INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The end-inspiratory airway occlusion method is the combination of the interrupter (1) and elastic subtraction (2) techniques, originally described by von Neergaard and Wirz in 1927. This method was reintroduced by Bates and coworkers recently, for the assessment of respiratory mechanics (3). The results of early studies in anesthetized paralyzed animals (4, 5) were interpreted in terms of a model which separates the flow-resistive component (R_int) from the viscoelastic component (R) of the total respiratory system resistance (Rrs) (6).

Hitherto, the end-inspiratory airway occlusion method has been used in otherwise healthy adults under general anesthesia and in adults with respiratory insufficiency who were on mechanical ventilation (7). We have previously reported values of intrinsic flow resistance and viscoelastic resistance in healthy children under general anesthesia undergoing minor elective surgery (15). Measurements in newborn infants have been presented by other investigators (16). Additionally, in several studies in children with and without lung disease, the midexpiratory airway occlusion technique has been used (17).

In the present study, we applied the end-inspiratory occlusion technique to critically ill, mechanically ventilated pediatric patients with and without primary lung disease. The working hypothesis was that children with parenchymal lung disease would demonstrate an increase in viscoelastic tissue resistance (R) but without changes in airway resistance, as represented by R_int, as seen in adult patients (10). Further, children with lower airway obstruction would have increased R_int and R (14), whereas those on mechanical ventilation without prior lung disease would show essentially normal R_int and R. This study also compared the flow and volume dependence of Rrs and its subdivisions to those found in adult patients.

Technical and theoretical features of the end-inspiratory occlusion method have been extensively reviewed in the literature (3, 7, 21) and are briefly summarized here. After a rapid end-inspiratory occlusion, there is an initial rapid drop in airway pressure proximal to the endotracheal (ET) tube (Pao), representing intrinsic flow resistive component (R_int), mostly caused by airway resistance (Raw). The subsequent slower decrease in Pao is the result of intrinsic viscoelastic resistance (R) of lung and chest wall tissues. Both R_int and R change with changes in constant inspiratory flow (I) and tidal volume (VT); but the directions of their volume and flow dependence are opposite of each other (7). R has been found to be the major component of Rrs, especially at I less than 1 L/s in adults (7). Consequently, the direction of flow and volume dependence of Rrs follows changes of R, not those of R_int, as has been assumed in the past (21).

The viscoelastic property of the respiratory system can be explained by the "spring-and-dashpot" model, in which a Maxwell body is added in parallel to the standard (flow) resistance (dashpot or shock absorber, R_int) and static elastance of the total respiratory system (spring, Est) (6, 7). The Maxwell body consists of a second spring (E₂) and dashpot (R₂) in series. After end-inspiratory airway occlusion, E₂ dissipates its stored energy on R₂, representing the existence of an additional non-Newtonian resistance (effective resistance, R). R is responsible for the slow decrease in Pao after the end-inspiratory airway occlusion. As the inspiratory time decreases (with high frequency, high inspiratory flow, or smaller VT), most energy offered to the Maxwell body is stored in E₂, because R₂ has less time to move and dissipate energy. This means lower effective tissue resistance (R) and higher effective tissue elastance (E) for the total respiratory system (21). Viscoelasticity (Maxwell body) explains the frequency dependence of resistance and elastance as well as the stress relaxation properties of the respiratory system (22).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Subjects

We studied two groups of children on mechanical ventilation in the pediatric intensive care unit (ICU) of the Children's Hospital of Pittsburgh between July and December of 1996. The first group (Group 1) consisted of seven patients who were mechanically ventilated for respiratory failure caused by lung disease. One of the Group 1 patients presented with lower airway obstruction; the remaining six patients had parenchymal lung disease. The subjects in Group 2 were six children without primary respiratory disorder but who were ventilated for decreased levels of consciousness and/or inability to protect their airways from gastric regurgitation. Patient characteristics for each group are summarized in Table 1. Results from these individuals were compared with the data obtained previously from eight healthy children, mechanically ventilated under general anesthesia for elective dental procedures (15).

All patients were intubated transorally with cuffed ET tubes, sedated with fentanyl, paralyzed with vecuronium administered intravenously, and were placed in the supine position. The Servo 900C ventilator (Siemens-Elema, Sweden) was used. The mode of ventilation was volume control with constant inspiratory flow and with the ventilator settings selected by the pediatric ICU patient care team. All subjects were on positive end-expiratory pressure (PEEP: range, 3 to 10 cm H₂O; median, 5 cm H₂O).

The study was approved by the institutional review board for human experimentation. Informed consent for participation in the study was obtained from the parents or legal guardians.

Equipment, Procedures, and Data Analysis

Airway opening pressure (Pao) was measured through a side port connected proximal to the ET tube using a pressure transducer (model 267BC; Hewlett-Packard, Palo Alto, CA). I was measured with a heated pneumotachograph (Fleisch No. 1; Fleisch, Lausanne, Switzerland), inserted between the ventilator circuit and the ET tube, using a Hewlett-Packard 47304A unit. The software used was the ANADAT-LABDAT (RHT InfoDat, Montreal, Canada). VT measurements were obtained by computer integration of flow signals.

The technical aspects of the end-inspiratory airway occlusion technique have been described in detail elsewhere (4, 7). For the period of each experiment, the standard ventilator circuit with the heater and the humidifier was replaced with a short (60 cm) low compliance ventilator tubing (Tygon; Norton Performance Plastics, Akron, OH) connecting the patient directly to the ventilator. This replacement minimized the effects of the compliance and resistance of the ventilator circuit on the mechanics measurements (23). To prevent possible hypoxemia during the experiment, the fraction of inspired oxygen (FI_O2) was increased to 10% above the established setting for each patient. Each subject was monitored continuously with a pulse oximeter (Nellcor, Palo Alto, CA), a capnometer (Datex Capnomac Ultima, Helsinki, Finland), and an indwelling arterial catheter (for blood pressure measurements). The continuous gas sampling port, proximal to the pneumotachograph, was closed intermittently during the period of pressure and flow measurements. Before any measurements were carried out, the ventilator circuit and ET tube were examined to eliminate any air leaks.

The rapid end-inspiratory airway occlusion was accomplished by simultaneous closure of both inspiratory and expiratory valves, located within the ventilator, and was maintained for approximately 5 s. The effects of the closure time of the Servo ventilator valve on respiratory mechanics measurements have been previously presented by Sly and coworkers (23). Because no valve closes instantaneously, a small gas volume (V) passes into the respiratory system while the valve is moving to closure, leading to underestimation of the initial and subsequent drops in Pao (P_max  P₁, P_max  P₂, respectively) and, therefore, underestimation of R_int [(P_max  P₁)/I] and Rrs [(P_max  P₂)/ I] (4, 23). For the purposes of the present study, the gas volume that passed through the valve during its closure was estimated by integration of the flow signal with respect to time from the moment that flow started to decline until it ceased. Then a correction factor based on V was added to R_int and Rrs (23).

The first assessment of respiratory mechanics with the end-inspiratory occlusion was performed at the baseline ventilator setting. To study the flow dependence of resistance and elastance while the end-inspiratory volume was kept constant (isovolume condition), I was changed to one level below and one above the baseline setting. To evaluate the volume dependence of respiratory mechanics using a constant inspiratory flow (isoflow condition), VT was varied to one level below and one level above the baseline setting. At each ventilator setting the measurement was repeated three times. Thus, for each individual, a total of five sets of triplicate measurements were obtained. The different flow and volume settings were tried in random order. Patients received at least five regular mechanical breaths between end-inspiratory occlusions.

At the conclusion of each experiment, the flow resistance of the equipment was measured in triplicate at the different I settings used for the patient. During data analysis the equipment resistance was subtracted from the uncorrected data to obtain R_int and Rrs.

Static intrinsic PEEP (PEEP_i) was determined by the end-expiratory airway occlusion technique (24). Est is provided (21) by the equation:

(1)

To compare patients of different size, all the resistance and elastance measurements were normalized for body weight. Data were presented as mean ± SEM. Resistance and elastance values, under baseline flow and volume conditions, were compared to the values previously obtained from children under general anesthesia (15).

Statistical Analysis

Least square regression analysis (linear or nonlinear) was used to correlate resistance and elastance measurements with changes in VT (under constant I) and I (under constant VT) for each subject, as was performed in previous studies (7, 11). The computer statistical package Sigma Stat, version 1.0 (Jandel Scientific, San Rafael, CA) was used. In all cases, p < 0.05 was considered as statistically significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Adverse Effects

Data collection for each subject was completed in less than 40 min (median, 30 min; range, 25 to 39 min). During and after the completion of the procedure, none of the patients had significant changes in Sa_O2, end-tidal CO₂, or blood pressure.

Variations in VT and I

The median percent change in VT used to assess the volume dependence of Rrs and its components was 21% (range 12 to 35%) below and 33% (range 12.5 to 100%) above the baseline VT setting in each patient. The median percent change in I used to evaluate flow dependence was 34% (range 16 to 34%) below and 65% (range 32 to 65%) above the baseline I.

Static PEEP_i

Five of 13 patients (Subjects 4, 7, 9, 11, and 13) had static PEEP_i exceeding 1 cm H₂O (median, 2.2 cm H₂O; range 1.3 to 6.0 cm H₂O).

R_int

The mean values of R_int at baseline ventilator settings are presented in Table 2. Two of 7 patients from Group 1 and five of six patients from Group 2 had R_int values, determined at baseline ventilator settings, higher than the 95% confidence limits which had been obtained in a previous study for healthy children under general anesthesia (15). All individuals with PEEP_i had increased R_int.

Figure 1 shows values of R_int for each patient in relation to changes in I or VT. Results under the isovolume, variable inspiratory flow setting were as follows. In Group 1 (with primary lung disease), R_int increased significantly (p < 0.05) with increasing I in two of seven patients. R_int decreased in one patient (p < 0.05) and did not change in four patients. In Group 2 (without primary lung disorder), R_int significantly increased (p < 0.05) in four of six patients, decreased in one patient (p < 0.05), and remained statistically unchanged in one patient (Figure 1A).

View larger version (18K): [in this window] [in a new window]   View larger version (17K): [in this window] [in a new window]   Figure 1.   (A) Rint in relation to changes in I while VT was kept constant (isovolume conditions) in children on mechanical ventilation with primary lung disease (Group 1: closed circles) and those without primary lung disorder (Group 2: open circles). Data from healthy children (closed squares, dashed line; mean ± SEM) (15) are also included for comparison. (B) Rint in relation to changes in VT while I was kept constant (isoflow conditions) in the same groups of patients.

Findings under the isoflow, variable VT setting were as follows. In Group 1, R_int did not change with increasing VT in five of seven subjects and significantly decreased (p < 0.05) in two subjects. In Group 2, R_int did not change in four of six subjects and decreased (p < 0.05) in two subjects (Figure 1B).

R

The mean values of R at baseline ventilator settings are presented in Table 2. All patients in Group 1 and four of six patients in Group 2 had R values above the 95% confidence limits for healthy anesthetized children (15). Figure 2 illustrates R in each patient in relation to changes in I or VT.

View larger version (18K): [in this window] [in a new window]   View larger version (17K): [in this window] [in a new window]   Figure 2.   (A) R in relation to changes in I while VT was kept constant (isovolume conditions) in children on mechanical ventilation with primary lung disease (Group 1: closed circles) and those without primary lung disorder (Group 2: open circles). Data for healthy children (closed squares, dashed line; mean ± SEM) (15) are also included for comparison. (B) R in relation to changes in VT while I was kept constant (isoflow conditions) in the same groups of patients.

Under isovolume, variable inspiratory flow settings, R decreased exponentially with increasing I in 11 of 13 patients, according to the equation R = A + B (1  e^C/I), where A is a constant attributed to time constant inequalities within the lung and/or chest wall (11) and B and C are constants that describe the viscoelastic properties of the respiratory system (11) (Figure 2A). Constant C is also affected by the VT used. In all of our children, A was not statistically different from 0; this was true even for Patient 7 who had lower airway obstruction, clinically. These findings indicate that the time constant inequality of the lung and chest wall tissues (pendelluft) did not importantly contribute to R (11). In two of 13 patients, R did not change with I. Under isoflow, variable VT settings, R increased linearly with VT in 12 of 13 patients, whereas it did not change in one patient (Figure 2B).

Rrs

Values for Rrs under baseline ventilator settings are provided in Table 2. All patients had values above the 95% confidence limits of Rrs for healthy anesthetized children (15). Under the constant VT, variable I setting, Rrs decreased significantly (p < 0.05) with increasing I in six of 13 patients, following the direction of change of R. In four of these six patients both R and R_int decreased with increasing I. In three of 13 patients, Rrs increased with increasing I (p < 0.05) because the increase in R_int was more prominent than the exponential decrease in R. In the remaining four of 13 patients, Rrs did not change with the change in I. Under the constant I, variable VT setting, Rrs increased with increasing VT in 11 of 13 subjects following the direction of change of R. In the two remaining subjects, Rrs did not change with increasing VT.

Est

Values of Est at baseline ventilator settings are provided in Table 2. Compared with healthy anesthetized children who were studied previously (15), five of seven patients in Group 1 and two of six patients in Group 2 had increased Est. Patient 1 (Group 1), who was undergoing extracorporeal membrane oxygenation for acute lung injury (ALI) secondary to meningococcal sepsis, had the highest Est value. Figure 3 illustrates Est in relation to changes in I or VT. With increasing I under constant VT, Est decreased slightly but significantly (p < 0.05) in 11 of 13 patients and did not change in the remaining two (Figure 3A). With increasing VT under constant I, Est decreased (p < 0.05) in 10 of 13 subjects and remained unchanged in the rest of them (Figure 3B).

View larger version (15K): [in this window] [in a new window]   View larger version (15K): [in this window] [in a new window]   Figure 3.   (A) Est in relation to changes in I while VT was kept constant (isovolume conditions) in children on mechanical ventilation with primary lung disease (Group 1: closed circles) and those without primary lung disorder (Group 2: open circles). Data for healthy children (closed squares, dashed line; mean ± SEM) (15) are also included. (B) Est in relation to changes in VT while I was kept constant (isoflow conditions) in the same groups of patients.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The constant flow, end-inspiratory occlusion technique has not been used for the evaluation of children in respiratory insufficiency in the ICU. The present investigation has demonstrated the feasibility and safety of this method applied on critically ill patients who are relatively stable hemodynamically. The analysis of Rrs and its flow-resistive (R_int) and viscoelastic (R) components, as well as their flow and volume dependences, can lead to a better understanding of the pathophysiologic nature of respiratory insufficiency. Based on such information, intensive care physicians can more rationally formulate a physiologically sound ventilator treatment strategy.

Methodological Consideration

In the present study, the apparent valve closure time, as judged from the flow signal, was used to correct resistance measurements for the error introduced by the valve closure time (4). The apparent valve closure time is affected by the inspiratory flow and the resistance of the respiratory system and the equipment (including connectors, a pneumotachograph, and an ET tube) (23). Therefore, in the present study, valve closure time might have caused a slight underestimation of R_int and Rrs. Bates and coworkers have shown that R_int may be underestimated by as much as 7% if the valve closure time is 12 ms (25). Sly and coworkers have estimated the mean time for valve closure of the Siemens 900C ventilator to be between 22 and 229 ms; but R_int was underestimated by a mean of 17% using the standard ventilator tubing (23). Kochi and associates have proposed a formula to correct for the underestimation of R_int resulting from gas flowing into the lungs while the valve is being closed (4).

For flow measurements we used a Fleisch No. 1 pneumotachograph, inserted between a Y connector and an ET tube. It has been shown that this connection can lead to an overestimation of flow. A flow of 1.5 L/s can be overestimated by 5% (26); the overestimation becomes progressively greater with increasing flow. The range of I used in this investigation was 0.06 to 1.1 L/s.

An error can also be introduced when the expiratory pause button of the ventilator is used for the measurement of static PEEP_i. This error (underestimation of PEEP_i with resultant overestimation of Est) increases when the patient's Est increases or the equipment elastance decreases (27). The high elastance of the ventilator tubings that we used in the present study probably minimized the underestimation of PEEP_i.

In our previous study of healthy children under general anesthesia, standard ranges of I and VT were used and the results were normalized for body weight (15). The same methodological strategy, however, could not be applied to children in the present study. This is because the children were on mechanical ventilation, having a variety of respiratory disorders. Consequently, the ranges of I and VT studied in each patient varied considerably. In the present study, data from the children with a wide age range (0.5 to 17.7 yr) were compared with the data from healthy children under general anesthesia with a narrower age range (2.3 to 6.5 yr). It should be emphasized that the data of the latter study were used as a general guide for comparison, rather than as control values; one reason being that elastance corrected for lung volume (specific elastance) increases with age or height (17). Additional potential problems with the use of the data of healthy children under general anesthesia as controls for patients in the ICU include the possible effects of anesthetics (bronchodilation) and different gas densities between the two settings. These problems, however, were relatively minor with the children in the previous study, in whom the anesthetic technique used was intravenous propofol with FI_O2 of 0.3 in nitrogen. Propofol, unlike inhaled anesthetics, has minimal bronchodilator effect (28); the gas mixture is similar between the two studies. Finally, the patients on mechanical ventilation would undergo changes in respiratory mechanics unrelated to their primary respiratory problems. These changes include autonomic responses to the presence of an ET tube, inflammatory processes secondary to infection, hyperoxia, volutrauma caused by high positive pressure ventilation, and the presence of airway secretions (29).

Special Characteristics of Respiratory Mechanics in Children

Postnatally the lungs develop and grow at an extremely rapid pace in human infants. Although at birth, terminal air sacks are composed mostly of primitive saccules, alveolar formation in the lungs is completed by 18 mo of age (30). Postnatal development of lung parenchyma, however, continues throughout the first decade of life (31). It has been known that, in the neonate, the specific elastance of both the lungs and chest wall are lower (compliance, higher) than those of older children and adults. Furthermore, in the neonate the outward recoil of the chest wall is disproportionately lower than the inward recoil of the lungs (32). Papastamelos and coworkers (33) recently demonstrated that by 9 to 12 mo of age chest wall elastance rapidly increases to the level of lung elastance. Thereafter, elastances of both the chest wall and the lungs continue to increase at the same pace (33).

In awake infants, the extremely compliant chest wall resists the inward recoil of the lungs by tonic contractions of inspiratory muscles that maintain FRC (33). When the sustained tension of the inspiratory muscles of the chest wall is abolished, either by general anesthesia or by muscle relaxants, the balance is shifted and FRC (or relaxation volume, V_r) in infants decreases to 15 to 20% of TLC and results in concomitant airway closure and atelectasis (12, 32). By contrast, in adults, such a reduction in V_r with anesthesia or muscle relaxants amounts to only 9 to 25% of FRC in the awake state (33, 35).

In a recent study involving eight healthy children undergoing general endotracheal anesthesia, all measurements were made after "sighs" (three slow inflations to TLC) to minimize airway closure or atelectasis and to set a consistent volume history (15). In this study, R_int decreased with increasing VT when the inspiratory flow was kept constant (isoflow condition) as seen in studies on anesthetized adults. Unlike in adult studies, however, R_int did not change with I (isovolume conditions). Flow and volume dependences of R were similar to those of adult patients and their directions were opposite to changes of R_int: R decreased with increasing I (isovolume studies) whereas it increased with VT (isoflow experiments) (15). Unlike in adults, Rrs did not change with changes in R (15). More important clinically, we found evidence that even under PEEP (5 cm H₂O), V_r of the lungs was still well below the level of physiologic FRC or the steepest portion of the pressure-volume (P-V) curve of the respiratory system. It is interesting to note that in a separate study of healthy children under general anesthesia, a PEEP value of 12 cm H₂O yielded the maximal value of respiratory system compliance (36). These differences in the characteristics of R_int, R, and Rrs and their flow and volume dependence between adults and children are apparently due to developmental characteristics of the lungs and chest wall in children during their postnatal development and growth.

R_int, R, and Rrs in Children with Respiratory Insufficiency

In the present study, all 13 children with respiratory insufficiency of diverse etiology demonstrated increased Rrs. The end-inspiratory occlusion technique, rarely used to study children in clinical settings, allowed us to further analyze the pathophysiology of respiratory insufficiency.

Most of our patients with primary lung disease and respiratory failure (Group 1) did not show increased R_int, indicating that their primary lung parenchymal pathology did not affect airway patency. Exceptions in this group are two of seven patients with increased R_int (one with status asthmaticus), both of whom also had increased PEEP_i, evidence of dynamic hyperinflation (24). High R_int has been documented in some of the studies on adult patients with adult respiratory distress syndrome (ARDS) (11, 12) but not in others (10).

It is important to note that nearly all (5 of 6) children in Group 2, who were without primary respiratory disorder and were mechanically ventilated, demonstrated increased R_int. At the time of the study, none of these patients had clinical evidence of lower airway obstruction. These findings indicate that even the patients who are intubated and put on mechanical ventilation without lung disease in the ICU settings develop lower airway obstruction after an extended period. The only patient in this group who did not have increased R_int (Patient 8) was studied only a few hours after the initiation of mechanical ventilation. Increased R_int may be secondary to increased bronchomotor tone in response to the presence of the ET tube, accumulation of airway secretions, and airway inflammation caused by hyperoxia (37) and volutrauma from lung hyperdistention (38).

In our previous study of healthy children under general anesthesia (15), R_int decreased significantly with increasing VT (isoflow settings), reflecting an increasing airway caliber with VT. On the other hand, R_int did not change significantly with I (isovolume conditions). Flow and volume dependence of R_int in the present study were variable, reflecting diverse conditions and causes of respiratory insufficiency.

In the present study, R was markedly increased in all patients in Group 1 with acute lung injury/pulmonary edema from the range found in healthy children under general anesthesia. Similar findings were reported in adults with ARDS (10, 11, 13). R was also increased in four of the six patients in Group 2 without primary pulmonary disorders. Of these children, three were also found to have increased R_int. Therefore, time constant inhomogeneity resulting from bronchospasm (39) or increased airway secretions, cell debris, etc. most likely contributed to the increased R (11). The remaining patient (Patient 8) had an elevated R without clinical signs of parenchymal lung disease and with a normal R_int. Because the chest wall importantly contributes to the viscoelastic properties of the respiratory system (8), it is likely that the thoracotomy, which the patient had undergone recently, might have altered the value of R. R decreases with increasing flow; R also decreases with decreasing VT (7). In patients with increased R the work of breathing would be the lowest with rapid shallow respirations. It is of clinical interest to note that the adult patients with ARDS, who failed attempted weaning from mechanical ventilation, developed rapid shallow breathing immediately after the discontinuation of a ventilator, whereas those who survived did not (40, 41). However, tachypnea was not a predictor of success and failure of weaning from mechanical ventilation in a recent study involving 208 children with respiratory failure, although 42% of those who failed the attempted weaning did develop tachypnea (42). The exact mechanism of rapid shallow breathing in ARDS is not well understood (41).

Est in Children with Respiratory Insufficiency

Est represents the elastic property of the respiratory system. Four of six children with parenchymal lung disease had increased Est. A similar observation was made on adult patients with ARDS (10, 11). In the remaining two patients, however, Est was not elevated; these patients were recovering from respiratory insufficiency and were weaned from ventilatory assistance shortly after the measurements. Patient 1, who was undergoing extracorporeal membrane oxygenation (ECMO) for acute lung injury secondary to meningococcal sepsis, had the highest Est value, probably reflecting the severity of his parenchymal lung disease. Patient 7 with status asthmaticus also exhibited increased Est. This patient also had elevated PEEP_i, a sign of dynamic hyperinflation. This finding could mean that FRC of this patient was on the upper, stiff portion of the P-V curve of the respiratory system, where Est was increased. Patient 8 also had an elevated Est, most likely as a result of the recent thoracotomy.

In healthy anesthetized adults or adult patients with ARDS, Est did not change with increasing I (isovolume conditions) but decreased with increasing VT (isoflow conditions) (7, 11). In healthy children under general anesthesia (15) and in most pediatric patients of the present study, Est decreased with increasing VT. These findings probably indicate that V_r was below the steep midportion of the P-V curve of the respiratory system. Est also decreased with increasing I; the reason is not clear. We speculate that lung inflation with increasing I may lead to progressive recruitment of previously closed airways or air spaces and result in decreased Est.