European Heart Journal (1996) 17, 1723-1728

Dynamics of the QT interval during and after exercise

in healthy children
M. Viitasalo, L. Rovamo, L. Toivonen, E. Pesonen and J. Heikkila
Cardiovascular Laboratory, Division of Cardiology, Department of Medicine, and Children's
Helsinki University Central Hospital, Helsinki, Finland

Introduction
The dynamic nature of the relationship between duration of repolarization and cardiac cycle length has
become increasingly interesting in cardiac electrophysiology, and may be used as an indicator of susceptibility
to ventricular arrhythmias. Theoretically, changes in the
duration of ventricular action potentials seem to underlie corresponding changes in the QT interval1'1. Both
ambulatory and exercise electrocardiograms have been
used to record changes in QT intervals.
Recently, it has been suggested that measuring
dynamic changes in the QT interval may be useful in
identifying patients with congenital and acquired forms
of the long QT syndrome'21 as well as for assessing
the risk of arrhythmia in patients treated with some
antiarrhythmic agents'31. Repolarization has been quantified in detail in computerized analyses of resting
electrocardiograms (ECG)141. Experimentally, the late

Revision submitted 30 January 1996, and accepted 15 February

1996.
Correspondence: Matti Viitasalo, MD, Cardiovascular Laboratory,
Division of Cardiology, Department of Medicine, University
Central Hospital, Haartmaninkatu 4, 00290 Helsinki, Finland.
0195-668X/96/I11723+06 $25.00/0

proportion of total QT, increased at high heart rates. Rate

correction using Bazett's method gave abnormal total
QT values (>440ms) in 12 children (67%) whereas linear
correction gave values below 440 ms only. In conclusion,
the relationship between QT and heart rate is linear and
differs during exercise and recovery. Inhomogeneity of
repolarization increases at high heart rates. Linear correction of total QT and early QT intervals improves the
evaluation of repolarization duration in exercise testing
in children.
(Eur Heart J 1996; 17: 1723-1728)
Key Words: Electrocardiography, exercise testing, QT
interval.

portion of the QT interval from the apex to the end of

the T wave correlates with dispersion of ventricular
action potential duration'51. However, in studies using
ambulatory and exercise electrocardiograms, automatic
analysing methods have only been able to generate a
linear regression slope relating the early portion of
repolarization duration (time interval between beginning
of Q wave and apex of T wave) to cycle length.
The purpose of the present study was to determine the principles of the relationship of the QT interval
and its early and late portions to heart rate in healthy
children during and after exercise.

Methods
Subjects
The subjects were 18 healthy children of hospital staff, 4
to 15 years old (12 female, 6 male), recruited via their
parents. None of the children had any evidence of heart
disease on the basis of history, physical examination
or resting electrocardiogram, or had experienced an
episode of syncope. None of the children participated
in regular athletic training or took any medication.
Characteristics of the subjects are listed in Table 1.
1996 The European Society of Cardiology

Downloaded from http://eurheartj.oxfordjournals.org/ by guest on November 26, 2015

The dynamics and homogeneity of the QT interval have

been used as indicators of susceptibility to ventricular
arrhythmias. We determined the relationship between QT
intervals and heart rate during exercise testing and subsequent recovery in 18 healthy children. The QT intervals
were measured to the apex (early QT), to the end (total
QT), and from the apex to the end of the T wave (late QT)
(inhomogeneity of repolarization) at heart rates from 60
by steps of 10 to 180 beats . min ~ ' . Group mean total
QT and early QT exhibited better linear correlations with
heart rate (r 0-998 and 0-999) than with cardiac cycle length
(r 0-954 and 0-959). The slope relating total QT to heart rate
was 1 -30 during exercise and - 1-42 during recovery
(P<005). The corresponding slopes relating early QT to
heart rate were 111 and 1-30 (/ > <005). Late QT, as a

Hospital,

1724 M. Viitasalo et al.

The QT interval was measured manually from the

beginning of the Q wave to the apex of the T wave (early
QT) and to the end of the T wave (late QT). The end of
the T wave was determined as the intercept of the
tangent drawn to the deepest downslope of the T wave
and the PQ line. The tangent method was selected
because at fast heart rates the end of the T wave cannot
be determined as it merges with the baseline. Possible U
waves were not included in the measurement. The late
duration of the QT interval (late QT) was calculated as
total QT early QT. The paper speed was 50 mm . s ~ '
for all measurements. The intervals were recorded to the
nearest 10 ms as arithmetic means of three consecutive
measurements. All measurements were carried out by
the same experienced reader, and the intra-reader variability tested by duplicate measurements. The coefficient
of variability of total QT measurements was 003 and of
early QT measurements 0035.
The measurements were taken from leads II and
V5 at heart rates of 60, 70, 80, 90, 100, 110, 120, 130,
140, 150, 160, 170, and 180 beats. min"' during exercise and at 160, 150, 140, 130, 120, 110, 100, 90, and
80 beats . min ~ ' during recovery. The heart rate was
read from four consecutive RR intervals, and it was
Eur Heart J, Vol. 17, November 1996

Results
Relationship of total QT and early QT
intervals to heart rate and cycle length
The measured values of total QT and early QT intervals
at the various heart rates during both exercise and
recovery are shown in Table 2. The group mean total
QT and early QT intervals measured at different specified heart rates are plotted against corresponding heart
rates in Fig. 1. There were high linear correlations for
total QT and early QT with heart rate both during
exercise and recovery, the correlation coefficient ranging
from 0-982 to 0999. The group mean linear regression
slope relating total QT to heart rate was 1-30 during
exercise and 1 -42 during recovery (/ ) <005). The
corresponding slopes relating early QT to heart rate
were - 111 and - 1-30 (/><005). Only in four subjects
were the slopes of total QT and early QT intervals
steeper during exercise than recovery. The slopes relating total QT and early QT to heart rate differed from
each other during exercise ( - 1-30 vs - 111, P<005),

Downloaded from http://eurheartj.oxfordjournals.org/ by guest on November 26, 2015

Table 1 Clinical, electrocardiographic, and exercise test- accepted to deviate by 2 beats. min ' from the taring characteristics of 18 study subjects
get. The differences between mean values measured from
leads II and V5 were not significant at any specified heart
Mean SD Range
rates. In the analysis, the data from the two leads were
combined as the arithmetic mean values.
The mean total QT and early QT intervals at
4-15
10
3
Age (years)
146
18 109-172
Height (cm)
different specified heart rates were plotted against corre41
11
Weight (kg)
18-62
sponding rates to examine their relationship. The best
Resting ECG
fitting
equations, relating QT intervals and heart rates,
78
11
Heart rate (beats . min ~ ')
63-98
were
computed.
The relationship of total QT and early
140
18 116-172
PR interval (ms)
QT to heart rate were compared between the exercise
81
8
QRS duration (ms)
68-96
24 316-412
369
QT interval (ms)
and recovery phases of the exercise test using linear
410
QTc interval (Bazett's method)
25 346_447
models. Furthermore, the relationship of total QT and
Exercise testing
early QT intervals to cardiac cycle length were studied
11
4
Exercise duration (min)
4-16
Maximal loading (W)
118
46
40-200 by plotting mean total QT and early QT values
measured at different heart rates against the correspond186
15 155-208
Peak heart rate (beats . min ~ ')
140
16 114-167
ing RR intervals. The best fitting first-, second- and
Peak systolic blood pressure (mmHg)
75
16
Peak diastolic blood pressure (mmHg)
49-97
third-degree polynom curves relating total QT and early
QT to RR intervals were computed.
The behaviour of the late duration of repolarization was studied by plotting late QT values against
heart rate. In addition, late QT was related to repolarExercise testing
ization time by plotting late QT/total QT values against
Exercise testing was performed on an electrically braked heart rate.
bicycle ergometer (Rodby Electronic AB) connected to
an electrocardiographic recorder (Marquette Electronics
Inc. Case 15) with a frequency response of 001-100 Hz.
Statistics
The load was increased by 10 to 20 W . min ~ ', depending on the age of the subject, until exhaustion. Details of The data are expressed as means plus standard deviexercise testing are shown in Table 1. The ECG was ations. The differences between mean values were tested
recorded continuously on chart paper using Mason- using the two-tailed Student's t-test. Correlation coefLikar lead connections.
ficients were evaluated using tests for linear regression.
The differences between the slopes of the linear
regression equations were tested with the regression
method
in groups using the chi squared test'61.
QT analysing system

The QT interval in healthy children

1725

Table 2 Measured QT intervals at different specified heart rates during exercise and
recovery in 18 children
Heart
rate
(beats . min ')

QTo

QTm

385 24
369 25
356 16
345 1 8
335 14
324 14
304 19
295 11
275 13
263 18
252 10
245 10
231 12

31013
303 20
291 18
279 18
270 16
256 14
242 15
236 15
220 15
209 17
200 16
192 13
180 10

TmTo
75 14
65 14
65 9
668
65 10
67 1 3
62 16
59 7
55 16
54 15
51 9
547
49 8

Recovery QT intervals (ms)

(mean SD)
QTo

QTm

TmTo

355 30
348 25
320 21
313 33
295 27
274 23
265 23
258 19
246 15

290 26
271 17
259 21
244 17
228 18
214 18
201 16
198 17
18615

65 10
77 15
61 4
67 28
67 23
6016
64 13
60 12
607

QTm = time interval between beginning of Q wave and apex of T wave; QTo = time interval between
beginning of Q wave and end of T wave; TmTo = time interval between apex of T wave and end of
T wave (calculated as QTo-QTm).

400
EXQTD y=462-1.30x

r=0.998
RecQTb y=468-1.42x
r=0.990

rates (< 100 beats. min ') and 0-38 at fast rates
(> 100 beats, m i n " ' ) (P<0-001). The corresponding
slopes relating early QT to RR intervals were 0 1 2 and
0-33 (P<0-001).

300

Late duration of QT interval

The terminal portion of repolarization, late QT
(measured as total QT early QT), decreased from a
mean value of 75 ( 14) ms at 60 beats. min ~' to a
200
mean value of 49 ( 8) ms at 180 beats . min " ' during
ExQTmy=379-l.llx
exercise (Table 2). The group mean late QT correlated
r=0.999
with heart rate during both exercise (late QT =
RecQTm y=389-1.30x
QTm
8 2 - 0 1 7 6 x heart rate, r=0-90, / > <001) and recovery
r=0.982
(lateQT = 7 8 - 0 1 1 9 x heart rate, r=0-58,/ ) <005). Late
100
50
100
150
200
QT as a proportion of the total repolarization time (late
Heart rate (beats.min )
QT/total QT) increased at high heart rates and was
Figure 1 Relationship of group mean total QT (QTo) greater during recovery than exercise at the same heart
and early QT (QTm) intervals to heart rate during rates (/><001) (Fig. 3).
exercise (Ex, open symbols) and recovery (Rec, closed
symbols). QTm = time interval between the beginning of
the Q wave and the apex of the T wave; QTo=time
interval between the beginning of the Q wave and the end
of the T wave.

Evaluation of exercise and postexercise

QT durations

but not during recovery. Only two children exhibited

steeper early QT than total QT slope during exercise.
Neither QT intervals nor slope values correlated with the
age or sex of the study subjects.
The best fitting equations relating exercise group
mean total QT and early QT intervals to RR intervals
were third-degree polynoms (Fig. 2). In linear regression
models, during exercise, the group mean slope relating
total QT to RR intervals was 014 at slow heart

When the measured QT values were normalized to the

heart rate of 60 beats. min~' by using Bazett's correction formula, QTc=QT/VRR, 12 out of 18 children
(67%) during exercise and five out of 18 (28%) during
recovery exhibited QTc values longer than 440 ms, in at
least one measured heart rate. The mean error of the
'corrected QTc' values ranged from 16 ms at a heart rate
of 180 beats . min" 1 to 52 ms at 110 beats . min" 1 .
Using the linear correction equations derived
from those presented in Fig. 1, all normalized total
Eur Heart J, Vol. 17, November 1996

Downloaded from http://eurheartj.oxfordjournals.org/ by guest on November 26, 2015

60
70
80
90
100
110
120
130
140
150
160
170
180

Exercise QT intervals (ms)

(mean SD)

1726 M. Viitasalo et al.

400

0.26

Rec Y=O.14+O.64xlO"3X
r=0.84 p<0.01

Rec

0.24

0.22
Ex
0.20

0.18

100
200

400

1000

Figure 2 Relationship between group mean total QT

(QTo) and early QT (QTm) to RR intervals during exercise (open symbols) and recovery (closed symbols). The best
fitting linear regressions (QTo=177+0-23RR, r=0-954;
QTm = 133+0-20RR, r=0-959), second-degree polynoms
(QTo=51+0-67RR-0-34x 10~3RR2, r=0-993; QTm =
22+0-59RR-0-30 x 10 3RR2, r=0-998), and third-degree
polynoms (QTo=96 + 1-4RR-1 -60 x 10 " 3RR2 + 0-64 x
10 " 6RR3,
r=0-998;
QTm = - 38+0-90RR-0-82 x
10"3RR2 +0-26 x 10"'RR3, r=0-999) for total QT and
early QT intervals during exercise are shown in the graph.
QTm=time interval between the beginning of the Q wave
and the apex of the T wave; QTo=time interval between
the beginning of the Q wave and the end of the T wave.

0.16

50

100

150

200

Heart rate (beats.min

Figure 3 Proportion of late QT interval from the total

repolarization time (TmTo/QTo) in relation to heart rate
during exercise (Ex, open circles) and recovery (Rec,
closed circles). QTo=time interval between beginning of
Q wave and end of T wave; TmTo=time interval between
apex of T wave and end of T wave.

QT to RR intervals in computerized linear regression

models'21. In each individual the dynamics have been
expressed with one linear regression slope relating repolarization duration to cycle length, irrespective of
heart rate'21. The present results in healthy children show
that the relationship of total QT and early QT to heart
rate are linear (Fig. 1) but to RR intervals curvi linear
QT intervals were shorter than 440 ms. The correction (Fig. 2) during exercise. If the dynamics of total QT and
equations are as follows: heart-rate adjusted total early QT intervals are expressed as linear in relation to
QT = measured total QT+1 -30 x (heart rate - 60) and cycle length the results depend on the prevailing heart
heart rate adjusted early QT = measured early rate in the data. This is because heart rate and cycle
QT+1-11 x (heart rate 60) during exercise; adjusted length are inversely related.
total QT = measured total QT+1-42 x (heart rate-60)
In the present method the QT intervals were
and adjusted early QT = measured early QT+l-30x measured during periods with opposite trends in heart
(heart rate 60) during recovery. Corresponding early rate. As a consequence, the group mean slopes relating
QT values were shorter than 350 ms.
total QT and early QT to heart rate during exercise
could be differentiated from the corresponding slopes
during recovery. The method presented by Merri et al.
Discussion
using Hotter recordings does not take into account that
QT intervals may
These results in healthy children show that it is more the heart rate preceding the measured
121
accelerate,
decelerate
or
vary
.
These
two methods
practical to correlate exercise QT values to heart rate than
may
therefore
reveal
different
information
about the
to cycle length. During exercise and subsequent recovery,
total QT, early QT or late QT intervals evolve in different dynamics of the QT interval.
It was shown previously, using resting ECGs in
ways. The late stage of the QT interval increases propordifferent
individuals, that the cycle length dependency of
tionally during early recovery indicating non-homogeneous
the
repolarization. A simple principle for evaluation of exercise repolarization duration is concentrated mainly in 471
first portion of the QT interval (early QT interval)' .
QT interval durations in children is presented.
Consequently, this principle has also been adapted in
studies of the QT interval at different heart rates in the
same individuals during interventions'2'. The present
Relationship of total QT and early QT
results show that the slopes relating total QT and early
intervals to cycle length and heart rate
QT intervals to heart rate differ from each other during
There is increasing interest in studying the dynamics of exercise (Fig. 1). Therefore, in intervention studies conelectrocardiographic repolarization duration, relating cerning the relationship of QT intervals to heart rate the
Eur Heart J. Vol. 17, November 1996

Downloaded from http://eurheartj.oxfordjournals.org/ by guest on November 26, 2015

600
800
RR interval (ms)

ExY=0.17+0.25xl0~3X
r=0.82 p<0.01

The QT interval in healthy children

total QT interval cannot be replaced by the early QT

interval without distorting the results.

Late duration of QT interval

Evaluation of exercise and postexercise

QT interval durations
Bazett's method for relating QT and RR intervals in
resting ECGs by QT = kVRR[9] and for correcting the
QT values for heart rate using the equation
QTc=QTVRR, has also been widely used when evaluating exercise QT intervals. If the congenital long QT
syndrome is suspected in children, exercise testing is
recommended to clarify the diagnosis110'"1. The present
results show that the heart rate dependency of Bazett's
'corrected QTc' values leads to marked error which is
greatest close to the rate of 110 beats. min ~ '. If exercise
testing is applied to aid diagnosing the long QT syndrome in children, the use of Bazett's method thus leads
to false-positive diagnosis.
The exercise QT intervals in children can be
normalized to heart rate with the aid of simple linear
regressions. By using this method erroneously long
rate-corrected QT intervals are avoided. Equations
which correct QT intervals to heart rates can be created
for different populations and exercise protocols separately. It seems reasonable to adjust total QT and early
QT values differently.

Limitations
This methodological study evaluates the behaviour of
the QT interval and its early and late portions during

exercise and recovery in a group of healthy children. To

provide reference data for slopes relating QT intervals
to heart rate and rate adjustments of QT values, large
numbers of healthy children and children with the
congenital long QT syndrome using different exercise
protocols should be studied. For such population
studies, accurate computerized methods to determine
both total QT and early QT intervals should be available. In this study, we used the generally accepted value
of 440 ms as the upper normal limit for the heart rate
normalized QT interval.

Conclusions
In studies concerning the dynamics of the QT interval it
is more practical to correlate QT values to heart rate
than to cycle length. The principles of the behaviour of
the QT interval portions during exercise and recovery
presented in this article can be used when developing
programs for computerized QT analysing systems. The
suggested model for evaluating the inhomogeneity of
repolarization warrants further investigation in different
heart diseases as well as in terms of the influence of
antiarrhythmic agents. The proposed approach for
evaluation of exercise and recovery QT interval durations may improve the diagnosis of the congenital long
QT syndrome in children.
Dr Viitasalo was supported by a grant from the Paavo Nurmi
Foundation, Helsinki, Finland.

References
[1] Seed WA, Noble MIM, Oldershaw Per al. Relation of human
cardiac action potential duration to the interval between
beats: implications for the validity of rate corrected QT
interval (QTc). Br Heart J 1987; 57. 32-7.
[2] Merri M, Moss AJ, Benhorin J, Locati EH, Alberti M,
Badilini F. Relation between ventricular repolarization duration and cardiac cycle length during 24-hour Holter recordings. Findings in normal patients and patients with long QT
syndrome. Circulation 1992; 85: 1816-21.
[3] Kadish AH, Weisman HF, Veltn EP, Epstein AE, Slepian MJ,
Levine JH. Paradoxical effects of exercise on the QT interval in patients with polymorphic ventricular tachycardia
receiving type la antiarrhythmic agents. Circulation 1990; 81:
14-9.
[4] Merri M, Benhonn J, Alberti M, Locati E, Moss AJ. Electrocardiographic quantitation of ventricular repolarization.
Circulation 1989; 80: 1301-8.
[5] Zabel M, Portnoy S, Franz MR. Electrocardiographs indexes
of dispersion of ventricular repolarization: an isolated heart
validation study. J Am Coll Cardiol 1995; 25: 746-52.
[6] Armitage P, Berry G. Statistical methods in medical research.
Trowbridge, Wiltshire: Blackwell Scientific Publications, 1987:
275.
[7] Benhorin J, Merri M, Alberti M et al. Long QT syndrome.
New electrocardiographic characteristics. Circulation 1990;
82: 521-7.
[8] Autenrieth G, Surawicz B, K.uo CS. Sequence of repolarization on the ventricular surface in the dog. Am Heart J 1975;
89: 463-9.
Eur Heart J, Vol. 17, November 1996

Downloaded from http://eurheartj.oxfordjournals.org/ by guest on November 26, 2015

In resting ECGs it has been shown that the time interval

between the T wave maximum and the end of the T wave
is independent of heart rate in different individuals'4'.
There seems to be great inter-individual variation in the
late duration of the T wave (Table 2), but in the same
individuals late QT correlates with heart rate when the
rate increases.
The T wave is a reflection of regional differences
in repolarization times of the ventricular myocardium.
Experimentally, the late duration of the QT interval
correlates with the dispersion of repolarization15'81. The
present finding, that the proportion of late QT in the total
QT interval, and thus possibly the dispersion of repolarization, increases at high heart rates and particularly during early recovery (Fig. 3), may be of clinical importance.
The susceptibility to ventricular arrhythmias associated
with exercise and early recovery may therefore be increased in cases when repolarization dispersion plays an
important role in the genesis of the arrhythmia, as occurs
in patients with the congenital long QT syndrome or in
patients using certain antiarrhythmic drugs. The explanation for the increase in late QT/total QT, particularly
during early recovery, remains unclear.

1727

1728 M. Viitasalo et al.

[9] Bazett HC. An analysis of the time relations of electrocardiogram. Heart 1920; 7: 353-70.
[10] Weintraub RG, Gow RM, Wilkinson JL. The congenital long
QT syndromes in childhood J Am Coll Cardiol 1990; 16:
674-80

[11] Vincent GM, Jaiswal D, Timothy KW. Effects of exercise on

heart rate, QT, QTc and QT/QS2 in the Romano-Ward
inherited long QT syndromes. Am J Cardiol 1991; 68498-503.

Downloaded from http://eurheartj.oxfordjournals.org/ by guest on November 26, 2015

Eur Heart J, Vol 17, November 1996