Performance Characteristics of Five New Anesthesia Ventilators and Four Intensive Care Ventilators in Pressure-support Mode: A Comparative Bench Study

The five anesthesia ventilators evaluated were the Felix (Taema, Antony France), Kion (Siemens AG, Munich, Germany), Fabius GS (Drägerwerk AG, Lübeck, Germany), Primus (Drägerwerk AG), and Avance workstation (GE-Datex-Ohmeda, Munchen, Germany) equipped with the model 7900 ventilator. This last ventilator can also be found in the GE-Datex-Ohmeda Aestiva, Aisys, and Aespire anesthesia workstations. The four ICU ventilators tested were the Servo 900C (Siemens), Servo 300 (Siemens), Horus (Taema), and Evita 4 (Drägerwerk AG). The main characteristics of anesthesia and ICU ventilators tested are presented in table 1.

The machines were provided by the manufacturers after a full revision had been made just before our investigation. All machines were stock, no modification was performed, and all were tested in operating conditions conforming to the manufacturer’s specifications.

All ventilators were connected to a classic, validated two-compartment lung model (Pneu View AI 2601I TTL; Michigan Instruments, Grand Rapids, MI) which has been described in detail in previous studies.17,20Briefly, the model consists of two separate chambers linked by a rigid metal strip. One chamber is connected to an ICU ventilator (Evita 4; Drägerwerk AG), which is set in volume control mode to mimic patient inspiratory effort (fig. 1). The magnitude and duration of the latter can thus be adjusted by changing the settings on this “driving” ventilator. The two chambers being linked, inflation of the first necessarily inflates the second, which is connected to the ventilator being tested. The onset of passive inflation is therefore detected as an “inspiratory” effort by the tested device, which triggers a pressure-support response. The elastance (E) and airway resistance (R) of each compartment can be adjusted separately.

Thus, the model allows simulation of various magnitudes of inspiratory effort, types of respiratory mechanics, and tested ventilator settings. The ventilator circuits connected to each chamber were equipped with a pneumotachograph and pressure transducer (Biopac Systems, Goleta, CA). Data were acquired online via  an analog–digital converter (MP100; Biopac Systems), sampled at 500 Hz, and stored in a laptop computer for subsequent analysis (Acqknowledge software; Biopac Systems).

All measurements were performed in ambient temperature and pressure-saturated conditions. Automatic body temperature and pressure-saturated compensation was disabled on the Evita 4, and all other devices were calibrated in ambient temperature and pressure-saturated conditions. Gas compressibility was not accounted for, given its negligible quantitative contribution in the conditions of the current tests.21 

Inspiratory trigger and pressurization ramp were evaluated as previously reported.16,17,20 Figure 2shows the method used to calculate the trigger characteristics and the pressurization phase during PSV based on the airway pressure–time curve.

At each sensitivity condition tested, triggering performance was assessed according to three criteria: the time delay, the pressure fall, and the airway pressure–time product per cycle.

• Triggering delay (DT): time between the onset of inspiratory effort and that of detectable pressurization.

• Pressure fall (DP): the maximal decrease in airway pressure measured from its baseline value. DP reflects in such way the inspiratory work required to trigger the ventilator; therefore, the lower its value, the smaller the work required of inspiratory muscles.22 

• Airway pressure–time product per cycle (PTP, cm H²O × ms) during the trigger phase, defined as the area under the P^awsignal during the DT interval (computed as DP × DT).

The pressure–time products at 300 and 500 ms for each respiratory cycle (PTP³⁰⁰and PTP⁵⁰⁰) are computed as the area under the time–pressure curve 300 and 500 ms after the onset of inspiratory effort. These two parameters reflect the speed of pressurization and the device’s capacity to maintain the set pressure during inspiratory effort. They depend both on the ventilator’s performance and the magnitude of inspiratory effort, the former being determined by the pressurization ramp and the flow generated by the device’s bellows or piston. PTP³⁰⁰and PTP⁵⁰⁰are expressed in cm H²O · s.

DT, DP, PTP, PTP³⁰⁰, and PTP⁵⁰⁰were measured as described above and in figure 2at three successive levels of PSV: 10, 15, and 20 cm H²O.

To mimic normal and strong inspiratory efforts by patients, the tidal volumes of the driving ventilator were set at 220 and 440 ml, respectively. These efforts were actually associated with pressures 100 ms after occlusion (P0.1) of 2 cm H²O (normal effort) and 4 cm H²O (strong effort), respectively, as measured on the bench.16,19The duration of inspiratory effort on the driving ventilator was set at 1 s for all tests. Inspiratory trigger was set at the maximum sensitivity without the presence of auto-triggering. The pressurization slope was set to its steepest value. When the inspiratory:expiratory cycling criteria was adjustable, it was maintained at its default value.

During the tests, E and R of the “driving” chamber were set to normal (E = 20 cm H²O · l⁻¹, R = 5.6 cm H²O · l⁻¹· s).

For all ICU and anesthesia ventilators, the measurements were performed at positive end-expiratory pressure (PEEP) of 0 and 5 cm H²O for each of the two different efforts (normal and strong) and for the three PSV levels (10, 15, and 20 cm H²O).

For the five anesthesia ventilators, two levels of fresh gas flow were tested: 1 and 10 l/min. Thus, 12 conditions were evaluated for each ICU ventilator and 24 conditions were tested for each anesthesia ventilator.

All parameter values represent the average of three to five breaths obtained during steady state. All results are expressed as mean ± SD or median with 95% confidence interval, depending on the normal or nonnormal distribution of the variables. Comparative statistics relied on the Kruskal-Wallis one-way analysis of variance on ranks. Post hoc  analysis was performed with the Scheffé test if analysis of variance reached significance. Significance was set at P < 0.05.

One hundred sixty-eight conditions were evaluated, 120 for the anesthesia ventilators and 48 for the ICU ventilators. None of the ventilators mistriggered, nor did any ventilator prematurely cycle to expiration.

DT, DP, and PTP values measured in zero end-expiratory pressure (ZEEP) and with PEEP for all studied ventilators at a level of P0.1 = 2 and P0.1 = 4 cm H²O are presented in figures 3 and 4, respectively.

On ZEEP, the inspiratory trigger time delay was significantly shorter with the ICU ventilators compared with all of the anesthesia ventilators except for the Primus and the Avance. PEEP had no impact on the inspiratory time delay in all ICU ventilators, whereas for anesthesia ventilators, it influenced the performance of the trigger system for two of the ventilators (Felix and Kion) whatever the level of PSV studied (10, 15, or 20 cm H²O). For the Kion, changes in pressure and trigger time delay required to open the inspiratory valve significantly increased with PEEP compared with ZEEP. The opposite was observed with the Felix, in which DT was significantly shorter in PEEP than in ZEEP (figs. 3 and 4).

For all machines, DT was not affected by the magnitude of inspiratory effort, except for the Kion, whose TD significantly increased with PEEP (figs. 3 and 4) as inspiratory effort increased.

For all anesthesia ventilators tested, increased fresh gas flow from 1 to 10 l/min did not significantly modify triggering performance on ZEEP or PEEP.

PTP³⁰⁰and PTP⁵⁰⁰values measured on both ZEEP and PEEP for all studied ventilators according to a level of P0.1 = 2 cm H²O (normal effort) are presented in figure 5and to a level of P0.1 = 4 cm H²O (strong effort) are presented in figure 6.

At all levels of PSV studied (10, 15, and 20 cm H²O), the pressurization capacity of all ICU ventilators was comparable, whereas it varied among the anesthesia ventilators, the difference being more marked with PEEP. At 300 ms in PEEP, the values obtained with the Felix and the Kion were half those obtained with the Fabius, Primus, and Avance.

For all ICU ventilators except the Servo 900, PTP³⁰⁰was not affected by the magnitude of inspiratory effort. However, with the anesthesia ventilators, PTP³⁰⁰tended to decrease as inspiratory effort increased (figs. 5 and 6).

The current study is the first to provide a strictly protocoled bench test evaluation of the performance in delivering pressure support of five new-generation anesthesia ventilators. The major findings of this trial can be summarized as follows: (1) For the five tested anesthesia ventilators, the PSV modality functions correctly; (2) performance was more homogeneous among the modern ICU ventilators than among the anesthesia ventilators; (3) the use of PEEP modified the quality of delivered pressure support in two anesthesia ventilators (Kion and Felix) but not in the three others (Fabius, Primus, and Avance); and (4) increasing fresh gas flow (1 to 10 l/min) in the internal circuit did not influence the PSV performance of the anesthesia ventilators.

This bench test study also showed that triggering delay is less than 100 ms for all ICU ventilators except in the older Servo 900C as reported by previous studies 15–17,19and is less than 100 ms only for only two anesthesia ventilators (Primus and Avance) (figs. 3 and 4).

The inspiratory workload required to trigger the ventilators (i.e. , PTP) is very low for the modern ICU ventilators, in line with previous studies of these machines.17,20The most recently developed anesthesia ventilators exhibited comparable performance, the highest PTP being measured in the Kion, which is the first anesthesia ventilator with PSV mode and the oldest of the anesthesia ventilators tested. The best characteristics of the pressurization phase for the anesthesia ventilators were obtained with the Fabius, Primus, and Avance under all tested conditions and were comparable with those of obtained with the ICU ventilators. The Fabius, Primus, and Avance are “piston ventilators,” which use an electric motor to compress gas in the breathing circuit, creating the driving force for mechanical insufflation to proceed. Therefore, they use no driving gas and may be used without depleting the oxygen cylinder in case of oxygen pipeline failure. These features may explain in part that these more recent anesthesia ventilators have comparable performance to modern ICU ventilators.

It is important for the users to know that the Fabius GS and the Fabius Tiro have the same ventilator (electrical piston) and use the same software management that regulates the ventilator. Similarly, the Primus (worldwide outside of the United States) and the Apollo (United States) anesthesia stations have the same ventilator (electrical piston). But the implemented software that regulates the ventilator is different between the Fabius GS/Tiro and Primus/Apollo. This difference may explain in part the differences in performances obtained with the two ventilators (Fabius vs.  Primus).

The newer technologies used by manufacturers, i.e. , microprocessors, servo valves, and fast and potent turbines, have substantially improved both modern anesthesia and ICU ventilators regarding global trigger response.

It seems that the industry has so far chosen not to invest heavily in the development of PSV on anesthesia ventilators. This might seem surprising, because PSV has been available on ICU ventilators for more than 20 yr. Two main factors probably account for this. The first is of a technical nature. Indeed, an anesthesia ventilator is composed of two circuits, one for driving gas, the other for the patient circuit with the anesthetic gases, with independent bellows, and the resultant large internal volume makes it more difficult to implement fast-responding and efficient triggering mechanisms. The second factor is mainly clinical, i.e. , that whereas ICU ventilators need to provide a mode of partial ventilatory support tailored to the patient’s breathing pattern during weaning, the need for such a mode in anesthesia has only become apparent in recent years. The need is probably linked to the use of laryngeal masks, which is a spontaneous-assisted mode with leaks resembling noninvasive ventilation in some aspects, and the increasing use of local–regional anesthesia combined with light sedation during which spontaneous breathing is maintained.

Pressure-support ventilatory modes have recently been introduced as readily available options on newer anesthesia ventilators. Unlike ICU ventilators, which vent exhaled gases to the atmosphere and directly release gas from the wall outlet into the circuit, anesthesia ventilators recirculate exhaled gases into the inspiratory limb. It has been suggested but never evaluated that in pressure-support mode, this feature mandates a large internal volume, which may in turn alter the performance of triggering and pressurization systems.5,23In the current study, we did not find a significant difference for the performance of triggering and pressurization systems between a low (1 l/min) and a high (10 l/min) fresh gas flow for all anesthesia ventilators and all tested dynamic conditions.

We tested the influence of PEEP on the quality of the trigger and pressure delivering because recent studies suggested that use of PEEP may have some benefits on pulmonary function.24–28 

The most important limitation of this study is the fact that it was performed on a lung model instead of in patients. It is possible that performance in patients may differ greatly from the performance demonstrated here. The advantage of the model is that mechanical characteristics can be standardized and reproduced. In addition, the test lung was modified to simulate spontaneous breathing. Hence, the different machines were tested under similar conditions during dynamic experiments. However, it is clear that these laboratory conditions are not real life and, therefore, that the results of these bench studies should be extrapolated to patients with caution. Therefore, a clinical study evaluating these characteristics and other aspects of the performance of anesthesia ventilators, e.g. , on gas exchange and comfort in the operating room, should be performed. In addition, and perhaps more importantly, little is known about the validated indications of PSV in anesthetized patients. Further exploration of this topic is clearly warranted.

For the five tested anesthesia ventilators, PSV functioned correctly. The efficiency of delivering PSV for the anesthesia ventilators is acceptable, comparable to older-generation ICU ventilators (i.e. , Servo 900C); however, it did not reach the level of performance of the new-generation ICU ventilators for three of the five tested anesthesia ventilators.

The use of PEEP modified the quality of delivered pressure support for two anesthesia ventilators (Kion and Felix). Increasing fresh gas flow (1 to 10 l/min) in the internal circuit did not influence PSV performance of the tested anesthesia ventilators. Regarding trigger sensitivity and the system’s ability to meet inspiratory flow during pressure-supported breaths, the most recent anesthesia ventilators have performances comparable to those of the modern ICU ventilators. Further clinical studies should now be conducted to better define the indications of PSV during anesthesia.

The authors thank Jérôme Pigeot (Biomedical Engineer, Fisher-Paykell, Courtaboeuf, France) for his technical assistance during the bench study.