OUP user menu

Changes of the corrected QT interval in healthy boys and girls over day and night

Thomas Krasemann , Christina Strompen , Jasmin Blumenberg , Josef Gehrmann , Gerhard Burkhardtsmaier , Johannes Vogt
DOI: http://dx.doi.org/10.1093/eurheartj/ehn452 202-208 First published online: 2 October 2008


Aims The study was designed to detect changes in corrected QT intervals over day and night in both sexes in healthy children.

Methods and results The corrected QT interval was calculated from 24 h ECGs obtained from 282 healthy children aged 6 months to 18 years. The QTc interval as measured by the 24 h recording differs to the standard ECG measurement which is in average of 40–50 ms shorter. The QTc interval changes little over a 24 h period and is remarkably constant despite significant heart rate changes in healthy children.

Conclusion The routine ECG—even if the calculated values differ markedly from those obtained over 24 h—seems to be a good screening method for the measurement of corrected QT intervals, because the corrected QT interval is kept constant over the whole day in healthy children.

  • QT-interval
  • 24-hour-ECG
  • Children


More than 100 years ago, in 1903, William Eindhoven described the electrocardiogram; and prolonged recordings were first made almost 60 years later by Holter in 1961.1

It is now known that during growth from the neonate to the adolescent, the size and position of the heart changes in relation to the whole body as well as the right ventricular pressure. Thus, the normal ECG of a small child might be considered to be pathological in adulthood. It is therefore essential to have normal values (or at least guide values) for each age group. Davignon et al.,2 Garson,3 Stoermer and Heck,4 or Rijnbeck et al.5 performed several 1000 routine ECGs in healthy children and published their normal values of the routine ECG.

Measurement of the QT interval is difficult even with high quality electrocardiography. It is usually carried out in lead II to avoid U waves. However, James cited in Garson3 stated 40 years ago: ‘In electrocardiography there is no more nebulous measurement than the QT interval.’ The QT interval, however, is an important measurement because QT prolongation is associated with severe ventricular arrhythmias in the congenital and drug induced long QT syndromes.

There are only very few studies on QT interval in children out of the neonatal age where measurements are particularly difficult to interpret, for example the one conducted by Eberle et al.6 Their youngest child was 5.2 years and their oldest 16.5 years. They only looked at the QT interval on the routine ECG, but did not perform a 24 h study. The very few Holter ECG studies on children focus on the heart rate.711 Until now, there has been no data on the QT interval in children at different times of the 24 h cycle, which have been shown in adults.12,13

In 1920, Bazett14 corrected the QT interval to make it in different heart rates comparable. Several formulas have been developed since then,1518 but worldwide the commonest in use is still Bazett's formula. Interestingly, in comparison to other formulas, when group-based regression parameters were applied to individuals, no formula had a clear advantage over Bazett's formula.19

Our study was designed to detect any differences of the corrected QT intervals during sleep and being awake in healthy children of both genders, and differences between the sexes, too.


We recruited healthy children aged 6 months to 18 years from kindergartens and schools between October 2000 and March 2002. An information leaflet was distributed to all children, whose parents could then decide whether their children were allowed to participate. The information leaflet contained information about the inclusion criteria, including the fact that we were seeking healthy children free of chronic or acute diseases. The included children were healthy with an unremarkable physical examination, free from any chronic diseases, and were not on any medication. If a heart murmur was present, echocardiography confirmed that this was innocent. In all cases, weight and length were measured. Adult studies suggested that excessive body weight may have an influence on the QT duration.2022 Even if this was not proven for children at the time of the recruitment, only children with a weight between the 10th and 90th percentile were included. The study population was divided into different age groups similar to those used by Davignon et al.2 To reach a power of 80%, we aimed for 17 participants of each gender in each age group. Table 1 shows the different age groups, the number of children in each group, and the sex ratio. Table 2 shows the weight distribution.

View this table:
Table 1

Number of participants

GroupAge (years)Number ♀Number ♂Number overall
16 months–<1358
View this table:
Table 2

Age and weight

GroupAge (years)ParameterWeight ♂ (kg)Weight ♀ (kg)
16 months–<1Mean8.157.9
Standard deviation1.560.85
Standard deviation1.572.3
Standard deviation2.82.09
Standard deviation8.744.73
Standard deviation8.529.76
Standard deviation13.129.8
Standard deviation6.8112.31

Children with known frequent ectopics were excluded. If the routine-ECG showed rhythm disturbances or the corrected QT interval exceeded 480 mL/s or rhythm disturbances were visible, the children were excluded from the study.

A routine 12 lead ECG was performed at a paper speed of 50 mm/s and 10 mm/mV using a Megacart Elima AB, model 9657578, Siemens. Two independent investigators measured the heart rate, rhythm, and QT interval, which were then averaged. To minimize the problems of distinguishing between T- and U-waves16,23, the QT-interval was measured in lead II; if there were too many artefacts (i.e. T-wave amplitude very low) either lead III or V5 were used. The corrected QT interval was calculated with the use of Bazett's formula.14

Holter ECG was performed using hydrogel electrodes (Blazoner Medical) and digital recorders (Seer MC, GE Medical Systems, Milwaukee, WI, USA). The three leads were defined as shown in Figure 1. The sampling rate was 128 Hz. If the child did not tolerate the electrodes for the whole of 24 h, this was taken into account (see below). On 24 h ECG, heart rate and QT intervals were measured in 15 s intervals. Parents reported the time when the children were asleep and when they woke up. We defined night as the time when all children were asleep and day as the time when all children were awake. Thus, night time was between midnight and 5 am, while day time was 9 am to 7 pm. It is noteworthy that due to the definition of daytime (9.00–19.00 o'clock) and nighttime (0.00–5.00 o'clock), the sum of both does not equal 24 h.

Figure 1

Lead position for 24 h ECG.

Analysis of the Holter ECGs was performed with the software Mars PC version 6.0 (GE Medical Systems). This software allows a QT measurement. The QT intervals were measured in 15 s intervals. Bazett's formula then was applied. All Holter ECGs were edited manually to rule out artefacts.

Statistical analysis

For each age group and sexes, average and standard deviation of heart rate, QT interval and corrected QT interval were calculated for each of the Holter channels over the 24 h period and for the day and night periods. All channels were compared with each other. After a two-sided analysis of variance showed differences, the student's t-test for related groups was used to compare the mean values of the day and night intervals, and between the sexes the student's t-test for non-related groups was used. The QTc on the routine ECG was compared to the mean of the QTc over 24 h in each channel and for the day and night periods using student's t-test as well. A P-value of <0.05 was considered to be significant.

Statistical analysis was performed with software SPSS version 11.5 for windows (SPSS Incorporated, Chicago, USA) and Microsoft Excel 2003 (11.56, Microsoft Corporation USA).

The Ethical Committee of the Westphalian Wilhelms University and the Aerztekammer Westfalen-Lippe approved the study on 17 May 2000.


Of 290 children screened for inclusion, two were excluded because their bodyweight exceeded the 90th percentile. Five children were on medication and were excluded as well. One boy was excluded for known frequent ectopics. None was excluded for an abnormal corrected QT-interval. Thus, 282 children were included in the study.

The baseline ECG data are shown in Table 3.

View this table:
Table 3

Routine ECG

GroupParameterHR ♂ (bpm)QT ♂ (ms)QTc ♂ (ms)HF ♀ (bpm)QT ♀ (ms)QTc ♀ (ms)
Standard deviation13102040.456040
Standard deviation16.4330209.352010
Standard deviation10.57202014.922010
Standard deviation14.15202011.82020
Standard deviation12.11202014.42020
Standard deviation11.59302014.673020
Standard deviation15.7220309.882010

Over the 24 h period, an average of 5422 15 s intervals (theoretical maximum of 5760) was obtained for heart rate with a range from 4624 to 5601 for the different age groups. Fewer 15 s intervals were satisfactory for QTc-calculation with an average of 4692 in channel 1 (range 4342–5528), 3830 in channel 2 (range 2883–4891), and 4821 in channel 3 (range 4108–5392). The quality and number of artefacts in 15 s-intervals were evenly distributed over the 24 h period. One boy in group 1 did not tolerate the leads for 24 h, thus his night recording was not complete.

Table 4 shows the average and standard deviation of the heart rate and QTc intervals in different channels. The night time heart rate was lower in all age groups. There was no significant difference between the sexes for the QTc in any period (Table 5). There was a suggestion that older girls had longer QTc at night in channel 3. The comparison of day and night as well as both sexes concerning the parameters heart rate and QT in all channels are depicted in Table 5.

View this table:
Table 4

24 h ECG

DaytimeGenderAge (years)Heart rate (bpm, mean/STD)QTc1 (ms, mean/STD)QTc2 (ms, mean/STD)QTc3 (ms, mean/STD)
24 h6 months–<1117 ± 18439 ± 14438 ± 12441 ± 11
1–<393 ± 16443 ± 19444 ± 18437 ± 15
3–<596 ± 13430 ± 17431 ± 19424 ± 19
5–<887 ± 11438 ± 15439 ± 21436 ± 16
8–<1283 ± 9441 ± 16442 ± 17441 ± 16
12–<1682 ± 7441 ± 12440 ± 15441 ± 13
16–1877 ± 7424 ± 21422 ± 17421 ± 20
6 months–<1108 ± 30448 ± 14449 ± 20443 ± 20
1–<3109 ± 9441 ± 11440 ± 18438 ± 10
3–<5105 ± 10433 ± 13437 ± 16427 ± 13
5–<892 ± 8440 ± 15438 ± 19435 ± 15
8–<1285 ± 8440 ± 15444 ± 16439 ± 15
12–<1684 ± 8444 ± 17442 ± 18444 ± 17
16–1880 ± 2448 ± 19454 ± 29447 ± 18
Day6 months–<1128 ± 16439 ± 13440 ± 15442 ± 9
1–<3106 ± 21445 ± 20445 ± 18441 ± 18
3–<5110 ± 14434 ± 17433 ± 22426 ± 19
5–<898 ± 14438 ± 15440 ± 21434 ± 15
8–<1293 ± 11440 ± 16442 ± 18438 ± 16
12–<1693 ± 8441 ± 12440 ± 16441 ± 12
16–1886 ± 12425 ± 20425 ± 15422 ± 18
6 months–<1115 ± 33442 ± 18447 ± 18438 ± 25
1–<3123 ± 9443 ± 12442 ± 20439 ± 11
3–<5119 ± 9436 ± 13439 ± 18426 ± 15
5–<8105 ± 10440 ± 14439 ± 20433 ± 15
8–<1295 ± 11438 ± 13445 ± 14435 ± 14
12–<1695 ± 10443 ± 15443 ± 19440 ± 17
16–1889 ± 5446 ± 19453 ± 30443 ± 18
Night6 months–<197 ± 20441 ± 20440 ± 16442 ± 18
1–<380 ± 15444 ± 22445 ± 17438 ± 18
3–<579 ± 11423 ± 24427 ± 24422 ± 22
5–<872 ± 9439 ± 17439 ± 23440 ± 19
8–<1268 ± 9444 ± 20445 ± 25446 ± 20
12–<1664 ± 7442 ± 174391 ± 21443 ± 18
16–1862 ± 9421 ± 21415 ± 25420 ± 24
6 months–<1102 ± 24453 ± 9453 ± 25451 ± 15
1–<393 ± 9440 ± 10442 ± 194404 ± 13
3–<589 ± 9431 ± 14437 ± 18427 ± 12
5–<876 ± 8432 ± 51432 ± 52431 ± 51
8–<1271 ± 8441 ± 22443 ± 25443 ± 20
12–<1670 ± 8448 ± 24441 ± 21447 ± 20
16–1863 ± 1451 ± 19455 ± 29451 ± 18
  • STD, standard deviation; day, night.

View this table:
Table 5

Differences of heart rate and QTc between day and night, and between the genders (significancant values are boldfaced)

Age (years)ParameterDay vs. night (♂) P =Day vs. night (♀) P =♀ vs. ♂ (24 h) P =♀ vs. ♂ (day) P =♀ vs. ♂ (night) P=
6 months–<1HF0.00180.14860.69390.57070.7708
6 months–<1QTc10.66660.18390.47380.84720.3169
6 months–<1QTc20.76670.27250.44720.58680.4842
6 months–<1QTc30.89120.17650.88480.83930.4915

Ninety-five per cent confidence intervals were calculated and are depicted in the online Supplementary figures 2–15. For a better overview, the defined night and day times are highlighted.

The QTc calculated from routine ECG was approximately 40–50 ms shorter than QTc measurements in each channel over 24 h (P < 0.001). These differences were similar when comparing the baseline ECG QTc to any of the channels in the day and night periods alone.


In 1956, Jervell and Lange-Nielsen24 described the long QT syndrome. Since then, several types of long QT syndrome have been described. These might lead to fatal Torsade de pointes and ventricular fibrillation.23

An association of QT prolongation with sudden infant death syndrome has been suspected, but is yet to be proven conclusively.2528

Typically, a routine ECG is obtained and the QTc calculated. For this, usually Bazett's formula is used.14 As early as 1920, Bazett showed that the QT interval is dependent on the heart rate. To compare the QT interval of different patients, he described his formula, which remains the most commonly used.19,29 Several other formulas have been developed.15 One advantage of Bazett's formula is its simplicity.30 Davignon et al.2 used Bazett's formula for the QTc calculation to establish normal values for the paediatric ECG.

To establish or rule out the diagnosis of a long QT syndrome in adults, a stress ECG is used as well, because it is known that the corrected QT interval prolongs under physical activity.23 This method is not applicable to small children. A 24 h ECG might be useful in these age groups if normal values for the QTc are available.

Normal values for QT intervals and corrected QT intervals in routine ECGs2 are not directly applicable to the 24 h ECG. Our data show that there is a significant difference between the QTc obtained from routine ECG and any of the channels of the 24 h ECG. According to our data, the average corrected QT interval over 24 h is significantly longer than in the routine ECG. This is independent of the time of day. Differences between standard ECGs and ambulatory ECGs have been described previously, but not in the paediatric population.31 In a recent paper, Dalla Pozza et al.32 recently found longer maximum QTc intervals on Holter ECGs than in routine ECGs in children with Turner syndrome. It is noteworthy that in our study, the confidence interval in the groups with small numbers of children is relatively large, showing more the intraindividual changes (online Supplementary figures 2–15). Thus, comparing routine ECG and short periods of the 24 h tape will mislead. Unfortunately, measurements of the QT interval were not carried out exactly at the same time in routine and ambulatory ECG.

As only children with a normal QTc duration on the routine ECG were included in our study, our data do not allow us to answer if a shorter QTc during the 24 h study means an abnormal foreshortened QT in routine ECGs or what the cut-off for a pathologically prolonged QT-interval is. Further studies on affected individuals with a short QT or long QT syndrome are needed to answer these questions.

It has to be taken into account that the lead positions in the ambulatory ECG are different from the routine-ECG, and ideally a routine ECG is obtained at rest.

We think that it is essential for the comparison of QT values measured in Holter ECG to have defined leads, as the leads are defined in the routine ECG. With our defined lead position, we could rule out that non-identical lead positions have an influence on the QT measurement.

In our study, we focused on the QTc differences between day and night, representing the time of day when all children were either awake or asleep. We can assume that during day time, children are physically more active than during night time. In adult studies, QT intervals during day time and night time were compared as well.13,33,34 The QT interval is dependent on several physiological variables, such as heart rate, sex, and age.22 Earlier studies on corrected QT intervals in all the ECGs either had very few participants or showed methodological weaknesses.7,10,11,35

About 25 years ago, three studies on the circadian pattern of the heart rate in healthy children were carried out by Scott et al.,10 Southall et al., 11 and Lindinger and Hoffmann.7 These showed a nocturnal heart rate that was lower than the diurnal heart rate. This was comparable to our study. There were only two age groups in which we did not find a significant difference between day and night heart rate, these were the female infants aged 6 to 12 months and the male adolescents aged 16 to 18 years. Probably, the low number of participants did not allow the difference to become significant.

As expected, heart rate slowed with increasing age. This is in concordance with earlier studies.7,8,10,11 Different from other investigations,5,6,3638 we did not find significant difference of the heart rate between both sexes in all age groups.

There is only one published study to date focusing on the corrected QT intervals in 24 h ECGs in healthy children.35 This study included only 20 boys and 12 girls aged 6 to 11 years. Children aged below 6 years (in our study 54%) and all children above 11 years (in our study 64 children, 23%) were not included. The study has another disadvantage compared to our study: the Holter recorder only had two channels, and it is not clear whether the different leads were positioned in a defined manner. Historical studies have used tape recorders,35,39 while we used digital data storage to avoid undulations.

Our age groups were based on the study of Davignon et al.2 While most of our age groups contained a reasonable number of children, a few did not: for the 16 to 18 year old children, we can assume the same normal values as for adults, as typically these adolescents already finished their puberty. Thus, we were not forced to increase the numbers of subjects. The small number of participants in these groups might be the reason why we did not observe shorter QT intervals in the post puberty male group as described by Rautaharju et al.40

It is less satisfactory to have only a small number of children in the infant age groups (6 months to 3 years). To have a statistical power of 80% each group should have contained a minimum of 17 children. Unfortunately, there is no other study on the subject in these age groups available. Especially, the first year of life might be of interest because the relationship of prolonged QT interval and sudden infant death syndrome has been assumed.27 The values obtained from the other groups can be used as guide values to interpret QTc measurements in Holter ECGs obtained from children (Table 4).

There were only a few significant differences of the QTc interval between night and day time, but interestingly these were not significant in all groups, or not in all leads (Table 5). One has to take into account that mathematically significant differences here show an absolute maximum of 10 ms. The clinical bearing of these differences is not yet clear.

While the heart rate showed typical differences between night and day, we could not find this for the QTc interval. Molnar et al.33 showed a marked change in the QTc interval over 24 h in 21 adults. The QTc was shorter while the adults were awake. It is not clear whether this is a statistical problem (small number of participants) or is caused by methodological weakness (lead position not defined).

Our definition of day and night was based on parental observations and therefore we could be certain when all children were either asleep or awake during the defined times.

In adults, the QTc interval differs between genders in the routine ECG.12,33,36,38 Our study did not show any significant differences between both sexes in any age group.

More recent studies have shown that the obesity leads to prolonged QT intervals and QTc prolongation as well.41 To rule out the influence of abnormal body weight, we only accepted children with a body weight between the 10th and 90th percentile.

With increased number of participants in the different age groups, the confidence interval becomes smaller. One could assume that rapid QT changes might be overseen in a 24 h ECG, but it is much more likely to miss them in a routine ECG.

The routine ECG seems to be a good screening method for the measurement of corrected QT intervals, because the corrected QT interval is kept constant over the whole day.

We hope to add some sufficient data for the interpretation of the paediatric 24 h and routine ECG, because, as Frank Wilson pointed out:3 ‘There are comparatively few people who are not in greater danger of having their peace and happiness destroyed by an erroneous diagnosis of cardiac abnormality based on a faulty interpretation of an electrocardiogram than being injured or killed by an atomic bomb.’


This study has demonstrated that the QTc interval changes little over a 24 h period and displays remarkably constant despite significant heart rate changes in healthy children. The QTc interval as measured by the standard ECG differs to the 24 h recording measurement which is in average of 40–50 ms longer. There were no significant differences between the different lead positions, but certain lead positions were more susceptible to artefacts.


We would like to thank Peter Kuras for his enormous help with the data-extraction.

Conflict of interest: none declared.


View Abstract