Diaphragmatic
muscle fatigue
in the newborn
N. MULLER, G. GULSTON, D. CADE, J. WHITTON, A. B. FROESE, M. H. BRYAN, AND A. C. BRYAN Respiratory Physiology, Research Institute, Hospital for Sick Children, Toronto, Ontario M5G 1X8, Canada
MULLER, N., G. GULSTON, D. CADE, J. WHITTON, A. B. FROESE, M. H. BRYAN, AND A. C. BRYAN. Diaphragmatic muscle fatigue in the newborn. J. Appl. Physiol.: Respirat. Environ. Exercise Physiol. 46(4): 688-695, 1979.-In nine newborn infants we looked for evidence of diaphragmatic muscle fatigue by analysis of the EMG frequency spectrum. We measured the diaphragmatic EMG with surface electrodes, the motion of the rib cage and abdomen with magnetometers, and monitored the sleep state using EEG, EOG, and behavioral criteria. Spectral frequency analysis of EMG signals, uncontaminated by ECG artifact, was subsequently performed by digital computer. In six normal infants, when we compared breaths with rib cage distortion to undistorted breaths we found the characteristic changes of fatigue in the EMG spectrum: a fall in the high-frequency (>160-640 Hz) power and an increase in the low-frequency (lo-40 Hz) power. The fall in the high-to-low frequency ratio was progressive with increasing distortion (all r > 0.60, P < 0.01) and with increasing amplitude of the diaphragmatic EMG (all r > 0.65, P < 0.05). Three other infants had just come off the ventilator and were on continuous positive end-expiratory pressure at the time of the study. Two of these deteriorated and when they again required assisted ventilation, the EMG spectrum showed a significant drop in the high-tolow frequency ratio (P < 0.005). The one infant that did well off the ventilator showed no evidence of diaphragmatic fatigue. We conclude that respiratory muscle fatigue does occur even in normal infants whenever there is significant distortion of the rib cage, i.e., during active sleep, and that it plays a significant role when there is superimposed lung disease. spectral analysis; newborn
respiratory
failure; work of breathing
MUSCLES of the newborn appear histochemically to be poorly equipped to sustain high work loads. Keens et al. (10) have shown that, compared to adults, infants have far fewer high-oxidative fibers, which are very resistant to fatigue (1, 3). We have therefore looked for evidence of respiratory muscle fatigue by analysis of the diaphragmatic EMG. The changes in the EMG of skeletal muscles with fatigue are well established. Once a muscle force is exerted that will lead to fatigue there is a very rapid decrease in the high-frequency power with a progressive increase in the lowfrequency power (8). Similar changes are seen with both needle and surface electrodes. Gross et al. (5) have applied this method to the diaphragm using high- and lowbandpass filters and have shown that fatigue patterns can be produced in adults by loaded breathing. They reported similar frequency changes using either bipolar esophageal electrodes or surface electrodes. Because we THE RESPIRATORY
688
were concerned that the EMG frequency spectrum of a newborn might be different from that of an adult, we chose to use spectral frequency analysis to compare it with the spectrum described by Kadefors et al. (8). METHODS
We studied six infants free of cardiopulmonary disease and three newborns who were being weaned off the ventilator at the time of the study. Of these three infants, infant 7 had patent ductus arteriosus, heart failure, and patchy atelectasis; infant 8 had bronchopulmonary dysplasia and patent ductus arteriosus; and infant 9 had respiratory distress syndrome and pneumonia. The infants ranged in gestational age from 26 to 40 wk (mean t SD, 31.3 t 4.8 wk) and were studied 3 days to 5 wk (mean t SD, 2.8 t 1.4 wk) after birth. The study was done in a controlIed thermal environment while the infants were sleeping. Sleep state was monitored in the six normal newborns by a two-lead encephalograph, an electroculograph, and behavioral criteria. It was not monitored in the other infants. The diaphragmatic EMG (Edi) was recorded using surface electrodes in the right sixth and seventh interspaces between the midclavicular and midaxillary line. In three infants we also simultaneously recorded the Edi using right subcostal electrodes, in the nipple line and 1 in. medially or laterally. The intercostal EMG was recorded in the second interspace parasternally. Abdominal and rib cage motion was measured using anteroposterior magnetometer pairs in the midline at the level of the nipple line and just above the umbilicus. Magnetometer pairs were chosen so that there was a linear voltagedistance relationship over the operational range. However, with movement it is difficult to keep the signal on the record without occasional alteration of the zero suppression on the magnetometers, thus changing the voltage-distance calibration. Therefore, the magnetometer output is reported in arbitrary units of pen deflection for segments during which the zero suppression and gain were constant. No sedation was used. All data were stored on a tape recorder which has a signal-to-noise ratio of 47 dB. The tape was later played back into an IBM 1800. The analogto-digital convertor (ADC) in this computer has a quantization of 14 bits, which is, in practice, equivalent to a signal-to-noise ratio of 84 dB. Its Nyquist frequency was 2,000 Hz. Before analog-to-digital conversion the signal was low-pass filtered with a cutoff frequency of 1,500 Hz.
0161-7567/79/0000-0000$01.25
Copyright
0 1979 the American
Physiological
Society
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (130.070.008.131) on November 11, 2018. Copyright © 1979 American Physiological Society. All rights reserved.
DIAPHRAGMATIC
MUSCLE
FATIGUE
IN
THE
689
NEWBORN
Spectral analysis was obtained using the fast Fourier transform. The greatest problem in the analysis of the diaphragmatic EMG in the newborn is the presence of the cardiac artifact. To eliminate the QRS artifact we used two input channels on the computer: into one we played back the filtered ECG signal obtained from the intercostal EMG leads (ECG into channel 1) and into the other the diaphragmatic EMG (Edi into channel 2). As shown in Fig. 1, the peak of the R in channel one triggered the computer to disregard the next 50 ms and analyze the following 200 ms of the Edi. Each 200 ms of analysis constituted one file section (FS). After sampling one FS the computer stopped until triggered again by another cardiac artifact. TRIGGER
EKG
Diaphragmatic EMG
FIG. 1. Peak of cardiac artifact in channel 1 triggers computer, which disregards following 50 ms, and analyzes next 200 ms (= one fde section) of diaphragmatic EMG.
f IO FIG. 2. Comparison between normal frequency spectrum of diaphragmatic EMG in 9 infants with spectrum of adult biceps brachii muscle during an isometric-isotonic contraction as described by Kadefors et al. (8). This normal frequency spectrum was observed when
Because we gated the ECG signal, conventional criteria to assess the amplitude of the diaphragmatic EMG could not be used. We estimated the total power of a 200-ms FS and used this as an approximation of the force of the diaphragmatic contraction. This allowed us to differentiate FS occurring during inspiration from those occurring during expiration. A quantitative criterion for this distinction was obtained by the analysis of FS during an apneic period, defined as an end-expiratory pause of more than 3 s. The total power during apnea was computed and only FS with a power 3.16 times greater than the upper 95% confidence limit during apnea were accepted as being inspiration. This gives a signal-to-noise ratio of 10 dB. We studied periods lasting 20-40 s. Depending on the relationship between the ECG artifact and inspiration we were able to analyze 5-20 breaths in each period. It is recommended (18) that when using a computer frequency analysis the input frequency be at least twice the highest frequency of interest. Our Nyquist frequency was 2,000 Hz and we analyzed up to 640 Hz. We divided the frequency spectrum of the Edi into octave bands from 10 to 640 Hz, with center frequencies of 14,28, 56, 113, 226, and 452 Hz. The center frequency is defined as fo = m, where fg = 2fi. We expressed the amplitude per frequency band as a percentage of the total power for the file section, except in Figs. 2 and 3 where we transformed it into decibels to compare the normal frequency spectrum of the newborn diaphragm, and its changes with fatigue, with the spectrum observed by Kadefors et al. (8) in the adult biceps brachii muscle. In our study the high-to-low frequency ratio is obtained by dividing the percentage of the total power between 160 and 640 Hz by the percentage of the total power between 10 and 40 Hz, i.e.,
1 100
FREQUENCY
there was no distortion immediately following newborns.
(Hz)
1 1000
of rib cage in 6 normal infants and in breaths discontinuation of IMV or CPAP in 3 other
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (130.070.008.131) on November 11, 2018. Copyright © 1979 American Physiological Society. All rights reserved.
690
MULLER
l NO
ET
AL.
DISTORTION
o MARKED
DISTORTION
lb
IhO FREQUENCY
lOb0
(HZ)
FIG. 3. Changes in diaphragmatic EMG power spectrum with distortion of rib cagein infant 2. Points represent mean and SE of power of different frequency bands expressed as dB. Breaths occurring during
maximal changes increase
high-to-low frequency ratio
requiring artificial ventilation, but did stabilize once back on CPAP. In these infants, the breaths immediately following the cessation of IMV or CPAP were used as control.
power between 160 and 640 Hz = power between 10 and 40 Hz The amplitude of the Edi is expressed as a ratio of the mean power observed in a particular section over the mean power during apnea. This permits a better correlation of the different total powers observed in the various patients. In the six normal newborns, a period with minimal or no distortion of the rib cage was used as control and the other sections compared to it. It should be noted that we use the word distortion in a different sense from Konno and Mead (13). They call departure from the relaxation characteristics of the whole rib cage, distortion. We define distortion as independent motion of the upper and lower rib cage. We quantified the degree of distortion by measuring the maximal point of inward motion of the upper rib cage during inspiration. Inspiration was considered to start with the initial outward movement of the abdomen and end when this reached its peak. Three other infants were studied at a point where the decision to wean them off the ventilator had been made on clinical grounds. Intermittent mandatory ventilation (IMV) was discontinued, but a positive end-expiratory pressure (CPAP) of 3-5 cmHz0 was maintained. The IMV rates were, respectively, 7,7, and 36 beats/min prior to weaning. Within 35 min, infant 7 had frequent apneic spells, “looked sick,” and had to be ventilated again. Infant 8 had a severe apneic spell that necessitated return to artificial ventilation 4 min later. Infant 9, who had been on an IMV of 36 beats/min, did well on CPAP alone and was maintained on it. Twenty-four hours later, when he was taken off CPAP, he had severe apneic spells
distortion of chest wall in REM sleep, show characteristic that occur with muscle fatigue: a fall in high frequency and an in low frequency part of spectrum.
RESULTS
NormaL infants. Figure 2 shows that the frequency spectrum analysis of the Edi during undistorted breaths is similar to the adult biceps brachii muscle during an isotonic-isometric contraction as reported by Kadefors et al. (8). Kadefors and co-workers used octave band filters with center frequencies that were slightly different from the ones we used. In all infants the nondistorted control breaths were observed in non-REM sleep; in the infants old enough to have a clear-cut quiet sleep the undistorted breaths occurred during that period. The degree of distortion observed in REM sleep was variable, but distortion was always present when the infant was in REM sleep. The changes in the frequency spectrum when there is marked distortion of the rib cage can be seen in Fig. 3. There is a fall in the high-frequency (160-640 Hz) power and an increase in the low-frequency (lo-40 Hz) power. When we plotted the change in the high-to-low frequency ratio, i.e., amplitude of the signal between 160 and 640 Hz over the amplitude between 10 and 40 Hz, against distortion of the rib cage we observed a statistically significant correlation in all infants (all r > 0.60, P < 0.01). Figure 4 shows the correlation observed in infant 1; Table 1 shows the summary of the results in the six normal infants. In three of these infants we also analyzed separately the first and the last 100 ms of file sections during breaths with and without marked distortion of the rib cage. There
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (130.070.008.131) on November 11, 2018. Copyright © 1979 American Physiological Society. All rights reserved.
DIAYHRAGMATIC
MUSCLE
FATIGUE
IN
THE
691
NEWBORN
was not a significant difference between the frequency spectrum of the first and second half of a file section in any of these infants (P > 0.10 in all the frequency bands). In the six normal infants we also observed a statistically significant correlation between the power of the Edi and the fall in the high-to-low frequency ratio (all r > 0.65, P < 0.05). Figure 5 shows that mean correlation and 06
.5-
r=
1
0.90
*
I 6
I 3
0
I 9
DISTORTION
1 12
I 15
[Arbitrary
FIG. 4. Correlation between distortion to-low frequency ratio in infant 1. Each breaths. Lines represent mean correlation
I 18
I 21
1 24
Units]
of chest wall and fall in highdot represents mean of 5-20 and 95% confidence limits.
n
No.
1 2 3 4 5 6 n Represents including 5-20
the number breaths.
P
r
15 12 18 10 10 10
-0.90 -0.85 -0.61 -0.82 -0.90 -0.81 of periods
that
muscle fatigue
in the newborn
N. MULLER, G. GULSTON, D. CADE, J. WHITTON, A. B. FROESE, M. H. BRYAN, AND A. C. BRYAN Respiratory Physiology, Research Institute, Hospital for Sick Children, Toronto, Ontario M5G 1X8, Canada
MULLER, N., G. GULSTON, D. CADE, J. WHITTON, A. B. FROESE, M. H. BRYAN, AND A. C. BRYAN. Diaphragmatic muscle fatigue in the newborn. J. Appl. Physiol.: Respirat. Environ. Exercise Physiol. 46(4): 688-695, 1979.-In nine newborn infants we looked for evidence of diaphragmatic muscle fatigue by analysis of the EMG frequency spectrum. We measured the diaphragmatic EMG with surface electrodes, the motion of the rib cage and abdomen with magnetometers, and monitored the sleep state using EEG, EOG, and behavioral criteria. Spectral frequency analysis of EMG signals, uncontaminated by ECG artifact, was subsequently performed by digital computer. In six normal infants, when we compared breaths with rib cage distortion to undistorted breaths we found the characteristic changes of fatigue in the EMG spectrum: a fall in the high-frequency (>160-640 Hz) power and an increase in the low-frequency (lo-40 Hz) power. The fall in the high-to-low frequency ratio was progressive with increasing distortion (all r > 0.60, P < 0.01) and with increasing amplitude of the diaphragmatic EMG (all r > 0.65, P < 0.05). Three other infants had just come off the ventilator and were on continuous positive end-expiratory pressure at the time of the study. Two of these deteriorated and when they again required assisted ventilation, the EMG spectrum showed a significant drop in the high-tolow frequency ratio (P < 0.005). The one infant that did well off the ventilator showed no evidence of diaphragmatic fatigue. We conclude that respiratory muscle fatigue does occur even in normal infants whenever there is significant distortion of the rib cage, i.e., during active sleep, and that it plays a significant role when there is superimposed lung disease. spectral analysis; newborn
respiratory
failure; work of breathing
MUSCLES of the newborn appear histochemically to be poorly equipped to sustain high work loads. Keens et al. (10) have shown that, compared to adults, infants have far fewer high-oxidative fibers, which are very resistant to fatigue (1, 3). We have therefore looked for evidence of respiratory muscle fatigue by analysis of the diaphragmatic EMG. The changes in the EMG of skeletal muscles with fatigue are well established. Once a muscle force is exerted that will lead to fatigue there is a very rapid decrease in the high-frequency power with a progressive increase in the lowfrequency power (8). Similar changes are seen with both needle and surface electrodes. Gross et al. (5) have applied this method to the diaphragm using high- and lowbandpass filters and have shown that fatigue patterns can be produced in adults by loaded breathing. They reported similar frequency changes using either bipolar esophageal electrodes or surface electrodes. Because we THE RESPIRATORY
688
were concerned that the EMG frequency spectrum of a newborn might be different from that of an adult, we chose to use spectral frequency analysis to compare it with the spectrum described by Kadefors et al. (8). METHODS
We studied six infants free of cardiopulmonary disease and three newborns who were being weaned off the ventilator at the time of the study. Of these three infants, infant 7 had patent ductus arteriosus, heart failure, and patchy atelectasis; infant 8 had bronchopulmonary dysplasia and patent ductus arteriosus; and infant 9 had respiratory distress syndrome and pneumonia. The infants ranged in gestational age from 26 to 40 wk (mean t SD, 31.3 t 4.8 wk) and were studied 3 days to 5 wk (mean t SD, 2.8 t 1.4 wk) after birth. The study was done in a controlIed thermal environment while the infants were sleeping. Sleep state was monitored in the six normal newborns by a two-lead encephalograph, an electroculograph, and behavioral criteria. It was not monitored in the other infants. The diaphragmatic EMG (Edi) was recorded using surface electrodes in the right sixth and seventh interspaces between the midclavicular and midaxillary line. In three infants we also simultaneously recorded the Edi using right subcostal electrodes, in the nipple line and 1 in. medially or laterally. The intercostal EMG was recorded in the second interspace parasternally. Abdominal and rib cage motion was measured using anteroposterior magnetometer pairs in the midline at the level of the nipple line and just above the umbilicus. Magnetometer pairs were chosen so that there was a linear voltagedistance relationship over the operational range. However, with movement it is difficult to keep the signal on the record without occasional alteration of the zero suppression on the magnetometers, thus changing the voltage-distance calibration. Therefore, the magnetometer output is reported in arbitrary units of pen deflection for segments during which the zero suppression and gain were constant. No sedation was used. All data were stored on a tape recorder which has a signal-to-noise ratio of 47 dB. The tape was later played back into an IBM 1800. The analogto-digital convertor (ADC) in this computer has a quantization of 14 bits, which is, in practice, equivalent to a signal-to-noise ratio of 84 dB. Its Nyquist frequency was 2,000 Hz. Before analog-to-digital conversion the signal was low-pass filtered with a cutoff frequency of 1,500 Hz.
0161-7567/79/0000-0000$01.25
Copyright
0 1979 the American
Physiological
Society
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (130.070.008.131) on November 11, 2018. Copyright © 1979 American Physiological Society. All rights reserved.
DIAPHRAGMATIC
MUSCLE
FATIGUE
IN
THE
689
NEWBORN
Spectral analysis was obtained using the fast Fourier transform. The greatest problem in the analysis of the diaphragmatic EMG in the newborn is the presence of the cardiac artifact. To eliminate the QRS artifact we used two input channels on the computer: into one we played back the filtered ECG signal obtained from the intercostal EMG leads (ECG into channel 1) and into the other the diaphragmatic EMG (Edi into channel 2). As shown in Fig. 1, the peak of the R in channel one triggered the computer to disregard the next 50 ms and analyze the following 200 ms of the Edi. Each 200 ms of analysis constituted one file section (FS). After sampling one FS the computer stopped until triggered again by another cardiac artifact. TRIGGER
EKG
Diaphragmatic EMG
FIG. 1. Peak of cardiac artifact in channel 1 triggers computer, which disregards following 50 ms, and analyzes next 200 ms (= one fde section) of diaphragmatic EMG.
f IO FIG. 2. Comparison between normal frequency spectrum of diaphragmatic EMG in 9 infants with spectrum of adult biceps brachii muscle during an isometric-isotonic contraction as described by Kadefors et al. (8). This normal frequency spectrum was observed when
Because we gated the ECG signal, conventional criteria to assess the amplitude of the diaphragmatic EMG could not be used. We estimated the total power of a 200-ms FS and used this as an approximation of the force of the diaphragmatic contraction. This allowed us to differentiate FS occurring during inspiration from those occurring during expiration. A quantitative criterion for this distinction was obtained by the analysis of FS during an apneic period, defined as an end-expiratory pause of more than 3 s. The total power during apnea was computed and only FS with a power 3.16 times greater than the upper 95% confidence limit during apnea were accepted as being inspiration. This gives a signal-to-noise ratio of 10 dB. We studied periods lasting 20-40 s. Depending on the relationship between the ECG artifact and inspiration we were able to analyze 5-20 breaths in each period. It is recommended (18) that when using a computer frequency analysis the input frequency be at least twice the highest frequency of interest. Our Nyquist frequency was 2,000 Hz and we analyzed up to 640 Hz. We divided the frequency spectrum of the Edi into octave bands from 10 to 640 Hz, with center frequencies of 14,28, 56, 113, 226, and 452 Hz. The center frequency is defined as fo = m, where fg = 2fi. We expressed the amplitude per frequency band as a percentage of the total power for the file section, except in Figs. 2 and 3 where we transformed it into decibels to compare the normal frequency spectrum of the newborn diaphragm, and its changes with fatigue, with the spectrum observed by Kadefors et al. (8) in the adult biceps brachii muscle. In our study the high-to-low frequency ratio is obtained by dividing the percentage of the total power between 160 and 640 Hz by the percentage of the total power between 10 and 40 Hz, i.e.,
1 100
FREQUENCY
there was no distortion immediately following newborns.
(Hz)
1 1000
of rib cage in 6 normal infants and in breaths discontinuation of IMV or CPAP in 3 other
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (130.070.008.131) on November 11, 2018. Copyright © 1979 American Physiological Society. All rights reserved.
690
MULLER
l NO
ET
AL.
DISTORTION
o MARKED
DISTORTION
lb
IhO FREQUENCY
lOb0
(HZ)
FIG. 3. Changes in diaphragmatic EMG power spectrum with distortion of rib cagein infant 2. Points represent mean and SE of power of different frequency bands expressed as dB. Breaths occurring during
maximal changes increase
high-to-low frequency ratio
requiring artificial ventilation, but did stabilize once back on CPAP. In these infants, the breaths immediately following the cessation of IMV or CPAP were used as control.
power between 160 and 640 Hz = power between 10 and 40 Hz The amplitude of the Edi is expressed as a ratio of the mean power observed in a particular section over the mean power during apnea. This permits a better correlation of the different total powers observed in the various patients. In the six normal newborns, a period with minimal or no distortion of the rib cage was used as control and the other sections compared to it. It should be noted that we use the word distortion in a different sense from Konno and Mead (13). They call departure from the relaxation characteristics of the whole rib cage, distortion. We define distortion as independent motion of the upper and lower rib cage. We quantified the degree of distortion by measuring the maximal point of inward motion of the upper rib cage during inspiration. Inspiration was considered to start with the initial outward movement of the abdomen and end when this reached its peak. Three other infants were studied at a point where the decision to wean them off the ventilator had been made on clinical grounds. Intermittent mandatory ventilation (IMV) was discontinued, but a positive end-expiratory pressure (CPAP) of 3-5 cmHz0 was maintained. The IMV rates were, respectively, 7,7, and 36 beats/min prior to weaning. Within 35 min, infant 7 had frequent apneic spells, “looked sick,” and had to be ventilated again. Infant 8 had a severe apneic spell that necessitated return to artificial ventilation 4 min later. Infant 9, who had been on an IMV of 36 beats/min, did well on CPAP alone and was maintained on it. Twenty-four hours later, when he was taken off CPAP, he had severe apneic spells
distortion of chest wall in REM sleep, show characteristic that occur with muscle fatigue: a fall in high frequency and an in low frequency part of spectrum.
RESULTS
NormaL infants. Figure 2 shows that the frequency spectrum analysis of the Edi during undistorted breaths is similar to the adult biceps brachii muscle during an isotonic-isometric contraction as reported by Kadefors et al. (8). Kadefors and co-workers used octave band filters with center frequencies that were slightly different from the ones we used. In all infants the nondistorted control breaths were observed in non-REM sleep; in the infants old enough to have a clear-cut quiet sleep the undistorted breaths occurred during that period. The degree of distortion observed in REM sleep was variable, but distortion was always present when the infant was in REM sleep. The changes in the frequency spectrum when there is marked distortion of the rib cage can be seen in Fig. 3. There is a fall in the high-frequency (160-640 Hz) power and an increase in the low-frequency (lo-40 Hz) power. When we plotted the change in the high-to-low frequency ratio, i.e., amplitude of the signal between 160 and 640 Hz over the amplitude between 10 and 40 Hz, against distortion of the rib cage we observed a statistically significant correlation in all infants (all r > 0.60, P < 0.01). Figure 4 shows the correlation observed in infant 1; Table 1 shows the summary of the results in the six normal infants. In three of these infants we also analyzed separately the first and the last 100 ms of file sections during breaths with and without marked distortion of the rib cage. There
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (130.070.008.131) on November 11, 2018. Copyright © 1979 American Physiological Society. All rights reserved.
DIAYHRAGMATIC
MUSCLE
FATIGUE
IN
THE
691
NEWBORN
was not a significant difference between the frequency spectrum of the first and second half of a file section in any of these infants (P > 0.10 in all the frequency bands). In the six normal infants we also observed a statistically significant correlation between the power of the Edi and the fall in the high-to-low frequency ratio (all r > 0.65, P < 0.05). Figure 5 shows that mean correlation and 06
.5-
r=
1
0.90
*
I 6
I 3
0
I 9
DISTORTION
1 12
I 15
[Arbitrary
FIG. 4. Correlation between distortion to-low frequency ratio in infant 1. Each breaths. Lines represent mean correlation
I 18
I 21
1 24
Units]
of chest wall and fall in highdot represents mean of 5-20 and 95% confidence limits.
n
No.
1 2 3 4 5 6 n Represents including 5-20
the number breaths.
P
r
15 12 18 10 10 10
-0.90 -0.85 -0.61 -0.82 -0.90 -0.81 of periods
that
