Introduction

Different ventilatory modes are used for non-invasive mechanical ventilation (NIMV). Volume-targeted ventilation is characterized by the delivery of a fixed, predetermined tidal volume (VT). Thus, the main advantage of assist-control/volume-targeted (AC/VT) ventilation is that a guaranteed minimal VT is delivered, but this can result in detrimentally high inspiratory airway pressures causing discomfort and poor tolerability. AC/VT ventilation is suitable for patients with neuromuscular diseases where the ventilator acts as a substitute for the weakened respiratory muscles, which are unable to trigger the ventilator [1]. However, a relatively high back-up rate (2–3 breaths lower than the spontaneous respiratory rate of the patient) is required to avoid nocturnal desaturation and, as a consequence, many patients adopt a controlled mode without triggering the ventilator.

Another more recent mode of ventilation is pressure support (PS). PS ventilation is pressure-targeted and each breath is triggered and terminated by the patient and supported by the ventilator [2]; the patient can control his respiratory rate, duration of inspiration (Ti) and VT. This explains the relative ease in adapting to, and the greater comfort and synchrony of, this mode. In contrast to volume-targeted ventilation, VT is not predetermined but depends on the level of PS, the inspiratory effort of the patient and the mechanical properties of the patient’s respiratory system. During this mode, since there are no mandatory breaths present, an in-built low frequency back-up rate is used to prevent episodes of apnea. Furthermore, because the breaths are triggered by the patient, the sensitivity of the trigger is crucial.

Although studies have compared these two different modes [3, 4, 5, 6], the influence of the back-up rate during PS and AC/VT has not been thoroughly investigated. On the one hand, in the case of a low back-up rate, the patient can select the respiratory rate, on the other hand, a high back-up rate reduces the respiratory effort required to trigger the ventilator. We recently observed, in a study comparing PS and AC/VT ventilation in children with cystic fibrosis (CF), that the adoption of a controlled breathing pattern during AC/VT ventilation was present in almost a third of the patients [7]. In addition, this controlled pattern of breathing was associated with a greater decrease in the work of breathing and also a greater subjective comfort rating [7]. The beneficial effect of an increased back-up rate could be particularly important if the trigger sensitivity of the ventilator is poor.

In the present study, we wanted to address two questions: does an increase in back-up rate result in an improvement in respiratory effort and alveolar ventilation? And, second, is the influence of the back-up rate on respiratory effort affected by trigger sensitivity? To address these questions we first performed an in vitro laboratory-based study comparing the quality of the inspiratory trigger in seven domiciliary ventilators able to deliver PS (n=4) or AC/VT ventilation (n=3). Following this, we performed a clinical study in young patients with CF, comparing changes in respiratory effort that occur with differing frequencies of back-up rate in both PS and AC/VT modes.

Material and methods

Experimental in vitro study

The characteristics of the ventilators that were tested are shown on Table 1. Each ventilator examined was tested at its most sensitive pressure and/or flow trigger in a situation mimicking a normal (P0.1 of about 1.8 cmH2O and an airway pressure [Paw] swing of 11 cmH2O) or a greater (P0.1 =3.2 cmH2O, Paw swing =18 cmH2O) breathing effort, as previously described [8] (Fig. 1). The sensitivity of the triggers was analyzed as previously described [8, 9] (see online supplement).

Table 1 Characteristics of the ventilators evaluated during the bench study
Fig. 1
figure 1

Two-chamber Michigan test lung

Clinical study

Patients

Patients with CF were enrolled if they were naive for NIMV and fulfilled the criteria for NIMV according to the consensus conference [10]. Inclusion and exclusion criteria were those previously described [7] (see online supplement). Written informed consent was obtained from each patient and his/her parents.

Experimental apparatus and adjustment

As the bench study showed that all the AC/VT ventilators shared very poor trigger systems, the ventilator that was most commonly used for domiciliary ventilation was chosen for the physiological study (Eole 3, Saime, Savigny le Temple, France). For the PS mode, the ventilator with the best performance was chosen for the physiological study (Onyx, Mallinckrodt, Les Ulis, France). No additional oxygen was used during the study.

During the 2 days prior to the protocol, a training session was performed as previously described [7]. In brief, for PS ventilation, inspiratory pressure was increased to the maximum level tolerated; for AC/VT ventilation, VT was increased up to the maximum VT tolerated (not exceeding 15 ml . kg−1) [10]. For the two modes, the triggers were pressure triggers, set to the lowest value which did not induce auto-triggering. For the AC/VT mode, the inspiratory/expiratory time ratio (I/E) was set at its usual value i.e. 1/2. For the two modes, the back-up rate represented the minimal number of breaths delivered by the ventilator per minute. The lowest back-up rates were 4/min and 10/min for PS and AC/VT ventilation, respectively. The back-up rate was set two to three breaths below the spontaneous respiratory rate [10] and was then increased until the patient adopted a completely controlled pattern of breathing or could not tolerate a higher back-up rate. This individually determined maximally tolerated ventilatory rate was defined as Fmax.

Protocol

All studies were performed in the afternoon, about 1 h after the second chest physiotherapy session of the day. Throughout the study, patients were studied in the semi-recumbent position and were instructed to close their mouths to limit air leaks. The nasal masks were industrial masks (Respironics, Murrysville, PA) or custom-made masks. Three separate protocols were performed: first, a period of spontaneous breathing (SB), followed, in a random order, by a period each of AC/VT and PS ventilation. In between each period of either AC/VT or PS ventilation there was a period of SB. For each AC/VT and PS mode, three 20-min periods of either Fmax, Fmax minus 5 breaths . min-1 (F-5) and Fmax minus 10 breaths . min-1 (F-10) were performed, again in a random order.

Measurements

Respiratory flow, airway pressure (Paw), pulse oximetry (SaO2) end-tidal CO2 (PetCO2), esophageal (Pes) and gastric pressure (Pga) were recorded during the last 5-min of each session, as previously described, ensuring that a stable steady state breathing pattern was established [7, 11] (see ESM supplement).

Data analysis and assessment of patient’s effort

Breathing parameters were determined from flow tracing. Esophageal and diaphragmatic pressure/time products (PTPes and PTPdi) were measured as previously described and were expressed per breath and per minute [7].

Statistical analysis

The data are given as means ± SD. Since the three basal conditions were statistically identical (ANOVA), the first basal condition was used as the SB period in the following statistics. Comparison during PS and AC/VT ventilation at Fmax and SB was made with a repeated measures ANOVA with one factor. Differences between ventilatory rates with the two modes were assessed by repeated measures analysis of variance ANOVA with two factors, i.e. the ventilatory mode (PS vs AC/VT) and the back-up rate. When ANOVA appeared appropriate (F test with a p value below 0.05), pairwise comparisons were performed using Bonferroni test. A p value below 0.05 was considered as the limit of significance.

Results

Experimental in vitro study

Important differences in terms of trigger pressure, trigger time delay, trigger pressure/time product (PTP) and the slope between flow and pressure from the time of the initial negative pressure up to this time +0.03 s (SBFP) were observed between the different ventilators (Fig. 2). Only three ventilators had adequate triggers in the different conditions (Onyx, Neftis, Home 2). The best performing PS ventilator was the Onyx, which is a PS device that had the shortest time delay in all conditions. Another PS device, Neftis, was slightly slower (ΔT <0.15 s). A volumetric ventilator, the Home 2, had the longest time delay (ΔT >0.7 s): the high ΔT implies that this ventilator is unable to pressurize the airway rapidly. The analysis of the trigger PTP yielded similar results (data not shown). Helia 2, Helia S and Achieva with volume triggering were not able to trigger during the highest positive end-explicatory pressure (PEEP) value, whilst Eole 3 and Achieva with pressure triggering were not able to detect the lowest value of breathing effort without PEEP. The Achieva, which can function as a volume- or a pressure-targeted ventilator, can be used with a pressure or a flow trigger. But the pressure trigger time delay of this device was two to three times longer than the Onyx and, in fact, this device was not able to detect a breathing effort without PEEP.

Fig. 2 a
figure 2

Trigger pressure (ΔP), b Trigger time delay (ΔT) c The slope between the flow and pressure (SBFP) from the time of the initial negative pressure up to this time +0.03 s for the seven ventilators that were tested in the bench study. PEEP positive end-expiratory pressure, NP the ventilator was not manufactured to generate PEEP, ND the ventilator did not detect the breathing effort in the considered condition, P-Tr pressure trigger, V-Tr volume trigger

Clinical in vivo study

Patients

Ten patients (7 girls, mean age 15.0±4.4 years, range 9–20 years) were included in the study. Mean vital capacity and forced expiratory volume in 1 s (FEV1) were 44.6±15.1 and 27.5±11.0% of the predicted value, and mean partial arterial oxygen and carbon dioxide pressures were 61.3±10.7 and 43.1±3.4 mmHg, respectively. All the patients tolerated the gastroesophageal catheter and the two NIMV sessions. All the patients had excessive daytime fatigue with sleep disturbance. Mean PS was 14±2 cmH2O and mean VT during AC/VT ventilation was 0.72±0.22 l. Because of the absence of intrinsic PEEP in our population, PEEP was not added [12]. Except for one patient, for whom Fmax was 10 cycles lower during the AC/VT mode than during the PS mode, there were less than 5 cycles . min−1 difference for Fmax between the two modes for all the patients (Table 2).

Table 2 Breathing pattern, gas exchange and respiratory effort during spontaneous breathing and the two ventilator frequencies (Fmax maximal tolerated back-up rate, F-5 Fmax minus 5) during pressure-support (PS) and volume-targeted ventilation (AC/VT) in the ten patients

Six patients did not tolerate F-10 during AC/VT. The repeated measures ANOVA with two factors was thus performed with only two conditions of back-up rate (Fmax and F-5).

Breathing pattern and gas exchange

No patient-ventilator trigger asynchrony, defined as a decrease in Pes not followed by a ventilator-triggered breath, was observed during the two ventilatory modes at any frequency. During the controlled breaths, a respiratory effort was observed, but not before or after the patient’s inspiration was visualized on the flow trace. For the whole group, when compared to SB, there was a significant increase in VT and minute ventilation (VE) during the two ventilatory modes at Fmax (Table 2). Respiratory rate decreased significantly only during AC/VT ventilation. When comparing Fmax and F-5 (two-way ANOVA), there was a moderate, but significant, decrease in VE that was explained by a smaller VT during PS ventilation and a lower respiratory rate during AC/VT ventilation (Table 2). The only difference between the two modes was a larger VT during AC/VT ventilation compared to PS ventilation (p=0.03).

It is important to note that the decrease of the back-up rate was associated with an significant increase in Ti (from 1.13±0.29 to 1.51±0.56 s during Fmax and F-5, respectively, p=0.02) and a decrease in mean inspiratory flow (VT/Ti) (from 0.62±0.07 to 0.49±0.08 l . s−1 during Fmax and F-5, respectively, p=0.01) and maximal inspiratory flow (Vimax) (from 0.87±0.11 to 0.72±0.11 l . s−1 during Fmax and F-5, respectively, p=0.01) during the AC/VT mode. Curiously, for all the patients, Ti during AC/VT did not correspond to [(1/3)×60/respiratory rate] as expected according to the ventilator adjustment but to [(1/3)×60/back-up rate]. A significant increase in SaO2 and a significant decrease in PetCO2 were observed during both ventilatory modes at Fmax compared to SB. At F-5, this improvement of gas exchange was observed for SaO2 but not for PetCO2 during both PS and AC/VT ventilation.

Respiratory effort

For the whole group, all indices of respiratory effort were significantly reduced during PS and AC/VT compared to SB (Table 2). The two-way ANOVA analysis revealed that these indices were significantly lower during Fmax than during F-5 (Table 2). With regard to the indices of respiratory effort, no differences were noted between the two ventilatory modes (Table 2).

Individual values for PTPes/min and PTPdi/min are presented in Fig. 3. The four patients who tolerated AC/VT at F-10 had higher PTPes/min during this back-up rate than during F-5. In addition, three of them had higher PTPes/min during AC/VT at F-10 than during SB and, clearly, decreasing the back-up rate was systematically associated with an increase in PTPes/min during AC/VT (Fig. 3). This effect was less obvious during PS (Fig. 3). However, no differences in the PTPes and PTPdi per breath and per minute were observed between PS and AC/VT at Fmax and F-5 (Table 2).

Fig. 3
figure 3

Individual values of the esophageal pressure time product (PTPes/min) of the ten patients during spontaneous breathing (SB), the maximal tolerated back-up rate (Fmax), Fmax minus 5 breaths (F-5) and Fmax minus 10 breaths (F-10) during pressure support (PS) ventilation and assist-control/volume-targeted (AC/VT) ventilation. At F-10, two and six patients did not tolerated PS or AC/VT ventilation, respectively (open symbols), the patients who tolerated the three conditions are represented by closed symbols

Effect of the increase in the back-up rate on the inspiratory triggering

No change was observed in Ti and the Ti/Ttot ratio with the different back-up rates during PS ventilation. Unexpectedly, during AC/VT ventilation Ti/Ttot failed to be equal to 1/2 according to the I/E ratio setting at 1/2 (see the Method section). In fact, as underlined above, for each patient Ti did not depend on the effective respiratory rate but on the back-up rate setting, i.e. was inversely correlated to the back-up rate. Indeed, Ti increased during the decrease in back-up rate from 1.13±0.29 during Fmax to 1.51±0.56 during F-5, p=0.007, and the Ti/Ttot increased from 36.8±2.3% during Fmax to 41.4±3.9% during F-5, p=0.01.

Table 3 compares the inspiratory triggers of the PS ventilator and the volume-targeted ventilator at Fmax. All trigger characteristics (ΔP, ΔT and PTP) were significantly better for the PS ventilator.

Table 3 Characteristics of the inspiratory trigger of the pressure-support (PS, Onyx) and the assist-control/volume-targeted (AC/VT, Eole) devices used for the clinical study in the ten patients

The relationship between the percentage of triggered cycles and the change in back-up rate is shown in Fig. 4. At Fmax during PS ventilation, four patients triggered more than 80% of the breaths, three patients triggered between 40 and 80% and three patients triggered less than 40%. During AC/VT ventilation, the corresponding patient numbers were seven, one and two, respectively (Fig. 4). No clinical or functional parameters, such as the spontaneous respiratory rate, the severity of the lung disease assessed by FEV1 or the baseline arterial blood gases, were associated with the adoption of a controlled pattern of breathing in either the PS or AC/VT mode.

Fig. 4
figure 4

Percentage of assisted breaths during the maximal tolerated back-up rate (Fmax), Fmax minus 5 breaths (F-5) and Fmax minus 10 breaths (F-10) during pressure support (PS) ventilation and assist-control/volume-targeted (AC/VT) ventilation

For all the patients who had at least 40% of controlled breaths at Fmax, a mean of 20 controlled breaths per patient was compared to a similar number of assisted breaths. There were no significant differences in VT, respiratory rate and VE between controlled and assisted breaths during either PS or AC/VT ventilation (data not shown). Despite the fact that the PTPs were lower in the controlled, rather than assisted, mode during both pressure- and volume-cycled ventilation, this difference was only significant for PTPes/breath during AC/VT ventilation (with a mean PTPes/breath of 6.5±2.3 and 3.9±2.3 cmH2O.s during assisted and controlled breaths, respectively, p=0.05).

Discussion

The main finding of this study is that the level of back-up rate during NIMV is important in determining the respiratory effort reduction, as judged by the PTP in young patients with CF. Furthermore, a low back-up rate was either not tolerated or associated with a significant increase of the respiratory effort. This negative effect was mainly observed with the AC/VT ventilator and was explained by the negative effect of the I/E ratio setting in this mode. Indeed, our study confirmed that, during the AC/VT mode, the I/E ratio was not calculated on the patient’s respiratory rate but on the back-up rate setting. In the AC/VT mode, a decrease in the back-up rate induced an increase in Ti and, consequently, a decrease in Vimax and VT/Ti, which may contribute to the increase in the respiratory effort and the intolerance of the AC/VT mode at a low back-up rate. In addition, the in vitro study suggests that the sensitivity of the inspiratory trigger in the AC/VT ventilator is, in general, less effective than that of a PS device. These findings support the assumption that careful adjustment of the back-up rate during the AC/VT mode and an improvement in the quality of the inspiratory triggers could enhance the efficacy of NIMV in young patients with CF. Finally, because of an easier set-up and better tolerance, the PS mode should be the first option in these patients.

In vitro study

As previously observed by Lofaso et al. [8], the in vitro study demonstrated the marked heterogeneity of trigger characteristics in the different ventilators tested. The optimum ventilator in terms of trigger efficacy was the Onyx, with an excellent performance being observed both in the in vitro and the in vivo study. In contrast, the Achieva ventilator, which uses a flow trigger, appeared to have a problem with the resistance of the inspiratory line. Indeed, the resistance of the inspiratory line has to be as low as possible to reach the trigger volume with the minimum patient effort. The SBFP of the Achieva obtained with PEEP (>20 cmH2O . l-1 . s−1) suggests more a closed than open system, which probably explains the poor results of the Achieva-VT with PEEP (no detection or a very long trigger time delay and a large PTP). In contrast, when the SBFP is smaller (approximately 6.5 cmH2O . l-1 . s−1), i.e. when there is no PEEP, the Achieva-VT performs very well. The same problem was observed with the volume-targeted device, Eole. Its 11 cmH2O . l-1 . s−1 SBFP was the highest one observed without PEEP. This probably explains its detection problem with the “normal breathing effort”.

In conclusion, this in vitro study demonstrated that the performance of inspiratory triggers in the PS ventilators was generally better than the AC/VT ventilators. PEEP may influence the trigger sensitivity of some devices, probably by modifying the impedance of the inspiratory circuit at the end of the expiration. Some manufacturers may have appreciated this impedance when designing inspiratory triggers based on flow detection. Clearly, this type of detection needs an open inspiratory circuit in order to obtain a flow at the initiation of the inspiratory effort, whereas a pressure trigger needs to be rather a closed inspiratory circuit.

In vivo study

A decrease in respiratory effort, as assessed by PTPes and PTPdi, was observed in young patients with CF during the application of PS and AC/VT ventilation, when compared to SB. Furthermore, the increase of the back-up rate resulted in a greater decrease in the respiratory effort. This effect was more pronounced during AC/VT ventilation because of the automatic changes of I/E ratio that caused different inspiratory flows. Thus, attention should be paid when setting up the ventilator, with particular focus on the I/E ratio. This, combined with an improved trigger performance, could improve the efficiency of long-term domiciliary NIMV delivered in the AC/VT mode in patients with CF.

In the clinical study we used the two PS and AC/VT ventilators that are most commonly used for home mechanical ventilation in France and which have been previously compared in a similar CF population [7]. The different setting of the I/E ratio represents the major difference between the AC/VT and PS modes. During PS, the positive pressure wave is synchronized with the inspiratory effort of the patient and is, thus, both patient-initiated and patient-terminated. This means that the patient can keep a certain control over the respiratory rate, the Ti and the expiratory time (Te). In contrast, during AC/VT ventilation, Ti is predetermined and fixed according to the back-up rate I/E ratio set on the ventilator. As a consequence, Ti increases when the back-up rate decreases and is independent of respiratory rate, which explains the delay in the pressurization of the airway at a low back-up rate and, thus, the inability of the ventilator to satisfy the patient’s demand.

It has been previously demonstrated that the flow rate at the start of inspiration affects respiratory output [13]. In a study of ICU patients, Bonmarchand et al. found that a high inspiratory flow rate at the start of inspiration during PS was associated with a greater decrease in diaphragmatic activity in adult patients with obstructive lung disease [13]. Similarly, in a recent study comparing several PS devices in critically ill patients undergoing weaning from mechanical ventilation, we found that the PS device associated with the highest initial flow rate was more efficient in reducing respiratory output than the PS device associated with the lowest initial flow rate [8]. In a similar population, Cinnella et al. compared an AC/VT mode and a pressure-targeted mode at similar VT and Ti [4]. They observed that pressure-targeted mode reduced the respiratory work more effectively than AC/VT mode, but only at moderate VT (=8 ml . kg−1). When VT was equal to 12 ml . kg−1, they did not find any differences between the pressure-targeted mode and the volume-targeted mode, which can be explained by the increase in the initial flow rate. This difference in initial inspiratory flow rate explains, in part, the similarity in respiratory effort observed between the AC/VT mode and the PS mode despite a significantly greater VT during AC/VT ventilation. The greater peak inspiratory flow during PS, compared to AC/VT ventilation, which was also observed during a previous study in young patients with CF [7], can explain why these two modes result in a similar reduction of the respiratory effort.

In conclusion, the setting of the Ti ratio according to the back-up rate set on the ventilator, and not according to respiratory rate, was a technical characteristic that was found on all the volume-targeted ventilators that were evaluated in the in vitro study. The effect of the back-up rate during AC/VT ventilation is, thus, principally explained by this technical limitation. In practice, clinicians must be aware of this ventilator property and should adapt the I/E ratio setting if using an AC/VT device with a low back-up rate.

The effect of the back-up rate was clearly different during PS ventilation. Although the trigger of the PS device was set at its most sensitive value, a significant proportion of the respiratory effort performed by the patient was related to the triggering of the ventilator. For the PS ventilator used in the clinical study, the inspiratory trigger was extremely sensitive and has previously been shown to have similar trigger sensitivity to an intensive care ventilator [8]. This high sensitivity of the inspiratory trigger explains, in part, why all the patients tolerated a low back-up rate (Fig. 3). Also, most patients maintained an assisted mode during PS and the effect of an increase in the back-up rate was moderate and of a similar magnitude to what has been reported in previous studies. Indeed, when increasing the back-up rate, PTPes/min decreased by approximately 5%. This reduction in the respiratory effort is slightly less significant than that observed when different trigger sensitivities were evaluated in adult patients in the intensive care [14, 15].

In conclusion, the main finding of this study is that increasing the back-up ventilator rate is the key factor to decreasing the respiratory effort and increasing alveolar ventilation in young patients with CF. Furthermore, the application of AC/VT ventilation is more difficult than PS ventilation due to the different algorithm of the I/E ratio. Indeed, the adjustment of the Ti on the back-up rate, and not the effective respiratory rate, of AC/VT devices represents an important technical limitation that should be appreciated. It should be less ambiguous to propose a Ti setting rather than a false I/E ratio if this parameter remains calculated on the back-up rate and not on the effective respiratory rate. Clearly, if no improvements in the setting of the I/E ratio and the sensitivity of the trigger are made on these AC/VT ventilators, it is important to use them with a high back-up rate to avoid increased respiratory effort and poor tolerability. The performance of PS ventilators is clearly better, but even these sophisticated triggering systems are still responsible for a significant respiratory effort in young patients with CF. An improvement in both the cycling pattern setting and the inspiratory triggers of these domiciliary ventilators should be able to improve the tolerance and the efficacy of NIMV in this group of patients.