Introduction

The aim of mechanical ventilation is to replace or assist the respiratory muscles in order to increase alveolar ventilation while reducing dyspnea and inspiratory effort. Noninvasive positive pressure ventilation (NPPV) is increasingly used in the paediatric intensive care unit (PICU) for the treatment of acute respiratory insufficiency [1, 2]. This larger use of NPPV contrasts with the limited number of clinical studies in this field [3–5]. Moreover, to our knowledge, no physiological studies have been reported in children hospitalised in the PICU for acute respiratory insufficiency. The situation is different in the chronic situation where short-term physiological studies have demonstrated that NPPV is able to unload the respiratory muscles and improve alveolar ventilation in patients with severe upper airway obstruction and sleep apnea, as well as cystic fibrosis [6–10].

The first aim of this physiological study was to demonstrate the ability of NPPV to unload the respiratory muscles, improve gas exchange and clinical outcome in children with acute moderate hypercapnic respiratory insufficiency. The second aim was to compare a clinical setting of NPPV, based on clinical noninvasive parameters, with an invasive setting, based on the recording of the respiratory muscle output by means of an oeso-gastric pressure mounted catheter.

Materials and methods

Patients

The study was performed from December 2004 to January 2007 in a tertiary care university PICU. The study was approved by the local institutional board. Written informed consent was obtained from all the parents and the patients when available.

Criteria for enrolment were as follows: admission in the PICU for acute moderate hypercapnic respiratory insufficiency defined by a respiratory rate (RR) ≥ 97th percentile for age [11] associated with a transcutaneous carbon dioxide pressure (PtCO2) ≥ 40 mmHg [2], age between 1 and 18 years, and weight ≥ 10 kg. Only patients treated with intermittent NPPV [with a minimum of 2 continuous hours of NPPV separated by at least 2 h of spontaneous breathing (SB)] for less than 12 h were included.

Criteria for exclusion were: unstable clinical condition (necessity of immediate intubation because of cardiac arrest, hemodynamic instability, major metabolic acidosis defined by a pH < 7.2), major alteration of consciousness (Glasgow coma score <12), ineffective cough, inability to cooperate or to tolerate the interface, or enrolment in another research protocol.

NPPV equipment

NPPV was delivered via a standard ICU ventilator (Evita 2 dura Neoflow; Dräger, Lübeck, Germany) and industrial nasal or facial masks (see online supplement).

Measurements

Pulse oximetry (SaO2, HP omnicare M1165/66A, Hewlett Packard, Böblingen, Germany) and PtcCO2 (Tina TCM 4/40, Radiometer Medical ApS, Brønshøj, Denmark) were recorded continuously. Respiratory flow was measured using a pneumotachograph (Hans Rudolph, Kansas City, MI) inserted between the mask and the ventilatory circuit, connected to a pressure transducer (MP 45 model, Validyne, Northridge, CA). Airway pressure (Paw) was measured with a differential pressure transducer (MP 45 model, Validyne, Northridge, CA) on the mask. Oesophageal (Pes) and gastric pressures (Pga) were measured as previously described [9, 10, 12] (see online supplement).

All the signals were digitised at 128 Hz, sampled for analysis using an analogical/numeric acquisition system (MP 100, Biopac Systems, Goletta, CA) and run to a computer with Acknowledge software.

Experimental protocol

After a minimum of 2 continuous hours of NPPV, and within the first 12 h of NPPV, the patients were studied. In practice, all patients were recorded between the 2nd and 4th NPPV session, in semi-recumbent position. After insertion of the oeso-gastric catheter, the first period was a SB period with additional oxygen delivered by nasal prongs to achieve a SaO2 ≥ 94%. All data, breathing pattern and respiratory muscle output, were recorded during a 2-min period following a 3-min period of stabilisation.

This SB period was followed by a period of NPPV with positive end expiratory pressure (PEEP) without a back-up rate. The initial NPPV settings were a PS at 6 cmH2O (defined as the inspiratory pressure delivered above the PEEP level), a PEEP at 4 cmH2O, an inspiratory trigger set at its most sensitive value, and an inspiratory flow set at the highest ramp. These settings were then adapted by the physician on clinical parameters such as the disappearance of retractions and the decrease of RR. Once this clinical setting was obtained, a first clinical NPPV period named NPPVClin was recorded for 2 min after a 10-min stabilisation period.

Then the NPPV settings were adjusted to obtain first, the optimal unloading of the respiratory muscles, as reflected by the normalisation (i-e 5–8 cmH2O) or the maximal decrease in Pes and transdiaphramatic pressure (Pdi) swings, and second, the best patient-ventilator synchronisation [15−18] (see online supplement). This physiological NPPV period named NPPVPhys was then recorded for 2 min after a 10-min stabilisation period. During the two NPPV settings, the inspired concentration of oxygen (FiO2) was maintained constant to obtain a SaO2 ≥ 94%.

After the measurements, NPPV was continued in all the patients, for at least 2 h, at least 2 times a day, until improvement or cure of the respiratory insufficiency.

Data analysis

Respiratory flow was integrated to yield tidal volume (V t) during SB. Because of abrupt change of flow delivered by ICU ventilator, the flow trace was not analysable during NPPV. The expiratory tidal volume (V t e) measured by the ventilator was thus retained for analysis. Pdi was obtained by subtracting the Pes signal from the Pga signal. Total inspiratory work of breathing (WOBtot), elastic (WOBel) and resistive work of breathing (WOBres), as well as dynamic lung compliance (CLdyn) and total pulmonary resistance (R L) were calculated as previously described [10] (see online supplement). The diaphragmatic (PTPdi/breath) and oesophageal pressure-time products per breath (PTPes/breath) and per minute (PTPdi/min and PTPes/min) were also calculated as previously described [13–15] (see online supplement).

Statistical analysis

Data are presented as mean ± SD. Repeated measures analysis of variance (ANOVA) was used to compare the different conditions (SB, NPPVClin and NPPVPhys) on the different variables. A P-value <0.05 was considered statistically significant.

Results

Characteristics of the population

During the period study, 233 patients (21% of the total number of patients admitted to the PICU) were treated either with noninvasive continuous positive airway pressure (CPAP) delivered by nasal prongs (n = 131) or with NPPV delivered by a face mask (n = 102). Only 24 out of the 102 patients (24%) were potentially eligible for the study (Fig. 1). Eleven patients were excluded because of parental refusal (n = 4), language problems (n = 2), or the impossibility to obtain written consent from the two parents (n = 5). Thirteen patients were thus included in the study but 12 patients were studied. The first patient did not complete the study because of the failure to introduce the oeso-gastric catheter. All the other 12 patients tolerated the oeso-gastric catheter and completed the study protocol. None of the patients required sedation for NPPV or the protocol.

Fig. 1
figure 1

Recruitment of the patients for the study. PICU pediatric intensive care unit, NPPV noninvasive positive pressure ventilation, CPAP continuous positive airway pressure

Full size image

The clinical characteristics of the patients are presented in Table 1. The mean age of the population was 71 ± 56 months and the mean weight was 22.5 ± 16.8 kg. The patients belonged to four main diagnostic groups: acute chest syndrome (ACS), acute respiratory insufficiency after liver transplantation, community acquired pneumonia, and acute asthma. The two remaining patients were admitted for acute respiratory insufficiency after abdominal trauma or due to a neuromuscular disease.

Table 1 Clinical characteristics and outcome
Full size table

Effects of NPPV on respiratory muscle output, breathing pattern and gas exchange

Spontaneous breathing

All the patients presented a moderate hypercapnic respiratory insufficiency with an increase in RR (mean 47.6 ± 12.8 breaths/min) and in PtcCO2 (mean 48.1 ± 8.7 mmHg) (Table 1). SaO2 was normal (97 ± 2.5%) with oxygen therapy (flow rate 2–10 L/min, mean 4.5 L/min) and mean PtcO2 was 69 ± 15.9 mmHg. Patients requiring 10 L/min of oxygen were those admitted for ACS. Patient 12, who had a neuromuscular disease, was the only patient to have a normal SaO2 under room air. The analysis of the respiratory mechanics of the patients showed a decrease in Cldyn and an increase in the different components of the work of breathing (Table 2).

Table 2 Mean values ± SD and range for the measured variables during spontaneous breath in 12 children with acute moderate hypercapnic respiratory insufficiency
Full size table

Clinical setting of NPPV (NPPVClin)

NPPVClin was associated with a significant improvement in breathing pattern, gas exchange and respiratory muscle output as shown in Table 3 and Fig. 2. Indeed, during NPPVClin, V t and V e increased by 33 and 17%, and PTPes/min and PTPdi/min decreased by 49 and 56%, respectively. This improvement in alveolar ventilation translated into a significant reduction in mean PtcCO2 from 48 ± 5 to 40 ± 8 mmHg. During NPPVClin, mean FiO2 was 0.47 ± 0.22 with a mean PtcO2 of 89 ± 10 mmHg. Patient 6, who developed a respiratory insufficiency after a third liver transplantation, was the only patient who increased his RR and respiratory muscle output during NPPV. During the physiological recording, an important increase in Pes swing was observed without a concomitant increase in Pdi swing reflecting diaphragmatic dysfunction. In this patient, NPPV was stopped after 6 h and he recovered from his respiratory insufficiency without ventilatory support after 4 days (Fig. 3).

Table 3 Effects of NPPV on breathing pattern, gas exchange and respiratory muscle output
Full size table
Fig. 2
figure 2

A representative sample of tracing from patient 3 during spontaneous breathing (SB) and noninvasive positive pressure ventilation (NPPV) showing the decrease in oesophageal (Pes) and transdiaphragmatic (Pdi) pressure swings during NPPV compared to SB. Pga gastric pressure, Paw airway pressure

Full size image
Fig. 3
figure 3

Individual variations of a oesophageal pressure time product per minute (PTPes) and b diaphragmatic pressure time product per minute (PTPdi) during spontaneous breathing (SB), noninvasive positive pressure ventilation set on clinical parameters (NPPVClin) and noninvasive positive pressure ventilation set on the recording of oesophageal and gastric pressures (NPPVPhys). P < 0.0001 compared to SB with no difference between NPPVClin and NPPVPhys

Full size image

Physiological setting of NPPV (NPPVPhys)

NPPVPhys was associated with a comparable improvement in breathing pattern and respiratory muscle output as shown in Table 3. During this physiological setting, a similar improvement in gas exchange was observed with a constant FiO2.

During this setting, the NPPV settings changed for 8 patients. The level of PS was increased by 2 cmH2O in patient 8 and could be decreased by 1–3 cmH2O in 5 other patients. The level of PEEP was increased by 1–4 cmH2O in 2 patients (patient 5 and 7) and could be decreased by 1 and 2 cmH2O in 2 other patients. However, the mean PS and PEEP settings did not differ between NPPVClin and NPPVPhys (Table 2). The setting of the inspiratory trigger between NPPVClin and NPPVPhys did not differ. Auto-triggering, with 10, 23 and 38% of auto-triggered breaths was observed in 3 patients during NPPVClin. During NPPVPhys, auto-triggering disappeared in patient 3 after a 1 cmH2O decrease of PS but remained stable in the two other patients. Another patient had 20% ineffective triggering efforts during NPPVClin, which disappeared during NPPVPhys after a 2 cmH2O decrease of PS.

Outcome of the patients

The median duration of NPPV was 3 days (range 1–28 days). All the patients were discharged from the PICU without NPPV except patient 12 who continued NPPV at home. The median length of stay in the PICU was 8 days (range of 3–131 days). Only patient 11 needed an intubation. NPPV failure was explained by a multiple organ failure after 3 liver transplantations and this patient remained in the PICU for 131 days.

Four patients developed transient skin irritation without skin necrosis which resolved after the application of colloid dressing on the facial pressure points. None of the patients developed complications related to NPPV such as gastric distension, oesophageal reflux or barotrauma. Finally, none of the patients died during the PICU stay.

Discussion

This is the first physiological study that demonstrates that NPPV is able to unload the respiratory muscles in a group of selected children admitted to the PICU for acute moderate hypercapnic respiratory insufficiency. This unloading of the respiratory muscles was associated with a significant improvement in alveolar ventilation and gas exchange. Importantly, the efficacy of NPPV set on clinical noninvasive parameters was comparable to a setting based on invasive measurements. The clinical outcome of this group of selected patients was also favourable with no death.

Efficacy of pressure support ventilation

The clinical benefit of NPPV in the PICU in the present study is in agreement with previous clinical studies [3, 4]. More recently, we reported our own 5-year experience of NPPV in a large group of 114 children [2]. The overall NPPV success rate was high (between 67 and 100%), except in the patients with the acute respiratory distress syndrome (22%). The present study included patients with similar primary diagnoses than our descriptive study [2]. We acknowledge that the patients included in the present study represent a small group of patients presenting with moderate hypercapnic respiratory insufficiency due to various diseases. However, almost the uniform benefit of NPPV in this heterogeneous population has to be underscored. Our findings may not be extrapolated to patients admitted with more severe hypercapnic respiratory failure or with acute hypoxemic respiratory failure. Indeed, the benefit of NPPV has been shown to be less obvious in children with severe status asthmaticus [16, 17]. Of note, our results are in agreement with the studies performed in a more stable situation in children with hypercapnic respiratory failure due to cystic fibrosis or severe upper airway obstruction [6, 8–10, 18–20].

Setting of NPPV in the PICU

The ventilatory mode used was PS with PEEP. The levels of PS and PEEP used in our study are comparable to those used in other studies, both in the acute and the chronic setting [3, 4, 9, 17]. In 10 adults presenting an acute lung injury, L’Her et al. [21] found that noninvasive CPAP improved gas exchange but had a minimal effect on respiratory effort compared to PS with PEEP. Other clinical trials suggested that PS might be superior over CPAP in selected patient with acute lung injury with regard to the clinical response and/or the decrease in respiratory effort [22–24].

The sensitivity of the inspiratory and expiratory triggers is of great importance in children, in particular in case of “leak” ventilation such as NPPV. Indeed, the inability of the ventilator to detect the patient’s respiratory effort leads to patient-ventilator asynchrony. In the present study, patient-ventilator asynchrony was more common, occurring in 33% (4/12) of the patients. In adults, ineffective inspiratory efforts and double-triggering are the most common types of asynchrony during PS, with auto-triggering representing less than 1% of the asynchrony events. This was clearly different in our study, with auto-triggering representing the most frequent (3/4) type of patient-ventilator asynchrony. Ineffective triggering may be associated with a high inspiratory pressure, intrinsic PEEP, or a high V t [25, 26]. Auto-triggering may be caused by a too sensitive inspiratory trigger, air leaks, or cardiac oscillations, particularly in young children [27]. In this physiological study, we used the Pes tracing to detect the patient’s inspiratory effort and his synchronisation with the ventilator, as has been performed in a series of previous studies [6, 9, 28]. Airway pressure (Paw) has also been used in some more recent studies [26, 29]. We recorded the Pes trace to evaluate the benefit of NPPV on the patient’s respiratory muscle output. Because of the accuracy and validity of the Pes signal, we also used this variable to assess the patient-ventilator synchrony. However, in the only patient presenting ineffective inspiratory efforts, ineffective triggering detected by the Pes signal correlated with the Paw signal, as reported by Thille et al. [26].

Most ICU ventilators have a cycling mechanism based on the achievement of a preset flow threshold. Leaks around the mask may prevent the airflow reaching the preset expiratory flow which leads to an abnormal prolongation of the inspiratory time and patient-ventilator mismatching. The excellent sensitivity of the ICU ventilator used in our study allowed the use a spontaneous mode, which contributed to the good adaptation and tolerance of the NPPV.

In practice, NPPV is set on clinical noninvasive parameters such as a decrease in RR and an improvement in gas exchange. These parameters have been shown to be of clinical value. Indeed, in our experience, the decrease in RR and PtcCO2 after 2 h of NPPV was significant predictors of the success of NPPV [2]. In another study, the FiO2 required to maintain SaO2 > 94% after 1 h of NPPV, was also predictive of the clinical outcome with NPPV [1]. In the present study, a clinical setting of NPPV was as efficacious as a physiological setting, as has been previously shown in stable patients with advanced cystic fibrosis lung disease [9]. However, as has been observed in children with cystic fibrosis [9], the synchronisation of the patient with the ventilator was generally better with the physiological setting. As such, a more in-depth physiological measurement could be proposed for patients who have difficulties to tolerate or acclimatise to NPPV.

Tolerance of NPPV

The tolerance of NPPV was excellent and only one patient failed NPPV. First, after a careful acclimatisation, NPPV was applied intermittently for a minimum of 2–4 h, two to four times per day. In our study, four patients developed skin injury, which resolved after the application of colloid dressing on the facial pressure points. Patient 10 used alternatively a face mask and the Helmet which has been shown to be an interesting interface in four children with acute leukaemia [30]. Gastro-oesophageal reflux is commonly observed during acute respiratory insufficiency in children and has also been reported during NPPV [31]. Our patients did not complain of gastric distension or emesis which may be explained by the systematic use of a naso-gastric tube connected to room air.

Limitation of the study

The present study included only 12 patients admitted to the PICU for acute moderate respiratory insufficiency due to heterogeneous diseases. However, these patients represent a small but real activity of a polyvalent PICU and it seems that despite the relative heterogeneity of primary diagnosis and age, these patients had an almost uniform increase in respiratory muscle output, which decreased significantly during NPPV. This observation is important for clinicians and probably rather an advantage than a limitation, by showing that NPPV may be a safe and efficient ventilatory support in various cases of moderate hypercapnic respiratory insufficiency in children.

As this study was a physiological study, no control group was included. This precludes a definite conclusion about the potential effectiveness of NPPV to ameliorate the outcome of these patients.

The recruitment of patients was limited by the difficulty to include the patients within the 12 h after initiation of NPPV, only one physician was able to record the physiological parameters, and written consent from the two parents was sometimes unavailable. Also, for technical reasons, we took the option not to include patients with a weight <10 kg.

Because oxygen therapy was systematically added to maintain a target SaO2 of at least 94%, the benefit of NPPV on oxygenation was difficult to establish. However, a beneficial effect of NPPV on oxygenation has been observed in previous studies [1, 3].

Another limitation of this study was the difficulty to record the flow trace during NPPV. Because of abrupt change of flow delivered by ICU ventilator, the flow trace was not analysable during and we took thus the option to use the V t e measured by the ventilator. However, this volume measured by the ventilator and the volume measured at the Y-piece seem to be well correlated in patients [32].

Conclusion

This study is the first study showing that NPPV is able to unload the respiratory muscles in young patients with acute moderate hypercapnic respiratory insufficiency, improving alveolar ventilation and gas exchange. Our data, together with the results of previous clinical and physiological studies, support the benefit of PS and PEEP in this acute situation. Moreover, a noninvasive setting of NPPV was as effective at unloading the respiratory muscles as an invasive approach based on the measurement of respiratory muscle unloading.