Respiratory depression is a common complication in animals undergoing general anesthesia and is a particular concern in horses because atelectasis formation in their dependent lungs may lead to development of hypercapnia, hypoxemia, and acid-base imbalances.1 Although mechanical ventilation is commonly used to prevent or treat lung collapse and related pulmonary gas exchange disturbances in large animals, it may also cause various complications, particularly when the ventilator setting is incorrect, the device malfunctions, or the VT delivery is inaccurate. To avoid hypo- or hyperventilation and potentially associated ventilation-induced lung injury, it is critical to deliver an adequate and accurate VT to a patient.2,3 Even during volume-control ventilation, the VT that actually reaches a patient's lungs might become smaller than the volume set on a ventilator for various reasons, including losses by compliance of the breathing system and hoses, leaks in the breathing system or around a cuffed endotracheal tube, and gas compression.4,5 Gas compression might be of greater importance in large animals, compared with small animals, because higher peak inspiratory pressures are achieved in breathing circuits during anesthesia of large animals. In addition, the VT actually delivered might become higher than the preset VT, such as when high FGF rates are used.4
In piston-driven ventilators, the area of the piston is fixed. Thus, the volume of gas delivered by the piston is directly related to the linear movement of the piston in the piston chamber, and the VT delivery should be most accurate. A recent study5 revealed that an electrically powered and microprocessor-controlled piston ventilator for use in large-animal veterinary medicine delivers a VT with a δVT of only 5% above or below the equipment preset VT and that a calibration factor could be determined to further increase accuracy.5 In contrast, most pneumatically powered ventilators for use in large animals appear to be less accurate, despite the possibility of semiquantitatively estimating the VT by observing movement of the bellows in a transparent cylinder housing marked with a graduated scale in liters. Therefore, pneumatic ventilators may require closer monitoring of the truly delivered VT.6
To our knowledge there has not been a systematic evaluation of the accuracy of VT delivery with commonly used pneumatic, large-animal ventilators. Therefore, the goal of the study reported here was to determine the accuracy of VT delivery among 5 different models of large-animal ventilators (4 pneumatically driven ventilators and 1 electrically driven piston ventilator) when tested at various settings for VT delivery, PIF rate, and FGF rate. We hypothesized that, because of their design, pneumatic ventilators would be less accurate in delivering a desired (ie, preset or dialed) VT than would the piston ventilator and that the magnitude of δVT would be affected by the PIF and FGF rates, preset VT, and compliance of the breathing hoses used in the anesthetic circuit system.
Materials and Methods
Included in the study were 5 different models of ventilators frequently used in clinical practice for anesthesia of large animals at the New Bolton Center of the University of Pennsylvania's School of Veterinary Medicine and at the Rood and Riddle Equine Hospital in Lexington, Ky. The anesthesia ventilators tested included 4 pneumatically driven ventilators equipped with bellows (ventilators A through D) and 1 electrical motor-driven piston ventilator (ventilator E).
Ventilator Aa was classified as a dual-circuit, time-cycled, pneumatically powered, and electronically controlled ventilator. The ventilator operated with hanging (descending) bellows suspended in a plexiglass cylinder marked with a graduated scale in liters (range, 0.5 to 15 L). The VT was set by manually adjusting the maximum descent of the bellows and, hence, their filling volume, thereby allowing for volume-targeted ventilation. To do so, an anesthetist turned a crank that was at the front and bottom of the bellows housing. This set a stamp in motion within the bellows chamber so that the bellows would fill only to the desired VT (range, 0.5 to 15 L). The control unit was equipped with an on-off switch, PIF rate regulator, RR dial (range, 1 to 99 breaths/min), and inspiratory-to-expiratory ratio dial (range, 1:1 to 1:4.5).
Ventilator Bb was also a dual-circuit, time-cycled, pneumatically powered, and electronically controlled respirator, designed with hanging bellows. Somewhat similar to ventilator A, the anesthetist could rotate a crank on the side of the control unit that set in motion a wire rope that was fixed to a metal plate at the bottom of the bellows. By shortening this wire rope, the maximum descent of the bellows and thus their filling volume were adjusted to a desired volume between 0.5 and 15 L, thereby allowing for volume-targeted ventilation. The control dials included an on-off switch and allowed for adjustments to the RR, Tinsp, and PIF rate. The VT was controlled by adjusting the PIF rate and Tinsp until the desired VT (range, 0.5 to 15 L) was obtained. The VT was estimated by visualizing the movement of the bellows in their cylinder housing (graduated in liters), thereby allowing for volume-targeted ventilation.
Ventilator Cc was a pneumatically powered, time-cycled, and electronically (microprocessor) controlled dual-circuit ventilator with standing (ascending) bellows. The control dials allowed adjustment to the PIF rate (range, 10 to 600 L/min), Tinsp, and RR (range, 2 to 15 breaths/min). In addition, the control unit included a power switch and a manual ventilation button. By adjusting the PIF rate and Tinsp, the operator could preset the desired VT, which was estimated by visualizing the movement of the bellows in their cylinder housing (graduated in liters), thereby allowing for volume-targeted ventilation. The inspiratory-to-expiratory ratio, calculated by the microprocessor and based on the selected Tinsp and RR, was digitally displayed.
Ventilator Dd was a dual-circuit, time-cycled, and solely pneumatically powered ventilator that was without any electronic controls and free of any ferrous metals. This ventilator was MRI compatible, and adjustable settings included only the on-off knob; PIF rate, Tinsp, and Texp knobs; and a manual ventilation button. By adjusting the PIF rate (range, 10 to 600 L/min) and Tinsp, the operator preset the desired VT, which was, as with the aforementioned ventilators, estimated by visualizing the movement of the bellows in their cylinder housing (graduated in liters) and thereby allowed for volume-targeted ventilation. By adjusting the Tinsp and Texp, an operator set the RR.
Ventilator Ee was an anesthesia unit equipped with a single-circuit, piston ventilator and was volume cycled. Instead of the bellows or rebreathing bag used in pneumatic ventilators, ventilator E had a stainless steel piston with 2 rolling diaphragms housed in a stainless steel cylinder. The control dials of ventilator E allowed setting the VT (range, 0.1 to 20 L), RR (range, 1 to 20 breaths/min), Tinsp (range, 0.5 to 4.0 seconds), maximum work pressure limit (range, 10 to 80 cm H2O), continuous positive airway pressure or positive end-expiratory pressure (range, 0 to 50 cm H2O, in 1-cm H2O increments), and trigger sensitivity (when the ventilator was operated in the assisted ventilation mode). The ventilator's microprocessor automatically determined the Texp and PIF rate on the basis of these preset parameters.7 Ventilator E was tested with 2 different circuit configurations, one with elastic black rubber breathing hoses and the other with largely noncompliant, transparent PVC breathing hoses, to assess the impact of breathing circuit compliance on the accuracy of VT delivery.
None of the tested ventilators compensated for variable compliance of its anesthetic circuit system setup, and none was equipped with a fresh gas-decoupling valve. In modern anesthetic systems made for use in humans, a fresh gas-decoupling valve is positioned between the fresh gas source and the ventilator and, during patient inspiratory phase, diverts the FGF to a reservoir bag, thereby eliminating any impact by FGF rate on the delivered VT.6–10
Experimental design
For each anesthetic ventilator, a gas leak check of the anesthetic circuit system setup was performed before experimenting with the ventilator. For the gas leak check, the Y-piece of the anesthetic circuit was closed with a rubber plug, the adjustable pressure relief valve of the anesthesia circuit was closed, and the circuit was pressurized with 100% O2. The leak was considered acceptable if an FGF rate < 500 mL/min was needed to maintain a constant pressure of 20 cm H2O. After results of a gas leak check were acceptable, the rubber plug was removed from the Y-piece of the anesthetic circuit, and a TMFVM,f with an 8-L calibration syringeg attached, was connected to the Y-piece as previously described.5 The recently validated5 TMFVM was used as the gold-standard method to measure the delivered VT. The calibration syringe was ventilated at 5 levels of VT (1.0, 2.5, 4.0, 5.5, and 7.0 L), each with 3 different preset PIF rates (approx 100, 150, and 200 L/min). In addition, for each ventilator, each VT and PIF setting combination was performed at 4 different FGF rates of 100% O2: 3, 5, and 7 L/min and the leak flow rate for the given anesthetic ventilator setup. After each VT delivery, the syringe plunger was manually pushed back to empty the syringe. Each test was performed at room temperature and dry gas conditions. Each VT administration was repeated 10 times, and the delivered VT was measured with a TMFVM, then recorded. All manually logged data were entered into spreadsheetsh for subsequent statistical analysis.
Statistical analysis
All data recorded were tested for normal distribution with the Shapiro-Wilk test and visual analysis of Q-Q plots. For each experimental setting, the mean ± SD was calculated for all 10 repetitive measurements of delivered VT. With those means for each tested ventilator and experimental setting combination, the mean δVT was calculated as a percentage of the preset VT for each combination. Next, for each ventilator, the complete set of mean δVT data recorded under the various experimental setting combinations was pooled, and ANOVA, with α set to 5% and values of P < 0.05 considered significant, was used to compare results across ventilators to determine which ventilators performed the best with respect to having had the lowest pooled mean δVT. Any relationship detected between the δVT and combinations of preset VT, PIF rate, and FGF rate was analyzed by determining the Pearson correlation coefficient and the coefficient of determination (r2), with the α set to 5% and values of P < 0.05 considered significant. Data analysis was performed with available software.h,i
Results
Overall, the leak rate for anesthetic ventilator setups was similar, with a mean ± SD leak rate of 0.261 ± 0.04 L O2/min. Therefore, an FGF rate of 0.25 L O2/ min was used as the leak rate when testing each ventilator setup at the various PIF rate and VT settings.
For each ventilator and each experimental setting combination tested, the mean ± SD delivered VT was calculated on the basis of results from 10 repetitive measurements. All results for delivered VT were smaller than the preset VT. Because all results for SD of the mean delivered VT were ≤ 0.75% of the mean VT, only the mean delivered VT results were used for further data analysis. With the mean delivered VT for each ventilator and experimental setting combination, the difference between the preset (dialed) VT and the mean delivered VT was calculated as a percentage of the preset VT to yield the respective mean δVT (Table 1). Overall, the mean δVT ranged from 1.2% (ventilator E, with a PIF rate of 200 L O2/ min, FGF rate of 3 L O2/min, and use of transparent PVC breathing hoses) to 22.2% (ventilator D, with a PIF rate of 200 L O2/min, FGF rate of 0.25 L O2/min, and use of transparent PVC breathing hoses).
Summary of the mean δVT (as a percentage of the preset VT) calculated from results of 10 repetitive measurements for each combination of experimental settings (VT [1.0, 2.5, 4.0, 5.5, and 7.0 L O2], PIF rate [approx 100, 150, and 200 L O2/min], and FGF rate [0.25 {leak rate}, 3, 5, and 7 L O2/min]) in each of 5 different models of large-animal ventilators (4 pneumatic [ventilators A and B with hanging bellows and ventilators C and D with standing bellows] and 1 piston-driven [ventilator E]).
| Ventilators* | ||||||||
|---|---|---|---|---|---|---|---|---|
| E | ||||||||
| PIF rate (L O2/min) | FGF rate (L O2/min) | VT (L O2) | A | B | C | D | BR | pvc |
| 100 | 0.25† | 1.0 | 5.3 | 12.5 | 15.6 | 20.9 | 11.0 | 8.5 |
| 2.5 | 5.1 | 10.2 | 14.2 | 18.5 | 7.2 | 5.3 | ||
| 4.0 | 4.5 | 9.5 | 13.5 | 17.2 | 5.3 | 4.7 | ||
| 5.5 | 3.4 | 9.1 | 13.6 | 17.5 | 4.7 | 4.2 | ||
| 7.0 | 2.8 | 8.7 | 12.1 | 16.3 | 3.1 | 3.5 | ||
| 3 | 1.0 | 4.8 | 10.4 | 13.6 | 17.3 | 8.9 | 7.4 | |
| 2.5 | 4.3 | 9.4 | 12.5 | 15.5 | 7.3 | 4.8 | ||
| 4.0 | 4.6 | 9.5 | 11.9 | 13.2 | 4.9 | 4.1 | ||
| 5.5 | 5 | 8.3 | 11.4 | 13.4 | 5.5 | 3.4 | ||
| 7.0 | 3.2 | 7.6 | 12 | 11.2 | 4.3 | 3.1 | ||
| 5 | 1.0 | 3.2 | 9.4 | 11.6 | 15.8 | 4.9 | 5.6 | |
| 2.5 | 3.9 | 8.1 | 10.2 | 14.2 | 4.6 | 3.2 | ||
| 4.0 | 3.4 | 7.1 | 9.9 | 13.3 | 3.3 | 3.6 | ||
| 5.5 | 2.4 | 6.5 | 8.7 | 11.9 | 3.9 | 2.7 | ||
| 7.0 | 2.8 | 6.1 | 8.9 | 10.2 | 3.4 | 2.2 | ||
| 7 | 1.0 | 3.6 | 8.6 | 10.1 | 14.5 | 4.5 | 3.5 | |
| 2.5 | 3.3 | 5.5 | 9.2 | 13.2 | 3.1 | 2.1 | ||
| 4.0 | 3.7 | 6.2 | 9.3 | 12.9 | 2.7 | 1.9 | ||
| 5.5 | 3.1 | 4.9 | 8.4 | 11.4 | 2.2 | 2.1 | ||
| 7.0 | 2.7 | 4.7 | 9.2 | 9.3 | 1.5 | 1.8 | ||
| 150 | 0.25† | 1.0 | 6.5 | 14.3 | 18.6 | 18.7 | 14.2 | 11.0 |
| 2.5 | 6.5 | 13.1 | 17.2 | 16.8 | 7.2 | 7.2 | ||
| 4.0 | 6.1 | 12.5 | 17.1 | 15.3 | 6.2 | 5.9 | ||
| 5.5 | 6.2 | 11.1 | 16.4 | 14.2 | 5.5 | 4.7 | ||
| 7.0 | 5.4 | 9.8 | 14.8 | 12.9 | 6.5 | 4.2 | ||
| 3 | 1.0 | 7.6 | 13 | 17.5 | 19.4 | 12.2 | 9.1 | |
| 2.5 | 7.2 | 11.4 | 15.9 | 17.2 | 6.7 | 4.3 | ||
| 4.0 | 7.6 | 10.8 | 14.3 | 16.3 | 4.6 | 3.2 | ||
| 5.5 | 6.8 | 9.3 | 13.9 | 14.9 | 4.4 | 3.6 | ||
| 7.0 | 6.2 | 8.1 | 13.1 | 12.1 | 4.5 | 2.2 | ||
| 5 | 1.0 | 6.2 | 11.9 | 15.4 | 17.5 | 10.9 | 7.6 | |
| 2.5 | 5.8 | 10.6 | 15.3 | 14.6 | 7.3 | 4.9 | ||
| 4.0 | 5.7 | 9.4 | 14.1 | 15.2 | 6.4 | 3.3 | ||
| 5.5 | 5.7 | 9.6 | 13.6 | 13.2 | 5.8 | 3.7 | ||
| 7.0 | 5.3 | 9.5 | 11.8 | 11.7 | 3.4 | 2.4 | ||
| 7 | 1.0 | 4.4 | 11.2 | 13.4 | 18.2 | 10.8 | 8.0 | |
| 2.5 | 4.1 | 9.3 | 12.3 | 16.4 | 6.4 | 5.4 | ||
| 4.0 | 3.8 | 10.3 | 11.9 | 13.0 | 5.5 | 3.8 | ||
| 5.5 | 3.3 | 8.4 | 11.2 | 12.4 | 3.9 | 2.3 | ||
| 7.0 | 3.1 | 7.6 | 12.1 | 9.8 | 2.3 | 1.2 | ||
| 200 | 0.25† | 1.0 | 15.2 | 17.8 | 19.8 | 22.2 | 16.7 | 13.5 |
| 2.5 | 13.6 | 15.3 | 17.9 | 18.7 | 9.2 | 6.7 | ||
| 4.0 | 13.1 | 14.6 | 18.1 | 17.2 | 5.2 | 4.4 | ||
| 5.5 | 12.4 | 12.3 | 16.8 | 16.4 | 4.5 | 4.9 | ||
| 7.0 | 11.5 | 10.2 | 16.3 | 14.9 | 4.7 | 3.9 | ||
| 3 | 1.0 | 13.2 | 16.2 | 17.2 | 18.9 | 16.4 | 8.3 | |
| 2.5 | 11.5 | 14.9 | 16.1 | 17.1 | 8.3 | 6.4 | ||
| 4.0 | 10.8 | 15.1 | 15.3 | 16.5 | 5.4 | 3.9 | ||
| 5.5 | 9.3 | 13.5 | 14.1 | 15.5 | 5.9 | 3.3 | ||
| 7.0 | 9.4 | 12.6 | 13.5 | 14.8 | 4.1 | 2.1 | ||
| 5 | 1.0 | 10.6 | 14.4 | 15.9 | 17.5 | 13.4 | 11.0 | |
| 2.5 | 10.2 | 13.2 | 14.3 | 15.4 | 8.6 | 5.9 | ||
| 4.0 | 9.3 | 13.6 | 14.6 | 16.3 | 4.4 | 3.9 | ||
| 5.5 | 8.8 | 11.9 | 13.7 | 14.7 | 5.6 | 4.1 | ||
| 7.0 | 8.4 | 12.1 | 13.1 | 13.9 | 3.7 | 2.8 | ||
| 7 | 1.0 | 9.3 | 13.4 | 16.2 | 16.9 | 12.3 | 7.4 | |
| 2.5 | 8.6 | 11.3 | 15.2 | 15.5 | 7.6 | 5.9 | ||
| 4.0 | 8.7 | 10.5 | 13.8 | 14.3 | 5.4 | 3.3 | ||
| 5.5 | 8.2 | 9.4 | 14.1 | 14.8 | 6.9 | 3.7 | ||
| 7.0 | 7.7 | 8.9 | 11.5 | 13.2 | 6.5 | 4.1 | ||
Values reported as the mean δVT, each as a percentage of the respective preset VT.
Leak rate for all tested ventilator circuit setups was similar, with a mean leak rate of 0.261 ± 0.04 L O2/min.
BR = Black rubber breathing hoses used. PVC = Transparent PVC breathing hoses used.
For each ventilator, the complete set of the mean δVT data recorded under the various experimental settings was pooled, then ANOVA was used to compare results across ventilators (Table 2). Ventilator E (an electrically powered and microprocessor-controlled piston ventilatore), when used with an anesthetic breathing circuit of transparent PVC breathing hoses (like those used during testing of all the pneumatically powered ventilatorsa–d), was the most accurately operating ventilator in that it had the lowest pooled mean δVT (4.8 ± 2.5%). In addition, the pooled mean δVT was significantly (P < 0.001) lower (better) for ventilators A (6.6 ± 3.2%) and B (10.6 ± 2.9%) that had hanging bellows, compared with ventilators C (13.8 ± 3.0%) and D (15.2 ± 2.6%) that had standing bellows and consistently produced the largest mean δVT values. Of note, the mean δVT values were similar for ventilator A and ventilator E tested with PVC breathing hoses as long as the preset VT was ≤ 4 L and the PIF rate was ≤ 100 L O2/min. For ventilator E, the mean δVT was consistently larger when operated with breathing hoses made of more elastic black rubber material (6.4 ± 3.4%), compared with less elastic PVC breathing hoses (4.8 ± 2.5%), but ventilator E still performed overall with greater accuracy than any of the pneumatic ventilators (A through D).
Results of ANOVA to identify differences in pooled mean δVT values among the ventilators described in Table 1.
| E | |||||
|---|---|---|---|---|---|
| Ventilator | B | C | D | BR | PVC |
| A | 3.94 ± 0.56* | 7.11 ± 0.54* | 8.56 ± 0.58* | −0.21 ± 0.61 | −1.88 ± 0.52* |
| B | — | 3.17 ± 0.51* | 4.63 ± 0.51* | −4.15% ± 0.57* | −5.83 ± 0.49* |
| C | — | — | 1.45 ± 0.49* | −7.32 ± 0.56* | −9.01 ± 0.48* |
| D | — | — | — | −8.77 ± 0.55* | −10.45 ± 0.47* |
| E (BR) | — | — | — | — | −1.67 ± 0.54* |
Values reported as the differences in means ± SDs of the pooled δVT for each row and column pairwise comparison between ventilators. Negative values indicated that the ventilator listed at the top of the column had a lower (better) VT than did the ventilator listed at the far left of the row.
The pooled mean δVT differed significantly (P < 0.005) between the ventilator listed at the top of the column and the ventilator listed at the far left of the row.
See Table 1 for remainder of the key.
The δVT significantly correlated with the rates of PIF (r2 = 0.69; P = 0.003) and FGF (r2 = 0.62; P < 0.001). With all ventilators tested, an increase in the PIF rate resulted in a decrease in the delivered VT and hence an increase in the δVT. In contrast, with an increase in the FGF rate, the delivered VT increased and, thus, the δVT decreased. Furthermore, the preset VT correlated significantly (r2 = 0.58; P < 0.001) with the δVT in that an increase in the VT resulted in a decrease of the δVT.
Discussion
Results of the present study confirmed our hypothesis that pneumatically powered ventilators would be less accurate in delivering a desired (ie, preset or dialed) VT than would electrical motor-driven piston ventilators and that the magnitude of the δVT would be affected by the PIF and FGF rates, preset VT, and compliance of the breathing hoses used in the anesthetic circuit system. These results also corroborated findings from a study8 of anesthesia ventilators widely used in human medicine that shows the smallest δVT with a piston-driven ventilator.
Unless a ventilator unit is equipped with a fresh gas-decoupling valve, fresh gas will flow continuously throughout the respiratory cycle and therefore increase the VT delivered by the ventilator.9 A higher FGF rate and longer Tinsp will result in a larger δVT. To prevent such an additive impact on the delivered VT by the FGF rate, contemporary anesthesia ventilators used in human medicine have been equipped with a fresh gas-decoupling mechanism.6–10 The impact of the FGF rate on VT delivery is most critical in patients requiring ventilation with small VT or periods of high-flow anesthesia.
Because the ventilators tested in the present study each lacked a fresh gas-decoupling mechanism, one would expect an increase in delivered VT when the FGF was increased. Additionally, the finding that every ventilator in the present study had delivered VT less than the preset VT at all experimental setting combinations suggested that the FGF rate compensated only to some extent for the device-specific δVT and, thus, artificially reduced the δVT under test conditions with higher FGF rates. Although not observed in our study, this effect in very precisely operating ventilators could potentially lead to hyperventilation and alveolar overinflation if patients are small or suffer from parenchymal lung disease that would require ventilation with small VT.1
The compliance of the breathing circuit has been recognized to have a marked impact on VT delivery.11 Every anesthetic breathing circuit has its own specific compliance that is determined by the stretchiness of its elastic components, such as bellows (or bags in older models of pneumatic ventilators) and breathing hoses.11 A previous study12 shows that the magnitude of volume losses and the efficiency of an anesthetic circuit are dependent on the material of the breathing system components. Similarly, our results indicated that the mean pooled δVT for ventilator E was larger when tested with the more compliant rubber breathing hoses, compared with the less compliant PVC breathing hoses. This observed difference in the δVT on the basis of the type of breathing hoses used could have underestimated what would be commonly encountered under clinical conditions in large animals. Ideally, to measure compliance losses properly, the experimental setup should have mimicked an airway pressure of 20 to 30 cm H2O that is usually observed during mechanical ventilation in horses.
In addition, the engineering design of ventilators determines their magnitude of δVT,13 and we confirmed this notion. The δVT was the smallest for ventilator E (a piston [noncompliant steel] ventilator) when PVC breathing hoses were used, whereas ventilators A through D (rubber bellows-equipped, pneumatically powered ventilators) connected to the PVC breathing hoses all produced a δVT that was substantially larger than that produced by ventilator E. In pneumatically powered ventilators, movement of the bellows is controlled by the driving gas that enters the bellows chamber, pushes the bellows upwards in ventilators with hanging bellowsa,b or downwards in ventilators with standing bellows,c,d and thereby displaces a volume of breathing gas into the breathing circuit equal to the volume of driving gas that entered the bellows chamber.13 However, the pressure that builds in the breathing circuit because of the bellows movement and constitutes the driving force for delivery of the VT to the patient (to the calibration syringe in our study) is determined by the resistance and compliance of the breathing circuit components, including the bellows themselves.13 The more compliant those components are, the lower the pressure is within the breathing circuit and, consequently, the less compressed the gas is in the bellows chamber, all leading to a lower delivered VT. Variable compression of the driving gas is a fundamental obstacle to accurate VT delivery by any pneumatically powered, bellows-type ventilator.13 To reduce compliance-related δVT, some microprocessor-controlled ventilators perform an automated compliance test and then compensate for any compliance-related loss of delivered VT by deviating from the preset VT, which eventually allows for greater precision of VT delivery.1,5,8
Our findings further suggested that design differences in the bellows configuration accounted for differences in the δVT because both ventilators that operated with hanging bellows (ventilators A and B) produced a substantially smaller δVT than the 2 units with standing bellows (ventilators C and D). Furthermore, even in the case of the same type of bellows (eg, hanging bellows), differences in bellowed-ventilator designs will influence the accuracy of volume delivery. For instance, ventilator A was filled, as the manufacturer claimed,13 only to the desired VT that was preset by the operator who manually adjusted a plunger within the hanging bellows cylinder that limited the descent of the bellows. Provided the operator set a sufficient PIF rate and Tinsp, this mechanism would then ensure that the bellows emptied completely with each inspiration of the patient. This also explains why the impact of the bellows compliance was less in ventilator A (hanging bellows), compared with ventilators C and D (standing bellows), as long as the bellows were stretched less (as with smaller preset VT settings) and lower PIF rates (≤ 100 L/min) were applied, and why the δVT (under our testing conditions) was smaller for ventilator A, compared with ventilators C and D. Furthermore, if a ventilator with bellows begins the inspiratory phase with the bellows at maximum volume, as occurred with ventilators C and D, then compliance of the bellows is also at its maximum, and this alone compromises accurate VT delivery.
Another cause for a δVT might be a leak in the rebreathing circuit system. Because of the linearity between flow and pressure in a closed system,14 a higher inspiratory flow would result in a higher system pressure. The leak test was performed with a standard method of only 20 cm H2O pressure application on the basis of current recommendations.7,15,16 A higher test pressure during the PIF rate testing could have resulted in a different total leakage. Repeating the leak tests at various pressures could have allowed better estimation of losses in VT delivery when working with lower or higher peak inspiratory pressures. Consistent with a previous study,17 findings in the present study indicated that the δVT decreased when larger VT presets were used. Also, the volume of gas lost by expansion of the anesthetic circuit system during the inspiratory phase produces a δVT that is fixed and therefore relatively higher when small VT presets are used.12–18
There are a number of limitations to the present study that need to be considered when drawing conclusions for clinical settings. First, we did not measure the pressure in the circuit system during testing. It has been shown that volume losses and hence accuracy of VT delivery by ventilators are correlated with the maximum pressure achieved in the breathing system during inspiration.12 The questions of whether and to what extent peak anesthetic circuit system pressures exceeded 20 cm H2O in our study under the different testing conditions cannot be answered. Consequently, we do not know whether higher circuit system pressures could have accounted for volume losses and impacted the δVT during ventilation. Second, VT delivery by the ventilators was tested at room temperature and dry gas conditions. However, in clinical situations, when a delivered VT reaches a patient's lungs, the gas becomes saturated with water vapor and becomes warmer. On the basis of physical gas laws, when gases warm, their volumes increase.17 Likewise, any temperature increase promotes greater elasticity of components of the circuit system, particularly rubber bellows and potentially the breathing hoses, and results in greater compliance of the anesthetic circuit system. Whether these changes are clinically relevant can be answered only in follow-up studies performed with live animals under conditions similar to clinical situations. Third, although a variety of gases, including O2, N2, N2O, He, and medical air (21% O2 and 79% N2), are used in large-animal veterinary medicine and particularly in anesthesia of horses, O2 was the only gas used in the present study. These gases differ in their physical properties, such as viscosity, density, and heat capacity, and accordingly differ in their gas flow patterns.19,20 A recent study5 shows that a piston-driven ventilator delivers VT accurately and independently of the gas mixture used. In contrast to bellows-driven ventilators (defined as dual-circuit ventilators), piston-driven ventilators operate with a single circuit and work basically like a plunger within a syringe that is driven by a microprocessor-controlled electric motor. This unique design allows for very precise VT delivery because it does not involve an elastic bellows and because it uses a piston instead of a compressible gas to deliver the preset VT. Further studies are needed to evaluate the accuracy of bellows-driven ventilators when different gas mixtures are used. Fourth, the routine leak test that was performed with each pneumatically powered anesthesia ventilator at the beginning of each experiment did not test the bellows housing for the presence of any gas leak, which, depending on magnitude, could have affected the driving gas flow rate and pressure and thus the delivered VT. Fifth, for the present study, we used a calibration syringe with a graduated scale up to only 7 L. This limited the maximum VT we could test. From a statistical perspective and for a better extrapolation of our findings to clinical situations, it would have been valuable to have also tested larger volumes.
Another limitation, common to all bench-type studies, was the difficulty to extrapolate results from the laboratory to in vivo and then to clinical situations in which lung compliance and airway resistance also affect accuracy of VT delivery. For a given preset VT, the pressure that results in the breathing circuit is determined by the resistance and compliance of the breathing circuit and the patient's lungs. This applies much more so with pneumatic ventilators with bellows.13 Because the pressure in the bellows compartment will vary between patients and can vary between breaths in a patient, the gas driving the bellows will be subject to varying degrees of compression that cannot be predicted. Variable compression of the driving gas is a fundamental obstacle to accurate VT delivery by bellows ventilators. This is particularly true for small VT deliveries and high inspiratory pressures, the latter of which often occur in horses undergoing colic surgery in dorsal recumbency. The observed inaccuracies in actual VT delivery and the potential complications associated with inadequate VT delivery call for routine VT monitoring in clinical practice. A Pitot tube-based flow sensor is currently the only validated method for accurately measuring inspired and expired VT in anesthetized horses.21 Volume-controlled ventilation should ideally be adjusted on the basis of expired VT.1,21
All tested large-animal ventilators tested in the present study had a smaller delivered VT than preset VT, and at times the δVT exceeded 10%, which in human medicine is considered the upper limit of what is clinically acceptable.11 Still, the piston-driven ventilator (ventilator E) performed in this regard substantially better than the pneumatically powered ventilators (ventilators A through D). Ventilator E repetitively surpassed the 10% threshold for δVT only when black rubber breathing hoses and a preset VT of 1 L were used. Among the pneumatic ventilators, those with hanging bellows (ventilators A and B) performed superior to those with standing bellows (ventilators C and D), with the latter performing consistently with a δVT in excess of 10%, even when tested at larger VT settings. In all tested ventilators, the δVT depended on settings for VT and for PIF and FGF rates. Therefore, close monitoring of VT with external flow and volume meters is warranted, particularly when pneumatic ventilators are used or when very precise VT delivery is required.
Acknowledgments
Funded by a Raymond Firestone Trust and Tamworth Research grant. The authors declare that there were no conflicts of interest.
We thank Dr. John A.E. Hubbell and the leadership at Rood & Riddle Equine Hospital for hosting us and allowing us to test their anesthesia ventilators.
ABBREVIATIONS
| δVT | Volume error (difference between preset and actually delivered tidal volume) |
| FGF | Fresh gas flow |
| PIF | Peak inspiratory flow |
| PVC | Polyvinyl chloride |
| RR | Ventilation rate |
| Texp | Expiratory time |
| Tinsp | Inspiratory time |
| TMFVM | Thermal mass flow and volume meter |
| VT | Tidal volume |
Footnotes
Large Animal Control Center with AVE respirator, serial No. 291, Dräger Inc, Telford, Pa.
Surgivet LDS 3000 large animal anesthesia machine with DHV 1000 large animal ventilator, Smiths Medical, Minneapolis, Minn.
Model 2800 large animal anesthesia ventilation system, Mallard Medical, Redding, Calif.
Model 2800C large animal anesthesia ventilation system, Mallard Medical, Redding, Calif.
Tafonius Junior, Hallowell Engineering and Manufacturing Corp, Pittsfield, Mass.
SFM3000 Mass Flow Meter, Sensirion AG, Stäfa, Switzerland.
Hallowell Engineering and Manufacturing Corp, Pittsfield, Mass.
Excel 2016, Microsoft Corp, Redmond, Wash.
SAS, version 9.3, SAS Institute Inc, Cary, NC.
References
1. Moens Y. Mechanical ventilation and respiratory mechanics during equine anesthesia. Vet Clin North Am Equine Pract 2013;29:51–67.
2. Dreyfuss D, Saumon G. Ventilator-induced lung injury: lessons from experimental studies. Am J Respir Crit Care Med 1998;157:294–323.
3. Gajic O, Dara SI, Mendez JL, et al. Ventilator-associated lung injury in patients without acute lung injury at the onset of mechanical ventilation. Crit Care Med 2004;32:1817–1824.
4. Gravenstein N, Banner MJ, McLaughlin G. Tidal volume changes due to the interaction of anesthesia machine and anesthesia ventilator. J Clin Monit 1987;3:187–190.
5. Hopster K, Bertone C, Driessen B. Evaluation of the effects of gas volume and composition on accuracy of volume measurement by two flow sensors and delivery by a piston-driven large-animal ventilator. Am J Vet Res 2019;80:135–143.
6. Patil VP, Patil VP, Shetmahajan M, Divatia JV. The modern integrated anaesthesia workstation. Indian J Anaesth 2013;57:446–454.
7. Mosley CA. Anesthesia equipment. In: Grimm KA, Lamont LA, Tranquilli WJ, et al, eds. Lumb & Jones veterinary anesthesia and analgesia. 5th ed. Ames, Iowa: Wiley-Blackwell, 2015;23–85.
8. Wallon G, Bonnet A, Guérin C. Delivery of tidal volume from four anaesthesia ventilators during volume-controlled ventilation: a bench study. Br J Anaesth 2013;110:1045–1051.
9. Dorsch JA, Dorsch SE. Anaesthesia ventilators. In: Dorsch JA, Dorsch SE, eds. Understanding anaesthesia equipment. 5th ed. Philadelphia: Lippincott Williams and Wilkins, 2008;310-372
10. Jain RK, Swaminathan S. Anaesthesia ventilators. Indian J Anaesth 2013;57:525–532.
11. Bachiller PR, McDonough JM, Feldman JM. Do new anesthesia ventilators deliver small tidal volumes accurately during volume-controlled ventilation? Anesth Analg 2008;106:1392–1400.
12. Coté CJ, Petkau AJ, Ryan JR, et al. Wasted ventilation measured in vitro with eight anesthetic circuits with and without inline humidification. Anesthesiology 1983;59:442–446.
13. Dräger. The anesthesia ventilator. Dräger marketing communication. Lübeck, Germany: Drägerwerk AG and Co. KGaA, 2010.
14. Krope J, Dobersek D, Goricanec D. Flow pressure analysis of pipe networks with linear theory method, in Proceedings. 2006 WSEAS/IASME Int Conf Fluid Mech 2006;59–62.
15. Bednarski RM. Anesthetic equipment. In: Muir WW, Hubbell JAE, eds. Equine anesthesia monitoring and emergency therapy. St Louis: Mosby, 1991;325–351.
16. Association of Anaesthetists of Great Britain and Ireland, Hartle A, Anderson E, et al. Checking anaesthetic equipment 2012: Association of Anaesthetists of Great Britain and Ireland. Anaesthesia 2012;67:660–668.
17. Lyazidi A, Thille AW, Carteaux G, et al. Bench test evaluation of volume delivered by modern ICU ventilators during volume-controlled ventilation. Intensive Care Med 2010;36:2074–2080.
18. Forbat AF, Her C. Correction for gas compression in mechanical ventilators. Anesth Analg 1980;59:488–493.
19. Compressed Gas Association Inc. Handbook of compressed gases. 3rd ed. New York: Chapman & Hall, 1990;1–593.
20. Driessen B, Nann L, Klein LV. Use of a helium/oxygen carrier gas mixture for inhalation anesthesia during laser surgery in the airway of the horse. In: Steffey EP, ed. Recent advances in anesthetic management of large domestic animals. Ithaca, NY: International Veterinary Information Service, 2003;A0618.04032003.
21. Moens YP, Gootjes P, Ionita JC, et al. In vitro validation of a Pitot-based flow meter for the measurement of respiratory volume and flow in large animal anaesthesia. Vet Anaesth Analg 2009;36:209–219.
