Pediatric Pulmonology 21 :35-41 (1996)

Alvaro Gonzalez, MD, Luca Tortorolo, MD, Tilo Gerhardt, MD, Mario Rojas, MD, Ruth Everett, RN, and Eduardo Bancalari, MD

Summary. This study set out to describe the variability and assess the reproducibility of repeated pulmonary function measurements in ventilated preterm infants. We measured tidal volume (VT), lung compliance (CJ, and resistance (R,) in 16 infants (mean I SO: birthweight 1222 I 343 g) during spontaneous breathing and during mechanical ventilation, suppressing breathing efforts by mild hyperventilation. CL and RL were calculated from the equation of motion using linear regression analysis (LR), and by the Mead and Wittenberger method (MW). Flow and transpulmonary pressure were recorded for at least two consecutive periods, after which the esophageal tube was removed and replaced 1 hour later for a second set of recordings. The mean percent change (YO A) between the initial and the repeated measurements with their respective 95% confidence intervals were calculated. Reproducibility was assessed by the intraclass correlation coefficient (ICC) (total agreement = 1, good reproducibility 20.75). The mean Yo A between initial and repeat measurements during spontaneous breathing ranged from 11% to 14% for CL and Vh and from 22% to 32% for RL. The variation for R, was even higher when the analysis was done separately for the inspiratory and expiratory phase. CL and V, had good reproducibility (ICC >0.9), while RL was significantly less reproducible (ICC <0.75). Measurements obtained from mechanical breaths had less variability than from spontaneous breaths, ranging from 8% to 15% for CL and V , and from 13% to 21% for R,. Reproducibility assessed by the ICC was good for most measurements during mechanical breaths. The variability and reproducibility of measurements were similar for both methods of analysis during mechanical ventilation, but during spontaneous breathing variability was larger with the MW method than with LR analysis. We concluded that VT and Cl were reproducible during spontaneous and mechanical breathing. However, RL measurements were reproducible only during mechanical ventilation. The high variability of Rl in spontaneously breathing preterm infants may reduce the clinical usefulness of this measurement for individual patients. Pediatr Pulmonol. 1996; 21 :35-41. P 1996 Wiley-Liss. Inc

Key words: Pulmonary compliance, pulmonary resistance, pulmonary function testing, mechanical ventilation, Infants.

INTRODUCTION

The measurement of pulmonary mechanics in newborn infants has been used to describe and understand the pathophysiology of respiratory failure and has been used to document the beneficial effects of medications such as bronchodilators, surfactant, diuretics, and steroids.'" Computerized pulmonary function systems have made the measurements of lung mechanics easy and and the procedure is now being used in the management of individual neonates with respiratory failure. However, the experience of the trained investigator in recognizing artifacts and avoiding the many pitfalls associated with lung function measurements in the neonate has not been replaced by the c~mputer.'.~ The clinical usefulness of the measurements in documenting the response to therapeutic interventions in individual infants, and their reliability as diagnostic and prognostic tools, depends on the reproduc- ibility of the lung function measurements. 0 1996 Wiley-Liss, Inc.

A large inrrupatient variability of the measurements has been described in fullterm and large preterm infants,'.''but no systematic studies of the intrapatient variability of CL and R,, have been performed in small, ventilated preterm infants. Sleep state, position, relation to feeding, size of tidal volume, and degree of chestwall distortion influence the measurements of lung function and contribute to their

From the University of Miami Schml of Medicine, Department of Pediatrics, Division of Neonatology, Miami, Florida.

Received July 8, 1995; (revision) accepted October 21. 1995.

Supported by the University of Miami Project: New Born

Address correspondence and reprint requests to Tilo Gerhardt. M.D., University of Miami School of Medicine. Division of Neonatology, Department of Pediatrics ( R - I 3 I ) , P.O. Box 016960, Miami, FL 33101.

Alvaro Gonzaler is presently at the Universidad Cntolica de Chile

36 Gonzalez et al.

variability. 11-14 Information about the intrapatient variabil- ity of pulmonary function measurements is of clinical importance because a large variability in the test results would preclude their use in the management of individ- ual patients.

The purpose of this study was to describe the variability and assess the reproducibility of measurements of pulmo- nary mechanics by the method of Mead and Wittenberger (MW)15 and by linear regression analysis based on the equation of motion (LR)5373169'7 in a population of ventilated preterm infants who were recovering from respiratory failure.

SUBJECTS AND METHODS Subjects

Twenty clinically stable ventilated preterm infants (13 males) with birthweights between 700 and 1,800 g born at the University of MiamUJackson Memorial Medical Center between April 1992 and April 1993 were studied. The studies were done during the first week of life, after the infants had recovered from respiratory distress syn- drome (RDS) and were on low ventilatory settings close to extubation (Fi0, 10.4, mechanical ventilator rate 1 2 0 breathdmin, peak inspiratory pressure 1 2 0 cmH20). In- fants with congenital malformations, a patent ductus arte- riosus or sepsis, or with metabolic or electrolyte imbal- ances were excluded. The protocol was approved by the Committee for the Protection of Human Subjects at the University of Miami, and signed informed consent was obtained from the parents.

Pulmonary Mechanics Measurement Respiratory flow was measured with a hot wire ane-

mometer (Neonatal Volume Monitor, Bear NVM-1). The flow signal was calibrated using known flows measured with an electronic flowmeter (Gas Products model 2800, Matheson, Secaucus, NJ). Tidal volume was obtained by digital integration of the flow signal. Airway pressure was measured with a transducer (Statham Model ID P23XL, Statham Instruments, Oxnard, CA) from a side port of the endotracheal tube connector. Esophageal pressure was measured through a size 8.0 Fr water-filled feeding tube connected to a pressure transducer (Statham Model ID P23XL, Statham Instruments, Oxnard, CA). The tube was introduced initially into the infant's stomach and flushed, then was pulled back into the lower portion of the esopha- gus, as indicated by the appearance of negative pressure deflections during inspiration. To test for full transmission of pleural pressure to the esophageal catheter, a standard occlusion test was p e r f ~ r m e d . ~ ~ ~ ' ~ The infants were briefly disconnected from the ventilator, and the endotracheal tube connector was manually occluded at the end of expi- ration. Proper function of the esophageal tube was con-

firmed when airway and esophageal pressures changes were less than 10% different from each other. All trans- ducers were powered, and their output was amplified with Gould Transducer couplers (Gould, Cleveland, OH). Calibration was done with a water manometer. There was no phase difference between flow and pressure signals, and no overdamping or underdamping of the signals up to 10 Hz. All signals were recorded on a Gould stripchart recorder, but were also digitized and collected at 100 Hz/ channel and stored on a PC computer for later analysis (PC 386 based system).

Protocol The infants were placed in the supine position with

the head in the midline, the endotracheal tube (Em) was suctioned, and the anemometer was placed between the ETT and the ventilator circuit, at least half an hour before the measurements were started. Continuous feedings were suspended. When the infants were on bolus feeding, they received their last meal 1 hour prior to the test. We placed and flushed the esophageal tube just before the pressure recording. Care was taken to make the infants comfort- able, to wait until they fell asleep, and to perform the recordings during quiet sleep. Sedation was not given. The measurements were performed in the neonatal inten- sive care unit with the infants in their incubators, continu- ously monitoring heart rate and 0, saturation. Pulmonary mechanics were recorded during two different ventilatory settings: (a) on endotracheal continuous positive airway pressure (CPAP +4 cm H20) to measure the mechanics of only spontaneous breaths, and (b) on intermittent man- datory ventilation (IMV), suppressing breathing effort by mild hyperventilation at a rate of 30-40 breathdmin in order to analyze only mechanical breaths. A first set of measurements (first study) consisted of at least two consecutive periods of spontaneous breathing of 1 minute duration each and two periods of similar duration during the IMV mode. After this, the esophageal tube and the anemometer were removed, but were replaced 1 hour later to perform a second set of measurements (second study) in the same manner as described earlier.

The computer was programmed to calculate compli- ance and resistance by the MW method,15 and by LR analysis of the equation of motion: P = VIC + F * R.7,11 In this equation P equals transpulmonary pressure, F equals tidal flow, and V equals tidal volume. P, F, and V are variables that change throughout the respiratory cycle, and C and R are assumed to be constant values for compli- ance and resistance, respectively. For the LR method P/F was plotted against V/F; the slope of this relationship was calculated by linear regression analysis and equals l/C. Pulmonary resistance was calculated from the slope of the relationship P N vs. FN. Outliers on the regression line were eliminated by only including measurement

Pulmonary Function in Ventilated Infants 37

TABLE 1-Pulmonary Mechanics

Spontaneous breaths Mechanical breaths First study Second study First study Second study

CL LR 1.01 t 0.57 0.94 t 0.49 0.87 f 0.58 0.87 t 0.53 CL m 0.97 t 0.56 0.90 t 0.48 0.86 2 0.51 0.85 t 0.48 RL total,, 108 t 49 128 t 64 154 I 1 8 155 F 83 RL totalhlw 113 ? 52 126 2 65 155 t 74 157 2 88 RL inspiratory, 82 f 277 85 t 39t 132 ? 671 134 t 70t RL inspiratorym 85 t 377 97 t 46 130 t 64t 128 F 62 RL expiratory,, 192 t 86 204 2 121 193 t 102 200 2 124 RL expiratorym 203 f 93 226 t 147 219 t 99 229 C 163

6.9 f 3.1* VT 4.0 f 1.6* 4.8 t 1.8*

Results are means t SD. Units: CL, ml/cmH,O/kg; RL, cmH,O/L/sec; VT, ml/kg. LR, linear regression method; Mw, Mead & Wittenberger’s method. *P < 0.01 mechanical vs. spontaneous breaths; t P < 0.05 mechanical vs. spontaneous breaths.

6.7 2 3.2*

points with F larger than 10% of peak flow. All breaths fulfilling the following criteria were analyzed for each of the recorded periods: (1) inspiratory and expiratory volume within 10% of each other, thus minimizing air- leaks around the ETT; (2) esophageal pressure returning to the same level at the beginning and end of a breath, thus ruling out the presence of esophageal peristalsis or spasm; (3) tidal volume within the normal range (3-8 mVkg), thereby avoiding the condition of pulmonary ov- erdistension; (4) regular phasic breathing with inspiratory and expiratory time in the proper ratio, thus avoiding sighs, expiratory braking, and breath holding; (5) the correlation coefficient (r) for the least mean square fit of the regression line 20.95; (6) pressure volume loops without inversion and figure eight crossover. However, breaths with banana-shaped loops, suggesting a decrease in CL during inspiration, were included in the analysis, unless the breath had a negative inspiratory resistance. As a consequence of these strict criteria, the number of breaths qualifying for analysis varied from patient to patient. The number was lower during spontaneous than during mechanical breathing. At least 10 breaths were analyzed for each period in each infant; however, the average number was approximately 20 breaths. The breaths qualifying for the LR analysis were also analyzed by the MW method. The mean and standard deviation of the measurements were calculated from the analyzed breaths for each recording period in each infant.

Statistical Analysis To assess the intrapatient variability for the recorded

measurements, the two-sided % A of the difference be- tween the measurements of the first and the second study, as well as between consecutive periods in the same study were calculated in each patient. The means of this % A for all patients with their respective 95% confidence intervals were calculated for each measurement, which gave a simple description of intrapatient variability. The

reproducibility of the repeated measurements was exam- ined by analysis of variance, utilizing the intraclass corre- lation coefficient of reliability (ICC).” This coefficient reaches a maximum value of 1.0 when initial and repeat measurements are equal. The corresponding 95% confi- dence limits for each coefficient were also calculated. A good reproducibility was considered when the lower confidence limit was 20.75. To test for differences in the measurements between the spontaneous and mechanical breaths, as well as to compare the measurements obtained by the two methods, the paired Student’s t test was used. A p value 50.05 was considered significant.

RESULTS

Recordings were obtained in 20 infants. Four of them could not be analyzed because of a large leak around the ETT,’ technical problems with the calibration and measurements,’ inability to obtain a stable quiet state during the recording causing motion artifacts and poor correlation in the linear regression analysis.’ Sixteen in- fants were therefore included in the analysis; their mean +: SD birthweight was 1,227 t 343 g; study weight was 1,129 ? 337 g; and age at time of study was 2.5 5 2.0 days.

The average CL, RL, and VT values for the whole group of infants were not different between the first and second study (Table 1). Similarly, no difference in the results obtained by the MW and the LR methods was found for any of the variables analyzed. The values obtained for VT and inspiratory resistance were significantly higher in the mechanical breaths than in the spontaneous breaths (Table 1).

Description of Variability The mean % A between the first and second study

during spontaneous breathing for the values of CL and VT ranged between 11% and 14%, with their upper 95%

38 Gonzalez et al.

0.5 + c! -

0.25

7 -

-

/ * L R *"v

t T

1

" C LTotal V, R LTotal R Imp. R EXP C L Total v T R Total R Insp. RLExp

Fig. 1. Variability of measurements of CL, RL, and VT during spon- taneous breathing. Comparison betweem the first and second study. LR, linear regression method; MR, Mead & Wittenberg- er's method.

T T

I _I

Fig. 2. Variability of measurements of CL, RL, and V, during me- chanical breaths. Comparison between the first and second study. LR, linear regression method; MW, Mead & Wittenberg- er's method.

confidence limit less than 22% (Fig. 1). The % A for RL values was larger than that of CL and ranged from 22% to 32%, with an upper 95% confidence limit of 53% in the case of inspiratory resistance. The two methods of analysis showed similar % A for CL; however, RL mea- surements showed a larger % A when analyzed by MW than when analysed by LR methods.

The mean % A of CL and VT during controlled mechani- cal ventilation was low and ranged from 8% to 15% (Fig. 2). The mean % A for RL measurements was higher than that of CL, but significantly lower than that observed during spontaneous breaths and ranged from 13% to 21%.

Fig. 3. Reproducibility of measurements of CL, RL, and V, during spontaneous breathing. Comparison of the intraclass correla- tion coefficient (ICC) and its respective 95% inferior confidence limit between the first and second study. LR, linear regression method; MW, Mead & Wittenberger's method.

The highest upper 95% confidence limit was 28% in the case of expiratory resistance. There were no significant differences in % A between the two methods of analysis during mechanical ventilation.

Assessment of Reproducibility The higher intrapatient variability for the measure-

ments of RL than for CL during spontaneous breathing was also reflected by the ICC of reliability. As shown in Figure 3, the measurements of V, and CL had a very good reproducibility (ICC >0.9) when the LR method was used. However, reproducibility of CL was lower with the MW method, with the lower 95% confidence limit being under 0.75. Measurements of RL showed significantly lower reproducibility than CL for both methods of analysis during spontaneous breathing (ICC C0.75).

As shown in Figure 4, the reproducibility of all the measurements was significantly better during mechanical breaths than during spontaneous breaths, and even the RL values showed satisfactory reproducibility during me- chanical ventilation. There were no significant differences in reproducibility between the two methods.

To assess the influence of time and the repositioning of the esophageal tube on the measurements, variability and reproducibility of CL and RL were also determined in consecutive files (recordings obtained a few minutes apart without changing esophageal catheter). The vari- ability of measurements was not significantly different from that observed between the first and second study. The mean % A for CL and VT ranged from 7% to 12%, and for RL from 21% to 31% during spontaneous breathing. During mechanical breaths the mean % A was <15% for all variables. Reproducibility of measurements within

Pulmonary Function in Ventilated Infants 39

confirmed by our study, in which variability in repeat measurements was very similar regardless of changing the esophageal tube or leaving it untouched. The results are further supported by Ratjen et al.," who found an even higher variability of dynamic compliance in sponta- neously breathing preterm infants than reported in the present study. The variability observed by Ratjen et a1.I0 was independent of the position of the esophageal balloon. In contrast to the spontaneously breathing infants, the investigators also found a significantly lower variability of measurements in mechanically ventilated preterm infants.

There are multiple factors contributing to this variabil- ity. Some of them may be related to the equipment itself: The frequency response of the transducers, a potential phase shift between the flow and pressure signals, a drift in calibration values, and the rate of digitization may affect the rneasurement~.'~~ Our system was tested at fre- quencies up to 2 Hz against a lung model with known compliance and resistance, and provided reproducible measurements with less than 2% and 4% A, respectively. Technical problems, therefore, are unlikely to be responsi- ble for the observed variability. The algorithm employed to analyze the recorded information is also important. The MW method15 is based on only two measurement points during the respiratory cycle to calculate compli- ance and resistance, and should be much more susceptible to artifacts in the recordings of flow and pressure than the LR method, which uses measurement points through- out the respiratory cycle for anal~sis.~" Indeed, the LR method had a somewhat lower variability than the MW method.

Both methods, however, assume that compliance is constant during the respiratory cycle. This is often not the case in infants with lung disease, in whom the linear portion of the pressure volume curve is short and compli- ance decreases toward the end of inspiration.6 Under this situation, when a single compliance value does not de- scribe the pressure-volume relationship accurately, inspi- ratory and expiratory resistance cannot be determined correctly. For example, with a compliance decreasing toward the end of inspiration, inspiratory resistance is underestimated, and low or even negative resistance val- ues may be obtained. In this range of low resistance values, small absolute changes in the measurements result in large relative changes when comparing two measure- ments. The erroneously low values for inspiratory resis- tance can, therefore, contribute significantly to the ob- served high variability of inspiratory resistance. This problem is less pronounced in mechanically ventilated infants in whom the ventilator generates higher tidal flows than during spontaneous breaths because the inspiratory pressure rises faster and reaches a higher peak. The higher flows cause turbulence and thus increase inspiratory resis- tance. Higher flows and higher resistance require a larger

t

T 1 I l r

C LTotal VT R TOM R Insp. R Exp

Fig. 4. Reproducibility of measurements of C , RL, and V, during mechanical breaths. Comparison of the intraclass correlation coefficient (ICC) and its respective 95% inferior confidence limit between the first and second study. LR, linear regression method; MW, Mead & Wittenberger's method.

consecutive files was unsatisfactory for RL measurements during spontaneous breathing, while CL measurements were well reproduced. All measurements showed good reproducibility during mechanical ventilation.

DISCUSSION

The results of the present study show an important intrasubject variability of CL and RL measurements in stable ventilated infants, which was more pronounced in spontaneous than in mechanical breaths. The measure- ments of CL and RL for the whole group of infants did not change significantly over time. However, when the repeated measurements in individual infants were ana- lyzed, CL and RL were much more variable than for the whole group, and showed a mean % A of up to 14% and 33%, respectively. Both the calculation of % A and the analysis of the ICC confirmed that especially the RL val- ues obtained in spontaneously breathing infants showed poor reproducibility. This was even more so when the resistance values were split into their inspiratory and expiratory components. In contrast to spontaneous breaths, the CL and RL values analyzed in mechanical breaths showed significantly lower variation and a much better reproducibility.

Our results are similar to those of Gupta et al.,' who examined the intrasubject variability of CL and RL mea- surements in spontaneously breathing full-term infants. The 95% confidence limits for dynamic compliance were ?28% and for total resistance 5 6 % of the mean, similar to the limits found in our preterm infants. The investiga- tors observed that repositioning of the esophageal balloon had no significant effect on the measurements, a finding

40 Gonzalez et al.

portion of the total pressure to overcome the resistive forces. This is reflected by a wider pressure volume loop during mechanical than during spontaneous breaths. The same absolute changes in measurements result in smaller relative changes when higher rather than lower resistance values are compared.

In addition to these methodological problems, there are patient-related factors contributing to the variability in measurements. Position of the infant, relation to feedings, sleep state, and changes in breathing pattern have all been found to influence the measurement~."-'~ In the present study, an effort was made to control these variables. All infants were studied in the supine position with their heads stabilized in the midline, measurements were done at least 1 hr after feeding when the stomach was empty, and only periods of regular breathing without myoclonic jerks or other movements were analyzed. Strict criteria were used in the selection of breaths for analysis, thus avoiding large leaks around the ETT. The buildup of secretions in the ETT was prevented by suctioning before each recording.

More difficult to control is the breathing pattern of the infants. An increase in respiratory rate could lead to a decrease in dynamic compliance because of frequency dependence. In our infants variability in compliance was not associated with changes in frequency, and respiratory rate varied less than tidal volume. An increase in tidal volume may expand the tidal range to the flatter part of the pressure volume curve, leading to a decrease in compliance. However, variability in compliance was not associated with changes in tidal volume size in spontane- ously breathing infants.

In preterm infants the degree of chestwall distortion varies largely with sleep state and changes in breathing strategy, and the presence of chestwall distortion may affect the measurement of pulmonary mechanic^.'^ In a recent study, however, we showed that moderately severe chestwall distortion has only a minimal effect on CL and RL measurements, and, therefore, is probably not an im- portant factor contributing to the observed variability of measurements .2 1,22

Functional residual capacity (FRC) is actively main- tained in the neonate through tachypnea, or laryngeal and diaphragmatic braking.23 Any period of apnea, or an attempt to cry, will reduce FRC and can lead to alveolar collapse. Therefore, changes in FRC can be associated with changes in pulmonary mechanics. Except for provid- ing a CPAP +4 cm H20, in the present study, FRC was not controlled in the spontaneously breathing infants.

The accuracy of esophageal pressure measurements and the assumption that esophageal pressure changes equal pleural pressure changes, are critical to the reliable determination of dynamic CL and RL. The problems asso- ciated with esophageal manometry have been reviewed recently by Coates and Stocks" and are related to the

position of the catheter or balloon in the esophagus, distor- tion of the pressure trace by cardiac artifacts, and condi- tions responsible for a damped signal. In a recent study we have shown that transmission of pleural pressure to the esophagus is highly variable over time and that a damping of the signal occurs frequently for a few breaths and then transmission returns to normaLZ2 These changes were not related to changes in chestwall distortion, but were possibly caused by changes in esophageal tone. Clearly, such changes can contribute to the variability in CL and RL measurements. This interpretation is supported by the observation that changes in CL and RL in individual infants were most frequently associated with changes in peak-to-peak esophageal pressure. Rarely was the vari- ability in repeated measurements of CL and RL associated with changes in tidal volume or frequency. However, true increases or decreases in CL and RL caused by a change in FRC can result in proportional changes in esophageal pressure deflections because an infant adjusts his respira- tory effort to maintain alveolar ventilation.

During mechanical ventilation most of the patient- related variability can be controlled. The frequency is fixed, inspiratory and expiratory flows and tidal volume are nearly identical from breath to breath, and FRC may be more stable because respiratory drive and periods of active expiration do not influence the FRC level. But most important, transpulmonary pressure changes are largely determined by airway pressure changes and only depend a little on the more variable esophageal pressure measure- ments. Consequently, variability of compliance and resis- tance measurements in mechanically ventilated infants is much lower than in spontaneously breathing neonates, indirectly supporting the earlier conclusion that the large intrapatient variability in spontaneously breathing infants stems from the limitations of esophageal manometry.

The following conclusions and recommendations can be drawn from the present study:

1. Strict criteria for data selection need to be defined for each test to exclude technically unsatisfactory re- cordings and to identify those physiological condition that may exceed the applicability of a certain test. 2. Until pulmonary function equipment and testing procedures become more standardized, each pulmo- nary laboratory should develop its own reference data and evaluate the reproducibility of its measurements. If changes between repeated measurements in individual infants are within the 95% confidence intervals of vari- ation for the reference variable, one cannot discard the possibility that the observed change is occurring by chance and may not reflect a true improvement or worsening in pulmonary status. 3. In intubated infants, variability in measurements can be reduced by controlling ventilation mechanically during the time measurements are made. This approach

Pulmonary Function in Ventilated Infants 41

measurements of lung mechanics in infants. Pediatr Pulmonol. 1992; 12:146-152.

7. Gerhardt T, Bancalari E. Measurement and monitoring of pulmo- nary function. Clin Perinatol. 1991; 18581409.

8. Gerhardt T. Measurement of pulmonary mechanics in the NICU: Limitations to its usefulness. Neonatal Resp Dis. 1995; 5:2.

9. Gupta SK, Wagener JS, Erenberg A. Pulmonary mechanics in healthy term neonates: Variability in measurements obtained with a computerized system. J Pediatr. 1990; 117:603406.

10. Ratjen FA, Wiesemann HG. Variability of dynamic compliance measurements in spontaneously breathing and ventilated newborn infants. Pediatr Pulmonol. 1992; 12:73-80.

11. Mortola JP, Saetta M. Measurements of respiratory mechanics in the newborn: A simple approach. Pediatr Pulmonol. 1987;

12. Car10 WA, Beoglos A, Siner BS, Martin RJ. Neck and body position effects on pulmonary mechanics in infants. Pediatrics. 1989; 843670474.

13. Beardsmore CS, MacFadyen UM, Moosavi SH, Wimpress SP, Thompson J, Simpson H. Measurement of lung volume during active and quiet sleep in infants. Pediatr Pulmonol. 1989; 7:71-77.

14. Le Souef P, Lopes JM, England SJ, Bryan MH, Bryan AC. Influ- ence of chest wall distortion on esophageal pressure. J Appl Phys- iol. 1983; 55:353-358.

15. Mead J, Wittenberger JL. Physical properties of human lungs measured during spontaneous respiration. J Appl Physiol. 1953;

16. Uhl RR, Lewis FJ. Digital computer calculation of human pulmo- nary mechanics using aleast squares fit technique. Comput Biomed Res. 1974; 4:489-498.

17. Lorino N, Lorina AM, Harf D, Atcan G, Laurent D. Linear model- ing of ventilatory mechanics during spontaneous breathing. Com-

3:123-130.

5:779-796.

may be helpful when short-term responses to medica- tion (diuretics, bronchodilators) are to be tested. 4. The observed variability does not occur over sec- onds but rather over minutes. Frequently 10 consecu- tive breaths recorded during quiet and regular breathing show minimal variability. However, two of these re- cordings obtained within a few minutes of each other can show considerable variability, which does not in- crease much further when the tests are performed an hour apart. The variability could therefore be reduced by broadening the reference base. This might be done by recording several (5-10) short episodes of regular breathing over 20-30 minutes and averaging the re- sults. Repeating this procedure after therapeutic inter- vention or after a certain time has elapsed should allow a more reliable statement of whether pulmonary me- chanics truly changed or remained the same. Studies to validate this approach will be necessary.

ACKNOWLEDGMENTS

We thank Dr. Orlando Gomez-Marin for this statistical assistance and Mr. Nelson Claure for his technical &s- sistance.

REFERENCES

1. Buthani VK, Sivieri E, Abbassi S, Shaffer TH. Evaluation of neonatal pulmonary mechanics and energetics. Pediatr Pulmonol. 1988; 4:150-158.

2. Davis JM, Veness-Meehan K, Notter RH, Bhutani VK, Kendig JW, Shapiro DL. Changes in pulmonary mechanics after the admin- istration of surfactant to infants with respiratory distress syndrome. N Engl J Med. 1988; 319:476-479.

3. Gomez del Rio M, Gerhardt T, Hehre D, Feller R, Bancalari E. Effect of a beta-agonist nebulization on lung function in neonates with increased pulmonary resistance. Pediatr Pulmonol. 1986;

4. Englehardt B, Elliott S, Hazinski T. Short- and long-term effects of furosemide on lung function in infants with bronchopulmonary dysplasia. J Pediatr. 1986; 109:1034-1039.

5. Bhutani VK, Sivieri E, Abassi S, Shaffer T. Evaluation of neonatal pulmonary mechanics and energetics: A two factor least mean square analysis. Pediatr Pulmonol. 1988; 4150-158.

6. Silva Net0 G, Gerhardt T, Silberberg A, Gerhardt T, Claure N, Duara S, Bancalari E. Nonlinear pressure/volume relationship and

2287-29 1.

18.

19.

20.

21.

22.

23.

24.

put Biomed Res. 1982; 15:129-1%. Baydur A, Behrakis PK, Zin WA, Jaeger M, Milic-Emili G. A simple method for assessing the validity of the esophageal balloon technique. Am Rev Resp Dis. 1982; 126:788-791. Asher MI, Coates AL, Collinge Jh4, Milk Emili J. Measurement of pleural pressure in neonates. J Appl Physiol. 1982; 52:491494. Fleiss JL. Reliability of measurements. In: Fleiss JL, ed. The Design and Analysis of Clinical Experiments. New York: John Wiley, 1986:l-15. Silva Net0 G, Gerhardt T, Claure N, Duara S, Bancalari E. Effect of chestwall distortion on the measurements of pulmonary mechan- ics in preterm infants. Pediatr Pulmonol. 1995; 20:34-39. Silva Net0 G, Gerhardt T, Claure N, Duara S, Bancalari E. Influence of chest wall distortion and esophageal catheter position on esopha- geal manometry in preterm infants. Pediatr Res. 1995; 37:617-622. Kosch PC, Stark AR. Dynamic maintenance of end-expiratory lung volume in full-term infants. J Appl Physiol. 1984; 57:11261133. Coates AL, Stocks J. Esophageal pressure manometry in human infants. Pediatr Pulmonol. 1991; 11:350-360.