- Email: [email protected]
Maturation of cardiac control in full-term and preterm infants during sleep
Early Human Development, 1980,4/2, 145-159 o Elsevier/North-Holland Biomedical Press
145
Maturation of cardiac control in full-term and preterm infants during sleep PETER G. KATONA, ALAN FRASZ and JOHN EGBERT
Schools of Medicine and Engineering, Case Western Reserve University, Cleveland, Ohio 44106, U.S.A.
Accepted for publication 26 February 1980
SUMMARY
ECG and respiration were recorded from 13 full-term and 8 preterm infants at 2-&week intervals during sleep at home in the first year of life. Average pulse interval in both quiet and active sleep was minimum at postnatal ages between 4 and 10 weeks, with the minimum during quiet sleep being significantly smaller for preterm (409 f 22 (SE) msec) than for full-term (445 1 9 msec) infants. The minimum average pulse interval of preterm infants was smaller than the pulse interval of full-term infants at any postconceptional age, and a smaller average pulse interval and smaller variations in pulse interval in preterm infants during quiet sleep persisted until a postnatal age of 7 months. The pulse interval variations attributable to respiration varied substantially with age. The results indicate that developmental changes in cardiac rate control are functions of both postnatal and postconceptional age, with the post-conceptional age at birth setting the mean level of pulse interval, a level which is then altered by development linked to postnatal age. sympathetic control; parasympathetic control; maturation; heart rate; respiratory sinus arrhythmia; sleep; infants
INTRODUCTION
Although the development of autonomic control mechanisms has been extensively investigated in the newborn period, only incomplete information is available about the continuing rapid development that occurs during the first year of life. Lipton and Steinschneider were among the first to draw attention to this development by describing the changes in heart rate response to an external stiiulus at ages 0, 2.5 and 5 months [13]. The need to better
146
understand this development has been recently underscored by suggestions that sudden infant death syndrome (SIDS) may be a manifestation of malfunctioning autonomic control mechanisms, and that defects in autonomic control mechanisms might be detected on the basis of respiratory and/or heart rate patterns [ 17,201 . It seems especially desirable to compare developmental patterns of fullterm and pretenn infants because SIDS is more common in preterm infants, and because the difference in both heart rate and respiratory rate between full-term and preterm infants is maximum at a postnatal age of about 2 months [lo], when SIDS is also most prevalent. The purpose of our work is to directly compare the longitudinal development of cardiac control in normal full-term and preterm infants during the first year of life using measurements obtained in the infants’ homes. The few previous reports dealing with maturational development of the cardio-respiratory system past the immediate neonatal period are generally concerned with either full-term or preterm infants, and the results are based on measurements performed in the laboratory or clinic [6,12,27,28]. METHODS Normal full-term and preterm infants were recruited on the basis of expected cooperation from their parents. The study was initiated for a total of 27 infants; data is presented for the 21 subjects who complied with the study protocol. The final study group consisted of 13 full-term infants with average birth weight of 3248 f 439 (SD) g, and 8 preterm infants with average gestational age and birth weight of 33.3 f 2.2 weeks and 1874 f 362 g, respectively. Table I shows the gestational age and birth weight of the subjects. Electrocardiogram and respiration were recorded in each infant’s home at approximately a-week intervals until age 5 months, and at 2-5-week intervals up to 12 months. Recordings of l-3 hours duration were obtained while the infant was asleep as judged behaviorally. The environment was generally quiet and the mean temperature of the rooms of full-term infants (23.7 ? 0.9 (SD) “C) was essentially the same as the temperature of rooms of preterm infants (24.1 * 0.9 “C). ECG and respiration were obtained using a Beckman VSM-100B vital signs monitor. The monitor, which was designed to indicate average heart rate and to derive the respiratory waveform from changes in impedance across the chest, was modified to also provide a beat-by-beat indication of the pulse intervals. The two active electrodes were placed on the mid-axillary lines in the 3rd-5th interspace. The data were recorded on a Wolff 4channel cassette tape-recorder, and displayed on a 2-channel chart recorder for immediate verification. The recordings were first scored for quiet and active sleep. These two states, which are precursors of the NREM and REM states in adults, can be consistently identified after a gestational age of 36 weeks [ 15 ] ; only a few
i
x x x x x xxxxx x x x xxx x x x x x x x x x xx x x x xx xx x x x xx xx xx PXXXX x x x x x x x x I xxx xx x x x xx xxx* x
20
2360 1518 1717 2280 2020 1900 1290 1910
34 32 32 36 34 34 29 35
Pl P2 P3 P4 P5 P6 P7 P8
10 xx x x x x x x x xxx xxx x x x x x x x x x x x x xxx x xx xx x x x x xxx x xxxxx xxxxx x x x *xxx* x x x x x x x x * x xx P xxx xx x xx x x x I xxxxx x x x
40 39 40 40 41 41 40 39 40 39 40 40 40
Fl F2 F3 F4 F5 Fti F7 F8 F9 FlO Fll Fl2 F13
1
Age (in wk) at which recordings were taken
2722 2667 3667 2976 3585 3972 3570 3788 2778 3175 3405 3036 2892
Gestational age (wk) Birth weight (g)
Infant
Full-term (Fl-F13) and preterm (Pl-P8) subjects studied ~~
TABLE I
x x
x
x x
x x
x
x x x x x x xx x x x x x x x x xx x
x
XXX
x xx x x xx x x
x
xx x x
x x x x x x
..___ xx
x
x x
x
x
x
x
x
x
x x x x x x x
x
x
x x x
148
recordings in some preterm infants were obtained before this age. The identification of states is traditionally based on recordings of EEG, EOG and muscle potentials, but polygraphic studies have shown that during quiet sleep respiration is regular and body movements are absent except for occasional sighs, while during active sleep respiration is irregular and there are frequent body movements [15,16,22]. Classifying sleep states as quiet or active on the basis of the regularity or irregularity of respiration, and the absence or presence of motion artifact in the impedance tracing, respectively, minimized the disturbance and discomfort for the infants during the repeated recordings. Sleep states similarly defined have been termed State 1 and State 2, respectively, and their close correspondence with the EEG-based quiet and active states has been discussed [16]. Table I shows the ages for which records were accepted for quantitative analysis. One episode each of quiet and of active sleep was selected from the accepted records. The quiet sleep selected was the first one of at least 10 minutes duration, while the active period selected was the first episode which was both preceded and followed by quiet sleep. Selection of the first episodes of clearly identifiable quiet and active sleep periods was motivated by previously reported differences among different epochs of the same state [l] . In view of the uniformity of quiet sleep during a single episode, a 3-minute record of the selected episode not containing any sighs was played into a PDP-12 computer for quantitative analysis. For active sleep, the entire episode or its beginning 30-minute segment, whichever was shorter, was subjected to computer analysis. For both quiet and active sleep the following quantities were determined: mean pulse interval, RR; standard deviation of the pulse interval, SRR; and coefficient of variation of pulse interval, SRR/RR. For quiet sleep, the coefficient of variation of pulse interval divided by the mean duration of the respiratory cycle was also computed. The pulse intervals were measured with a l-msec resolution. In order to determine how much of the variation in pulse interval could be attributed to respiration, a normalized correlation function [14] between the pulse interval and respiration was also computed using the formula
c_ x(t) y(t + T) dt r(7) =
where x(t) and y(t) represent the respiratory waveform and a pulse interval function, respectively. The continuous pulse interval function was constructed using linear interpolation between the values of pulse interval obtained at the occurrence of each QRS complex. For each record rmax was computed by replacing integration by finite summation, and by choosing the shift r which maximized r(7). The maximum correlation coefficient ranges between -1 and 1, being 0 if the waveshapes do not correlate, and 1 if they are identical after the appropriate tie shift.
149
The results were analyzed by dividing age into 2-, 4- or 8-week intervals, and then averaging the obtained variables for all infants with an accepted recording in that interval. If an infant had more than one recording in an interval, only a single (averaged for that infant) value was entered into the computation of the overall average. Quiet-sleep variables were derived for the entire 12-month period using all 21 subjects, active-sleep variables were computed for the first 6 months using the first 10 full-term and all 8 preterm infants. RESULTS
Figure 1 shows a typical recording of quiet sleep obtained from a full-term infant at a postnatal age of 8 weeks. Respiration is uniform except for occasional sighs [Zl] , occurring at 4-6-minute intervals, Heart rate is also regular, excluding the slight changes associated with the sigh. Figure 2 illustrates quiet sleep for the same infant at a postnatal age of 44 weeks. Apart from the sigh, respiration is again uniform, but now the pulse interval shows pronounced, regular variations in synchrony with respiration. Figure 3 and Table II show the average quiet-sleep pulse interval as a function of postnatal age for both full-term and preterm infants. The pulse interval decreases (heart rate increases) during the first weeks of life, attaining a minimum at 4-6 weeks in full-term and at 8-10 weeks in preterm infants. The minimum is 445 msec for full-term and 409 msec for preterm infants, the difference being statistically significant (P < 0.001, Student’s t-test). A difference in the average pulse interval persists throughout the ensuing rapid increase in pulse interval until at age of about 26 weeks, but by 36 weeks the pulse interval for the two groups becomes the same. Figure 4 shows the standard deviation of pulse interval as a function of %I
go-
-667 MSFC
IZO-
-500
180-
-333
1 MIN Fig. 1. Quiet sleep in &week full-term infant (top: pulse interval; bottom: respiration). Uniform respiration is interrupted by a sigh. Respiratory variations in pulse interval are small.
150 go-
-667 MSEC
B/M 120-
Loo
180-
-333
1 MIN
Fig. 2. Quiet sleep in 44week full-term infant (top: pulse interval; bottom: respiration). Pronounced respiratory variations in pulse interval are evident.
1i so -
600
MEAN PULSE INTERVAL IMSEC)
550
450. F U L L TERM 0 PRE7ERM 400-
POSTNATAL
AGE
(W
350 0
10
20
30
40
E E K S)
, 50
Fig. 3. Mean pulse interval during quiet sleep as a function of postnatal age. Vertical bars indicate standard error. Number of measurements at each age is given in Table II. Points on the vertical axis indicate quiet-sleep data reported for the newborn period by Theorell et al. [Zl] for normal full-term and by Watanabe et al. [27] for low birth weight (primarily preterm) infants.
postnatal age during quiet sleep. The pronounced increase of pulse interval fluctuations during most of the first year is apparent for both full-term and preterm infants. Since the variability is expected to increase with an increased pulse interval and respiratory duration, the variability normalized with respect to both of these variables was also explored. An increase in
151
variability during the lo-32-week period persists when the standard deviation divided by the mean pulse interval (i.e. the coefficient of variation) is considered. When the coefficient of pulse interval variations is further normalized by the duration of the respiratory cycle, an increase in variability with time still remains, but the difference between full-term and preterm infants disappears. In order to assess the component of pulse interval variations that can be attributed to respiration, the maximum correlation coefficient, rmax, was computed as a function of age. Figure 5 illustrates that for both full-term and preterm infants rmax decreases for several weeks after birth, and then starts a rapid rise at 8-12 weeks. Even though the onset of rise appears to be delayed in preterm infants, the range of rmax for full-term (0.22-0.49) and preterm infants (0.20-0.50) is remarkably similar. Since the larger the rmax the more the variations in pulse interval may be attributed to respiration, the graphs illustrate that the pulse interval variations observed at a postnatal age of about 10 weeks are related to respiration to a lesser extent than the variations at either younger (2 weeks) or older (>14 weeks) ages. It is conceivable that the developmental patterns for full-term and preterm infants become even more similar if the gestational age as opposed to the postnatal age is considered. Consequently, the same 3 variables that were plotted as a function of postnatal age in Figures 3-5 are also plotted in Figures 6-8 as a function of gestational (post-conceptional) age. While, in TABLE II Heart period in full-term and preterm infants during sleep as a function of age _-__ Active sleep Quiet sleep__-___-. Age (days) PretWm Preterm Full-term Full-term 7-20 21-34 35-48 49-62 63-76 77-90 91-104 105-118 119-132 140-167 168-195 196-223 224-251 252-279 280-307 308-335 336-365
452 f 445 f 445 + 457 r 4791. 489 t 501 t 503 k 624 i 523 + 540 t 539 * 542 + 521 t 546 * 537 + 564 f
S(11) Q(10) 6(12) 10 (10) S(13) 12 (10) Q(11) ll(12) 16 (10) Q(12) ll(11) 14 (10) 13 (9) 23 (10) 24 (6) 14 (8) 15 (7)
433 + 15 (7) 421 t 17 (7) 414 2 14 (7) 409 k 22 (6) 4 1 0 t 8(6) 427 t 16 (6) 4 4 0 t 6(7) 453 i 16 (7) 469 i 14 (8) 497 + 19 (6) 5121 9 (8) 511 i- 16 (8) 513 * 16 (8) 548* 5 (6) 549 ?r 16 (5) 559 I 12 (6) 577 142 (4)
431 f 426 f 441 + 447 t 461 f 454 f 469 t 483~ 484 k 498* 529 +
8 (8) 9 (7) 8 (8) 11 (7) 10 (8) 18 (7) 12 (8) 8(10) 11 (6) 7 (10) 12 (9)
411+ 398 f 422 + 401 * 420+ 426 + 445 r 444 f 462 + 489 f 483 i
12 (7) 8(6) 22 (4) 23 (6) 7(5) 18 (6) 14 (6) 12 (7) 13 (8) 16 (6) 6 (7)
The figure in parentheses indicates the number of infants contributing to the average. Standard errors are indicated.
162 .6
30
FULL TERM
PULSE INTERVAL VARIABILITY (MSECI
25 I
POSTNATAL AGE I
(WEEKS) I
hr 2cI-
I!
-FULL
TERM
.1
0 PRETERM ICI-
PRETERM
’ c
5L 0
10
POSTNATAL AGE I 1 20 30
(WEEKS) 1 1 . 40 SO
P O S T N A T A L AGE
10
20
30
(WEEKS) 40
Fig. 4. (left) Standard deviation of pulse interval during quiet sleep as a function of postnatal age. Vertical bars indicate error. Number of measurements at each age is given in Table II. Fig. 5. (right) The maximum correlation between respiration and pulse interval during quiet sleep as a function of postnatal age. Respiratory sinus arrhythmia accounts for a small fraction of pulse interval variations at age 8 weeks for full-term and 6-12 weeks for preterm infants.
general, the curves for full-term and preterm infants overlap over most of the age range considered, differences at early post-conceptional ages persist. The difference for pulse interval variability is not statistically significant at any age, but it is significant (P > 0.05, t-test) for both mean pulse interval and rmax at a post-conceptional age of 41. weeks. During active sleep the pulse interval is irregular, characterized by tachycardia during movements, and short, transient episodes of bradycardia occur-
I sa
153
L% + a
450-
5 w I
400-
??
3503 0
Full term Preterm
0
1
4 0
50
6 0
90
7 0
f
90
G E S T A T I O N A L A G E Lweeks)
Fig. 6. Mean pulse interval during quiet sleep as a function of post-conceptional (gestational) age. Vertical bars indicate standard error.
25P i m $2
LL 4 >
20-
0
3 0
I
4 0
I
5 0
I
60
GESTATIONAL AGE
Preterm
70
8 0
I
9 0
(weeks)
Fig. 7. Standard deviation of pulse interval during quiet sleep as a function of postcon. ceptional age. Vertical bars indicate standard error. ring apparently randomly, but often coinciding with short (2-5 set) apnea (Fig. 9). The average active sleep pulse interval for the first 6 months of postnatal age is given in Table II. The pulse interval of preterm infants is again consistently smaller than that of full-term infants throughout this age range.
154 .6-
I +
30
I
4 0
1
50
I
60
.
Full term
0
Preterm
,
70
1
00
r
90
GESTATIONAL AGE (weeks) Fig. 8. The maximum correlation between respiration and pulse interval during quiet sleep as a function of post-conceptional age.
go%I
-667 MSEC
120-
-500
180-
-333
1 MIN Fig. 9. Active sleep in 2%week full-term infant (top: pulse interval; bottom: respiration). Respiration and pulse interval are irregular.
The pulse interval in active sleep is lower than in quiet sleep for both fullterm and preterm infants, with the difference averaging 24.1 + 8.5 (SD) msec for full-term and 8.0 + 9.5 msec for preterm infants. The variability measures of active sleep pulse interval show more fluctuations than the variability measures observed for quiet sleep. The mean values of SRR and SRR/RR over the entire g-month interval are given in Table III, together with the corresponding values for quiet sleep. SRR and SRR/RR are approximately 3 times larger for active sleep than for quiet sleep. It
155 TABLE III Pulse interval variability (SD and coefficient of variation) averaged over the first 6 months for quiet and active sleep for full-term (N = 10) and preterm (N = 8) infants _ _ _ _ Full-term
-~ Active Quiet __-
Preterm
SRR (msec)
15.3 (6.7-23)
12.7 (7.7-18.9)
36.1 (29.3-50.2)
SRR/RR (W)
2.96 8.50 (1.9-3.7) (7.6-9.5)
2.67 (1.9-3.7)
8.03 (7.3--10.3)
__._~~~_
40.1 (35-47)
Quiet
Active
The range refers to a-week intervals.
should be noted that the active-sleep variability includes heart rate fluctuations that occur over a time period longer than 3 minutes. Such variations are generally absent in a single epoch of quiet sleep. DISCUSSION
To our knowledge, other investigators have not used recordings obtained in the homes to characterize autonomic development of infants. The use of home recordings is expected to reduce the influence of an artificial laboratory environment [2] on autonomic function, and it may thus give a more valid picture of autonomic function in the infant’s normal environment. Home monitoring also allows relatively frequent measurements, and thus facilitates the identification of developmental changes which are determined by postnatal age, post-conceptional age, or by a combination of both. On the other hand, the environment in the home cannot be as precisely controlled as in the laboratory. This could hinder comparisons between individual infants, but should not distort an assessment of normal development as long as grouped averages of longitudinally followed infants are considered. For both full-term and preterm infants the quiet sleep pulse interval first decreased and then increased (heart rate increased and then decreased) as a function of postnatal age, with preterm infants showing a higher heart rate until an age of about 7 months. These results are consistent with the observation that full-term newborns have a slower heart rate than premature infants at 38-40 weeks postconceptional age [3,19], and also agree with the observations of Harper et al. and Watanabe et al., who studied, respectively, full-term infants for 6 months [6], and low birth weight infants for 30-40 days [ 281. Comparing the plot of mean pulse interval against postnatal age (Fig. 3) with the plot against post-conceptional age (Fig. 6) leads to 3 conclusions. First, the initial decline in pulse interval is linked to postnatal and not to post-conceptional age, a finding in agreement with that of Watanabe et al.
166
[ 281, Second, the rapid increase in pulse interval after attaining its minimum is likewise strongly influenced by postnatal age, since the onset of rise in preterm infants in Figure 3 is not delayed by 7 weeks, the mean difference in post-conceptional age for the full-term and preterm infants. Correspondingly, the rapid rise in pulse interval in Figure 6 occurs sooner for preterm than for full-term infants. Third, a difference in mean pulse interval persists regardless of whether the pulse interval is considered as a function of postnatal or post-conceptional age; in fact, the pulse interval in preterm infants for several weeks after birth is lower than pulse intervals reported for fetuses [ll,lB] or term infants at any gestational age. The mean level of pulse interval in the early neonatal period is therefore dependent on the conceptional age at birth. The development of unnormalized variability in pulse interval (SRR) followed the general pattern of development of mean pulse interval as a function of postnatal age: minimum between 4 and 10 weeks, rapid rise in the remainder of the first half year, and a levelling off or more gradual rise in the second half year. A striking difference, however, is the considerably larger change in pulse interval variability than in the mean pulse interval during the first 6 months. By 30 weeks of postnatal age, the pulse interval increased approximately 30% from its minimum value, while the corresponding increase in SRR was approximately 100%. This indicates a rise in the pulse interval variability even when normalized with respect to the mean pulse interval. Harper et al. [6] noted a decrease in quiet sleep heart rate variability in full-term infants during the first month of life, but their reported essential constancy of heart rate variability in the following 5 months is in apparent disagreement with our findings. The contradiction is probably due to the different measures used to characterize the developmental patterns. We chose mean pulse interval to characterize chronotropic control, and used the standard deviation and the coefficient of variation of the pulse interval as descriptors of variability. Harper et al. utilized modal heart rate and the interquartile range of heart rate, respectively, for the same purpose. Our choice was made on the grounds that neural impulses modulate heart rate by determining the firing times of the pacemaker cells through altering the slope of the depolarizing voltage, rendering the pulse interval, rather than the heart rate, as the physiologically more meaningful parameter when considering variability in cardiac rhythm. Comparing on the basis of postnatal age, pulse interval variability is higher in full-term than in preterm infants for about 7 months after birth (Fig. 4). On the basis of post-conceptional age, however, a larger variability in fullterm infants is suggested only at a post-conceptional age of 40-44 weeks; the difference at older ages disappears (Fig. 7). It thus appears that, even though the pulse interval variability in newborn full-term infants is smaller than in preterm infants at a similar gestational age [19], pulse interval variability becomes primarily dependent on postconceptional age after a gestational of age of about 44 weeks.
157
Although the variability of pulse interval has at least two frequency components [24] , respiratory modulation appeared to dominate the variations in most of the recordings. The fraction of variability attributable to respiration was quantitatively assessed by rmax, which was found to be a function of age. The curves for rmax (Fig. 5) show a similar pattern of postnatal decrease followed by a sharp increase in both full-term and preterm infants, but the increase does not occur at the same postnatal ages in the two groups. Neither do the curves overlap at less than 43 weeks considering rmm as a function of post-conceptional age (Fig. 8). These results indicate that both types of developmental mechanisms are involved in determining rmm. It should be noted that rmax is only one of several possible measures of coupling between respiration and variations in pulse interval, and that a similar age-dependent pattern of respiratory modulation of pulse interval has been reported by Harper et al. in full-term infants monitored in the laboratory [ 71. Since the set of curves in Figure 4 indicating the degree of pulse interval variation and the set of curves in Figure 5 indicating the fraction of variation readily attributable to respiration both show a decline after birth and a rapid increase after 8-13 weeks, the time course of the magnitude of variations attributable to respiration will follow a similar course, and will have an even more prominent minimum. Changes in the respiratory frequency cannot account for these prominent age-related changes in respiratory pulse interval variations. The pattern of average pulse interval development in active sleep during the first 6 months was generally similar to the pattern found for quiet sleep. The average pulse interval during active sleep was slightly lower, and variability much higher than in quiet sleep. These results agree with those of others [6,28], but we attach limited significance to single overall measures of variability of pulse interval during active sleep. These variations are due not so much to respiration, but to spontaneous and apparently random increases and decreases, often during and after movements, respectively. We believe that a complete description of heart rate patterns during active sleep requires that one characterize the number and types of phasic changes, rather than lumping them into a single quantity when an overah measure of variability is considered. The observed developmental changes in pulse interval and its variability may reflect alterations in the intrinsic rate of the denervated heart, in the level of adrenergic control, and in the level of cholinergic control. The potential importance of the intrinsic heart rate in explaining the observed changes is suggested by recent experiments in which developing bradycardia in the awake neonatal lamb could not be related to changing autonomic tone [29], by findings that the rate of contraction of atrial tissue from perinatal and adult animals were different [26] , by reports of profound maturational changes in the conduction system of the neonatal heart [8], and by our own experiments, suggesting that in the sleeping kitten the pharmacologically blocked heart rate is maximum at about 14 days of age [4]. In addition to the intrinsic pulse interval, adrenergic effects, most likely
158
due to circulating catecholamines, may also contribute to the higher heart rate in prematures, and to the increasing heart rate in the first several weeks of life. The sympathetic nervous system, although rapidly developing, is relatively immature around the tie of birth [ 5,251, and circulating catecholamines seem to play a major role in the adrenergic control of the immature cardiovascular system [ 231. Since choline@ mechanisms are known to dominate in young adults, and appear also to dominate in the fetus at term [18] , it is expected that the parasympathetic component of the autonomic nervous system is at least partially responsible for the observed development patterns. Specifically, since the respiratory heart rate variations are primarily mediated by vagal efferents and the degree of the variations is strongly correlated with vagal tone [9], the substantial increase in respiratory sinus arrhythmia after an age of 4-8 weeks is likely to reflect rapidly increasing vagal cardiac control.
ACKNOWLEDGEMENTS
We want to thank the parents of our subjects for their cooperation and patience, and Mrs. Carolyn Alpert and Mrs. Nancy Williams for obtaining most of the recordings. The study was supported by Public Health Service Grant HD-06314.
REFERENCES 1 Ashton, R. and Connolly, K. (1971): The relation of respiration rate and heart rate to sleep states in the human newborn. Dev. Med. Child Neural., 13,180-.187. 2 Bernstein, P., Emde, R. and Campos, J. (1973): REM sleep in four-month infants under home and laboratory conditions. Psychosom. Med., 35, 322- 329. 3 Dreyfus-Brisac, C. (1970): Ontogenesis of sleep in human prematures after 32 weeks of conceptional age. Dev. Psychobiol., 3, 91-112. 4 Egbert, J.R. and Katona, P.G. (1980): Development of autonomic heart rate control in the kitten during sleep. Am. J. Physiol., in press. 5 Friedman, W.F., Pool, P.E., Jacobowitz, D., Seagren, S.C. and Braunwald, E. (1968): Sympathetic innervation of the developing rabbit heart. Circ. Res., 23, 26-32. 6 Harper, R.M., Hoppenbrouwers, T., Sterman, M.B., McGinty, D.J. and Hodgman, J. (1976): Polygraphic studies of normal infants during the first six months of life. I. Heart rate and variability as a function of state. Pediatr. Res., 10, 945-951. 7 Harper, R.M., Walter, D.O., Leake, B., Hoffman, H.J., Sieck, G.C., Sterman, M.B., Hoppenbrouwers, T. and Hodgman, J. (1978): Development of sinus arrhythmia during sleeping and waking states in normal infants. Sleep, 1, 33- 48. 8 James, T.N. (1970): Cardiac conduction system: fetal and postnatal development. Am. J. Cardiol., 25, 213- 226. 9 Katona, P.G. and Jih, F. (1975): Respiratory sinus arrhythmia: non-invasive measure of parasympathetic cardiac control. Am. J. Physiol., 39, Sol- 805. 10 Katona, P.G. and Egbert, J.R. (1978): Heart rate and respiratory rate differences between preterm and full-term infants during quiet sleep: possible implications for sudden infant death syndrome. Pediatrics, 62, 91--95.
159
11 Larks, S.D. and Larks, G.G. (1972). The course of the fetal heart rate during pregnancy. Cardiology, 57, 20% 211. 12 Lewis, M., Wilson, C.D., Ban, P. and Baumel, M.H. (1970): An exploratory study of resting cardiac rate and variability from the last trimester of prenatal life through the first year of postnatal life. Child Dev., 41, 799-811. 13 Lipton, E.L., Steinschneider, A. and Richmond, J.B. (1964): Autonomic function in the neonate. VIII. Cardiopulmonary observations. Pediatrics, 32, 212. 14 Matusek, M. (1973): Frequency and correlation analysis. In: Handbook of Electroencephalography and Clinical Neurophysiology, Vol. 5, Part A, p. 5A-42. Editor: A. Remond. Elsevier, Amsterdam. 15 Parmelee, A.H. and Stern, E. (1972): Development of states in infants. In: Sleep and the Maturing Nervous System, pp. 199-228. Editors: C. Clemente, D.P. Purpura and F.E. Mayer. Academic Press, New York. 16 Prechtl, H.F.R., Akiyama, Y., Zinkin, P. and Grant, D.K. (1968): Polygraphic studies of the full-term newborn: I. Technical aspects and quantitative analysis. In: Studies in Infancy, Clin. Dev. Med., Vol. 27, pp. l-- 21. Editors: R. MacKeith and M. Bax. Heinemann, London. 17 Salk, L., Grellong, B.A. and Dietrich, J. (1974): Sudden infant death. Normal cardiac habituation and poor autonomic control. N. Engl. J. Med., 291, 219-222. 18 Schifferli, P-Y. and Caldeyro-Bar&, R. (1973): Effects of atropine and beta-adrenergic drugs on heart rate of the human fetus. In: Fetal Pharmacology, pp. 259. 279. Editor: L. Boreus. Raven Press, New York. 19 Siassi, B., Hodgman, J.E., Cabal, L. and Han, E.H. (1979): Cardiac and respiratory activity in relation to gestation and sleep states in newborn infants. Pediatr. Res., 13,1163-1170. 20 Steinschneider, A. (1974): The concept of sleep apnea as related to SIDS. In: SIDS 1974, pp. 177-190. Editor: R.R. Robinson. Canadian Foundation for the Study of Infant Deaths. 21 Thach, B.T. and Taeusch, H.W. (1976): Sighing in newborn infants: role of inflationaugmenting reflex. J. Appl. Physiol., 41, 502-507. 22 Theorell, K., Prechtl, H.F.R. and Vos, J.E. (1974): A polygraphic study of normal and abnormal newborn infants. Neuropadiatrie, 5, 279-317. 23 Truccone, N.J. and Levine, R. (1973): Cardiovascular effects of propranolol in intact puppies and adult dogs. Pediatr. Res., 7,931-936. 24 Viilimiiki, LA., Rautaharju, P.M., Roy, S.B. and Scott, K.E. (1974): Heart rate patterns in healthy term and premature infants and in respiratory distress syndrome. Europ. J. Cardiol., 1, 411-419. 25 Vapaavouri, E.K., Shinebourne, E.A., Williams, R.L., Heymann, M.A. and Rudolph, A.M. (1973): Development of cardiovascular responses to autonomic blockade in intact fetal and neonatal lambs. Biol. Neon., 22, 177-188. 26 Vlk, J. and Vincensi, F.F. (1977): Functional autonomic innervation of mammalian cardiac pacemaker during the perinatal period. Biol. Neon., 31, 19-26. 27 Walsh, S.Z. (1964): Comparative study of electrocardiograms of healthy premature and full-term infants of similar weight. Am. Heart J.. 68, 183-192. 28 Watanabe, K., Iwase, K. and Hara, K. (1973): Heart rate variability during sleep and wakefulness in low-birthweight infants. Biol. Neon., 22, 87 -98. 29 Woods, J.R., Dondavino, A., Brinkman, III, C.R. and Aasali, N.S. (1977): Autonomic control of cardiovascular functions during neonatal development and in adult sheep. Circ. Res., 40, 401-407.
145
Maturation of cardiac control in full-term and preterm infants during sleep PETER G. KATONA, ALAN FRASZ and JOHN EGBERT
Schools of Medicine and Engineering, Case Western Reserve University, Cleveland, Ohio 44106, U.S.A.
Accepted for publication 26 February 1980
SUMMARY
ECG and respiration were recorded from 13 full-term and 8 preterm infants at 2-&week intervals during sleep at home in the first year of life. Average pulse interval in both quiet and active sleep was minimum at postnatal ages between 4 and 10 weeks, with the minimum during quiet sleep being significantly smaller for preterm (409 f 22 (SE) msec) than for full-term (445 1 9 msec) infants. The minimum average pulse interval of preterm infants was smaller than the pulse interval of full-term infants at any postconceptional age, and a smaller average pulse interval and smaller variations in pulse interval in preterm infants during quiet sleep persisted until a postnatal age of 7 months. The pulse interval variations attributable to respiration varied substantially with age. The results indicate that developmental changes in cardiac rate control are functions of both postnatal and postconceptional age, with the post-conceptional age at birth setting the mean level of pulse interval, a level which is then altered by development linked to postnatal age. sympathetic control; parasympathetic control; maturation; heart rate; respiratory sinus arrhythmia; sleep; infants
INTRODUCTION
Although the development of autonomic control mechanisms has been extensively investigated in the newborn period, only incomplete information is available about the continuing rapid development that occurs during the first year of life. Lipton and Steinschneider were among the first to draw attention to this development by describing the changes in heart rate response to an external stiiulus at ages 0, 2.5 and 5 months [13]. The need to better
146
understand this development has been recently underscored by suggestions that sudden infant death syndrome (SIDS) may be a manifestation of malfunctioning autonomic control mechanisms, and that defects in autonomic control mechanisms might be detected on the basis of respiratory and/or heart rate patterns [ 17,201 . It seems especially desirable to compare developmental patterns of fullterm and pretenn infants because SIDS is more common in preterm infants, and because the difference in both heart rate and respiratory rate between full-term and preterm infants is maximum at a postnatal age of about 2 months [lo], when SIDS is also most prevalent. The purpose of our work is to directly compare the longitudinal development of cardiac control in normal full-term and preterm infants during the first year of life using measurements obtained in the infants’ homes. The few previous reports dealing with maturational development of the cardio-respiratory system past the immediate neonatal period are generally concerned with either full-term or preterm infants, and the results are based on measurements performed in the laboratory or clinic [6,12,27,28]. METHODS Normal full-term and preterm infants were recruited on the basis of expected cooperation from their parents. The study was initiated for a total of 27 infants; data is presented for the 21 subjects who complied with the study protocol. The final study group consisted of 13 full-term infants with average birth weight of 3248 f 439 (SD) g, and 8 preterm infants with average gestational age and birth weight of 33.3 f 2.2 weeks and 1874 f 362 g, respectively. Table I shows the gestational age and birth weight of the subjects. Electrocardiogram and respiration were recorded in each infant’s home at approximately a-week intervals until age 5 months, and at 2-5-week intervals up to 12 months. Recordings of l-3 hours duration were obtained while the infant was asleep as judged behaviorally. The environment was generally quiet and the mean temperature of the rooms of full-term infants (23.7 ? 0.9 (SD) “C) was essentially the same as the temperature of rooms of preterm infants (24.1 * 0.9 “C). ECG and respiration were obtained using a Beckman VSM-100B vital signs monitor. The monitor, which was designed to indicate average heart rate and to derive the respiratory waveform from changes in impedance across the chest, was modified to also provide a beat-by-beat indication of the pulse intervals. The two active electrodes were placed on the mid-axillary lines in the 3rd-5th interspace. The data were recorded on a Wolff 4channel cassette tape-recorder, and displayed on a 2-channel chart recorder for immediate verification. The recordings were first scored for quiet and active sleep. These two states, which are precursors of the NREM and REM states in adults, can be consistently identified after a gestational age of 36 weeks [ 15 ] ; only a few
i
x x x x x xxxxx x x x xxx x x x x x x x x x xx x x x xx xx x x x xx xx xx PXXXX x x x x x x x x I xxx xx x x x xx xxx* x
20
2360 1518 1717 2280 2020 1900 1290 1910
34 32 32 36 34 34 29 35
Pl P2 P3 P4 P5 P6 P7 P8
10 xx x x x x x x x xxx xxx x x x x x x x x x x x x xxx x xx xx x x x x xxx x xxxxx xxxxx x x x *xxx* x x x x x x x x * x xx P xxx xx x xx x x x I xxxxx x x x
40 39 40 40 41 41 40 39 40 39 40 40 40
Fl F2 F3 F4 F5 Fti F7 F8 F9 FlO Fll Fl2 F13
1
Age (in wk) at which recordings were taken
2722 2667 3667 2976 3585 3972 3570 3788 2778 3175 3405 3036 2892
Gestational age (wk) Birth weight (g)
Infant
Full-term (Fl-F13) and preterm (Pl-P8) subjects studied ~~
TABLE I
x x
x
x x
x x
x
x x x x x x xx x x x x x x x x xx x
x
XXX
x xx x x xx x x
x
xx x x
x x x x x x
..___ xx
x
x x
x
x
x
x
x
x
x x x x x x x
x
x
x x x
148
recordings in some preterm infants were obtained before this age. The identification of states is traditionally based on recordings of EEG, EOG and muscle potentials, but polygraphic studies have shown that during quiet sleep respiration is regular and body movements are absent except for occasional sighs, while during active sleep respiration is irregular and there are frequent body movements [15,16,22]. Classifying sleep states as quiet or active on the basis of the regularity or irregularity of respiration, and the absence or presence of motion artifact in the impedance tracing, respectively, minimized the disturbance and discomfort for the infants during the repeated recordings. Sleep states similarly defined have been termed State 1 and State 2, respectively, and their close correspondence with the EEG-based quiet and active states has been discussed [16]. Table I shows the ages for which records were accepted for quantitative analysis. One episode each of quiet and of active sleep was selected from the accepted records. The quiet sleep selected was the first one of at least 10 minutes duration, while the active period selected was the first episode which was both preceded and followed by quiet sleep. Selection of the first episodes of clearly identifiable quiet and active sleep periods was motivated by previously reported differences among different epochs of the same state [l] . In view of the uniformity of quiet sleep during a single episode, a 3-minute record of the selected episode not containing any sighs was played into a PDP-12 computer for quantitative analysis. For active sleep, the entire episode or its beginning 30-minute segment, whichever was shorter, was subjected to computer analysis. For both quiet and active sleep the following quantities were determined: mean pulse interval, RR; standard deviation of the pulse interval, SRR; and coefficient of variation of pulse interval, SRR/RR. For quiet sleep, the coefficient of variation of pulse interval divided by the mean duration of the respiratory cycle was also computed. The pulse intervals were measured with a l-msec resolution. In order to determine how much of the variation in pulse interval could be attributed to respiration, a normalized correlation function [14] between the pulse interval and respiration was also computed using the formula
c_ x(t) y(t + T) dt r(7) =
where x(t) and y(t) represent the respiratory waveform and a pulse interval function, respectively. The continuous pulse interval function was constructed using linear interpolation between the values of pulse interval obtained at the occurrence of each QRS complex. For each record rmax was computed by replacing integration by finite summation, and by choosing the shift r which maximized r(7). The maximum correlation coefficient ranges between -1 and 1, being 0 if the waveshapes do not correlate, and 1 if they are identical after the appropriate tie shift.
149
The results were analyzed by dividing age into 2-, 4- or 8-week intervals, and then averaging the obtained variables for all infants with an accepted recording in that interval. If an infant had more than one recording in an interval, only a single (averaged for that infant) value was entered into the computation of the overall average. Quiet-sleep variables were derived for the entire 12-month period using all 21 subjects, active-sleep variables were computed for the first 6 months using the first 10 full-term and all 8 preterm infants. RESULTS
Figure 1 shows a typical recording of quiet sleep obtained from a full-term infant at a postnatal age of 8 weeks. Respiration is uniform except for occasional sighs [Zl] , occurring at 4-6-minute intervals, Heart rate is also regular, excluding the slight changes associated with the sigh. Figure 2 illustrates quiet sleep for the same infant at a postnatal age of 44 weeks. Apart from the sigh, respiration is again uniform, but now the pulse interval shows pronounced, regular variations in synchrony with respiration. Figure 3 and Table II show the average quiet-sleep pulse interval as a function of postnatal age for both full-term and preterm infants. The pulse interval decreases (heart rate increases) during the first weeks of life, attaining a minimum at 4-6 weeks in full-term and at 8-10 weeks in preterm infants. The minimum is 445 msec for full-term and 409 msec for preterm infants, the difference being statistically significant (P < 0.001, Student’s t-test). A difference in the average pulse interval persists throughout the ensuing rapid increase in pulse interval until at age of about 26 weeks, but by 36 weeks the pulse interval for the two groups becomes the same. Figure 4 shows the standard deviation of pulse interval as a function of %I
go-
-667 MSFC
IZO-
-500
180-
-333
1 MIN Fig. 1. Quiet sleep in &week full-term infant (top: pulse interval; bottom: respiration). Uniform respiration is interrupted by a sigh. Respiratory variations in pulse interval are small.
150 go-
-667 MSEC
B/M 120-
Loo
180-
-333
1 MIN
Fig. 2. Quiet sleep in 44week full-term infant (top: pulse interval; bottom: respiration). Pronounced respiratory variations in pulse interval are evident.
1i so -
600
MEAN PULSE INTERVAL IMSEC)
550
450. F U L L TERM 0 PRE7ERM 400-
POSTNATAL
AGE
(W
350 0
10
20
30
40
E E K S)
, 50
Fig. 3. Mean pulse interval during quiet sleep as a function of postnatal age. Vertical bars indicate standard error. Number of measurements at each age is given in Table II. Points on the vertical axis indicate quiet-sleep data reported for the newborn period by Theorell et al. [Zl] for normal full-term and by Watanabe et al. [27] for low birth weight (primarily preterm) infants.
postnatal age during quiet sleep. The pronounced increase of pulse interval fluctuations during most of the first year is apparent for both full-term and preterm infants. Since the variability is expected to increase with an increased pulse interval and respiratory duration, the variability normalized with respect to both of these variables was also explored. An increase in
151
variability during the lo-32-week period persists when the standard deviation divided by the mean pulse interval (i.e. the coefficient of variation) is considered. When the coefficient of pulse interval variations is further normalized by the duration of the respiratory cycle, an increase in variability with time still remains, but the difference between full-term and preterm infants disappears. In order to assess the component of pulse interval variations that can be attributed to respiration, the maximum correlation coefficient, rmax, was computed as a function of age. Figure 5 illustrates that for both full-term and preterm infants rmax decreases for several weeks after birth, and then starts a rapid rise at 8-12 weeks. Even though the onset of rise appears to be delayed in preterm infants, the range of rmax for full-term (0.22-0.49) and preterm infants (0.20-0.50) is remarkably similar. Since the larger the rmax the more the variations in pulse interval may be attributed to respiration, the graphs illustrate that the pulse interval variations observed at a postnatal age of about 10 weeks are related to respiration to a lesser extent than the variations at either younger (2 weeks) or older (>14 weeks) ages. It is conceivable that the developmental patterns for full-term and preterm infants become even more similar if the gestational age as opposed to the postnatal age is considered. Consequently, the same 3 variables that were plotted as a function of postnatal age in Figures 3-5 are also plotted in Figures 6-8 as a function of gestational (post-conceptional) age. While, in TABLE II Heart period in full-term and preterm infants during sleep as a function of age _-__ Active sleep Quiet sleep__-___-. Age (days) PretWm Preterm Full-term Full-term 7-20 21-34 35-48 49-62 63-76 77-90 91-104 105-118 119-132 140-167 168-195 196-223 224-251 252-279 280-307 308-335 336-365
452 f 445 f 445 + 457 r 4791. 489 t 501 t 503 k 624 i 523 + 540 t 539 * 542 + 521 t 546 * 537 + 564 f
S(11) Q(10) 6(12) 10 (10) S(13) 12 (10) Q(11) ll(12) 16 (10) Q(12) ll(11) 14 (10) 13 (9) 23 (10) 24 (6) 14 (8) 15 (7)
433 + 15 (7) 421 t 17 (7) 414 2 14 (7) 409 k 22 (6) 4 1 0 t 8(6) 427 t 16 (6) 4 4 0 t 6(7) 453 i 16 (7) 469 i 14 (8) 497 + 19 (6) 5121 9 (8) 511 i- 16 (8) 513 * 16 (8) 548* 5 (6) 549 ?r 16 (5) 559 I 12 (6) 577 142 (4)
431 f 426 f 441 + 447 t 461 f 454 f 469 t 483~ 484 k 498* 529 +
8 (8) 9 (7) 8 (8) 11 (7) 10 (8) 18 (7) 12 (8) 8(10) 11 (6) 7 (10) 12 (9)
411+ 398 f 422 + 401 * 420+ 426 + 445 r 444 f 462 + 489 f 483 i
12 (7) 8(6) 22 (4) 23 (6) 7(5) 18 (6) 14 (6) 12 (7) 13 (8) 16 (6) 6 (7)
The figure in parentheses indicates the number of infants contributing to the average. Standard errors are indicated.
162 .6
30
FULL TERM
PULSE INTERVAL VARIABILITY (MSECI
25 I
POSTNATAL AGE I
(WEEKS) I
hr 2cI-
I!
-FULL
TERM
.1
0 PRETERM ICI-
PRETERM
’ c
5L 0
10
POSTNATAL AGE I 1 20 30
(WEEKS) 1 1 . 40 SO
P O S T N A T A L AGE
10
20
30
(WEEKS) 40
Fig. 4. (left) Standard deviation of pulse interval during quiet sleep as a function of postnatal age. Vertical bars indicate error. Number of measurements at each age is given in Table II. Fig. 5. (right) The maximum correlation between respiration and pulse interval during quiet sleep as a function of postnatal age. Respiratory sinus arrhythmia accounts for a small fraction of pulse interval variations at age 8 weeks for full-term and 6-12 weeks for preterm infants.
general, the curves for full-term and preterm infants overlap over most of the age range considered, differences at early post-conceptional ages persist. The difference for pulse interval variability is not statistically significant at any age, but it is significant (P > 0.05, t-test) for both mean pulse interval and rmax at a post-conceptional age of 41. weeks. During active sleep the pulse interval is irregular, characterized by tachycardia during movements, and short, transient episodes of bradycardia occur-
I sa
153
L% + a
450-
5 w I
400-
??
3503 0
Full term Preterm
0
1
4 0
50
6 0
90
7 0
f
90
G E S T A T I O N A L A G E Lweeks)
Fig. 6. Mean pulse interval during quiet sleep as a function of post-conceptional (gestational) age. Vertical bars indicate standard error.
25P i m $2
LL 4 >
20-
0
3 0
I
4 0
I
5 0
I
60
GESTATIONAL AGE
Preterm
70
8 0
I
9 0
(weeks)
Fig. 7. Standard deviation of pulse interval during quiet sleep as a function of postcon. ceptional age. Vertical bars indicate standard error. ring apparently randomly, but often coinciding with short (2-5 set) apnea (Fig. 9). The average active sleep pulse interval for the first 6 months of postnatal age is given in Table II. The pulse interval of preterm infants is again consistently smaller than that of full-term infants throughout this age range.
154 .6-
I +
30
I
4 0
1
50
I
60
.
Full term
0
Preterm
,
70
1
00
r
90
GESTATIONAL AGE (weeks) Fig. 8. The maximum correlation between respiration and pulse interval during quiet sleep as a function of post-conceptional age.
go%I
-667 MSEC
120-
-500
180-
-333
1 MIN Fig. 9. Active sleep in 2%week full-term infant (top: pulse interval; bottom: respiration). Respiration and pulse interval are irregular.
The pulse interval in active sleep is lower than in quiet sleep for both fullterm and preterm infants, with the difference averaging 24.1 + 8.5 (SD) msec for full-term and 8.0 + 9.5 msec for preterm infants. The variability measures of active sleep pulse interval show more fluctuations than the variability measures observed for quiet sleep. The mean values of SRR and SRR/RR over the entire g-month interval are given in Table III, together with the corresponding values for quiet sleep. SRR and SRR/RR are approximately 3 times larger for active sleep than for quiet sleep. It
155 TABLE III Pulse interval variability (SD and coefficient of variation) averaged over the first 6 months for quiet and active sleep for full-term (N = 10) and preterm (N = 8) infants _ _ _ _ Full-term
-~ Active Quiet __-
Preterm
SRR (msec)
15.3 (6.7-23)
12.7 (7.7-18.9)
36.1 (29.3-50.2)
SRR/RR (W)
2.96 8.50 (1.9-3.7) (7.6-9.5)
2.67 (1.9-3.7)
8.03 (7.3--10.3)
__._~~~_
40.1 (35-47)
Quiet
Active
The range refers to a-week intervals.
should be noted that the active-sleep variability includes heart rate fluctuations that occur over a time period longer than 3 minutes. Such variations are generally absent in a single epoch of quiet sleep. DISCUSSION
To our knowledge, other investigators have not used recordings obtained in the homes to characterize autonomic development of infants. The use of home recordings is expected to reduce the influence of an artificial laboratory environment [2] on autonomic function, and it may thus give a more valid picture of autonomic function in the infant’s normal environment. Home monitoring also allows relatively frequent measurements, and thus facilitates the identification of developmental changes which are determined by postnatal age, post-conceptional age, or by a combination of both. On the other hand, the environment in the home cannot be as precisely controlled as in the laboratory. This could hinder comparisons between individual infants, but should not distort an assessment of normal development as long as grouped averages of longitudinally followed infants are considered. For both full-term and preterm infants the quiet sleep pulse interval first decreased and then increased (heart rate increased and then decreased) as a function of postnatal age, with preterm infants showing a higher heart rate until an age of about 7 months. These results are consistent with the observation that full-term newborns have a slower heart rate than premature infants at 38-40 weeks postconceptional age [3,19], and also agree with the observations of Harper et al. and Watanabe et al., who studied, respectively, full-term infants for 6 months [6], and low birth weight infants for 30-40 days [ 281. Comparing the plot of mean pulse interval against postnatal age (Fig. 3) with the plot against post-conceptional age (Fig. 6) leads to 3 conclusions. First, the initial decline in pulse interval is linked to postnatal and not to post-conceptional age, a finding in agreement with that of Watanabe et al.
166
[ 281, Second, the rapid increase in pulse interval after attaining its minimum is likewise strongly influenced by postnatal age, since the onset of rise in preterm infants in Figure 3 is not delayed by 7 weeks, the mean difference in post-conceptional age for the full-term and preterm infants. Correspondingly, the rapid rise in pulse interval in Figure 6 occurs sooner for preterm than for full-term infants. Third, a difference in mean pulse interval persists regardless of whether the pulse interval is considered as a function of postnatal or post-conceptional age; in fact, the pulse interval in preterm infants for several weeks after birth is lower than pulse intervals reported for fetuses [ll,lB] or term infants at any gestational age. The mean level of pulse interval in the early neonatal period is therefore dependent on the conceptional age at birth. The development of unnormalized variability in pulse interval (SRR) followed the general pattern of development of mean pulse interval as a function of postnatal age: minimum between 4 and 10 weeks, rapid rise in the remainder of the first half year, and a levelling off or more gradual rise in the second half year. A striking difference, however, is the considerably larger change in pulse interval variability than in the mean pulse interval during the first 6 months. By 30 weeks of postnatal age, the pulse interval increased approximately 30% from its minimum value, while the corresponding increase in SRR was approximately 100%. This indicates a rise in the pulse interval variability even when normalized with respect to the mean pulse interval. Harper et al. [6] noted a decrease in quiet sleep heart rate variability in full-term infants during the first month of life, but their reported essential constancy of heart rate variability in the following 5 months is in apparent disagreement with our findings. The contradiction is probably due to the different measures used to characterize the developmental patterns. We chose mean pulse interval to characterize chronotropic control, and used the standard deviation and the coefficient of variation of the pulse interval as descriptors of variability. Harper et al. utilized modal heart rate and the interquartile range of heart rate, respectively, for the same purpose. Our choice was made on the grounds that neural impulses modulate heart rate by determining the firing times of the pacemaker cells through altering the slope of the depolarizing voltage, rendering the pulse interval, rather than the heart rate, as the physiologically more meaningful parameter when considering variability in cardiac rhythm. Comparing on the basis of postnatal age, pulse interval variability is higher in full-term than in preterm infants for about 7 months after birth (Fig. 4). On the basis of post-conceptional age, however, a larger variability in fullterm infants is suggested only at a post-conceptional age of 40-44 weeks; the difference at older ages disappears (Fig. 7). It thus appears that, even though the pulse interval variability in newborn full-term infants is smaller than in preterm infants at a similar gestational age [19], pulse interval variability becomes primarily dependent on postconceptional age after a gestational of age of about 44 weeks.
157
Although the variability of pulse interval has at least two frequency components [24] , respiratory modulation appeared to dominate the variations in most of the recordings. The fraction of variability attributable to respiration was quantitatively assessed by rmax, which was found to be a function of age. The curves for rmax (Fig. 5) show a similar pattern of postnatal decrease followed by a sharp increase in both full-term and preterm infants, but the increase does not occur at the same postnatal ages in the two groups. Neither do the curves overlap at less than 43 weeks considering rmm as a function of post-conceptional age (Fig. 8). These results indicate that both types of developmental mechanisms are involved in determining rmm. It should be noted that rmax is only one of several possible measures of coupling between respiration and variations in pulse interval, and that a similar age-dependent pattern of respiratory modulation of pulse interval has been reported by Harper et al. in full-term infants monitored in the laboratory [ 71. Since the set of curves in Figure 4 indicating the degree of pulse interval variation and the set of curves in Figure 5 indicating the fraction of variation readily attributable to respiration both show a decline after birth and a rapid increase after 8-13 weeks, the time course of the magnitude of variations attributable to respiration will follow a similar course, and will have an even more prominent minimum. Changes in the respiratory frequency cannot account for these prominent age-related changes in respiratory pulse interval variations. The pattern of average pulse interval development in active sleep during the first 6 months was generally similar to the pattern found for quiet sleep. The average pulse interval during active sleep was slightly lower, and variability much higher than in quiet sleep. These results agree with those of others [6,28], but we attach limited significance to single overall measures of variability of pulse interval during active sleep. These variations are due not so much to respiration, but to spontaneous and apparently random increases and decreases, often during and after movements, respectively. We believe that a complete description of heart rate patterns during active sleep requires that one characterize the number and types of phasic changes, rather than lumping them into a single quantity when an overah measure of variability is considered. The observed developmental changes in pulse interval and its variability may reflect alterations in the intrinsic rate of the denervated heart, in the level of adrenergic control, and in the level of cholinergic control. The potential importance of the intrinsic heart rate in explaining the observed changes is suggested by recent experiments in which developing bradycardia in the awake neonatal lamb could not be related to changing autonomic tone [29], by findings that the rate of contraction of atrial tissue from perinatal and adult animals were different [26] , by reports of profound maturational changes in the conduction system of the neonatal heart [8], and by our own experiments, suggesting that in the sleeping kitten the pharmacologically blocked heart rate is maximum at about 14 days of age [4]. In addition to the intrinsic pulse interval, adrenergic effects, most likely
158
due to circulating catecholamines, may also contribute to the higher heart rate in prematures, and to the increasing heart rate in the first several weeks of life. The sympathetic nervous system, although rapidly developing, is relatively immature around the tie of birth [ 5,251, and circulating catecholamines seem to play a major role in the adrenergic control of the immature cardiovascular system [ 231. Since choline@ mechanisms are known to dominate in young adults, and appear also to dominate in the fetus at term [18] , it is expected that the parasympathetic component of the autonomic nervous system is at least partially responsible for the observed development patterns. Specifically, since the respiratory heart rate variations are primarily mediated by vagal efferents and the degree of the variations is strongly correlated with vagal tone [9], the substantial increase in respiratory sinus arrhythmia after an age of 4-8 weeks is likely to reflect rapidly increasing vagal cardiac control.
ACKNOWLEDGEMENTS
We want to thank the parents of our subjects for their cooperation and patience, and Mrs. Carolyn Alpert and Mrs. Nancy Williams for obtaining most of the recordings. The study was supported by Public Health Service Grant HD-06314.
REFERENCES 1 Ashton, R. and Connolly, K. (1971): The relation of respiration rate and heart rate to sleep states in the human newborn. Dev. Med. Child Neural., 13,180-.187. 2 Bernstein, P., Emde, R. and Campos, J. (1973): REM sleep in four-month infants under home and laboratory conditions. Psychosom. Med., 35, 322- 329. 3 Dreyfus-Brisac, C. (1970): Ontogenesis of sleep in human prematures after 32 weeks of conceptional age. Dev. Psychobiol., 3, 91-112. 4 Egbert, J.R. and Katona, P.G. (1980): Development of autonomic heart rate control in the kitten during sleep. Am. J. Physiol., in press. 5 Friedman, W.F., Pool, P.E., Jacobowitz, D., Seagren, S.C. and Braunwald, E. (1968): Sympathetic innervation of the developing rabbit heart. Circ. Res., 23, 26-32. 6 Harper, R.M., Hoppenbrouwers, T., Sterman, M.B., McGinty, D.J. and Hodgman, J. (1976): Polygraphic studies of normal infants during the first six months of life. I. Heart rate and variability as a function of state. Pediatr. Res., 10, 945-951. 7 Harper, R.M., Walter, D.O., Leake, B., Hoffman, H.J., Sieck, G.C., Sterman, M.B., Hoppenbrouwers, T. and Hodgman, J. (1978): Development of sinus arrhythmia during sleeping and waking states in normal infants. Sleep, 1, 33- 48. 8 James, T.N. (1970): Cardiac conduction system: fetal and postnatal development. Am. J. Cardiol., 25, 213- 226. 9 Katona, P.G. and Jih, F. (1975): Respiratory sinus arrhythmia: non-invasive measure of parasympathetic cardiac control. Am. J. Physiol., 39, Sol- 805. 10 Katona, P.G. and Egbert, J.R. (1978): Heart rate and respiratory rate differences between preterm and full-term infants during quiet sleep: possible implications for sudden infant death syndrome. Pediatrics, 62, 91--95.
159
11 Larks, S.D. and Larks, G.G. (1972). The course of the fetal heart rate during pregnancy. Cardiology, 57, 20% 211. 12 Lewis, M., Wilson, C.D., Ban, P. and Baumel, M.H. (1970): An exploratory study of resting cardiac rate and variability from the last trimester of prenatal life through the first year of postnatal life. Child Dev., 41, 799-811. 13 Lipton, E.L., Steinschneider, A. and Richmond, J.B. (1964): Autonomic function in the neonate. VIII. Cardiopulmonary observations. Pediatrics, 32, 212. 14 Matusek, M. (1973): Frequency and correlation analysis. In: Handbook of Electroencephalography and Clinical Neurophysiology, Vol. 5, Part A, p. 5A-42. Editor: A. Remond. Elsevier, Amsterdam. 15 Parmelee, A.H. and Stern, E. (1972): Development of states in infants. In: Sleep and the Maturing Nervous System, pp. 199-228. Editors: C. Clemente, D.P. Purpura and F.E. Mayer. Academic Press, New York. 16 Prechtl, H.F.R., Akiyama, Y., Zinkin, P. and Grant, D.K. (1968): Polygraphic studies of the full-term newborn: I. Technical aspects and quantitative analysis. In: Studies in Infancy, Clin. Dev. Med., Vol. 27, pp. l-- 21. Editors: R. MacKeith and M. Bax. Heinemann, London. 17 Salk, L., Grellong, B.A. and Dietrich, J. (1974): Sudden infant death. Normal cardiac habituation and poor autonomic control. N. Engl. J. Med., 291, 219-222. 18 Schifferli, P-Y. and Caldeyro-Bar&, R. (1973): Effects of atropine and beta-adrenergic drugs on heart rate of the human fetus. In: Fetal Pharmacology, pp. 259. 279. Editor: L. Boreus. Raven Press, New York. 19 Siassi, B., Hodgman, J.E., Cabal, L. and Han, E.H. (1979): Cardiac and respiratory activity in relation to gestation and sleep states in newborn infants. Pediatr. Res., 13,1163-1170. 20 Steinschneider, A. (1974): The concept of sleep apnea as related to SIDS. In: SIDS 1974, pp. 177-190. Editor: R.R. Robinson. Canadian Foundation for the Study of Infant Deaths. 21 Thach, B.T. and Taeusch, H.W. (1976): Sighing in newborn infants: role of inflationaugmenting reflex. J. Appl. Physiol., 41, 502-507. 22 Theorell, K., Prechtl, H.F.R. and Vos, J.E. (1974): A polygraphic study of normal and abnormal newborn infants. Neuropadiatrie, 5, 279-317. 23 Truccone, N.J. and Levine, R. (1973): Cardiovascular effects of propranolol in intact puppies and adult dogs. Pediatr. Res., 7,931-936. 24 Viilimiiki, LA., Rautaharju, P.M., Roy, S.B. and Scott, K.E. (1974): Heart rate patterns in healthy term and premature infants and in respiratory distress syndrome. Europ. J. Cardiol., 1, 411-419. 25 Vapaavouri, E.K., Shinebourne, E.A., Williams, R.L., Heymann, M.A. and Rudolph, A.M. (1973): Development of cardiovascular responses to autonomic blockade in intact fetal and neonatal lambs. Biol. Neon., 22, 177-188. 26 Vlk, J. and Vincensi, F.F. (1977): Functional autonomic innervation of mammalian cardiac pacemaker during the perinatal period. Biol. Neon., 31, 19-26. 27 Walsh, S.Z. (1964): Comparative study of electrocardiograms of healthy premature and full-term infants of similar weight. Am. Heart J.. 68, 183-192. 28 Watanabe, K., Iwase, K. and Hara, K. (1973): Heart rate variability during sleep and wakefulness in low-birthweight infants. Biol. Neon., 22, 87 -98. 29 Woods, J.R., Dondavino, A., Brinkman, III, C.R. and Aasali, N.S. (1977): Autonomic control of cardiovascular functions during neonatal development and in adult sheep. Circ. Res., 40, 401-407.