Main

Despite advances in neonatal respiratory care, bronchopulmonary dysplasia remains a frequent complication in preterm infants, and invasive mechanical ventilation is a major risk factor (1,2). This led to the use of noninvasive respiratory support to reduce complications associated with invasive mechanical ventilation in preterm infants. Nasal continuous positive airway pressure (NCPAP) is the most common form of noninvasive respiratory support used in these infants. However, studies have shown important failure rates of NCPAP in preventing invasive ventilation or after extubation in the smaller infants (3–5). Efforts to reduce failure prompted the use of nasal ventilation that has been found to be more effective than NCPAP in preventing extubation failure (6–9) and as a primary mode of support in respiratory distress syndrome (RDS) (10–12).

The most commonly reported modes of nasal ventilation are nasal intermittent mandatory ventilation (NIMV) and synchronized NIMV (S-NIMV). However, the effects of nasal ventilation on respiratory function and the possible advantages of synchronization during nasal ventilation have not been fully evaluated.

The objective of this study was to evaluate the physiologic effects of synchronization during nasal ventilation in stable preterm infants. It was hypothesized that S-NIMV will improve tidal volume (VT) and minute ventilation (VE) when compared with NCPAP and nonsynchronized NIMV. Second, it will improve gas exchange, reduce patient effort and chest wall distortion, and decrease frequency of apnea and hypoxemia spells.

METHODS

Patient population.

Clinically stable preterm infants born at <32 wk of gestation and weighing under 1500 g at birth, who required NCPAP were eligible for the study. Infants with major congenital anomalies, requiring sedation, or with ongoing sepsis or meningitis were excluded. The study was conducted at the University of Miami/Jackson Memorial Hospital NICU. The study was approved by Institutional Review Boards of the University of Miami and Jackson Memorial Hospital. All infants were recruited and studied after written informed parental consent was obtained.

In the absence of preliminary data on the primary outcome measure of VT between NIMV and S-NIMV, data from Hummler et al. (13) comparing IMV and SIMV in intubated infants was used to estimate the sample size. Enrollment of at least 16 infants was estimated to detect a 24% within-subject difference in VT between S-NIMV and NIMV with a 0.9 power and 0.05 alpha.

Study protocol.

Infants were studied in their incubators, while they were left undisturbed and after standard nursing procedures were completed. The infants were in supine position throughout the study. During the study, infants underwent a sequence of five 1-h study periods: NCPAP, NIMV at 20/min, NIMV at 40/min, S-NIMV at 20/min, and S-NIMV at 40/min. The sequence of these five periods was determined at random for each infant. There were 10-min intervals for stabilization between study periods. Investigators were not blinded to the treatment mode.

NIMV and S-NIMV were provided through the same short binasal prongs used for NCPAP (INCA prongs; Ackrad Labs, Copper Surgical, Trumbull, CT). All infants had a NCPAP seal (Cannulaide; Beevers Manufacturing, McMinnville, OR). There were no steps taken to maintain the infant's mouth closed (e.g. use of chin straps) or restrain them. The position of the nasal prongs was continuously monitored for dislodgement. NCPAP, NIMV, and S-NIMV were provided by a time-cycled, pressure-limited neonatal ventilator (Infant Star 950; Infrasonics, Inc., San Diego, CA). Synchronization was achieved by a Graseby capsule taped on the abdominal wall. Position of the capsule and trigger sensitivity were adjusted in each patient to achieve minimal trigger delay without autotriggering by observing the timing of esophageal pressure (PES) deflection during spontaneous inspiration, capsule pressure, and the ventilator cycle. Positive end-expiratory pressure (PEEP) was set at the same level as NCPAP before the study, peak inspiratory pressure (PIP) was set at 10 cm H2O above the level of PEEP, and inspiratory time was set at 0.3 s. The ventilator pressure was measured at the connector between the circuit and the nasal prongs. If required, the ventilator's peak pressure or bias flow was adjusted to reach the set PIP. The fraction of inspired oxygen (FiO2) was adjusted to maintain arterial oxygen saturation (SpO2) between 88 and 93%.

Measurements.

Respiratory activity and chest wall instability were measured by respiratory inductance plethysmography (RIP) using two soft elastic bands placed around the abdomen and chest (Respitrace Plus; Sensormedics Corporation, Yorba Linda, CA). These bands were secured with tape. The relative contribution of the chest wall and abdominal signals was determined by the qualitative diagnostic calibration method. The RIP chest wall and abdominal signals were added to yield a third signal, denoted the sum signal that was used to obtain of VT, VE, and respiratory rate (RR).

PES was measured with a size 5-Fr water-filled feeding orogastric catheter with the tip placed in the lower esophagus. First, the insertion length to the stomach and lower esophagus was estimated, and the tip of the catheter was inserted into the stomach. This was confirmed by the positive deflection in pressure during inspiration in synchrony with the outward movement of the RIP band on the abdomen. Subsequently, the catheter was slowly withdrawn until the pressure deflection during inspiration became negative. In this position, the insertion length was noted and the catheter secured. If the inspiratory deflection in pressure was small, a brief airway occlusion was done to confirm transmission of pressure. Patency was maintained during the study by the flushing the catheter. An orogastric feeding catheter was in place throughout the study for feeding and gastric gas decompression. Airway pressure (PAW) was measured at the connection between the ventilator circuit and the nasal adaptor. Both PES and PAW were measured with pressure transducers (Sorenson Transpac 42586-1; Abbot critical care systems, North Chicago, IL) connected to medical grade transducer couplers (Gould Electronics, Inc., Valley View, OH) and calibrated by water manometry.

SpO2 was continuously measured by pulse oximetry (Radical; Masimo, Corp., Irvine, CA). Transcutaneous CO2 tension (TcPCO2) was measured using a transcutaneous carbon dioxide monitor (Microgas 7650; Sensormedics Corporation, Yorba Linda, CA). FiO2 was measured with an oxygen analyzer (O2000; Maxtec, Salt Lake City, UT). All signals were digitized at 100 samples per second and recorded on a personal computer (AT-CODAS; Dataq Instruments, Akron, OH).

Patient comfort was assessed at the beginning of each 1-h period and every 20 min by two unmasked researchers using a score system simplified from a validated scale (14). This score included sleep state (0, sleep and 1, awake), calmness/agitation (0, calm; 1, mildly agitated; and 2, very agitated), crying (0, no and 1, crying), movement (0, no; 1, slight; and 2, vigorous), and facial expression (0, normal and 1, contorted/grimacing). Abdominal circumference was measured before and at the end of each period.

Data analysis.

The mean VT from all spontaneous and ventilator cycles (S-NIMV or NIMV) during each mode of support was obtained from the RIP sum signal during each 1-h recording period using computerized analysis without operator intervention. VE was derived from the mean VT and RR of every minute for each mode of support. VT is reported in arbitrary units (AU), VE in AU per minute, and RR in breaths per minute. Apnea episodes, defined as respiratory pauses lasting >20 s in the RIP sum signal, were counted for each mode of support. Computerized analysis of the entire 1-h periods was used to determine mean TcPCO2, SpO2, and FiO2 as well as episodes with SpO2 <85%.

Data on parameters of breathing effort and chest wall distortion were calculated from all consecutive artifact-free breaths during 1 min and every 15 min throughout each 1-h period. These included nonassisted spontaneous breaths, ventilator cycles in synchrony with spontaneous inspiration, nonsynchronous ventilator cycles, and ventilator cycles during periods of apnea. Breaths were considered unsuitable for analysis if the PES or RIP signals drifted or if there was movement artifact. Per-breath spontaneous inspiratory effort was determined by peak negative PES during inspiration (PES peak inspiratory) and the area under the PES curve (PES inspiratory area). Minute effort (PES minute inspiratory area) was calculated as the product of PES inspiratory area and RR. Active spontaneous expiratory effort was determined by peak positive PES during exhalation (PES peak expiratory).

To analyze the per-breath effects of timing of ventilator cycle with respect to spontaneous breath, VT, PES inspiratory area, PES peak expiratory, and expiratory duration were obtained from NIMV cycles delivered during the first half of inspiration, second half of inspiration, first half of exhalation, or second half of exhalation. These parameters were compared with those obtained from S-NIMV cycles and from nonassisted spontaneous breaths during NCPAP, or between NIMV or S-NIMV cycles.

Thoracoabdominal motion was analyzed by determining the phase delay between abdominal and chest wall expansion expressed in degrees as a fraction of the full respiratory cycle represented by 360°. The ratio of the displacement of the ribcage and abdomen compartments to the net volume displacement was denoted as the total compartment displacement ratio (TCD ratio). TCD ratio was calculated as the sum of the absolute values of the abdominal and thoracic signal's area under the curve divided by their algebraic sum (sign included). In cases where no chest wall distortion is present, the TCD ratio is equal to 1. Increasing values >1 in TCD ratio reflected greater chest wall distortion with more displacement than volume achieved.

Within-subject comparisons of the five study periods was done using repeated measures ANOVA or repeated measures ANOVA on ranks when the data were not normally distributed. Post hoc analysis to determine between periods differences was done using the Student-Newman-Keuls and Holm-Sidak methods, respectively. A two-tailed p < 0.05 was considered significant.

RESULTS

Sixteen preterm infants were studied. Their birthweight (median and interquartile range) was 928 (812–1130) g and GA was 27.5 (25.5–30.0) wk. At the time of study, they were 10 (5–19) d old, 30 (29–31) wk postmenstrual age, and weighed 960 (870–1190) g. Of the 16 infants studied, 10 were male. Fifteen infants had been mechanically ventilated before the study, and 12 had received surfactant. Ten of them were on 0.21 FiO2 and six required supplemental oxygen ranging from 0.23 to 0.30 FiO2 at the time of study. All infants were receiving caffeine.

All 16 infants completed the five 1-h study periods, and there were no observed adverse events. Abdominal circumference did not change between study periods.

The mean VT obtained from all spontaneous breaths and ventilator cycles (NIMV or S-NIMV), RR and the resulting VE did not differ significantly between the NCPAP, NIMV, and S-NIMV periods. Measures of gas exchange including mean levels of TcPCO2, SPO2, and FiO2 did not differ between periods (Table 1).

Table 1 Ventilation and gas exchange
Full size table

Spontaneous inspiratory effort was reduced during S-NIMV compared with NCPAP and NIMV (Table 2). This was reflected in a reduction in per-breath PES peak inspiratory during S-NIMV at 40/min compared with NCPAP and a reduction in per-breath PES inspiratory area during S-NIMV at 40/min compared with all other study periods. Minute inspiratory effort evaluated by PES minute inspiratory area was reduced during S-NIMV at 20/min compared with NCPAP and NIMV at 20/min. The reduction in PES minute inspiratory area was more striking during S-NIMV at 40/min. In contrast, inspiratory effort during NIMV did not differ from NCPAP.

Table 2 Spontaneous inspiratory effort, active spontaneous expiratory effort, and chest wall distortion
Full size table

The reduction in inspiratory effort during S-NIMV is illustrated in Figure 1. This representative recording during S-NIMV at 40/min shows reductions in PES peak inspiratory and PES inspiratory area temporally associated with S-NIMV cycles compared with nonassisted spontaneous breaths. VT during the S-NIMV cycles also decreased slightly.

Figure 1
figure 1

Representative recording during S-NIMV at 40/min. Tracings of PES, PAW, and VT during S-NIMV at 40/min show reductions in inspiratory effort (smaller PES peak inspiratory and PES inspiratory area) temporally associated with S-NIMV cycles compared with nonassisted spontaneous breaths. VT during the S-NIMV cycles was decreased slightly.

Full size image

Active spontaneous expiratory effort evaluated by PES peak expiratory increased during NIMV at 20, S-NIMV at 20/min, and NIMV at 40/min compared with NCPAP. In contrast, active exhalation pressure did not increase during S-NIMV at 40/min compared with NCPAP (Table 2). Chest wall distortion indices of phase shift angle and TCD ratio did not differ between NIMV, S-NIMV, and NCPAP (Table 2).

Analysis of the timing of ventilator cycle with respect to the spontaneous breath showed that 88 and 89% of ventilator cycles were delivered in synchrony (defined as delivered within the first half of spontaneous inspiration) during S-NIMV at 20/min and 40/min, respectively. This compared with only 21 and 23% of ventilator cycles during NIMV at 20/min and 40/min, respectively. Asynchrony was frequent during NIMV at 20/min and 40/min with 56% of ventilator cycles delivered during spontaneous exhalation. In contrast, only 5 and 6% of ventilator cycles were delivered during exhalation during S-NIMV at 20/min and 40/min, respectively. Although there were no significant differences in the mean VT from spontaneous and mechanical cycles obtained during each mode of support, there was considerable between patient variability. In half of the infants, VT from S-NIMV and NIMV cycles delivered during inspiration was larger than the VT from spontaneous breaths.

There were important effects related to the timing of ventilator cycle during NIMV at 20 and 40/min. This is shown in Table 3 for NIMV at 20/min (data during NIMV at 40/min is not shown to avoid redundancy). NIMV cycles delivered late in spontaneous inspiration (second half) and during the spontaneous exhalation prolonged the duration of exhalation and delayed the onset of the next spontaneous inspiration compared with spontaneous breaths during NCPAP and nonassisted spontaneous breaths during NIMV.

Table 3 Per-breath effects of the timing of ventilator cycles during NIMV and S-NIMV at 20 cycles per minute
Full size table

Only NIMV cycles delivered within the first half of spontaneous inspiration produced a decrease in per-breath spontaneous inspiratory effort as shown by a significantly lower PES inspiratory area. As expected, this reduction was similar to that observed during cycles. VT obtained from spontaneous breaths during NCPAP, nonassisted spontaneous breaths between NIMV cycles, NIMV cycles delivered early or late in inspiration or during exhalation, nonassisted spontaneous breaths between S-NIMV cycles or between S-NIMV cycles did not differ significantly. However, there was considerable between-patient variability.

NIMV cycles delivered during exhalation produced an increase in expiratory effort to actively exhale against the ventilator, as reflected by a significantly greater PES peak expiratory (Table 3). This is illustrated in Figure 2 with a representative tracing during NIMV at 20/min showing the positive deflection in PES and a prolonged expiration with a slow decline in VT during NIMV cycles delivered in expiration.

Figure 2
figure 2

Representative recording during NIMV at 20 cycles per minute. Recordings of PES, PAW, and VT during nonsynchronized NIMV at 20/min showing increased positive pressure deflection in PES during exhalation and prolongation of the expiratory phase with NIMV cycles delivered at the beginning of exhalation compared with nonassisted spontaneous breaths. The prolonged exhalation is evidenced by a slower decline in volume.

Full size image

Episodes with SpO2 <85% and percent of time spent with SpO2 <85% did not differ between study periods. Episodes of apnea or bradycardia were not observed. The patient comfort scale did not show differences between NCPAP, NIMV, and S-NIMV.

DISCUSSION

Contrary to the hypothesis, S-NIMV did not increase the mean VT and VE compared with NCPAP or NIMV in this group of clinically stable preterm infants. Nonetheless, synchronization produced a significant reduction in spontaneous breathing effort compared with NCPAP that was more striking at the higher S-NIMV rate. In contrast, there was no reduction in breathing effort during nonsynchronized NIMV.

The lack of effect of S-NIMV on ventilation and gas exchange may be because these infants were stable on NCPAP. With spontaneous ventilation and lung recruitment maintained adequately during NCPAP, they only reduced their effort in response to the assistance provided by S-NIMV. Previous studies reported similar findings with synchronized nasal ventilation at similar cycling rates (15,16). Conversely, Moretti et al. (17) reported increased VT and VE with S-NIMV compared with NCPAP immediately after extubation. This can be in part explained by the greater ventilatory demands during the transition to NCPAP and the use of a relatively low PEEP (3 cm H2O) in that trial.

The effect of NIMV or S-NIMV cycles on VT varied considerably between infants. Compared with the VT from spontaneous breaths, VT from S-NIMV cycles and NIMV cycles delivered during inspiration was greater in 50% of the infants. The factors that influence the increase in VT remain to be determined.

In this study synchronization during nasal ventilation reduced spontaneous breathing effort. However, this effect was not observed during NIMV. Further analysis revealed that only NIMV cycles delivered early in the spontaneous inspiration decreased spontaneous inspiratory effort. These findings suggest that synchronization during nasal ventilation successfully unloads the infant's respiratory pump.

It can be argued that the smaller negative deflection in PES is because of the transmission of the positive pressure instead of a reduction in inspiratory effort. However, the magnitude of the pressure transmitted is smaller at lower GA because of the highly compliant chest wall (18). Thus, the pressure transmitted is likely smaller than the reduction in inspiratory pressure observed in this study during S-NIMV.

The lack of a consistent increase in ventilation was unexpected and did not fully agree with studies showing advantages of S-NIMV (with the same ventilator) compared with NCPAP immediately after extubation (6–9) or NIMV in infants with RDS (10,11). However, a reduced inspiratory effort with S-NIMV could facilitate the maintenance of ventilation shortly after extubation when preterm infants have to compensate for the withdrawn ventilatory support. Khalaf et al. (8) showed that the benefits of S-NIMV after extubation were greater among infants with worse lung mechanics in whom the respiratory pump has to overcome increased mechanical loads to generate the required VT. In contrast, the infants included in this study had already reached stability on NCPAP and did not need to overcome the challenges experienced by recently extubated infants.

The reduced inspiratory effort temporally associated with S-NIMV cycles was accompanied by a slightly reduced VT, suggesting the S-NIMV cycles were sensed by the infant but the transmitted pressure was not sufficient to increase VT. A similar effect was observed in NIMV cycles delivered during inspiration. However, NIMV cycles delivered in expiration prolonged the duration of expiration and often elicited an active effort to exhale.

The reduction in inspiratory breathing effort observed during S-NIMV cycles, also seen during NIMV cycles delivered in early inspiration, suggests that timely delivery of the positive pressure plays an important role in achieving respiratory unloading during nasal ventilation. The synchronous cycles did not increase VT; therefore, it is unlikely that the decreased inspiratory effort was related to stretch receptor activity. The reduction in inspiratory effort may be mediated by a reflex activated by a jet of gas flow into the nasal passages (19).

Although adequate synchrony between the ventilator cycle and the spontaneous inspiration was achieved for the purposes of the study using the Graseby capsule, similar rates of synchrony may not be achieved during routine use and without the guidance of the PES deflection during inspiration. Nonetheless, this method of synchronization was used in clinical trials that showed advantages of S-NIMV compared with NCPAP after extubation (6–8).

In this study, asynchrony during NIMV cycles delivered late in the spontaneous inspiration or during exhalation disrupted the infant's spontaneous breathing pattern. These elicited active expiratory efforts against the ventilator cycle, delayed the spontaneous exhalation, and consequently delayed the onset of the next inspiration. This disturbance of the infant's respiratory rhythm was likely mediated by pulmonary stretch receptor activity. Surprisingly, the better infant-ventilator interaction during S-NIMV did not improve gas exchange. However, the duration of each mode was relatively short, and the study has not captured important effects that may occur over longer periods.

Thoracoabdominal asynchrony is common in preterm infants, and synchronization can improve chest wall stability and reduce asynchrony (16,20). In this study, there was minimal chest wall distortion during NCPAP perhaps reflecting the stability of the infants or their postconceptional age and S-NIMV or NIMV had minimal effects.

In this group of infants, the frequency of apnea and hypoxemia spells was minimal during NCPAP and did not change between study periods. Therefore, the advantages of NIMV or S-NIMV need to be evaluated in infants showing more frequent spells of apnea or hypoxemia during NCPAP than the population included in this study.

A previous study reported discomfort in preterm infants receiving NIMV and NCPAP compared with spontaneous breathing (21). In this study, there were no differences between NCPAP, NIMV, and S-NIMV in terms of the patient comfort scale. Increased risk of developing gastric and intestinal perforation with nasal ventilation has also been reported (22). In this study, there were no differences in abdominal girth between or during NCPAP, NIMV, and S-NIMV. However, because of the brief duration of each mode, these data cannot rule out the risk of gas being pushed in to the stomach during longer periods of nasal ventilation.

There were minimal differences on ventilation or gas exchange between ventilator rates of 20 and 40 cycles per minute with NIMV or S-NIMV. However, this group of stable preterm infants was likely more capable to compensate for the reduced support than sicker or less stable infants. Moreover, the finding of reduced breathing effort with S-NIMV at 40 versus 20/min should be interpreted with caution until longer term data show that higher S-NIMV rates do not increase the risk of adverse effects.

An important limitation of this study is that it documented only short-term effects of NIMV and S-NIMV. It cannot be excluded that there may be more striking effects of synchronization during longer periods of support. Although PIP was kept at 10 cm H2O above PEEP during the study, it may have been insufficient because of the resistance of the upper airway or because variable gas leaks (e.g. opening of the mouth) can affect pressure transmission during nasal ventilation. A possible limitation of this study is the ability to detect small changes in VT by noninvasive measurements with RIP. However, Brooks et al. (23) showed a significant linear relationship between VT measured by pneumotachography and RIP over several hours, and this relationship was not affected by position, chest wall asynchrony, RR, or routine care.

In conclusion, ventilation and gas exchange did not improve during NIMV or S-NIMV compared with NCPAP and did not differ with synchronization in this group of clinically stable infants. Synchronized nasal ventilation reduced spontaneous breathing effort and prevented active exhalation. These effects were more striking at higher S-NIMV rates. Nasal ventilation without synchronization did not achieve a reduction in inspiratory effort and seemed to have no short-term advantages over NCPAP in relatively stable preterm infants. These findings cannot be extrapolated to sicker infants who could benefit more from nasal ventilatory support. The advantages of synchronization in such infants remain to be determined.