INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The main determinants of weaning success or failure are the balance between load and capacity of the respiratory system and the energy expenditure that the diaphragm and other inspiratory muscles must face once disconnected from the ventilator (1). The decision to liberate a patient from ventilatory aid is based mainly on the results of weaning trials performed with the two most clinically used techniques, the T-piece trial and pressure support ventilation (PSV), which are intended to mimic the work of breathing during unsupported breathing as closely as possible (2, 3). Another option has recently been proposed to shorten the duration of intubation in a selected population of patients affected by chronic respiratory pathologies. These patients, hemodynamically stable, with preserved sensorium and no fever, but not ready to sustain totally unsupported breathing, may benefit from a technique consisting of early extubation and immediate application of noninvasive mechanical ventilation as a "bridge" to unassisted ventilation (4). This mode of ventilation allows the patients to maintain satisfactory gas exchange without the encumbrance of an endotracheal tube. It is, however, unknown whether noninvasive ventilation can fully or partially substitute "traditional" ventilation in adequately sustaining the respiratory pump of a patient not yet capable of completely unassisted breathing.

We designed this prospective, physiological study in patients affected by acute respiratory hypercapnic failure and chronic pulmonary disease and not capable of sustaining autonomous breathing, to compare diaphragmatic function, mechanical properties of the respiratory system, and gas exchange during the application of the same level of ventilatory support delivered invasively and noninvasively. We also analyzed and compared the same physiological parameters during the unsuccessful T-piece trial preceding extubation and during a brief trial of spontaneous breathing, immediately after removal of the endotracheal tube.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The study protocol was approved by the Institutional Ethics Committee. Oral consent was obtained from the patients or their next of kin.

We studied 12 intubated patients affected by chronic respiratory disorders who were recovering from a severe episode of hypercapnic respiratory failure and in whom noninvasive ventilation was used as a weaning technique. (For additional information on the patients, including Table E1, see online data supplement.)

The patient's clinical condition was evaluated by the APACHE II score (5), and their neurological status was evaluated by the Kelly and Matthay scale (6). Subjective ratings of dyspnea were made during the various trials using a visual analog scale (VAS) (7). At the end of each trial the patients were asked to indicate their response to the question "How short of breath are you right now?" on a VAS.

Arterial blood gases were measured from the radial artery (ABL 550; Radiometer, Copenhagen, Denmark).

Lung volumes were assessed before the study by body plethysmography (MasterLab; Jaeger, Hochberg, Germany).

Flow was measured with a pneumotachograph (Screenmate Box 0586; Jaeger) inserted between the face mask and the ventilator or between the ventilator Y and the endotracheal tube. Expired tidal volume (VTe) was used for data analysis. We also measured the ratio of expired to inspired tidal volume and applied a correction factor equal to this ratio to the flow signal used to measure lung resistance (RL) (8).

Changes in esophageal (Pes), gastric (Pga), and transdiaphragmatic (Pdi) pressures were measured using the balloon-catheter technique (9). Pressure at the airway opening (Paw) was measured via a side port. The subtraction of Pes from Paw gave the measurement of transpulmonary pressure (PL).

Respiratory mechanics were assessed using Mead and Wittenberger's technique (10). Inspiratory pulmonary resistance (RL) and elastance (EL) were calculated by fitting the equation of motion of a single-compartment model using multilinear regression, as described in detail elsewhere (11).

Dynamic intrinsic positive end-expiratory pressure (PEEP_i,dyn) was measured according to Appendini and coworkers (12). The pressure time integrals of the diaphragm per breath (PTPdi/b) and per minute (PTPdi/min) were also measured (13). The ratio of the tidal PTPdi and the VTe was calculated as an index of efficiency (14). The maximal transdiaphragmatic pressure was measured using the sniff maneuver, shortly after the patients had been extubated (15). The number of ineffective efforts/minute was recorded during both invasive and noninvasive ventilation (16).

All the patients studied followed the protocol of "early extubation and application of noninvasive mechanical ventilation" (NIMV), described in detail elsewhere (4). (For additional information on this procedure, see online data supplement.) The physiological data were recorded during the following four periods, which were in the same sequential order in all cases:

1. T-piece trial (T-piece). During this period the patients breathed through the endotracheal tube, connected at its tip to a pneumotachograph at a fraction of inspired oxygen (FI_O2) able to maintain Sa_O2  93%.

2. PSV while still intubated (i-PSV) after the T-piece trial failure. The level of inspiratory support was 18.2 ± 2.4 cm H₂O, whereas the level of positive end-expiratory pressure was 3.6 ± 1.3 cm H₂O. FI_O2 was 35.2 + 2.4%.

3. Totally unsupported breathing (S.B.). Flow and airway pressure were recorded via a full face mask connected to a pneumotachograph. Additional oxygen was given through the nose at a mean flow rate of 2.4 ± 1.0 L/min, which was sufficient to maintain an Sa_O2  90%. Data were collected throughout this trial, which lasted at least 5 min.

4. Noninvasive PSV (n-PSV). The ventilator settings, including the FI_O2, were kept the same as for i-PSV.

Equipment dead spaces were quite similar in the four trials (134 ml for T-piece, 118 for i-PSV, 182 for SB, and 196 for n-PSV).

All signals were collected using a personal computer equipped with an A/D board, and stored at a sampling rate of 100 Hz. The mean value of each physiological variable during the last 5 min of recording was used for analyses. Results are presented as mean ± standard deviation (SD). Differences between treatments and within treatments were evaluated with a nonparametric Wilcoxon matched pairs test. A p value < 0.05 was chosen as the threshold of significance.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Extubation, the brief trial of spontaneous breathing, and the following application of NIMV were well tolerated by all the patients enrolled. (For additional information on the clinical outcome of the patients, see online data supplement.) Table 1 illustrates the changes in diaphragm function and PEEP_i,dyn. The diaphragmatic tidal effort and energy expenditure, expressed per minute or per breath, were significantly lower during i-PSV and n-PSV than during the T-piece trial or S.B., whereas PEEP_i,dyn was lower during n-PSV than i-PSV. Figure 1 is a box-whisker plot of PTPdi/min during the four trials. PTPdi/ min was significantly lower during the ventilator-assisted trials.

View larger version (20K): [in this window] [in a new window]   Figure 1.   Box-whisker plot of pressure time product of the diaphragm per minute (PTPdi/min) in the four different trials. i-PSV, invasive pressure support ventilation; n-PSV, noninvasive pressure support ventilation; T-piece, during T-piece trial; S.B., during spontaneous breathing; small square, mean; large square, ± 1 standard deviation; vertical line, ± 1.96 standard deviation; *p < 0.01, i-PSV and n-PSV versus T-piece and S.B.

Table 2 shows the respiratory pattern observed during the four experimental trials. The only significant variation was recorded in VTe, which was lower during S.B. and the T-piece trial than during PSV delivered either invasively or noninvasively. Interestingly, VTe during n-PSV was significantly higher than during i-PSV, so the PTPdi/VTe ratio during n-PSV was lower.

The ratio of expired to inspired VT (8), used in the calculation of resistances to correct the inspiratory flow for possible leaks, was 0.9 ± 0.06. Figure 2 is a box-whisker plot of airway resistance during the four trials. Mechanical ventilation, both invasive and noninvasive, significantly decreased (p < 0.007) the resistance compared with that recorded during spontaneous breathing and the T-piece trial, even though RL was higher during i-PSV than during n-PSV. The RL was significantly lower (p < 0.05) during totally unassisted breathing than during T-piece breathing. When RL was corrected for the size of the endotracheal tube and the corresponding flow the values recorded during the various trials were similar (8.9 ± 4.0 cm H₂O/L/s for i-PSV; 9.6 ± 4.2 for n-PSV, 12.2 ± 5.6 for T-piece, and 10.3 ± 2.0 for S.B.).

View larger version (12K): [in this window] [in a new window]   Figure 2.   Box-whisker plot of airway resistance (RL) in the four different trials. i-PSV, invasive pressure support ventilation; n-PSV, noninvasive pressure support ventilation; T-piece, during T-piece trial; S.B., during spontaneous breathing; small square, mean; large square, ± 1 standard deviation; vertical line, ± 1.96 standard deviation. *p < 0.05, T-piece versus S.B. §p < 0.01, n-PSV and i-PSV versus T-piece. +p < 0.05 n-PSV versus S.B.

Lung elastance did not vary significantly between the various trials (i-PSV = 16.0 ± 8.6 cm H₂O/L, n-PSV = 15.8 ± 8.5, T-piece = 17.5 ± 9.5, and S.B. = 16.3 ± 9.7).

With regard to the analysis of synchronization, seven patients showed ineffective inspiratory efforts of various entity during i-PSV and/or n-PSV. The amount of ineffective efforts did not vary according to the ventilatory modes (see Table E2 in the online data supplement). Figure 3 is a box-whisker plot showing subjective assessments of dyspnea, as recorded on a VAS, during the different trials. n-PSV was better tolerated than i-PSV (p < 0.01), whereas the T-piece trial was less well tolerated than the others.

View larger version (11K): [in this window] [in a new window]   Figure 3.   Box-whisker plot of dyspnea, assessed by a visual analog scale (VAS), in three different trials. i-PSV, invasive pressure support ventilation; n-PSV, noninvasive pressure support ventilation; T-Piece, during T-piece trial; small square, mean; large square, ± 1 standard deviation; vertical line, ± 1.96 standard deviation. *p < 0.05, T-piece versus n-PSV and i-PSV. §p < 0.05, i-PSV versus n-PSV.

Table 3 depicts the mean changes in arterial blood gas concentrations, which remained similar during mechanical ventilation, irrespective of the mode of ventilation. At the end of the T-piece trial, Pa_CO2 was significantly worse than during invasive or noninvasive PSV.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The use of early noninvasive ventilation in the weaning of intubated chronic obstructive pulmonary disease (COPD) patients represents a new approach to shortening the duration of intubation (4). We have fairly strong clinical evidence that this strategy can work in stable COPD patients with hypercapnic respiratory failure, not yet ready for fully autonomous breathing. The physiological response, in terms of diaphragm function, respiratory mechanics, and breathing pattern, to the "early" application of noninvasive ventilation is, however, unknown. Likewise it is not known whether the physiological responses to PSV differ according to whether it is delivered invasively or noninvasively.

In the present physiological study we have shown that when PSV is delivered with the same levels of inspiratory and expiratory support, the clinical and physiological responses, excluding small differences, are similar irrespective of the presence or nonpresence of the endotracheal tube.

The energy expenditures of the diaphragm, both per minute and per breath, were similar with the two modalities of PSV, clearly showing that when well applied, noninvasive mechanical ventilation may mimic the assistance of the traditional mode of ventilation in partially overcoming the inspiratory burden to respiration. These findings are of clinical relevance as they show that although providing satisfactory gas exchange, noninvasive ventilation may physiologically fully substitute the "traditional" role of ventilation delivered through the endotracheal tube in those patients whose airways no longer need to be protected.

It must be recognized that the addition of equipment dead space may have profoundly influenced the amount of effort performed by the inspiratory muscles. The measured dead spaces in the four different conditions were similar, although slightly higher using the full face mask. This may have led to a small overestimation of the effort during n-PVS and the spontaneous breathing trial.

Minute ventilation was similar for the same level of support, but VTe was significantly higher using n-PSV. However, because the dead space was also increased, gas exchange was similar during i-PSV and n-PSV. Great care was taken in the experimental procedure to avoid any leaks and to use the type of interfaces that were least likely to induce this problem, so the ratio between inspired and expired VT (8) was close to 1.

Interestingly, RL was also reduced, although not significantly so, during n-PSV. The relationship between the pressure drop and airflow is nonlinear both in endotracheal tubes and upper airways; ideally resistances should, therefore, be compared for the same flow characteristics (17). As shown in Table 2, despite the inspiratory flows being similar in the two conditions, it was difficult to determine whether the endotracheal tube resistance during i-PSV exceeded the upper airway resistance during n-PSV. We are well aware of the limitations of calculating RL from the recording of Paw at the distal end of the endotracheal tube and not directly in the carena. These resistance values do, however, represent the "real" resistive threshold that the patient has to face when he or she is intubated. This burden has been estimated to represent about 15% of the total lung resistance (18). As a matter of fact, when the RL were corrected for endotracheal tube size and flow, the RL_corr were almost identical to the values recorded during n-PSV and approximately 10% of uncorrected values, suggesting that the loss of VT during i-PSV was likely to be due to the resistive component of the endotracheal tube. Together, these events led to a significant decrease in the PTPdi/VTe ratio, suggesting that the patient's pump is more efficient at generating its output (tidal volume) during n-PSV.

Lastly, when we asked the patients to represent their sensation of breathlessness during the two modes of ventilation on a visual analog scale, both i-PSV and n-PSV were shown to cause less discomfort to the patients than the T-piece trial. This is not surprising, considering that the T-piece trial was unsuccessful and therefore may have stressed not only the cardiorespiratory system but also induced anxiety and fear. Nevertheless, the dyspnea score was significantly lower for noninvasive ventilation than for the traditional PSV.

The sensation of dyspnea has been linked to several factors. One of the most important of these is the burden that the inspiratory muscles (19) have to face, but this may not be the case in patients affected by chronic hyperinflation. Indeed, because the effort produced by the diaphragm was fairly similar during the two modalities of applying PSV, it is likely that the absence of the endotracheal tube and the sense of discomfort that it may produce were the major determinants of the difference recorded on the VAS.

This study largely confirms the findings already described by Straus and coworkers (2) in patients ready to be weaned. These authors showed that the work of breathing during a T-piece trial closely mimicked that during total unsupported breathing, while we demonstrated a similar diaphragm energy expenditure. The novelty of the present study is the confirmation of these findings also in patients not yet ready for fully autonomous breathing. The results obtained in our study may be explained by the shorter delay between extubation and measurements; in fact, for safety reasons, we made all our recordings within 10 min. The major difference between this study and that by Straus and coworkers (2) is the significant difference in inspiratory resistance, which was lower after extubation. This may be because our patients were intubated for about 48 h, a markedly shorter period than the 7.5 d of Straus' study. "Fluttering" or "saw tooth" patterns or upper airway damage with tissue edema, commonly present after extubation, tend to increase airway resistance. Our patients, with their much briefer intubation, were less likely to have developed these complications.

It could be argued that at least some of the patients enrolled in our study may have already been at a stage in which they could tolerate extubation, because during the T-piece trial the rapid-shallow breathing index was only slightly above the so-called "threshold of weaning failure" (20). The pH level of these patients (see Table 3) did, however, markedly deteriorate during the T-piece trial so reconnection to the ventilator was mandatory. Indeed the PTPdi/min, during both the T-piece trial and spontaneous breathing, lay in the range obtained in a group of patients defined as not ready to be weaned (21). Noninvasive mechanical ventilation was, therefore, maintained for the first few days on a continuous basis. Spontaneous breathing, for increasingly long periods, was not usually allowed until at least 48 h after extubation.

In this physiological study we show that in a selected population of stable COPD patients who fail a T-piece trial, removal of the endotracheal tube and switching to noninvasive mechanical ventilation are associated with maintenance of satisfactory gas exchange and similar diaphragmatic effort and respiratory mechanics. The sense of breathlessness and the efficiency of the diaphragm in generating tidal volume are better using noninvasive ventilation; this latter is likely to be due to the resistance dissipated to overcome the resistance of the endotracheal tube.

The results of the present study give new insights into the physiological rationale for using noninvasive ventilation as a weaning technique and provide evidence that when n-PSV is well set it may fully substitute for the traditional route of delivery of mechanical ventilation.