Functional differences in bi-level pressure preset ventilators

Noninvasive intermittent positive pressure ventilation (NIPPV) was first developed using volume preset devices with pressure triggering and no expiratory airway pressure 1, 2. Bilevel pressure preset flow triggered devices are now increasingly used for NIPPV. The superiority of these devices is unproven 3, 4 yet there are several theoretical advantages over volume preset ventilators. Pressure preset ventilation enables compensation for leaks to occur, whilst volume preset does not 5. Air leaks are common during NIPPV and may lead to inadequate treatment 6. A pressure preset ventilator should respond to increased patient effort during a breath with increased flow and a greater tidal volume (V_T) 7, in contrast to volume preset ventilators where V_T is fixed. The addition of expiratory positive airway pressure (EPAP) to the ventilator circuit may be useful in patients with dynamic hyperinflation, as it reduces the effort needed to trigger the ventilator 8, 9. Patients with ventilatory failure due to obstructive sleep apnoea may benefit from EPAP as it stabilises the airway in the same way as continuous positive airway pressure 10. Flow triggering has been shown to reduce patient effort compared to pressure triggering whilst maintaining equivalent levels of ventilation 11.

Previous studies comparing the performance of ventilators using lung models have all shown differences between the devices tested 5, 12, 13, but there has been little emphasis on the effects that patient effort might have on the ventilators, although many of the presumed advantages of bilevel pressure preset ventilators relate to patient effort. This paper has examined the responses of four bilevel pressure preset ventilators to leaks within the circuit and changes in simulated patient effort.

Methods

Experimental model

A patient simulator which has previously been described 5 was used to make all measurements (fig. 1⇓). It consists of a bellows type lung simulator (Ohmeda, Herts, UK) contained within a box, connected to a negative pressure pump (Negavent; Dima Italia, Milan, Italy). The elastance and the resistance of the lung simulator were set to 20 cmH₂O·L⁻¹ and 5 cmH₂O·L⁻¹·s respectively. The ventilators were connected to the simulator using identical circuits, incorporating a Whisper swivel II expiatory valve (Respironics Inc.). A connector with a leak of radius 2 mm, which can be opened or closed, and a pneumotachograph (PNT) (Si­Plan Electronics Research, Stratford­upon­Avon, UK) were placed between the expiratory valve and the lung simulator. The box was not airtight but had a leak sufficient to allow the expansion of the bellows, with no measurable rise in pressure within the box.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.—

Experimental set up. 1: data logger; 2: bilevel ventilator; 3: expiratory valve; 4: variable leak; 5: pneumotachograph; 6: pressure transducer; 7: lung simulator within a box; 8: negative pressure pump.

Pressure within the box and in the circuit, between the pneumotachograph and the lung simulator, was measured using calibrated pressure transducers (Vygon, Gloucestershire, UK). The inspiratory flow signal from the PNT was integrated to give a volume signal and was calibrated using a 1 L syringe (Vitalograph, Buckinghamshire, UK). All signals were sampled at 33 Hz and recorded digitally on a data­logging system (Cardas, Pilogic, Dyffed, UK).

The negative pressure pump generated pulses of negative pressure within the box to simulate patient effort, expand the bellows and trigger the ventilators. The software within the pump was modified to eliminate any interaction between the pump and the ventilators. A sensor within the hose of the negative pressure pump normally monitors pressure continuously. Following modification, this sensor operated initially ensuring that the correct preset pressure is delivered by the pump. However it was subsequently disabled, so that the pump continued to deliver identical “patient effort” irrespective of any changes in pressure created within the box by the action of the ventilator on the bellows. Thus the “patient effort” applied was consistent, irrespective of the action of the ventilator.

The ventilators were used in the S/T mode with the minimum EPAP. The f_R was set at twelve breaths·min⁻¹ with an inspiratory:expiratory ratio of 1:2. Where adjustable, inspiratory triggers were set at the most sensitive, and the fastest rise in inspiratory pressure was selected. When any variable was altered, ten breaths were allowed for equilibration and results quoted as means of the next twenty breaths.

Results

Leak in the patient circuit

The percentage change in V_T and IPAP for each ventilator with the leak opened are illustrated in figs 2⇓ and 3⇓. All four machines compensated well for the leak. Any falls in V_T were less than 10% and falls in IPAP were less than 8%.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.—

The effect on tidal volume when the leak was opened. ○: BiPAP ST 30; ▵: Nippy2; +: Quantum PSV; □: VPAP II ST.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.—

The effect on inspiratory positive airway pressure (IPAP) when the leak was opened. ○: BiPAP ST 30; ▵: Nippy2; +: Quantum PSV; □: VPAP II ST.

The change in flow, created by the leak, delayed expiratory cycling and prolonged t_I at all levels of IPAP with the BiPAP. The mean (sd) t_I increased significantly (unpaired t-test; p<0.01) from 1.56 (0.51) s to 2.06 (0.46) s. The prolongation of t_I was variable and correlated positively with a change in V_T (r=0.9), which increased during some tests. t_I is preset on the Nippy2 and for nontriggered breaths on the Quantum, and therefore t_I was unchanged with the leak open. The mean t_I for the VPAP increased significantly (p<0.01) from 1.1 (0.2) s to 1.61 (0.31) s with the leak open, and correlated with the change in the resulting V_T (r=0.75).

For the two ventilators that prolonged t_I with the leak open, the delivered V_T was calculated for the equivalent t_I when the leak was closed. There was a minimal fall in mean (sd) V_T of 5.0 (19.2) mL with the BiPAP and 2.1 (6.8) mL with the VPAP.

Changes in duration and frequency of patient effort

The mean (sd) V_T for each ventilator at baseline were: BiPAP 529 (9.6) mL; Nippy2 415 (6.8) mL; Quantum 514 (6.3) mL; and VPAP 457 (4.6) mL. The percentage increases in minute volume from baseline for durations of “patient effort” of 0.25 and 1 s are shown in figs. 4⇓ and 5⇓. To allow comparison of performance, the “predicted increase” in minute volume illustrated is calculated, assuming that the baseline tidal volume remains constant and that all patient efforts result in a ventilator breath.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.—

The effect of increasing respiratory frequency on minute volume with “patient efforts” of 0.25 seconds. •: predicted increase; ○: BiPAP ST 30; ▵: Nippy2; +: Quantum PSV; □: VPAP II ST.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.—

The effect of increasing respiratory rate on minute volume with “patient efforts” of 1 second. •: predicted increase; ○: BiPAP ST 30; ▵: Nippy2; +: Quantum PSV; □: VPAP IIST.

With the “patient effort” of 0.25 s the t_I of the BiPAP was variable with a range of 1.36–2 s (see also table 3⇓). At a rate of 24 breaths·min⁻¹ there are only 2.5 s tocomplete each respiratory cycle and there was insufficient time for complete expiration when t_I was prolonged. This caused a fall in expiratory volume, a rise in residual volume and failure of some of the simulated breaths to trigger a ventilator breath. Over the 5 min period of the test at a rate of 24 breaths·min⁻¹, 28% of the simulated breaths failed to trigger the BiPAP. Analysing triggered breaths, the mean (sd) EPAP measured at the onset of inspiration was 2.7 (0.69) cmH₂O. The t_E of the preceding breath, defined as the time difference between the start of expiratory flow and the onset of simulated effort was 0.54 (0.11) s. In comparison, where there was failure to trigger the ventilator, EPAP was 4.0 (0.5) cmH₂O and t_E was 0.37 (0.08) s. Trigger failure and the reduced V_T caused by hyperinflation explains the abrupt fall in minute volume seen at the highest rate. With the simulated breaths of 1 s duration, the expiratory cycling matched t_I to the “patient effort”, allowing adequate time for expiration at each f_R. There was therefore no hyperinflation or trigger failure and the simulated effort led to a consistent increase in tidal and minute volume at each frequency.

View this table:

• View inline
• View popup

Table 3—

Mean (sd) ventilator t_I, with varying duration of “patient effort” (PE)

The T_I for the Nippy2 is preset and was therefore unchanged for triggered and timed breaths. There was insufficient time for expiration at the highest f_R leading to hyperinflation and trigger failure with “patient efforts” of both 0.25 and 1 s. With a duration of 0.25 s, 26% of simulated breaths at a f_R of 24 failed to trigger the Nippy2. For triggered breaths, the mean (sd) EPAP measured at the onset of inspiration was 2.8 (0.18) cmH₂O and t_E was 0.47 (0.05) s. During episodes of trigger failure the measured EPAP was 3.3 (0.19) cmH₂O and t_E was 0.37 (0.08) s. There was 16% trigger failure with simulated breaths of 1 s at an f_R of 24. For triggered breaths EPAP was 3.1 (0.42) cmH₂O and t_E was 0.48 (0.05) s. During episodes of trigger failure EPAP was 3.9 (0.56) cmH₂O and t_E was 0.37 (0.04) s. A consistent rise in tidal and minute volumes was seen with the longer “patient effort” at rates below which trigger failure and hyperinflation occured.

The Quantum produced minute volumes less than “predicted” values during all tests, even though there were no episodes of trigger failure. Cycling from IPAP to EPAP occurs when flow within the ventilator circuit falls to 75% of the peak flow. With “patient effort” of 0.25 s and the minimum rise time, this point is reached quickly and results in short ventilator breaths (mean 0.58 s; see table 3⇑). The longer “patient effort” duration of 1 s delayed expiratory cycling and subsequently increased V_T. However expiratory cycling still occurs before patient effort is complete and subsequently leads to tidal and minute volumes below the expected values.

Triggered and timed breaths with the VPAP were of very similar duration and volume and thus the increase in minute volume was close to the expected values for “patient efforts” of 0.25 s. t_I was well matched to “patient effort” with simulated efforts of 1 s and produced the greatest increases in tidal and minute volumes from baseline. There were no episodes of trigger failure.

Discussion

The results of these bench assessments show that there are a number of dynamic changes in the behaviour of the ventilators in response to leaks and variations in imposed effort. Many of these changes cannot be predicted from prior knowledge of each ventilator's quoted performance characteristics.

The potential deterioration in V_T delivered by the ventilator during an air leak may lead to ineffective ventilation and subsequent arterial desaturation during NIPPV 14. This reduction in effective ventilation is theoretically minimised by the use of pressure­preset devices and the fall in V_T seen with the machines tested on the lung model was minimal in the face of a persistent leak. The effects of air leakage during NIPPV can be more complex than a fall in IPAP and V_T, and the changes in t_I seen with the BiPAP and the VPAP are examples of this. The prolongation of t_I leads to an unpredicted and unnecessary increase in V_T of up to 15% that may potentially be detrimental.

As illustrated with the lung model, prolongation of t_I when the patient's f_R is high may lead to insufficient time for expiration, hyperinflation and trigger failure. It would therefore seem to be a disadvantage to prolong t_I during an air leak at a high f_R. Leaks and subsequent prolongation of t_I while using the BiPAP can increase the number of arousals during light sleep and contribute to sleep fragmentation 15.

Failure to trigger the ventilator with subsequent wasted patient effort in up to 40% of breaths during pressure support ventilation has been shown to occur in 54–75% of stable intubated patients weaning from invasive ventilation 16, 17. The prevalence of such trigger failures during NIPPV is not known, but significant differences in the sensitivity of triggering of the four ventilators have been demonstrated. The maximum difference in trigger delay was 52 ms and in trigger pressure was 1.1 cmH₂O. The differences demonstrated represent an increase in work done by the patient that will be repeated during each triggered breath, and may be of particular importance in a breathless, tiring patient 12.

Cycling to expiration before the patient's inspiratory effort is complete leads to an increase in the inspiratory load imposed, whilst prolongation of inspiratory pressure beyond the duration of inspiratory effort made by the patient can lead to active expiration and an increase in the expiratory work 18. Under the specific test conditions employed, the Quantum cycled prematurely to expiration. The ventilator t_I is shorter when inspiration is terminated by flow falling to a higher fixed percentage of the peak value 13. In stable intubated patients on pressure support ventilation, ventilator t_I was significantly shorter when inspiration was terminated at 50% of peak flow compared with 25% 19. However, there was no significant effect on V_T, f_R, or IPAP delivered, and a higher percentage such as that employed on the Quantum (75%) was not examined.

Cycling to expiration with the Nippy2 is dependent solely on the t_I selected by the operator and the trigger failures that were observed at the highest rate on the lung model would have been avoided by the selection of a shorter t_I. The BiPAP and the VPAP cycle to expiration according to unquoted flow criteria, which matched t_I efficiently to a “patient effort” of 1 s, allowing adequate time for expiration even at the highest f_R. However, as was demonstrated on an earlier model of the BiPAP 5, t_I lengthens if patient effort is less than 0.5 s in duration, which may occur at particularly high f_R 16.

NIPPV can reduce measured inspiratory effort during the treatment of acute and chronic respiratory failure 9, 20, and during exercise 21 when the ventilatory demands of the patient are particularly high. However the response of the ventilator to an increase in inspiratory effort made by the patient has been little explored. Using simulated patient breaths on a lung model, changes in the inspiratory flow demand have been shown to alter a number of measured parameters 13. There were differences between the devices tested, but in general, increasing the ventilatory demand led to an increase in the peak flow delivered by the ventilator, but also increased the expiratory loading and reduced the sensitivity of the triggering. The degree to which this affected the actual ventilation delivered was not quantified.

This study found considerable differences in ventilation delivered by different ventilators using the same test conditions. At an equivalent IPAP of 20 cmH₂O there was a difference of up to 115 mL in the V_T delivered by the four ventilators. Similar differences have been illustrated in previous comparisons during bench testing 5, 12 and in patients with stable chronic respiratory failure 3. It has previously been argued that this difference in V_T delivered for a given peak IPAP is mainly due to the pressure waveform of the ventilator and that this needs to be considered in the selection of a machine for NIPPV 5.

With the introduction of simulated patient effort, further differences between the ventilation delivered by each of the ventilators become apparent. When a brief inspiratory effort is made by the simulated patient there may be a fall in V_T and indeed a fall in minute volume, despite an increase in f_R. For some of the ventilators there is a marked increase in tidal and minute volumes with a more prolonged patient effort over and above that expected by an increase in f_R alone. At a high f_R, there may be a drop in minute ventilation caused by trigger failures and hyperinflation. Variable respiratory frequencies and durations of respiratory effort would be expected in breathless patients with ventilatory failure. The unpredictable response of these ventilators under these circumstances has not previously been demonstrated.

There are limitations in applying the findings of a lung model study to the clinical situation. Any leaks, and the duration and degree of patient effort, will vary during each breath. In addition, the effect of airways resistance on the ventilation delivered has not been considered. In particular the glottis behaves as a variable resistance during noninvasive intermittent positive pressure ventilation, affecting the pressure waveform and tidal volume independent of inspiratory effort 22. However, we have illustrated differences in the responsiveness of apparently similar bilevel ventilators currently in use for noninvasive intermittent positive pressure ventilation. The importance of these differences will depend on the precise clinical situation, but may be sufficient to influence the choice of ventilator. These differences may be most significant when selecting a ventilator for the treatment of breathless patients with ventilatory failure.