Pediatric Pulmonology 1052-56 (1991)
Diagnostic Methods -
Effect of Infant Position on Breath Amplitude Measured by Transthoracic Impedance and Strain Gauges Terry M. Baird, MD,' and Michael R. Neuman, MD,PhD2 Summary. Continuous monitoring of respiration by transthoracic electrical impedance gives a signal that has certain not well understood irregularities. Among them is a change in the arnplitude of the signal when there is no apparent change in the infant's tidal respiration. One factor that could hypothetically account for alterations of the impedance signal is a change in current path through the thorax secondary to a change in body position. To test this hypothesis we have studied the relationships between breath amplitude measured by transthoracic impedance, one strain gauge on the chest and one on the abdomen, and tidal volume by integrated flow in four body positions. Median breath amplitude was found to vary significantly with body position according to the measuring device. The median impedance breath amplitude increased by 27% in the supine position compared with the prone position, with no associated change in tidal volume. Differences in the strain gauge signal amplitude for these positions were not statistically significant. Correlation between breath amplitude measured by impedance changes and tidal volume was minimal (r = 0.114). These results indicate that infant position affects impedance breath amplitude independently of changes in tidal volume. Pediatr Pulmonol 1991; 10:52-56. Key words: Pneumotachographictidal volume; prone, supine, left and right lateral positions, correlations between methods, positions; differences in mean median breath amplitudes.
INTRODUCTION The most widely used method for continuous monitoring of infant breathing is transthoracic electrical impedance. Despite the general acceptance and widespread use of this method, it is known to have significant limitations.' One of these limitations is the common observation that the impedance signal may change when there is no apparent change in the monitored infant's breathing. This is particularly true for alterations in the amplitude of the signal. The latter (i.e., the height of the wave from peak to trough), corresponds to the depth of breathing, and it has been shown that impedance breath amplitude and tidal volume are However, changes in impedance breath amplitude do not always correspond to changes in tidal volume. The factors responsible for these alterations in the impedance signal are not clearly defined. Infant body position is known to affect the ~ i g n a l but , ~ this phenomenon has not been systematically studied. To investigate the relationship of infant position to impedance breath amplitude, we have measured breath amplitude with three different sensors used in infant monitoring: transthoracic impedance and
0 1991 Wiley-Liss, Inc.
two strain gauges, one on the chest and one on the abdomen. A pneumotachograph attached to a sealed face mask was used to provide a reference signal. MATERIALS AND METHODS
Twenty-four full-term, healthy infants were studied. Mean birth weight was 3,450 5 474 (SD) g, and all infants were between 24 and 72 h of age at the time of From the Departments of Pediatrics' and Obstetrics and Gynecology,' MetroHealth Medical Center, Case Western Reserve University School of Medicine, Cleveland, Ohio. Received June 18, 1990; (revision) accepted for publication August 16, 1990. This study was supported in part by U.S. PHS grant MOI-RR00210. Presented in abstract form at the Seventh Conference on Apnca of Infancy, Rancho Mirage, CA, January 26-28, 1989. Address correspondence and reprint requests to Dr. T. M. Baird, Division of Neonatology, MetroHealth Medical Center, 3395 Scranton Road, Cleveland, OH 44109.
Impedance and Strain Gauge Breath Amplitude
study. Informed consent was obtained in writing from a parent of all infants. Breathing was recorded by transthoracic impedance and two strain gauges, one on the chest and one on the abdomen. Additionally, 13 of these infants were studied with a pneumotachograph utilizing a sealed face mask. Infants were monitored for approximately 2 h during quiet sleep. Each infant was placed in four positions (left, prone, right, supine), with two trial periods of at least 5 min in each position. The order of positions was randomly assigned. Twenty-five breaths spread equally over time were chosen from each trial for analysis. Irregular breaths such as sighs were excluded. The amplitude of these selected breaths was measured from the recorded chart and entered into a computer for analysis, using a digitizing pad (Hipad; Houston Instruments, Austin, TX). Transthoracic impedance5 was monitored using silversilver chloride electrodes (Sentry Silver Sircuit Premie; Sentry Medical Products, Irvine, CA) placed on the left and right upper thorax, at the midaxillary line. An Edentec 2000W monitor (Edentec, Eden Prairie, MN) was used to record impedance. Mercury in silicon rubber strain gauges (Parks Medical Electronics Inc., Aloha, OR) were used to record movement of the chest and abdomen; this device measures displacement as a change in its electrical resistance in response to stretching.6 The chest strain gauge spanned the midaxillary lines at the level of the midsternum, and the abdominal strain gauge was placed just above the umbilicus spanning about half the circumference of the abdomen. Strain gauges were secured to the infant by clear tape (Transpore; 3M, St. Paul, MN). A gas flow meter (pneumotachograph) was constructed using a conventional respiratory face mask fitted with a 200 mesh stainless steel screen (Cleveland Wire and Cloth, Cleveland, OH) with a port proximal to the screen. Pressure at this port was measured using a Validyne differential pressure transducer (model MP 4514; Validyne Engineering Corp., Northridge, CA). Volume was determined by integrating the flow signal using a Could Biophysical Integrator Amplifier (model 134615-70; Could Inc., Cleveland, OH). This system was calibrated by forcing a known volume of air from a syringe in line with the face mask past the screen. The response of this system was linear over the range of tidal volumes measured. All signals were recorded on a Hewlet-Packard 7754B strip chart recorder. (Spss Inc., Statistical analysis utilized SPSS/PC Chicago, IL). The nonparametric median test and the Mann-Whitney U test were used to determine significant differences between groups.
+
RESULTS Tidal volume measured by the pneumotachograph in 13 patients (1,300 breaths) ranged from 1.3 to 13.8 mL/
2
0
6
4
53
10
8
IMPEOANCE AWLITUDE IohmSl
T I D A L VDLUME
L----T----, 10
20
I
I
30
40
TIDAL VOLUME I m l . I
Fig. 1. Histograms of impedance breath amplitudes (top) and tidal volumes (bottom). Each histogram shows the distribution of 1,300 breaths from 13 patients in all positions.
kg; the median breath was 7.65 mL/kg, with an interquartile range from 4.39 to 7.7 1 mL/kg . The corresponding impedance breath amplitude signals ranged from 0.2 SZ to 10.17 R, with a median of 1.92 SZ and an interquartile range from 1.28 to 2.72 When impedance breath amplitude was measured without using the face mask (4,800 breaths), the median breath amplitude was with an interquartile range from 0.72 to 1.68 1.11 The distribution of breath amplitudes is affected by the device used to measure it; with impedance pneumography the distribution is skewed toward lower volume breaths, with a “tail” of large volume breaths, whereas the other sensors show breath amplitudes more normally distributed about the means. Figure 1 compares the distributions of 1,300 breaths (13 patients) as measured by impedance amplitude and tidal volume. Correlation between transthoracic impedance and tidal volume on a breath-to-breath basis was weak, with linear regression over all breaths yielding r = 0.114. Similarly, correlation between measurements by each of the two strain gauges and the tidal volume was poor. The correlation coefficient for the chest strain gauge was r =
a.
a,
a.
54
Baird and Neuman
TABLE 1-Linear Regressions of Breath Amplitude as Determined by Various Sensors Compared to Tidal Volume in 13 Infants (Supine position) Patient Impedance Slope (WmL) r
Chest strain gauge Slope (mmimL) r Abdominal strain gauge Slope (mm/mL) r
1
2
0.098 0.76
0.012 0.56
0.060 0.39
0.021 0.39
0.138 0.66
0.037 0.32
3 0.106 0.69 -0.003 -0.06 0.216 0.87
4
5
6
7
8
9
10
11
0.044 0.78
0.046 0.53
0.219 0.74
0.101 0.73
0.072 0.64
0.203 0.58
0.080 0.54
0.065 0.81
0.170 0.86
0.098 0.89
0.019 0.67
0.010 0.33
0.089 0.58
0.096 0.86
0.071 0.74
0.057 0.29
0.027 0.30
0.005 0.11
-0.014 -0.14
0.072 0.55
0.110
0.080 0.70
0.288 0.79
0.156 0.63
0.199 0.66
0.189
0.185 0.78
0.070 0.73
0.102 0.87
0.L63 0.78
0.63
0.126 and for the abdominal strain gauge r = 0.167. Examining individual trials, however, the correlation between impedance breath amplitude and tidal volume was occasionally very high. Individual correlation coefficients ranged from 0 to 0.93. Table 1 lists r values for all trials in the supine position. The abdominal strain gauge most consistently gave high correlation with tidal volume in all positions; the r values for individual trials (abdominal strain gauge only) ranged from 0.22 to 0.91; in half the trials they were greater than 0.67 and in 13% (7/52) greater than 0.80. Breath amplitude was found to differ with position for the transthoracic impedance and the chest strain gauge signals but not for the abdominal strain gauge and pneumotachographic tidal volume. Transthoracic impedance and the chest strain gauge showed decreased breath amplitude in the left and prone positions and the maximum in the supine (Table 2). (Median breath amplitude was used to describe an individual infant’s breaths, and the mean of these median breaths was used to describe the total sample of all infants’ breaths.) Figure 2 shows these data in box plots,7 with both examples of the strain gauge data plotted using the same scale for comparison. Breath amplitude by position measured without a face mask in 24 infants is shown in Table 3 . DISCUSSION Transthoracic electrical impedance is by far the most frequently applied technique for noninvasive monitoring of infants’ respiration. Our study demonstrates some of the limitations of this technique and compares it with a less commonly used method for measuring respiration, the liquid metal strain gauge. Pneumotachography, measuring air flow and volume, is widely accepted as a reference standard method, although it is not appropriate for routine monitoring. As seen from the data, none of the three signals from the impedance device or the chest or abdominal strain gauge correlated consistently with tidal volume.
0.52
12
13
Our study design included trials in which the infants breathed through a face mask and trials in which breath amplitudes were recorded only by impedance and strain gauge signals. This was done because using a face mask is known to alter breathing patterns in newborns.’ Specifically, it has been shown that breathing into a face mask results in increased depth of breathing (increased impedance amplitude) and decreased frequency. In our study, depth of respiration increased by 73% when the mask was used; this increase is consistent with but larger than previously reported.* The finding that impedance breath amplitude does not correlate with tidal volume is generally accepted but has not been previously investigated in detail.’ Although under controlled conditions impedance breath amplitude could be used to approximate tidal volume,233our results indicate that this relationship does not generally hold in infants. We have shown that impedance breath amplitude significantly varied with repeated measurements of breathing in the same infant and that impedance breath amplitude systematically varied with the position of the infant. Our data also demonstrate that recording breath amplitude from an abdominal strain gauge correlates with air flow as accurately as impedance pneumography. This finding is consistent with previous reports comparing these two devices when breathing rate is monit~red.’”~ The difference between measuring breaths and monitoring breathing should be stressed. In monitoring breathing, the question is whether or not a breath is taken; to establish a quantitative relationship between the size (tidal volume) of the breath and the breath amplitude as measured by the monitoring device is desirable but not essential. The problem encountered in monitoring is that of recording every breath regardless of any particular breath’s characteristics. So-called apnea monitors frequently accomplish this by various signal processing techniques that “decide” if a particular wave form is or is not a breath.” Monitoring is therefore dependent on accurate measurement of breaths, especially in clinical
Impedance and Strain Gauge Breath Amplitude TABLE 2-Mean
Median Breath Amplitude by Position in 13 Infants
(a)*
Impedance Chest strain gauge (mm)* Abdominal strain gauge (mm) Pneumotachograph (mL) *P
Left
Prone
Right
Supine
1.75 ? 0.86 0.97 t 0.64 5.31 t 1.65 20.6 ? 6.50
1.76 2 0.02 0.74 2 0.74 4.28 ? 1.47 21.0 t 7.06
2.34 f 1.06 1.44 k 0.81 6.19 t 1.82 19.3 ? 5.80
2.42 ? 1.25 1.52 t 0.84 4.55 f 1.63 19.5 t 6.03
< 0.05 by median test comparing all four positions. THORACIC IMPEDANCE
TIDAL VOLUME
i6
n
e
55
T
i 1
U U
i6
T
I
T
Q0 b I
i6
I
I
D S
U
m -
0
( v -
1I
I
d
-
#
1
L
1
P
R
S
L
s
CHEST STRAIN GAUGE
P
R
S
ABDOMINAL STRAIN GAUGE
T
Q 06 T
*
I
T
I I
#
I
E 3 1
NL 0
L
P
R
S
P
L
R
S
Fig. 2. Box plots showing the distribution of median breath amplitudes according to the device used to measure breaths. Each box represents the distribution of 13 median breaths (25 breathshubject) by position. L, left; P,prone; R, right; S, supine. (Asterisks represent outliers.) TABLE %Mean
Median Breath Amplitude by Position in 24 Infants
Impedance (f2) Chest strain gauge (mm) Abdominal strain gauge (mm)
Left
Prone
Right
Supine
1.40 t 0.85 1.13 t 1.06 3.88 t 1.60
0.98 t 0.55" 0.66 5 0.62* 2.97 1.68
1.19 2 0.57 0.86 t 0.59 3.43 t 1.75
1.46 ? 0.70" 1 . 5 4 t 1.15* 3.62 t 1.65
*
*P < 0.05 by Mann-Whitney U test comparing prone and supine
settings where a significant amount of irregular breathing may occur such as sighing andlor hypoventilation. Movement of the chest wall (as demonstrated by chest strain gauge) and changes in impedance are affected by position. For the chest strain gauge, one possible explanation for this is that when the infant is prone the weight of the infant's body is on the strain gauge, holding it in
a relatively fixed position between the infant and the mattress. However, this does not explain why the impedance signal amplitude also decreases in the prone position; we have shown that tidal volume is unchanged in this position. Others have shown that tidal volume is increased in the prone position.'* Since the impedance signal is thought to result from the summation of multi-
56
Baird and Neuman
ple current pathways through the chest, it is possible that a change in body position (with resultant changes in the anatomic relations of the thoracic contents), would give a different impedance signal (both baseline impedance and breathing-related impedance wave forms), Although we have demonstrated that infant position has an effect on the amplitude of a transthoracic impedance signal, it is not the only factor affecting this signal. This is evidenced by the variance in signal between patients and in the same patient at different times. We conclude that multiple factors affect the amplitude of the transthoracic impedance signal independently of tidal volume and that infant position affects the signal in a systematic way.
ACKNOWLEDGMENTS
We thank Boris Makovos and the nurses of the Perinatal Clinical Research Center for their assistance.
REFERENCES 1. Neuman MR. Apnea Monitoring: Technical Aspects. In: Infantile Apnea and Home Monitoring. Bethesda, MD: NIH Publication No. 87-2905, 1987.
2. Hamilton LH, Beard JD, Kory RC. Impedance measurement of tidal volume and ventilation. J Appl Physiol. 1965; 20565-568. 3. Allison RD, Holmes EL, Nyboer J. Volumetric dynamics of respiration as measured by electrical impedance plethysmography . J Appl Physiol. 1964; 19:166-173. 4. Bancalari E. Pulmonary Function Testing and Other Diagnostic Laboratory Procedures. In: Thibeault DW, Gregory GA, eds. Neonatal Pulmonary Care, 2nd ed. Norwalk, CT: Appleton-Century-Crofts, 1986:205. 5. Olsson T, Victorin L. Transthoracic irnpcdance: Theoretical considerations and technical approach. Acta Pediatr Scand. 1970; 207[Suppll:1-27. 6. Whitney RJ. The measurement of volume changes in human limbs. J Physiol. 1953; 121:l-27. 7. Williamson DF, Parker RA, Kendrick JS. The boxplot: A simple visual method to interpret data. Ann Intern Med. 1989; 110: 916-921. 8. Fleming PJ, Levine MR, Goncalves A. Changes in respiratory pattern resulting from the use of a facemask to record respiration in newborn infants. Pediatr Res. 1982; 16:1031-1034. 9. Yount JE, Newcomb J, Hammill D, et al. Thoracic impedanceIs there a better Signal of Respiratory Effort? Fifth Conference on Apnea of Infancy Rancho Mirage, CA, 1987 (abstract). 10. Polgar G . Comparison of methods for recording respiration in newborn infants. Pediatrics. 1965; 36:861-868. I . Neuman MR. Optimal Detection of Respiration and Apnea by Infant Monitors. In: Medical Technology for the Neonate. Arlington, VA: Association for the Advancement of Medical Instrumentation, 1984:49-54. 2. Wagaman MJ, Shutack JG, Moomjian AS, ei al. Improved oxygenation and lung compliance with prone positioning of neonates. J Pediatr. 1979; 94:787-791.
Diagnostic Methods -
Effect of Infant Position on Breath Amplitude Measured by Transthoracic Impedance and Strain Gauges Terry M. Baird, MD,' and Michael R. Neuman, MD,PhD2 Summary. Continuous monitoring of respiration by transthoracic electrical impedance gives a signal that has certain not well understood irregularities. Among them is a change in the arnplitude of the signal when there is no apparent change in the infant's tidal respiration. One factor that could hypothetically account for alterations of the impedance signal is a change in current path through the thorax secondary to a change in body position. To test this hypothesis we have studied the relationships between breath amplitude measured by transthoracic impedance, one strain gauge on the chest and one on the abdomen, and tidal volume by integrated flow in four body positions. Median breath amplitude was found to vary significantly with body position according to the measuring device. The median impedance breath amplitude increased by 27% in the supine position compared with the prone position, with no associated change in tidal volume. Differences in the strain gauge signal amplitude for these positions were not statistically significant. Correlation between breath amplitude measured by impedance changes and tidal volume was minimal (r = 0.114). These results indicate that infant position affects impedance breath amplitude independently of changes in tidal volume. Pediatr Pulmonol 1991; 10:52-56. Key words: Pneumotachographictidal volume; prone, supine, left and right lateral positions, correlations between methods, positions; differences in mean median breath amplitudes.
INTRODUCTION The most widely used method for continuous monitoring of infant breathing is transthoracic electrical impedance. Despite the general acceptance and widespread use of this method, it is known to have significant limitations.' One of these limitations is the common observation that the impedance signal may change when there is no apparent change in the monitored infant's breathing. This is particularly true for alterations in the amplitude of the signal. The latter (i.e., the height of the wave from peak to trough), corresponds to the depth of breathing, and it has been shown that impedance breath amplitude and tidal volume are However, changes in impedance breath amplitude do not always correspond to changes in tidal volume. The factors responsible for these alterations in the impedance signal are not clearly defined. Infant body position is known to affect the ~ i g n a l but , ~ this phenomenon has not been systematically studied. To investigate the relationship of infant position to impedance breath amplitude, we have measured breath amplitude with three different sensors used in infant monitoring: transthoracic impedance and
0 1991 Wiley-Liss, Inc.
two strain gauges, one on the chest and one on the abdomen. A pneumotachograph attached to a sealed face mask was used to provide a reference signal. MATERIALS AND METHODS
Twenty-four full-term, healthy infants were studied. Mean birth weight was 3,450 5 474 (SD) g, and all infants were between 24 and 72 h of age at the time of From the Departments of Pediatrics' and Obstetrics and Gynecology,' MetroHealth Medical Center, Case Western Reserve University School of Medicine, Cleveland, Ohio. Received June 18, 1990; (revision) accepted for publication August 16, 1990. This study was supported in part by U.S. PHS grant MOI-RR00210. Presented in abstract form at the Seventh Conference on Apnca of Infancy, Rancho Mirage, CA, January 26-28, 1989. Address correspondence and reprint requests to Dr. T. M. Baird, Division of Neonatology, MetroHealth Medical Center, 3395 Scranton Road, Cleveland, OH 44109.
Impedance and Strain Gauge Breath Amplitude
study. Informed consent was obtained in writing from a parent of all infants. Breathing was recorded by transthoracic impedance and two strain gauges, one on the chest and one on the abdomen. Additionally, 13 of these infants were studied with a pneumotachograph utilizing a sealed face mask. Infants were monitored for approximately 2 h during quiet sleep. Each infant was placed in four positions (left, prone, right, supine), with two trial periods of at least 5 min in each position. The order of positions was randomly assigned. Twenty-five breaths spread equally over time were chosen from each trial for analysis. Irregular breaths such as sighs were excluded. The amplitude of these selected breaths was measured from the recorded chart and entered into a computer for analysis, using a digitizing pad (Hipad; Houston Instruments, Austin, TX). Transthoracic impedance5 was monitored using silversilver chloride electrodes (Sentry Silver Sircuit Premie; Sentry Medical Products, Irvine, CA) placed on the left and right upper thorax, at the midaxillary line. An Edentec 2000W monitor (Edentec, Eden Prairie, MN) was used to record impedance. Mercury in silicon rubber strain gauges (Parks Medical Electronics Inc., Aloha, OR) were used to record movement of the chest and abdomen; this device measures displacement as a change in its electrical resistance in response to stretching.6 The chest strain gauge spanned the midaxillary lines at the level of the midsternum, and the abdominal strain gauge was placed just above the umbilicus spanning about half the circumference of the abdomen. Strain gauges were secured to the infant by clear tape (Transpore; 3M, St. Paul, MN). A gas flow meter (pneumotachograph) was constructed using a conventional respiratory face mask fitted with a 200 mesh stainless steel screen (Cleveland Wire and Cloth, Cleveland, OH) with a port proximal to the screen. Pressure at this port was measured using a Validyne differential pressure transducer (model MP 4514; Validyne Engineering Corp., Northridge, CA). Volume was determined by integrating the flow signal using a Could Biophysical Integrator Amplifier (model 134615-70; Could Inc., Cleveland, OH). This system was calibrated by forcing a known volume of air from a syringe in line with the face mask past the screen. The response of this system was linear over the range of tidal volumes measured. All signals were recorded on a Hewlet-Packard 7754B strip chart recorder. (Spss Inc., Statistical analysis utilized SPSS/PC Chicago, IL). The nonparametric median test and the Mann-Whitney U test were used to determine significant differences between groups.
+
RESULTS Tidal volume measured by the pneumotachograph in 13 patients (1,300 breaths) ranged from 1.3 to 13.8 mL/
2
0
6
4
53
10
8
IMPEOANCE AWLITUDE IohmSl
T I D A L VDLUME
L----T----, 10
20
I
I
30
40
TIDAL VOLUME I m l . I
Fig. 1. Histograms of impedance breath amplitudes (top) and tidal volumes (bottom). Each histogram shows the distribution of 1,300 breaths from 13 patients in all positions.
kg; the median breath was 7.65 mL/kg, with an interquartile range from 4.39 to 7.7 1 mL/kg . The corresponding impedance breath amplitude signals ranged from 0.2 SZ to 10.17 R, with a median of 1.92 SZ and an interquartile range from 1.28 to 2.72 When impedance breath amplitude was measured without using the face mask (4,800 breaths), the median breath amplitude was with an interquartile range from 0.72 to 1.68 1.11 The distribution of breath amplitudes is affected by the device used to measure it; with impedance pneumography the distribution is skewed toward lower volume breaths, with a “tail” of large volume breaths, whereas the other sensors show breath amplitudes more normally distributed about the means. Figure 1 compares the distributions of 1,300 breaths (13 patients) as measured by impedance amplitude and tidal volume. Correlation between transthoracic impedance and tidal volume on a breath-to-breath basis was weak, with linear regression over all breaths yielding r = 0.114. Similarly, correlation between measurements by each of the two strain gauges and the tidal volume was poor. The correlation coefficient for the chest strain gauge was r =
a.
a,
a.
54
Baird and Neuman
TABLE 1-Linear Regressions of Breath Amplitude as Determined by Various Sensors Compared to Tidal Volume in 13 Infants (Supine position) Patient Impedance Slope (WmL) r
Chest strain gauge Slope (mmimL) r Abdominal strain gauge Slope (mm/mL) r
1
2
0.098 0.76
0.012 0.56
0.060 0.39
0.021 0.39
0.138 0.66
0.037 0.32
3 0.106 0.69 -0.003 -0.06 0.216 0.87
4
5
6
7
8
9
10
11
0.044 0.78
0.046 0.53
0.219 0.74
0.101 0.73
0.072 0.64
0.203 0.58
0.080 0.54
0.065 0.81
0.170 0.86
0.098 0.89
0.019 0.67
0.010 0.33
0.089 0.58
0.096 0.86
0.071 0.74
0.057 0.29
0.027 0.30
0.005 0.11
-0.014 -0.14
0.072 0.55
0.110
0.080 0.70
0.288 0.79
0.156 0.63
0.199 0.66
0.189
0.185 0.78
0.070 0.73
0.102 0.87
0.L63 0.78
0.63
0.126 and for the abdominal strain gauge r = 0.167. Examining individual trials, however, the correlation between impedance breath amplitude and tidal volume was occasionally very high. Individual correlation coefficients ranged from 0 to 0.93. Table 1 lists r values for all trials in the supine position. The abdominal strain gauge most consistently gave high correlation with tidal volume in all positions; the r values for individual trials (abdominal strain gauge only) ranged from 0.22 to 0.91; in half the trials they were greater than 0.67 and in 13% (7/52) greater than 0.80. Breath amplitude was found to differ with position for the transthoracic impedance and the chest strain gauge signals but not for the abdominal strain gauge and pneumotachographic tidal volume. Transthoracic impedance and the chest strain gauge showed decreased breath amplitude in the left and prone positions and the maximum in the supine (Table 2). (Median breath amplitude was used to describe an individual infant’s breaths, and the mean of these median breaths was used to describe the total sample of all infants’ breaths.) Figure 2 shows these data in box plots,7 with both examples of the strain gauge data plotted using the same scale for comparison. Breath amplitude by position measured without a face mask in 24 infants is shown in Table 3 . DISCUSSION Transthoracic electrical impedance is by far the most frequently applied technique for noninvasive monitoring of infants’ respiration. Our study demonstrates some of the limitations of this technique and compares it with a less commonly used method for measuring respiration, the liquid metal strain gauge. Pneumotachography, measuring air flow and volume, is widely accepted as a reference standard method, although it is not appropriate for routine monitoring. As seen from the data, none of the three signals from the impedance device or the chest or abdominal strain gauge correlated consistently with tidal volume.
0.52
12
13
Our study design included trials in which the infants breathed through a face mask and trials in which breath amplitudes were recorded only by impedance and strain gauge signals. This was done because using a face mask is known to alter breathing patterns in newborns.’ Specifically, it has been shown that breathing into a face mask results in increased depth of breathing (increased impedance amplitude) and decreased frequency. In our study, depth of respiration increased by 73% when the mask was used; this increase is consistent with but larger than previously reported.* The finding that impedance breath amplitude does not correlate with tidal volume is generally accepted but has not been previously investigated in detail.’ Although under controlled conditions impedance breath amplitude could be used to approximate tidal volume,233our results indicate that this relationship does not generally hold in infants. We have shown that impedance breath amplitude significantly varied with repeated measurements of breathing in the same infant and that impedance breath amplitude systematically varied with the position of the infant. Our data also demonstrate that recording breath amplitude from an abdominal strain gauge correlates with air flow as accurately as impedance pneumography. This finding is consistent with previous reports comparing these two devices when breathing rate is monit~red.’”~ The difference between measuring breaths and monitoring breathing should be stressed. In monitoring breathing, the question is whether or not a breath is taken; to establish a quantitative relationship between the size (tidal volume) of the breath and the breath amplitude as measured by the monitoring device is desirable but not essential. The problem encountered in monitoring is that of recording every breath regardless of any particular breath’s characteristics. So-called apnea monitors frequently accomplish this by various signal processing techniques that “decide” if a particular wave form is or is not a breath.” Monitoring is therefore dependent on accurate measurement of breaths, especially in clinical
Impedance and Strain Gauge Breath Amplitude TABLE 2-Mean
Median Breath Amplitude by Position in 13 Infants
(a)*
Impedance Chest strain gauge (mm)* Abdominal strain gauge (mm) Pneumotachograph (mL) *P
Left
Prone
Right
Supine
1.75 ? 0.86 0.97 t 0.64 5.31 t 1.65 20.6 ? 6.50
1.76 2 0.02 0.74 2 0.74 4.28 ? 1.47 21.0 t 7.06
2.34 f 1.06 1.44 k 0.81 6.19 t 1.82 19.3 ? 5.80
2.42 ? 1.25 1.52 t 0.84 4.55 f 1.63 19.5 t 6.03
< 0.05 by median test comparing all four positions. THORACIC IMPEDANCE
TIDAL VOLUME
i6
n
e
55
T
i 1
U U
i6
T
I
T
Q0 b I
i6
I
I
D S
U
m -
0
( v -
1I
I
d
-
#
1
L
1
P
R
S
L
s
CHEST STRAIN GAUGE
P
R
S
ABDOMINAL STRAIN GAUGE
T
Q 06 T
*
I
T
I I
#
I
E 3 1
NL 0
L
P
R
S
P
L
R
S
Fig. 2. Box plots showing the distribution of median breath amplitudes according to the device used to measure breaths. Each box represents the distribution of 13 median breaths (25 breathshubject) by position. L, left; P,prone; R, right; S, supine. (Asterisks represent outliers.) TABLE %Mean
Median Breath Amplitude by Position in 24 Infants
Impedance (f2) Chest strain gauge (mm) Abdominal strain gauge (mm)
Left
Prone
Right
Supine
1.40 t 0.85 1.13 t 1.06 3.88 t 1.60
0.98 t 0.55" 0.66 5 0.62* 2.97 1.68
1.19 2 0.57 0.86 t 0.59 3.43 t 1.75
1.46 ? 0.70" 1 . 5 4 t 1.15* 3.62 t 1.65
*
*P < 0.05 by Mann-Whitney U test comparing prone and supine
settings where a significant amount of irregular breathing may occur such as sighing andlor hypoventilation. Movement of the chest wall (as demonstrated by chest strain gauge) and changes in impedance are affected by position. For the chest strain gauge, one possible explanation for this is that when the infant is prone the weight of the infant's body is on the strain gauge, holding it in
a relatively fixed position between the infant and the mattress. However, this does not explain why the impedance signal amplitude also decreases in the prone position; we have shown that tidal volume is unchanged in this position. Others have shown that tidal volume is increased in the prone position.'* Since the impedance signal is thought to result from the summation of multi-
56
Baird and Neuman
ple current pathways through the chest, it is possible that a change in body position (with resultant changes in the anatomic relations of the thoracic contents), would give a different impedance signal (both baseline impedance and breathing-related impedance wave forms), Although we have demonstrated that infant position has an effect on the amplitude of a transthoracic impedance signal, it is not the only factor affecting this signal. This is evidenced by the variance in signal between patients and in the same patient at different times. We conclude that multiple factors affect the amplitude of the transthoracic impedance signal independently of tidal volume and that infant position affects the signal in a systematic way.
ACKNOWLEDGMENTS
We thank Boris Makovos and the nurses of the Perinatal Clinical Research Center for their assistance.
REFERENCES 1. Neuman MR. Apnea Monitoring: Technical Aspects. In: Infantile Apnea and Home Monitoring. Bethesda, MD: NIH Publication No. 87-2905, 1987.
2. Hamilton LH, Beard JD, Kory RC. Impedance measurement of tidal volume and ventilation. J Appl Physiol. 1965; 20565-568. 3. Allison RD, Holmes EL, Nyboer J. Volumetric dynamics of respiration as measured by electrical impedance plethysmography . J Appl Physiol. 1964; 19:166-173. 4. Bancalari E. Pulmonary Function Testing and Other Diagnostic Laboratory Procedures. In: Thibeault DW, Gregory GA, eds. Neonatal Pulmonary Care, 2nd ed. Norwalk, CT: Appleton-Century-Crofts, 1986:205. 5. Olsson T, Victorin L. Transthoracic irnpcdance: Theoretical considerations and technical approach. Acta Pediatr Scand. 1970; 207[Suppll:1-27. 6. Whitney RJ. The measurement of volume changes in human limbs. J Physiol. 1953; 121:l-27. 7. Williamson DF, Parker RA, Kendrick JS. The boxplot: A simple visual method to interpret data. Ann Intern Med. 1989; 110: 916-921. 8. Fleming PJ, Levine MR, Goncalves A. Changes in respiratory pattern resulting from the use of a facemask to record respiration in newborn infants. Pediatr Res. 1982; 16:1031-1034. 9. Yount JE, Newcomb J, Hammill D, et al. Thoracic impedanceIs there a better Signal of Respiratory Effort? Fifth Conference on Apnea of Infancy Rancho Mirage, CA, 1987 (abstract). 10. Polgar G . Comparison of methods for recording respiration in newborn infants. Pediatrics. 1965; 36:861-868. I . Neuman MR. Optimal Detection of Respiration and Apnea by Infant Monitors. In: Medical Technology for the Neonate. Arlington, VA: Association for the Advancement of Medical Instrumentation, 1984:49-54. 2. Wagaman MJ, Shutack JG, Moomjian AS, ei al. Improved oxygenation and lung compliance with prone positioning of neonates. J Pediatr. 1979; 94:787-791.
