QT interval in 24-hour ambulatory ECG recordings from 60 healthy adult subjects

QT Interval in X-hour Ambulatory ECG RecordingsFrom 60 Healthy Adult Subjects

Vemer Rasmussen, MD, Gorm Jensen, MD, and J. Fischer Hansen, MD

Abstract: QT and RR intervals were measured in 24-hour electrocardiographic

(ECG) recordings from 60 healthy subjects randomly selected among participants in the Copenhagen City Heart Study. Five men and five women of each IO-year age group between the ages of 20 and 79 were included. The mean of three consecutive RR and QT intervals was calculated from ECG strips recorded at 1000, 1400, 1800, 2200, 0200, and 0600 hours in each subject. The RR interval varied between 728 ms at 1400 hours and 984 ms at 0600 hours, and the uncorrected QT interval between 358 ms at 1400 hours and 417 ms at 0200 hours. The QT interval corrected for heart rate by Bazett’s formula varied between 418 ms at 0600 hours and 428 ms at 1000 hours, and the QT interval corrected by the regression equation of this study varied between 396 ms at 1400 hours and 422 ms at 0200 hours. Multiple regression analysis of uncorrected data yielded a partial regression coefficient for heart rate influence of 0.14. After correction by Bazett’s formula, a statistically significant effect of heart rate remained (partial regression coefficient -0.08, t = - 9.93, p
The establishment in 1957l and the early 1960s of the long QT syndromes and their relation to sudden death in ventricular dysrhythmias renewed interest in the QT interval, which until then had remained a curiosity of dubious clinical significance. One reason for the reluctant attitude was that QT prolongation is regularly observed in patients with

certain disorders in which arrhythmias are particularly rare (e.g., hypothyroidism and hypocalcemia), and that a long QT interval is induced by some pharmacologic agents with prominent antiarrhythmic properties (e.g., quinidine and amiodarone) .’ The observation of prolonged QT intervals in patients with myocardial infarction and its association with sudden death3-* increases the clinical significance of QT measurements, prompting us to establish reliable normal values and to determine if a correction procedure is appropriate and, if so, which method could be recommended. The first correction formula to appear was that

From the Department of Cardiology, Municipal Hospital of Copenhagen. and the Copenhagen City Heart Study. Medical Department B. Rigshospitalet, Copenhagen, Denmark.

Supported by the Danish Medical Research Council. Reprint request: Vemer Rasmussen, MD, 14, Skovalleen, 2880 Bagsvaerd,

DK

Denmark

91

92

Journal of Electrocardiology

Vol. 24 No. 1 January 1991

proposed in 1920 by Bazett,6 by which means a corrected value of QT (QTc) corresponding to a heart rate of 60 beats/min could be calculated. Because of its simplicity this correction became popular and has remained so despite the proposal over the years of several alternative methods.6-” As has been demonstrated repeatedly, the RR interval varies more than the QT interval; 1oP’2and, as a consequence, inconsistent changes of the corrected QT interval occur during exercise and other physiologic maneuvers. For that reason it appears inappropriate to correct for heart rate influence only, as it gives misleading results, particularly when applied to dynamic electrocardiographic (ECG) recordings. Instead, it might prove more useful to collect data from Holter recordings from a sample of normal healthy subjects at various hours of the day and to present the uncorrected measurements together with some statistical parameters. The aim of this report is to present the results of a study of QT intervals in Holter recordings from a carefully selected group of healthy individuals.

Materials

and Methods

The study group consisted of 60 participants of the Copenhagen City Heart Study, which comprised 19,662 healthy persons above the age of 20 years selected from a well defined specific area of Copenhagen with a total adult population of 87,172.13 In connection with a 5-year follow-up examination of all subjects enrolled in the Copenhagen City Heart Study, a subset comprising five men and five women of each of the six IO-year age groups between the ages of 20 and 79 was randomly selected for Holter monitoring. The ECG recordings were made using six AM tape recorders (Hittman CompAct IV) with a tape speed of 2 mm/s. A Hittman Compu Scan III system was used for analysis. From each of sixty 24-hour ECG recordings from these subjects, six ECG segments comprising at least five QRS complexes were printed at a paper speed of 50 mm/s at the recording hours 1000, 1400, 1800, 2200, 0200, and 0600. In case of distortion of the ECG signal at the preset hour, the first succeeding technically satisfactory ECG segment was used. In eight cases, one measurement was lacking. In live, the T wave was poorly defined in both simultaneous recordings at the preselected recording time. In three cases, monitoring had to be stopped for technical reasons before a full 24-hour recording had been obtained.

On each ECG segment three consecutive RR and three QT intervals were measured to the nearest 10 ms, the latter from the beginning of the QRS complex to the point at which the T wave returned to the isoelectric line. When a U wave was present, the downslope of the T wave was completed by freehand drawing. The mean value of the three measurements was used to represent the RR interval and the uncorrected QT interval (QTm), respectively. Interobserver variation was determined by comparing measurements made by one of the authors in 10 of the 60 patients with those made in the same patients by a technician who was not informed of the purpose in advance and who did not have access to the results of the author. The interobserver variation was insignificant. In 42% of the measurements there was full agreement, in 81% the variation was less than 10 ms, and in 99% the variation was less than 20 ms.

Statistical Analysis The QT interval was studied by a multiple regression analysis, including the RR interval as covariate and time-point and persons (patients) as class factors. In an analysis of variance the patient effects were further described as a sum of a gender effect, an age effect for six lo-year intervals, and a random term due to individual variation. This technique means that the mean of the jth observation (at time tj) of QT for the ith person is assumed to be of the form (Y + BRR + yi + Aj The effect of time of day (Fig. 1) was further studied by a circadian model with a mean of the form 15

r x

0

x

x

x

-5-

-IO-

*

-15 o

Fig.

2

1.

model

4

6

2

10

12

14

18

18

20

22

I 24

Estimates of the effect of time of day in the general and the circadian model (dotted line).

(x )

QT Intervals

cx + PRR + yi &in (2ntj)/24 For computer analysis program on an IBM mainframe.

+

ECOS

(ZITtj)/24

system SAS was used

Results The mean values of measured RR and QT intervals (QTm) are presented in Table 1, and a plot of observed points of QTm versus RR for each of the 6 hours of the day is depicted in Figure 2. Multiple regression analysis demonstrated a strong relation between the QT interval and heart rate (RR interval), yielding a partial regression coefficient of 0.14. After correction for heart rate according to the principles of Bazett,6 using the formula QTb = QTm/V’&, a statistically significant effect of heart rate on QT remained, the partial regression coeffi-

ms

in Healthy

Adults

l

Rasmussen

et al.

cient in this case being -O.O8(t = -9.93, p
IO.00

22.00

Fig. 2.

Plot of measured QT interval (QTm) versus RR interval for men and women at 0600 hours. Where no number is indicated, the plot comprises 30 males and 30 females.

M = 30

, Males

93

o Females

F =2Q

94

Journal of Electrocardiology

Vol. 24 No. 1 January 1991

Table 1. Mean Values of RR Intervals, Uncorrected

QT Intervals, QT Intervals Corrected for Heart Rate Influence Using Bazett’s Formula, and the Regression Equation of the Present Study at Various Times

Table 3. Parameters in the Circadian Model Value (msl Parameter

Estimate

SE

(level at 1,000 ms) P (RR) D (sine) E (cosine)

284.9 0.142 7.1 8.3

9.2 0.008 1.4 1.4

Q

Hour of Day

No. of ECGs

1000 1400 1800 2200 0200 0600 All hours

(:)

Fmy

g;

(“m’ss

752 728 775 850 968 984 844

368 358 371 386 417 411 385

428 422 424 422 427 418 423

403 396 403 407 422 413 406

56 59 60 60 59 60 354

QTc = QTm + 0.14( 1,000 - RR) (regression equation of the present study). QTb = QTm/V% (Bazett’s formula). QTm = uncorrected QT; QTb = intervals corrected for heart rate influence.

Parameters specific to individuals are not shown.

circadian effect is F = (12,287/76,095) = 23.33, DF = 2, 289; p
x (289/2)

Discussion day, and gender, there was an individual residual ef feet, the variance of which amounted to 330. A normal range of QT interval length for each of the 6 hours of the day was established by correcting the mean values for heart rate influence (Table 2) at a basic heart rate of 60 beats/mm (RR = 1,000 ms), using the formula Mean QTc = mean QTm + 0.14 ( 1,000 - RR), and calculating 95% limits as 1.96 x dvariance

of person effect + residual variance

= 1.96 x d285.91

+ 257.14

With person effects and arbitrary effect of time of day, the residual sumof squares (RSS) is 73,799 (286 d.f.). In the circadian model, for which estimates are shown in Table 3, the RSS is 76,095 (289 d.f.). Thus the test for the circadian model is F = (2,296/73,799) x (286/3) = 2.97; DF = 3, 286; p CO.05 (Fig. 1). Under the hypothesis of no circadian effect, the RSS is 88,382 (291 DF), and thus the test for no

The establishment of a study group representative of the general healthy population is a major problem when determining normal values of physiologic parameters because in most cases available subjects are to some extent selected by symptoms, occupation, and so on. In addition, latent disease among the participants is often not quantitatively excluded owing to restricted follow-up periods. In the present case conditions were propitious. The study group was randomly selected from among the participants of the Copenhagen City Heart Study, which was longitudinal, cross-sectional, and nmning over a period of more than 5 years. The sample was large ( 19,662), age-stratified, and selected from a stable resident population living in a specific, well defined, urban area; moreover, the response rate was high (over 70%). Thorough examination of each patient at an interval of approximately 5 years ensured that the probability of latent disease was small at the time of inclusion. Because a standardized method of QT measurement has not yet been devised, the QT interval was measured in three consecutive cycles and the time

Table 2. QT Interval Corrected for Heart Rate Influence by the Regression Equation of the Present Study Men (n = 30) Hour of Day 1000 1400 1800 2200 0200 0600 All hours

Women (n = 30)

Mean QTc (ms)

Normal Range (ms)

Mean QTc (ms)

Normal Range (ms)

389 382 389 393 408 399 392

343-435 336-428 343-435 347-439 362-454 353-445 346-438

418 411 418 422 437 428 421

372-464 365-457 372-464 376-468 391-483 382-474 375-467

Normal range = mean QTc f 0.96 u285.91

+ 257.14.

QT

averaged according to the recommendations of Kenny and Sutton.7 Being the only available leads, two bipolar chest leads were used for the analysis. As the most distal of these electrodes were located close to the V2 and Vq positions, the condition of Campbell et al.’ to use centrally positioned chest leads was met. The mean value of the uncorrected QT interval (QTm) in men and women of the present study (392 and 42 1 ms, respectively) differed considerably from those reported by Bazett6 (0.326 and 0.342 second, respectively), probably due mainly to the different circumstances under which the recordings of the two studies were made. Inclusion of younger subjects in Bazett’s study group may also have tended to reduce the QT interval. However, the present results were similar to those of more recent reports.‘-’ ’ When corrected for heart rate, as proposed by Bazett, the mean QT (QTb) was 423 ms, which is longer than most previously reported values of corrected QT6,7,9but in close agreement with the findings of Browne et a1.,14 who studied 15 healthy persons using prolonged, continuous ECG recordings. The apparently systematic difference between the results from long-term monitoring and those obtained under usual standardized conditions for ECG recording supports the observation of experimental studies that the hypothesis of proportionality between the QT interval and the cycle length, on which the formula of Bazett is based, does not apply in various physiologic stress situations that are repeatedly encountered during daily life. Davidowski and Wolf15 demonstrated that major changes of heart rate during maneuvers such as breath holding, hyperventilation, dive reflex, Valsalva maneuver, and cold pressor test caused only minimal changes in the QT interval of 20 healthy subjects, in whom the well appreciated shortening of the QT interval was observed during exercise-induced tachycardia. Browne et a1.14found higher values of QTc during sleep than in the awake state, indicating involvement of independent mechanisms in the regulation of heart rate and repolarization duration. The present study supports this hypothesis by disclosing a genuine circadian variation, i.e., a variation that is independent of heart rate. As a consequence, in addition to correction for heart rate, the hour of day and gender must be taken into account before a certain measured QT interval can be classified as either normal or abnormal. The results of this study might be useful for establishing normal ranges of QT interval for each gender at various hours of the day. However, as the correction for the personal and circadian effects were derived retrospectively, the results must be replicated

values

Intervals in Healthy Adults

l

Rasmussen

et al.

95

in an independent test population before being considered for general application.

Acknowledgments The authors thank Philip Hougaard, cand. stat. et lit. scient., Novo Research Institute, who performed the statistical analysis.

References F: Congenital deafmutism, 1. Jervell A, Lange-Nielsen functional heart disease with prolongation of the QT interval, and sudden death. Am Heart J 54:59, 1957 repolari2. Somberg J, Tepper D, Wynn J: Prolonged zation: a historical perspective. Am Heart J 109:395, 1985 3. Schwartz PJ, Wolf S: QT interval prolongation as predictor of sudden death in patients with myocardial infarction. Circulation 57: 1074, 1978 in acute myocar4. Ahnve S: QT Interval prolongation dial infarction. Eur Heart J 6(suppl D):85, 1985 5. Dorogazi RM, Childers R: Time related changes in the QT interval in acute myocardial infarction: possible relation to local hypocalcemia. Am J Cardiol 41:684, 1976 6. Bazett HC: An analysis of the time relations of electrocardiograms. Heart 7:353, 1920 7. Kenny RA, Sutton R: The prolonged QT interval-a frequently unrecognized abnormality. Postgrad Med J 61:379, 1985 8. Campbell RWF, Gardiner P, Amos PA et al: Measurement of the QT interval. Eur Heart J 6(suppl D):81, 1985 9. Simonsen E, Cady LD, Woodbury M: The normal QT interval. Am Heart J 63:747, 1962 10. Ahnve S, Vallin H: Influence of heart rate and inhibition of autonomic tone on the QT interval. Circulation 3:435, 1982 11. Ahnve S: Correction of the QT interval for heart rate: review of different formulas and the use of Bazett’s formula in myocardial infarction. Am Heart J 109:568, 1985 12. Jackman WM, Friday KJ, Anderson JL et al: The long QT syndromes: a critical review, new clinical observations and a unifying hypothesis. Prog Cardiovasc Dis 31:115, 1988 13. Ostor E, Schnohr P, Jensen G et al: Electrocardiographic findings and their association with mortality in the Copenhagen City Heart Study. Eur Heart J 2:317, 1981 14. Browne KF, Prystowski E, Heger JJ, et al: Prolongation of the QT interval in man during sleep. Am J Cardiol 52:56, 1983 15. Davidowski TA, Wolf S: The QT interval during reflex cardiovascular adaptation. Circulation 69:22, 1984