Effect of mood states on QT interval and QT dispersion in eating disorder patients

Authors


*Yoshiyuki Takimoto, MD, PhD, Department of Stress Science and Psychosomatic Medicine, Faculty of Medicine, University of Tokyo, Hongo 7-3-1, Bunkyo-ku, Tokyo 113-8655, Japan. Email: taki-tky@umin.ac.jp

Abstract

Aim:  Prolonged QT interval and QT dispersion have been reported in patients with eating disorders. Although the factors that cause prolongation remain unclear, mood states such as anxiety have been reported to influence QT interval and dispersion, probably via the autonomic nervous system. Therefore the aim of the present paper was to investigate mood effect on prolonged QT interval and QT dispersion.

Method:  The subjects were 47 female anorexia nervosa (AN) and 48 female bulimia nervosa (BN) patients. In all of the patients, serum electrolyte levels were normal. QT interval and QT dispersion were measured from 12-lead electrocardiographic recordings. Mood states in each patient were measured using a Profile of Mood States (POMS) evaluation, and the patients were divided into high- and low-score groups for each POMS subscale. The differences in QT variables were compared between the two groups for each subscale.

Results:  In the BN group, QT interval and QT dispersion in the high depression score group were significantly longer than those in the low depression score group, and QT dispersion was significantly greater in the high anxiety score group than in the low anxiety score group. In addition, QT interval and QT dispersion were significantly correlated with depression scores. In the AN group there were no significant differences in QT interval or QT dispersion between the high- and low-score groups for any POMS subscale.

Conclusions:  BN patients with worse states of depression or anxiety had longer QT intervals and larger QT dispersion. In BN patients, mood disturbance might increase the risk of arrhythmias.

EATING DISORDERS (ED) have a high prevalence rate (1–5%) among young women in developed countries, and are important social problems.1 Long-term follow-up studies have shown that ED patients have a mortality rate of approximately 1–6%,2 and ventricular arrhythmias has been believed to be one of the major causes of sudden death in such patients.3

Long QT interval and large QT dispersion have been thought to be associated with ventricular arrhythmias,4 and previous studies have reported that the QT interval and QT dispersion are prolonged not only in anorexia nervosa (AN) patients5 but also in bulimia nervosa (BN) patients.6

Malnutrition and abnormal electrolyte levels have been thought to be important in the prolongation of QT interval and QT dispersion7 but the mechanism underlying this effect has not yet to be determined. Other factors, for example autonomic nervous dysfunction, have affected QT interval.8

In addition, it has been reported that mood states such as anxiety have influenced QT interval and QT dispersion.9–11 In ED, mood states also might be one of mechanisms for the prolongation of QT interval and QT dispersion because ED patients often have mood state disturbance.12 Therefore, the aim of the present study was to determine whether mood states in patients with ED could influence QT interval and QT dispersion.

METHODS

Patients

Patients >18 years of age who were referred to University of Tokyo Hospital between January 1999 and September 2002 were included for enrollment in the study. ED were diagnosed by well-trained clinicians according to DSM-IV criteria,13 and we initially enrolled 50 female AN patients and 49 female BN patients into the study. Patients with abnormal plasma electrolytes and those who were taking medication, such as psychotropic or cardiovascular drugs, that might influence the electrocardiogram (ECG) were excluded from the study. All data were collected prospectively. Written informed consent was obtained from all of the subjects following complete disclosure of the study methods.

Measurement of QT interval and QT dispersion

ECG were recorded and QTc intervals and QT dispersion were measured automatically using the QTD-1 (Fukuda Denshi, ECG Company, Tokyo, Japan).14

The corrected QT interval using Bazzet's formula was calculated as follows:

image

where RRI refers to the RR interval just before the QT interval.15 QT dispersion was measured as the difference between the longest and the shortest QT interval in the 12 leads.16

Psychological evaluation

The Profile of Mood States (POMS)17 was used to evaluate mood states. The POMS consists of 65 items, which are rated using five levels of severity. The mood state was assessed on a six-mood subscale: tension–anxiety (T-A), anger–hostility (A-H), depression (D), vigor (V), fatigue (F) and confusion (C). For each POMS subscale, AN and BN patients were further divided into two groups: patients with a score higher than the median score, and patients with a score lower than the median score for each subscale.

Statistical analysis

All results are shown as mean ± SD. Student's t-test was used to investigate differences in body mass index (BMI), psychological test scores, QTc and QT dispersion between each pair of higher and lower score groups. The relationships between the QT variables and the BMI and psychological parameters were studied using Pearson's correlation coefficient.

All analyses were performed with SAS version 6.12 (SAS Institute Japan, Tokyo, Japan). P < 0.05 was considered significant.

RESULTS

Demographic characteristics

The total number of eligible patients was 95, consisting of 47 AN patients and 48 BN patients with normal electrolytes (normal serum potassium and normal serum calcium corrected for serum albumin levels). Two AN patients and one BN patient were excluded due to low serum potassium levels, and one AN patient was excluded due to low serum levels of both potassium and calcium. For each pair of higher and lower score groups for a given POMS subscale, the scores of the higher score group were significantly higher than those of the lower score group (data not shown, except for POMS subscales T-A and D). BMI and age were not significantly different between the higher and lower score groups for each POMS subscale (data not shown, except for T-A and D).

Anorexia nervosa patients

There were no significant differences in QTc or QT dispersion between the high- and low-score groups for any mood subscale in the AN patient group (Tables 1,2). BMI or any mood subscales were not significantly correlated with QTc or QT dispersion.

Table 1.  QTc interval and QT dispersion according to T-A score (mean ± SD)
 ANhT-A
(n = 22)
ANlT-A
(n = 25)
BNhT-A
(n = 23)
BNlT-A
(n = 25)
  1. * P < 0.05, *** P < 0.001 (ANhT-A vs ANlT-A); ###P < 0.001 (BNhT-A vs BNlT-A).

  2. Student's t-test was performed for analysis of the T-A score and QTc interval and dispersion.

  3. ANhT-A, anorexia nervosa patients with a tension–anxiety score higher than the median; ANlT-A, anorexia nervosa patients with a tension–anxiety score lower than the median; BMI, body mass index; BNhT-A, bulimia nervosa patients with a tension–anxiety score higher than the median; BNlT-A, bulimia nervosa patients with a tension–anxiety score lower than the median; QTc, corrected QT interval; QTd, QT dispersion; T-A, tension–anxiety score.

Age (years)21.0 ± 4.823.0 ± 7.421.2 ± 3.623.2 ± 4.8
BMI (kg/cm2)15.5 ± 1.614.9 ± 1.919.9 ± 2.620.9 ± 4.1
T-A score24.6 ± 6.1***17.8 ± 4.8***28.0 ± 5.2###16.5 ± 5.0###
QTc (ms)466.6 ± 26.9458.0 ± 30.3470.2 ± 37.7457.0 ± 22.6
QTd (ms)58.3 ± 16.949.5 ± 15.270.7 ± 34.5*48.4 ± 10.0*
Table 2.  QTc interval and QT dispersion according to depression score (mean ± SD)
 ANhD
(n = 20)
ANlD
(n = 27)
BNhD
(n = 24)
BNlD
(n = 24)
  1. * P < 0.05, *** P < 0.001 (ANhD vs ANlD); ###P < 0.001 (BNhD vs BNlD).

  2. Student's t-test was performed for analysis of the D score and QTc interval and dispersion.

  3. ANhD, anorexia nervosa patients with a depression score higher than the median; ANlD, anorexia nervosa patients with a depression score lower than the median; BMI, body mass index; BNhD, bulimia nervosa patients with a depression score higher than the median; BNlD, bulimia nervosa patients with a depression score lower than the median; D score, depression score; QTc, corrected QT interval; QTd, QT dispersion.

Age (years)21.6 ± 6.022.3 ± 5.622.6.3 ± 2.122.3 ± 2.0
BMI (kg/cm2)15.2 ± 1.715.2 ± 1.921.3 ± 3.919.7 ± 2.8
D score45.6 ± 8.9***17.5 ± 13.6***47.8 ± 5.8###26.5 ± 9.8###
QTc (ms)470.1 ± 27.4459.9 ± 31.2480.1 ± 36.1*452.4 ± 22.1*
QTd (ms)55.1 ± 15.752.8 ± 18.873.6 ± 33.2*45.8 ± 12.5*

Bulimia nervosa patients

The group of BN patients with a high T-A score had a significantly larger QT dispersion than the group with a low T-A score (P < 0.05, Table 1). In the group of BN patients with a high D score, QTc and QT dispersion were significantly greater than in the group with a low D score (P < 0.05, Table 2). There were no significant differences in QTc or QT dispersion between the two groups in BN patients for the other subscales of POMS (data not shown). The D score was significantly correlated with QTc and QT dispersion (QTc, r = 0.41, P = 0.006; QT dispersion, r = 0.30, P = 0.046), while the other mood subscales were not significantly correlated with QTc or QT dispersion. Similarly, BMI was not significantly correlated with QTc or QT dispersion (Table 3; Fig. 1).

Table 3.  Pearson's correlation coefficient in BN patients
 QTcQT dispersion
  • *

    P < 0.05.

  • BMI, body mass index; BN, bulimia nervosa; D score, depression score; QTc, corrected QT interval; T-A score, tension–anxiety score; TMD score, total mood disturbance score.

T-A score0.280.28
D score0.41*0.30*
TMD score0.37*0.24
BMI−0.10−0.06
Figure 1.

(a) Corrected QT (QTc) interval and (b) QT dispersion vs D score for patients with bulimia nervosa. The regression lines are shown. QTc interval and QT dispersion were significantly correlated with D score (a, r = 0.41, P = 0.006; b, r = 0.30, P = 0.046).

DISCUSSION

In the present study, QTc and QT dispersion in BN patients without abnormal electrolyte levels were significantly greater in patients with higher T-A or D scores, compared to patients with lower scores. In particular, D scores in BN patients had significant positive correlations with QTc and QT dispersion.

Although patients with major depression have been reported to have a significant prolongation of QT dispersion compared to healthy controls,18 no studies have reported the relationship between depression and QT interval. In the present study, depression scores alone were associated with both QT interval and QT dispersion. This result is consistent with previous studies that have shown that the depressive state can alter autonomic tone,19 and that QT interval and QT dispersion are associated with autonomic nervous activity.20,21

In the present study, anxiety was also shown to be associated with QT interval and QT dispersion in BN patients. It has been reported that anxiety can change autonomic control,22 and that anxiety influences the QT dispersion in hypertensive patients23 and normotensive subjects.24 Therefore the present results are consistent with those of previous studies.

Other factors such as bodyweight control also influence autonomic function.25 But in the present study no significant differences were found in BMI between each pair of high- and low-score groups. Therefore, mood state might influence the QT interval and QT dispersion independently of bodyweight.

In the present study AN patients with high T-A and D scores did not have a significantly prolonged QT interval or QT dispersion. In AN patients, sustained extreme loss of weight has been reported to be a more influential factor in autonomic nervous function than anxiety or depression.26 And, in contrast to BN patients, other factors, such as myocardial hypotrophy caused by malnutrition, might play more important roles in the prolongation of QT interval and QT dispersion in AN patients, compared to changes in autonomic nervous function due to mood disturbance.27

There were some limitations in the present study. First, although every effort was made to examine only those subjects who were not taking medication, it is possible that some subjects may have failed to declare their medication usage and might have a history of medication use, while there are many drugs that could potentially influence the QT interval. Second, we did not measure serum magnesium levels, which could also influence the ECG. But the presence of low magnesium levels was thought to be comparatively unlikely in BN patients because such patients do not suffer from malnutrition. Third, other factors that might influence the QT interval, such as high free fatty acid (FFA) levels, hypoglycemia, and high estrogen levels, were not examined.28,29 However, abnormalities in FFA, glucose and estrogen levels are also comparatively rare in BN patients because BN patients do not have malnutrition or amenorrhea. Fourth, although purging may have an effect on QT variables, the differences between bulimia nervosa purging type and non-purging type patients were not examined in the present study because the sample size of BN non-purging type was too small. Finally, some previous studies reported the poor predictive value of QT dispersion in stratifying sudden death risk.30

In conclusion, the present results suggest that mood states such as depression and anxiety are associated with prolongation of QT interval and QT dispersion in BN patients, and this might increase the risk of ventricular arrhythmias.

ACKNOWLEDGMENT

This work was partly supported by a grant for AN research from Ministry of Health, Labor, and Welfare.

Ancillary