Effects of Sleep Stage and Age on Short-term Heart Rate Variability During Sleep in Healthy Infants and Children

Effects of Sleep Stage and Age on Short-term Heart Rate Variability During Sleep in Healthy Infants and Children* Maria Pia Villa, MD; Giovanni Calcagnini, PhD; Jacopo Pagani, MD; Barbara Paggi, MD; Francesca Massa, MD; and Roberto Ronchetti, MD

Study design: Power spectrum analysis of heart rate variability (HRV) is a noninvasive technique that provides a quantitative assessment of cardiovascular neural control. Using this technique, we studied the autonomic nervous system changes induced by sleep in 14 healthy subjects: 7 infants (mean age, 9.40 ⴞ 2.32 months) and 7 children (mean age, 8.93 ⴞ 0.65 years) during a standard all-night polysomnographic recording. Our primary aim was to assess the effect of sleep stage and age on short-term HRV during sleep in healthy infants and children. Power spectral density was estimated by autoregressive modeling over 250 consecutive R-R intervals. In this study, we mainly considered two spectral components: the high-frequency (HF) component (0.15 to 0.40 Hz), which reflects parasympathetic cardiovascular modulation; and the low-frequency (LF) component (0.04 to 0.15 Hz), generally considered due to both parasympathetic and sympathetic modulation. Results: Heart rate was higher (p < 0.01 in all sleep stages) and total power lower (p < 0.02) in infants than in children. HF power was higher in children than in infants (p < 0.05). In infants and children, the ratio between LF and HF powers changed with the various sleep stages (p < 0.02 in infants; p < 0.01 in children): it decreased during deep sleep and increased during rapid eye movement sleep. However, it was invariably lower in children than in infants. Conclusion: These findings show that the sleep stage and age both significantly influence short-term HRV during sleep in healthy infants and children. Hence, to provide unbiased results, HRV studies investigating the effects of age on autonomic nervous system activity should segment sleep into the five stages. In addition, despite a relatively small study sample, our data confirm greater parasympathetic control during sleep in children than in infants. (CHEST 2000; 117:460 – 466) Key words: autonomic nervous system; power spectrum analysis; sleep stages Abbreviations: ANOVA ⫽ analysis of variance; HF ⫽ high-frequency; HRV ⫽ heart rate variability; LF ⫽ lowfrequency; NS ⫽ not significant; REM ⫽ rapid eye movement (active sleep); Sao2 ⫽ arterial oxygen saturation; TP ⫽ total power; VLF ⫽ very low-frequency

in cardiovascular functions during sleep C hanges reflect changes in autonomic nervous system dominance. Sympathetic nerve activity, BP, and heart rate are lower in nonrapid eye movement sleep than in wakefulness. In rapid eye movement (REM) sleep, sympathetic nerve activity increases, reaching values greater than those measured in wakefulness.1 These data suggest that during REM sleep, sympathetic control of cardiovascular function increases. Because short-term oscillations of heart rate, ie, *From the II Department of Pediatrics (Drs. Villa, Pagani, Paggi, Massa, and Ronchetti), University of Rome “La Sapienza,” Rome, Italy; and the Department of Information and Systems Science (Dr. Calcagnini), University of Rome “La Sapienza,” Rome, Italy. Manuscript received January 25, 1999; revision accepted July 15, 1999. Correspondence to: Professor Roberto Ronchetti, MD, II Cattedra Clinica Pediatrica, Universita` “La Sapienza,” Viale Regina Elena, 324, 00161-Rome, Italy; e-mail: [email protected] 460

heart rate variability (HRV), reflect autonomic nervous system activity, they are useful for assessing autonomic control under various physiologic and clinical conditions. Spectrum analysis of HRV is a noninvasive procedure that provides quantitative information on sympathetic and parasympathetic control.2– 4 The short-term HRV spectrum distinguishes three main power components. The high-frequency (HF) component (range, 0.15 to 0.40 Hz), corresponding to heart rate and BP oscillations induced by respiratory activity, is mediated by the vagal branch of the autonomic nervous system and is therefore considered a marker of parasympathetic activity. The lowfrequency (LF) component (0.04 to 0.15 Hz), related to the baroreflex control of systemic BP, provides an index of sympathetic activity. Although the very low-frequency (VLF) compoClinical Investigations

nent (⬍ 0.04 Hz) is harder to attribute to a specific physiologic process, it probably reflects the involvement of long-term regulatory mechanisms, including humoral factors and temperature control. Some evidence implies that VLF power derives from the renin-angiotensin system.2,5,6 Using time-domain and frequency-domain analysis of the HRV signal, some researchers have demonstrated parasympathetic dominance during nonrapid eye movement sleep, and increased sympathetic activity in REM sleep.3,7 Studies assessing autonomic function according to sleep state and age in children have detected marked between-subject differences in the maturation of sympathetic and parasympathetic control.8 –10 In this study, our primary aim was to assess the effect of sleep stages and age on short-term HRV during sleep in healthy infants and children. We also wanted to find the most sensitive sleep stage for use in assessing maturation of the autonomic nervous system by spectral analysis. Last, we sought to assess whether infants and children had a similar pattern of autonomic nervous activity during the various sleep stages. For this purpose, we computed time-domain and frequency-domain variables of short-term HRV from all-night polysomnographic recordings.

Materials and Methods We studied 14 healthy subjects, including 7 infants (mean age, 9.40 ⫾ 2.32 months; range, 2 to 12 months) and 7 children (mean age, 8.93 ⫾ 0.65 years; range, 8 to 10 years) who had served as normal controls in a baseline sleep study. All of them were attending our outpatient clinic because their parents reported that they occasionally snored or breathed noisily, but they all had normal polysomnographic studies and normal findings from physical and neurologic examinations. The study was approved by the local ethical committee, and parents gave informed consent for their children to take part. Polysomnograms for spectrum analysis were recorded on a Medilog SAC 800 system (Oxford Instruments, Oxfordshire, UK) in a sound-proof sleep room. Each recording started at around 9:00 pm and finished when the children woke up in the morning. None of the subjects was sedated or had sleep deprivation. All polysomnographic recordings included scalp EEG; eye movements; chin electromyogram; ECG (2-lead); nasal and oral airflow; thoracic and abdominal respiratory effort; and arterial oxygen saturation (Sao2). Polysomnographic signals were simultaneously recorded, digitized, and stored on an optical disk. Sleep stages (1 to 4 and REM) were scored visually,11 according to the international criteria of Rechtschaffen and Kales’12 and according to the statement of the American Thoracic Society concerning standards and indications for cardiopulmonary sleep studies in children.13

the Medilog system had a low sampling rate (125 samples/s), we used an algorithm of interpolation (parabolic) to refine the QRS-complex fiducial point, in order to improve the accuracy of R-wave recognition.15 The respiratory signal was sampled once every R-wave peak (respirogram) from chest or abdomen movements, whichever had the greater signal-to-noise ratio. Each tachogram was then divided into segments according to the sleep stages, and each segment was then searched for artifacts and for missing or ectopic beats that could affect spectral estimation. To obtain appropriate intervals, a filtering process either deleted or inserted beats when needed. Because the tachogram is not an even sampled time series but a point process, spectral frequency are expressed as hertz equivalents.4,5 Each segment belonging to a given sleep stage lasting ⬎ 250 consecutive beats was considered eligible for spectral analysis. For each sleep stage, we considered a minimum of 5 segments and a maximum of 10. If ⬎ 10 segments were available, 10 segments were included on a random basis. The investigator who interpreted and chose the segments was not blinded to which group he examined (infants vs children). Blinding was unnecessary because the selection criteria were predefined and the investigator would inevitably have been aware of the age group on account of the average heart rate. Exclusion criteria included changes in the amplitude, frequency, or both ⬎ 20% of the average values in any respiratory trace for that subject and that sleep stage; mean Sao2 ⬍ 90% or Sao2 signal fluctuations ⬎ 4%, or both; and the presence of arousal or body movement and artifacts or ectopic beats ⬎ 1% of the total number of beats for each segment. Exclusion criteria resulted in an average segment dropout of 10%. HRV spectra were estimated on 250 beats by monovariate autoregressive modeling (Levinson algorithm).16 The model order was chosen according to the Anderson whiteness test and Akaike optimality test (order ranging from 8 to 16).16 No trend removal was used. Autoregressive spectral estimation is acceptably reliable, even on short segments of data, and allows automatic decomposition into the various spectral components. We then computed HF (range, 0.15 to 0.40 Hz), LF (range, 0.04 to 0.15 Hz), and VLF (ⱕ 0.04 Hz) powers in absolute units (s2) in percentage power (ie, percent of the total spectral power), and in normalized power (ie, percentage of the total spectral power minus the VLF component); and the ratio between LF and HF powers (LF/HF). As recommended in the guidelines,5 absolute units of LF and HF were calculated for purposes of comparison; absolute units were not used to evaluate the autonomic nervous system balance because the increased heart rate during sympathetic activation is usually accompanied by a reduction in total power (TP). Expressing spectral components in absolute units (s2) therefore causes the changes in TP to influence LF and HF powers in the same direction, thus significantly preventing the appreciation of the fractional distribution of the energy.5 Moreover, a respiratory rate of ⬍ 12 breaths/min causes the LF and HF components of the HRV signal to overlap, thus preventing correct evaluation of the sympathovagal balance.17 The tachogram and respirogram were therefore examined by cross-spectral Levinson Wiggins Robinson (bivariate) autoregressive analysis to assess the influence of breathing in the HRV spectrum. A spectral coherence of ⬎ 0.5 in the HF band and, conversely, ⬍ 0.5 in the LF band is required for reliable estimations.18 Software for ECG R-wave detection, tachogram and respirogram construction, and spectral analysis was developed in our laboratory. Figure 1 gives tachogram segments and the relative power spectra for a representative subject.

Power Spectrum Analysis of HRV The R-R series (interval tachogram) was obtained from the polysomnographic ECG recording, and computed with an algorithm of R-wave recognition (derivative ⫹ threshold).14 Because

Statistical Analysis Data are expressed as mean ⫾ SD. Data were analyzed in two steps. First, we evaluated autonomic control in each subject CHEST / 117 / 2 / FEBRUARY, 2000

461

Figure 1. An example of a tachogram (top) and its power spectrum (bottom) during REM sleep (left, A) and stage 4 sleep (right, B). Vertical lines in tachograms indicate the interval selected for spectral analysis. See text for further details.

1.16 ⫾ 1.1), mean Sao2 (infants, 96.2 ⫾ 0.8%; children, 96.4 ⫾ 4.5%), and the respiratory rate (infants, 25.6 ⫾ 4.34 breaths/min; children, 20.4 ⫾ 2.61 breaths/min) were all within the normal ranges for our laboratory.11 Infants had significantly slower respiration in deep sleep (p ⬍ 0.05; Table 1). Time-domain analysis showed that in infants and children, heart rate (assessed by the R-R interval) and TP did not differ significantly among the various sleep stages; infants had a significantly higher heart rate (p ⬍ 0.01 in all stages) and lower TP (p ⬍ 0.02 in all stages; Table 1). Occasional selected segments for the infant group contained a respiratory component that slightly exceeded the upper limit mentioned in the HRV guidelines.5 To avoid underestimating HF power, these respiratory components

during the various sleep stages. Data obtained for each sleep stage were averaged to provide a mean stage-related spectral pattern; statistical differences among stages were determined with analysis of variance (ANOVA; nonparametric repeated measures: Friedman test). The second step evaluated the differences between the two age groups. The mean sleep stage-related values for each subject were averaged within the two groups, and then analyzed with the nonparametric (Mann-Whitney) test for unpaired data. A p value ⬍ 0.05 was defined as statistically significant. Parents gave their informed consent for their children to participate in the study, and all study procedures had the approval of the local hospital ethics committee.

Results During the all-night polysomnographic study, the apnea indices19 (infants, 1.12 ⫾ 1.2; children,

Table 1—Cardiorespiratory Variables in Seven Healthy Infants and Seven Healthy Children During Each Sleep Stage* Spectral Variable R-R interval, s Infants Children p value† Total power, s2 Infants Children p value† Respiratory rate, breaths/min Infants Children p value†

Sleep Stages 1

2

3

4

REM

p Value‡

0.49 ⫾ 0.04 0.82 ⫾ 0.10 0.01

0.49 ⫾ 0.04 0.86 ⫾ 0.11 0.001

0.51 ⫾ 0.02 0.84 ⫾ 0.10 0.001

0.51 ⫾ 0.03 0.83 ⫾ 0.09 0.001

0.48 ⫾ 0.02 0.82 ⫾ 0.12 0.001

NS NS

0.001 ⫾ 0.001 0.010 ⫾ 0.007 0.02

0.001 ⫾ 0.002 0.008 ⫾ 0.006 0.01

0.0004 ⫾ 0.0003 0.007 ⫾ 0.006 0.001

0.0004 ⫾ 0.0002 0.006 ⫾ 0.004 0.001

0.0006 ⫾ 0.0005 0.011 ⫾ 0.012 0.001

NS NS

24.3 ⫾ 3.1 19 ⫾ 1.61 0.002

NS NS

23 ⫾ 3.03 18 ⫾ 1.98 0.001

25 ⫾ 2.94 18 ⫾ 2.62 0.0001

24 ⫾ 2.50 18 ⫾ 2.30 0.0001

25 ⫾ 2.34 18 ⫾ 2.44 0.0001

*Data are expressed as mean ⫾ SD. NS ⫽ not significant. †Mann-Whitney statistical analysis used to calculate p values for differences between infants and children. ‡ANOVA (Friedman) used to calculate p values for differences between sleep stages. 462

Clinical Investigations

were considered as HF, provided that they showed ⬎ 0.50 coherence with respiratory signal. In all spectra, respiration and the LF component showed low values of coherence (⬍ 0.3), suggesting that the LF band was scarcely influenced by respiration and therefore reliably estimated. Frequency-domain analysis showed that in both age groups, HF percentage values increased according to the depth of sleep, and reached maximum in stage 4 sleep (Table 2). Children had significantly higher HF percentage values than infants (p ⬍ 0.05 in all stages). Although the LF component showed a reciprocal behavior (Table 2), the difference between the two age groups reached statistical significance only in stage 3 and REM. LF and HF, expressed as normalized units, showed a similar trend as LF and percentage HF, but without reaching statistical significance (Table 3). The LF/HF ratio changed significantly with the sleep stages (p ⬍ 0.01 in children; p ⬍ 0.02 in infants): in deep sleep, it decreased, and in REM sleep, it increased (Table 3). Children invariably had a lower LF/HF ratio than infants in all stages (p ⬍ 0.05; Fig 2). The VLF percentage component differed significantly during sleep only in children; the differences between infants and children reached statistical significance in stages 2, 3, and 4 (p ⬍ 0.03; Table 2). Table 4 shows the data for the spectral components expressed in absolute values, merely for information. Discussion In this study, we assessed short-term HRV from all-night polysomnographic recordings in infants and

children. To delineate HRV more precisely during sleep, we segmented sleep into five stages instead of the two stages (active and quiet sleep) previously studied. Our findings showed that this approach is essential to obtain unbiased results. HRV and Sleep Stages In infants and children, the HF spectral power component of short-term HRV increased with the depth of sleep, reaching maximum values in stage 4 sleep. The LF component showed a reciprocal behavior, reaching minimal values in stage 4 sleep. These changes caused the LF/HF ratio to diminish during deep sleep and sharply increase during REM sleep (Tables 2, 3). Conversely, in agreement with previous studies, the time-domain variables, heart rate, and HRV remained practically unchanged during the various sleep stages (Table 1).3,20,21 Interestingly, HF and LF percentages and normalized values in both groups changed significantly during the various sleep stages. These changes emphasize the need to divide sleep into the five stages. Because an increased HF component is usually considered a marker of increased parasympathetic tone,5 our data suggest prevalent parasympathetic control of the heart during deep sleep. This confirms previous findings in children3 and suggests similar behavior in infants. Unlike the HF power of HRV, LF has a more controversial origin and meaning. The genesis and amplitude of LF power of HRV depend chiefly on baroreflex control of systemic BP. The baroreflex system has a minor role during sleep.22 Yet, few studies have addressed the maturation of baroreceptors, and hence of baroreflex sensitivity, in infants. In this study, VLF percentage power values dif-

Table 2—Change in Frequency-Domain Indexes (Percentage Values) in Seven Healthy Infants and in Seven Healthy Children During Each Sleep Stage* Sleep Stages Spectral Variable HF% Infants Children p value† LF% Infants Children p value† VLF% Infants Children p value†

1

2

3

4

REM

p Value‡

15.8 ⫾ 10.6 42.8 ⫾ 19.1 0.05

23.6 ⫾ 8.3 61.8 ⫾ 9.5 0.01

29.7 ⫾ 12.0 66.7 ⫾ 7.70 0.001

41.7 ⫾ 10.8 69.8 ⫾ 6.20 0.001

15.47 ⫾ 11.6 46.00 ⫾ 17.7 0.01

0.02 0.01

33.2 ⫾ 14.3 22.5 ⫾ 8.50 NS

22.1 ⫾ 18.9 13.9 ⫾ 6.50 NS

25.6 ⫾ 13.0 11.4 ⫾ 3.50 0.04

17.2 ⫾ 7.9 11.1 ⫾ 4.4 NS

33.3 ⫾ 13.6 20.8 ⫾ 3.00 0.04

NS 0.01

37.1 ⫾ 11.3 26.6 ⫾ 12.2 NS

44.9 ⫾ 12.8 16.7 ⫾ 9.80 0.001

32.6 ⫾ 15.5 14.1 ⫾ 10.5 0.03

29.9 ⫾ 4.46 9.36 ⫾ 3.97 0.001

46.9 ⫾ 20.0 24.4 ⫾ 19.1 NS

NS 0.02

*HF% ⫽ HF percentage values; LF% ⫽ LF percentage values; VLF% ⫽ VLF percentage values. See Table 1 for other abbreviation. Data are expressed as mean ⫾ SD. †Mann-Whitney statistical analysis used to calculate p values for differences between infants and children. ‡ANOVA (Friedman) used to calculate p values for differences between sleep stages. CHEST / 117 / 2 / FEBRUARY, 2000

463

Table 3—Changes in Frequency-Domain Indexes (Normalized Units) in Seven Healthy Infants and in Seven Healthy Children During Each Sleep Stage* Sleep Stages Spectral Variable HF, n.u. Infants Children p value† LF, n.u. Infants Children p value† LF/HF ratio Infants Children p value†

p Value‡

1

2

3

4

REM

52.6 ⫾ 35.7 56.5 ⫾ 19.7 NS

61.0 ⫾ 28.6 76.5 ⫾ 11.2 NS

61.4 ⫾ 25.1 80.7 ⫾ 8.70 NS

70.8 ⫾ 20.3 80.6 ⫾ 11.1 NS

43.8 ⫾ 30.3 58.8 ⫾ 12.8 NS

0.02 0.00

41.6 ⫾ 30.3 32.4 ⫾ 16.0 NS

29.6 ⫾ 25.4 15.7 ⫾ 9.40 NS

30.3 ⫾ 20.4 12.2 ⫾ 5.90 NS

19.9 ⫾ 12.7 10.7 ⫾ 6.10 NS

46.4 ⫾ 22.6 30.2 ⫾ 10.0 NS

0.02 0.01

2.8 ⫾ 1.7 0.7 ⫾ 0.6 0.05

1.6 ⫾ 2.4 0.2 ⫾ 0.1 0.01

1.5 ⫾ 1.5 0.1 ⫾ 0.05 0.04

0.5 ⫾ 0.40 0.1 ⫾ 0.08 0.04

4.4 ⫾ 2.4 0.6 ⫾ 0.4 0.001

0.02 0.01

*n.u. ⫽ normalized units (percent of the total spectral power minus the VLF component). See Table 1 for other abbreviation. Data are expressed as mean ⫾ SD. †Mann-Whitney statistical analysis used to calculate p values for differences between infants and children. ‡ANOVA (Friedman) used to calculate p values for differences between sleep stages.

fered during the various sleep stages in children but remained unchanged in infants (Table 2). Because VLF arises largely from as yet poorly defined physiologic mechanisms, its meaning remains unclear.5 VLF results, therefore, should be interpreted with caution. In addition, because the estimation of the VLF component is strongly influenced by the length of the tachogram examined, VLF power is unreliable in short-term HRV analysis. Our power spectral indexes of HRV in children are difficult to compare with findings published before the HRV guidelines were issued,5 partly because some investigators used an LF frequency range (0.02 to 0.20 Hz)23,24 that included part of the VLF band. In addition, normal reference ranges for LF and HF frequency bands have never been defined in a pediatric population. Whether adult ranges also apply to younger age groups is a question for further investigation.

Figure 2. LF/HF ratio values in healthy infants and children. In both age groups, as the sleep stage deepened, the ratio decreased. The ratio values differed most during REM sleep. Data are expressed as mean ⫾ SEM. Statistical analysis used: MannWhitney. Differences among sleep stages (ANOVA: Friedman): in infants, p ⬍ 0.02; in children, p ⬍ 0.01. 464

HRV and Age Our data showed that in all sleep stages, the mean heart rate and TP differed significantly in infants and children. Infants had a higher mean heart rate and lower TP. In all sleep stages, HF percentages in children were nearly twice those in infants. A correlation between HRV and age has been described in normal newborns,25–27 infants,10,28,29 and children.9,10,29 A positive correlation has also been found for TP, LF absolute values, and HF absolute values.10,29,30 In addition, these studies found that the LF/HF ratio decreased from infancy to childhood. Although we obtained similar results, they could partly depend on the fact that infants breathe faster than children. Infants had significantly higher breathing rates than children in all sleep stages. Unfortunately, data about the transfer function between respiration and heart rate are available only in adults.31 These data suggest that the differences in the breathing rate cannot account for the nearly twofold differences we observed in the HF percentages. In adults, the transfer function between respiration and HR is flat beyond 0.1 Hz. Assuming a similar behavior in infants and children, the high HF percentage values and the high LF/HF ratio, spectral measures that describe better the fractional distribution of power in the various bands, support the hypothesis of an autonomic balance characterized by increased sympathetic neural control accompanied by a reciprocal decrease in parasympathetic activity in the younger group. This difference in sympathetic dominance presumably reflects autonomic nervous system maturation that takes place between the first year of life and childhood.9,32,33 The notion of preClinical Investigations

Table 4 —Change in Frequency-Domain Indexes (Absolute Values) in Seven Healthy Infants and in Seven Healthy Children During Each Sleep Stage* Sleep Stages Spectral Variable HF, s2 Infants Children p value† LF, s2 Infants Children p value† VLF, s2 Infants Children p value†

1

2

3

4

REM

p Value‡

0.0004 ⫾ 0.00048 0.0039 ⫾ 0.00256 0.02

0.0017 ⫾ 0.00258 0.0069 ⫾ 0.00537 0.01

0.0005 ⫾ 0.00045 0.0072 ⫾ 0.00570 0.001

0.0010 ⫾ 0.00110 0.0063 ⫾ 0.00353 0.001

0.0002 ⫾ 0.00027 0.0084 ⫾ 0.00889 0.001

NS NS

0.0002 ⫾ 0.00040 0.0029 ⫾ 0.00355 0.02

0.0001 ⫾ 0.00016 0.0012 ⫾ 0.00100 0.01

0.0001 ⫾ 0.00007 0.0009 ⫾ 0.00094 0.001

0.00005 ⫾ 0.00005 0.0009 ⫾ 0.00084 0.001

0.0001 ⫾ 0.00016 0.0022 ⫾ 0.00253 0.001

NS 0.00

0.0005 ⫾ 0.00091 0.0024 ⫾ 0.00202 NS

0.0023 ⫾ 0.00358 0.0015 ⫾ 0.00211 NS

0.0001 ⫾ 0.00009 0.0008 ⫾ 0.00060 0.001

0.0001 ⫾ 0.00009 0.0005 ⫾ 0.00029 0.001

0.0003 ⫾ 0.00033 0.0016 ⫾ 0.00101 0.001

NS 0.01

*Data are expressed as mean ⫾ SD. Absolute values are given. See Table 1 for other abbreviation. †Mann-Whitney statistical analysis used to calculate p values for differences between infants and children. ‡ANOVA (Friedman) used to calculate p values for differences between sleep stages.

vailing sympathetic control in children aged ⬍ 1 year could be clinically important because sympathetic overbalance is accentuated during REM sleep, a particularly frequent sleep stage in young children. This study was also designed to seek the most sensitive sleep stage for assessing maturation of the autonomic nervous system by spectral analysis and to find out whether infants and children have a similar pattern of autonomic nervous system activity during the various sleep stages. In infants and children, the most sensitive index of autonomic nervous system balance (the LF/HF ratio) differed most significantly during REM sleep (Fig 2). Second, even though their spectral indexes differed, infants and children had a similar pattern of autonomic nervous system changes in the various sleep stages. Because it relies simply on HRV, power spectral density analysis is a noninvasive procedure for assessing the autonomic balance (sympathetic/parasympathetic) in normal children and in children with pathologic conditions. Hence, it overcomes the main drawback of classic autonomic tests34 in pediatric ages. Conclusion HRV strongly depends on the sleep stage. Studies conducted without sleep staging (eg, using 24-h ECG Holter monitoring),10 therefore, cannot take sleep-related changes into account. This is especially true in infants, who spend most of their time in sleep, and most importantly in REM sleep. In addition, although we studied a relatively small number of healthy infants, mainly because parents

are reluctant to allow healthy infants to undergo all-night polysomnography, our findings confirm an age-related change in the autonomic nervous system during childhood. References 1 Somers VK, Dyken ME, Mark AL, et al. Sympathetic nerve activity during sleep in normal subjects. N Engl J Med 1995; 328:303–307 2 Akselrod S, Gordon D, Ubel FA, et al. Power spectrum analysis of heart rate fluctuations: a quantitative probe of beat-to-beat cardiovascular control. Science 1981; 213:220 – 222 3 Baharav A, Kotagal S, Gibbons V, et al. Fluctuations in autonomic nervous activity during sleep displayed by power spectrum analysis of heart rate variability. Neurology 1995; 45:1183–1187 4 Pagani M, Lombardi F, Guzzetti F, et al. Power spectral analysis of heart rate and arterial pressure variabilities as a marker of sympatho-vagal interaction in men and conscious dog. Circ Res 1986; 59:178 –193 5 Heart rate variability: standards of measurement, physiological interpretation and clinical use; Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology. Circulation 1996; 93:1043– 1065 6 Fleisher LA, Frank SM, Sessler DI, et al. Thermoregulation and heart rate variability. Clin Sci (Colch) 1996; 90:97–105 7 Gaultier C. Cardiorespiratory adaptation during sleep in infants and children. Pediatr Pulmonol 1995; 19:105–117 8 Schechtman VL, Harper RM, Kluge KA. Development of heart rate variation over the first 6 months of life in normal infants. Pediatr Res 1989; 26:343–346 9 Yeragani VK, Pohl R, Berger R, et al. Relationship between age and heart rate variability in supine and standing postures: a study of spectral analysis of heart rate. Pediatr Cardiol 1994; 15:14 –20 10 Massin M, von Bernuth G. Normal ranges of heart rate variability during infancy and childhood. Pediatr Cardiol 1997; 18:297–302 CHEST / 117 / 2 / FEBRUARY, 2000

465

11 Villa MP, Piro D, Dotta A, et al. Validation of automated sleep analysis in normal children. Eur Respir J 1998; 11:1– 4 12 Rechtschaffen A, Kales A. Manual of standardized terminology, techniques and scoring systems for sleep stages of human subjects. Los Angeles, CA: Brain Information Service/Brain Research Institute, 1968 13 Standards and indications for cardiopulmonary sleep studies in children. Am J Respir Crit Med Care 1996; 153:866 – 878 14 Baselli G, Cerutti S, Civardi S, et al. Heart rate variability signal processing: a quantitative approach as an aid to diagnosis in cardiovascular pathologies. Int J Biomed Comput 1987; 20:51–70 15 Bianchi AM, Mainardi LT, Petrucci E, et al. Time variant power spectrum analysis for the detection of transient episodes in the HRV signal, Proc IEEE Trans Biomed Eng 1993; 40:136 –144 16 Kay SM, Marple SL. Spectrum analysis: a modern perspective. Proc IEEE Trans Biomed Eng 1981; 69:1380 –1419 17 Brown TE, Beightol LA, Koh J, et al. Important influence of respiration on human R-R interval power spectra is largely ignored. J Appl Physiol 1993; 75:2310 –2317 18 Bernardi L, Rossi M, Soffiantino F, et al. Cross correlation of heart rate and respiration versus deep breathing: assessment of new test of cardiac autonomic function in diabetes. Diabetes 1989; 38:589 –596 19 Marcus CL, Omlin KJ, Basinki DJ, et al. Normal polysomnographic values for children and adolescents. Am Rev Respir Dis 1992; 146:1235–1239 20 Baust W, Bohnert B. The regulation of heart rate during sleep. Exp Brain Res 1969; 7:169 –180 21 Pivik RT, Busby KA, Gill E, et al. Heart rate variations during sleep in preadolescents. Sleep 1996; 19:117–135 22 Patton DJ, Hanna BD. Postnatal maturation of baroreflex heart rate control in neonatal swine. Can J Cardiol 1994; 10:233–238 23 Mancia G. Autonomic modulation of the cardiovascular system during sleep. N Engl J Med 1993; 328:347–349

466

24 Dykes FD, Ahmann PA, Baldzer K, et al. Breath amplitude modulation of HR variability in normal full term neonates. Pediatr Res 1986; 20:301–308 25 Baldzer K, Dykes FD, Jones SA, et al. Heart rate variability analysis in full-term infants: spectral indices for study of neonatal cardio-respiratory control. Pediatr Res 1989; 26: 188 –195 26 Clairambault J, Curzi-Dascalova L, Kauffmann F, et al. Heart rate variability in normal sleeping full-term and pre-term neonates. Early Hum Dev 1992; 28:169 –183 27 Spassov L, Curzi-Dascalova L, Clairambault J, et al. Heart rate and heart rate variability during sleep in small-forgestational-age newborns. Pediatr Res 1994; 35:500 –505 28 Schechtman VL, Raetz SL, Harper RK, et al. Dynamic analysis of cardiac R-R intervals in normal infants and in infants who subsequently succumbed to the sudden infant death syndrome. Pediatr Res 1992; 31:606 – 612 29 Finley JP, Neugent ST. Heart rate variability in infants, children and young adults. J Auton Nerv Syst 1995; 51:103– 108 30 Goto M, Nagashima M, Baba R, et al. Analysis of heart rate variability demonstrates effects of development on vagal modulation of heart rate in healthy children. J Pediatr 1997; 130:725–729 31 Saul JP, Berger RD, Albrecht P, et al. Transfer function analysis of the circulation: unique insights into cardiovascular regulation. Am J Physiol 1991; 261(4 pt 2):H1231–H1245 32 Chatow U, Davidson S, Rechman BL, et al. Development and maturation of the autonomic nervous system in premature and full-term infants using spectral analysis of heart rate fluctuation. Pediatr Res 1995; 37:294 –302 33 Siimes ASI, Valimaki IAT, Antila KJ, et al. Regulation of HR variation by the autonomic nervous system in neonatal lambs. Pediatr Res 1990; 27:383–391 34 Ewing DJ, Clarke BF. Diagnosis and management of diabetic autonomic neuropathy. Br Med J 1982; 285:916 –918

Clinical Investigations