QTc interval prolongation in eating disorders and how electrolyte abnormalities and QTc-prolonging psychotropics contribute to this phenomenon as potential extrinsic factors. The analyses of the K⁺, Ca²⁺, and Mg²⁺ serum electrolyte values did not reveal any notable association with QTc prolongation. Importantly, a significant correlation between QTc interval prolongation and psychotropic drugs was identified in this population. Although causality cannot be inferred from these findings, the usage of these medications was associated with prolonged QTc intervals when compared with patients with eating disorders who were not using this medication class. Further analyses of these intergroup differences identified lower average Mg²⁺ values in patients who used QTc-prolonging psychotropics. Despite the statistical significance of this finding, the average serum values for both groups were within the normal range for Mg²⁺ values, which questions any potential clinical relevance. Although it is possible that these psychotropics may not induce electrolyte abnormalities within the context of eating disorders, further investigation is warranted to explain the mechanism of prolongation with the use of these medications in patients with eating disorders. Overall, these preliminary results suggest that eating disorders alone may not inherently prolong the QTc interval and instead point to QTc-prolonging psychotropic medications as a salient factor in the paediatric population with eating disorders.

Significance

This study is one of the largest known retrospective analyses of QTc interval prolongation in paediatric patients with eating disorders, which aimed to include a minimum of 194 patients.^38,39 This doubles the largest reported sample size from studies conducted in a mixed-sex population with participants under the age of 18.¹² During childhood, inclusion of both sexes is of particular importance because males are more likely to be diagnosed with an eating disorder earlier in their development when compared with adulthood.⁴⁰ Furthermore, most of the research studying youth eating disorders focuses solely on female participants with a general lack of male representation.^23,26,30,32 Importantly, established sex differences between male and female QTc intervals have been less clearly depicted during puberty and do not align with the development of secondary sex characteristics.⁴¹ Thus, the inclusion of both biological sexes in this study is necessary to understand the relationship between eating disorders and QTc prolongation in a paediatric study population.

In adolescents, conflicting results have been published, with some studies reporting longer,^23,30,32 normal,¹ or even shorter^12,26,38 QTc intervals in patients with eating disorders relative to control groups. The juxtaposition of these findings may be attributed to differences in study design, participant inclusion criteria, and whether any extrinsic factors were examined.³⁸ Prior analyses of electrolytes, including K⁺ and Mg²⁺, support our finding that there may not be a clear association between QTc length and electrolyte abnormalities.^21,38

Limitations

Because of the retrospective design of this study, several factors must be considered when interpreting and applying these findings. First, 12.5% of patients were missing laboratory data on MEDITECH for at least 1 serum electrolyte value. In the analysis of incomplete data sets, these patients were excluded from the respective statistical calculations. For the 87.5% of patients with available serum electrolyte values, laboratory results from the date closest to ECG testing were analysed. Thus, interpatient variation for the time interval between ECG acquisition and blood sample collection may affect the relationship between QTc intervals and serum electrolytes.

Second, any medical interventions introduced between the date of ECG testing and serum electrolyte assessment were unknown and thus could not be taken into account in the data analysis. Beyond psychotropic medication usage and a confirmed eating disorder diagnosis, any additional extrinsic factors or pre-existing medical conditions that may affect serum electrolyte values and QTc interval prolongation were also not included within statistical analysis. However, it is imperative to explore these factors in a future study.

Third, a retrospective chart review from this clinic was insufficient to determine the exact date when patients began taking their prescribed psychotropic medication, the length of psychotropic usage, a complete history of dose adjustments, psychotropic usage before the current regimen, and overall medication adherence. Statistical analysis of psychotropic usage did not consider the effect of being on multiple psychotropic medications. The individual effect of antidepressants, antipsychotics, or combined usage of these classes was not analysed because of the small sample sizes that would be generated on subdividing the group of 23 patients who used psychotropic medications. Collectively, these factors could introduce confounding variables that must be considered when assessing these results. Overall, the heterogeneity of these patients and study parameters enables generalizations to be more representative of the greater paediatric population with eating disorders. Moreover, we did consider assessing baseline QTc before the initiation of psychotropic medications, ECGs over time, and follow-up in these patients. This was deemed not feasible due to inconsistent patient follow-up, inaccessible appointments and test results at out-of-facility locations, and a significant change in our electronic medical records system.

Furthermore, this study defined QTc prolongation as ≥440 ms and used an adult population study, rather than paediatric, when referencing the incidence of QTc prolongation in the general population.³⁴ Sex differences and pubertal changes to QTc definitions were also not accounted for in this study. The definitions of QTc prolongation and the reported incidences of both congenital and acquired QTc prolongation among the general population exhibit significant variation in the literature or are subject to various limitations.¹⁷ For instance, in one large study, 33,051 children were screened for long QT syndrome, but only 6- and 12-year-olds were assessed, thereby limiting generalizability. In addition, inclusion criteria required a remarkably high QTc of >500 ms.³⁵ In another study, 152,332 school-aged children across very specific age groups were assessed for conduction disturbances, but their reported QT intervals lacked correction by any methodology.³⁶ Therefore, gender, prepubertal, and postpubertal differences in QTc prolongation definitions were not factored in, and our definition of QTc prolongation was used because of varying definitions in the literature and consistency with the clinical standard used among overseeing paediatric cardiologists at the MEDP and definitions used by Montanez et al.³⁴ Future studies are necessary to aid with standardization of QTc prolongation definitions and to study the incidence of QTc prolongation in healthy paediatric populations.

Lastly, the correction of QT intervals using standardized formulas, such as the Bazett formula, leads to inaccuracies at the extremes of heart rate that may lead to the overdiagnosis of long QT syndrome.¹³ Although the ease of calculation with these formulas explains their popularity in clinical settings, linear regression analyses are more accurate.^13,18 In this study, the Bazett formula was selected because of its widespread clinical application, but it may have improperly corrected QT intervals for bradycardic and tachycardic patients who were not well represented by the average heart rate of 65.6 beats per minute. This is especially important given the prevalence of bradycardia in eating disorders such as AN.⁵ In addition, standardized formulas and generalized paediatric ranges for QTc prolongation do not consider the impact of age and sex on QTc length.^18,19

To further elucidate the role of psychotropic medications within the paediatric population with eating disorders and enhance the generalizability of these findings, future studies should examine sex, psychotropic class, and age-dependent effects on QTc interval prolongation. This relationship should also be compared between specific populations with eating disorders, such as AN and BN, to determine if different clinical presentations alter prolongation incidence. Furthermore, alternative methods of QTc interval calculation should be explored to understand how correction formula usage may influence this phenomenon. Finally, future research should explore how the QTc interval changes across multiple time points throughout the treatment by conducting serial ECG tests for both inpatient and outpatient services.

Conclusions

This study suggests that QTc prolongation in paediatric patients with eating disorders may not be inherent to eating disorders but, instead, could be associated with external factors including the usage of known QTc-prolonging psychotropic medications. Therefore, QTc intervals should continue to be monitored in all patients with eating disorders with a particular focus warranted in the management of those known to have risk factors associated with QTc prolongation. Given the morbidity and mortality of cardiac complications in eating disorders, further research is needed to examine this correlation in the paediatric population and develop interventions that prevent adverse cardiovascular outcomes.

Disclosures

The authors have no conflicts of interest to disclose.