Arieh Oppenheim-Eden, MD; Yitzhak Cohen, MD; Charles Weissman, MD; and Reuven Pizov, MD

The Effect of Helium on Ventilator Performance* Study of Five Ventilators and a Bedside Pitot Tube Spirometer Arieh Oppenheim-Eden, MD; Yitzhak Cohen, MD; Charles Weissman, MD; and Reuven Pizov, MD Objective: To assess in vitro the performance of five mechanical ventilators Siemens 300 and 900C (Siemens-Elma; Solna, Sweden), Puritan Bennett 7200 (Nellcor Puritan Bennett; Pleasanton, CA), Evita 4 (Dragerwerk; Lubeck, Germany), and Bear 1000 (Bear Medical Systems; Riverside CA) and a bedside sidestream spirometer (Datex CS3 Respiratory Module; Datex- Ohmeda; Helsinki, Finland) during ventilation with helium-oxygen mixtures. Design: In vitro study. Setting: ICUs of two university-affiliated hospitals. Methods and measurements: Each ventilator was connected to 100% helium through compressed air inlets and then tested at three to six different tidal volume (VT) settings using various helium-oxygen concentrations (fraction of inspired oxygen [FIO 2 ] of 0.2 to 1.0). FIO 2 and VT were measured with the Datex CS3 spirometer, and VT was validated with a water-displacement spirometer. Main results: The Puritan Bennett 7200 ventilator did not function with helium. With the other four ventilators, delivered FIO 2 was lower than the set FIO 2. For the Siemens 300 and 900C ventilators, this difference could be explained by the lack of 21% oxygen when helium was connected to the air supply port, while for the other two ventilators, a nonlinear relation was found. The VT of the Siemens 300 ventilator was independent of helium concentration, while for the other three ventilators, delivered VT was greater than the set VT and was dependent on helium concentration. During ventilation with 80% helium and 20% oxygen, VT increased to 125% of set VT for the Siemens 900C ventilator, and more than doubled for the Evita 4 and Bear 1000 ventilators. Under the same conditions, the Datex CS3 spirometer underestimated the delivered VT by about 33%. Conclusions: At present, no mechanical ventilator is calibrated for use with helium. This investigation offers correction factors for four ventilators for ventilation with helium. (CHEST 2001; 120:582 588) Key words: artificial ventilation; helium; intensive care; mechanical ventilator; spirometry Abbreviations: Fio 2 fraction of inspired oxygen; PEEP positive end-expiratory pressure; Vt tidal volume Helium was introduced for the treatment of upper-airway obstruction and asthma before World War II; however, its use was later discontinued because of short supply. 1 3 In recent years, a *From the Departments of Anesthesia and Critical Care Medicine (Drs. Oppenheim-Eden and Pizov), Carmel Lady Davis Medical Center, Technion Medical School, Haifa; and Hadassah University Medical Center (Drs. Cohen and Weissman), the Hebrew University, Hadassah School of Medicine, Jerusalem, Israel. This work was presented in part in the European Society of Anaesthesia 7th annual meeting, May 29 June 1, 1999, Amsterdam, The Netherlands. Manuscript received May 25, 2000; revision accepted February 23, 2001. Correspondence to: Arieh Oppenheim-Eden, MD, Department of Anesthesia and Critical Care Medicine, Carmel Medical Center, 7 Michal St, Haifa, 34362 Israel; e-mail: galo@netvision.net.il renewed interest in the use of helium in asthmatic patients has arisen, and it has been shown to be effective both in spontaneously breathing patients and patients receiving mechanical ventilation who have increased airway resistance. 4 7 In such patients, the use of helium has led to improved carbon dioxide removal and reduced peak and mean airway pressures, as well as to improved oxygenation. Helium has a density one seventh of that of air and improves ventilation by facilitating the conversion of turbulent to laminar flow. 8 Although the effect of changing helium concentration on flow is linear, the use of low helium concentrations is associated with a reduced clinical effect. 9 Therefore, for helium to have a clinically significant effect, helium concentrations of 50% are probably required, since at lower concen- 582 Laboratory and Animal Investigations

trations the difference in densities becomes too small to exert a clinically significant effect (Fig 1). The use of helium is associated with erroneous and unpredictable delivery of tidal volume (Vt) by various mechanical ventilators, 10,11 and also may lead to inaccurate delivery of fraction of inspired oxygen (Fio 2 ), because of the difference in physical properties between helium and air, and different ventilator design. In previous studies, helium was delivered either in a fixed concentration, or when 20% oxygen in helium was connected to the air inlet of the ventilator. 10,11 To date, no mechanical ventilator has been produced that addresses the problem of delivering gases or gas mixtures with variable densities. In the present study, we examined the performance of five currently used mechanical ventilators while delivering various Vts at different heliumoxygen concentrations, with special emphasis on high helium concentrations: Siemens 300 and 900C (Siemens-Elma, Solna, Sweden), Puritan Bennett 7200 (Nellcor Puritan Bennett, Pleasanton, CA), Evita 4 (Dragerwerk, Lubeck, Germany), and Bear 1000 (Bear Medical Systems, Inc., Riverside CA). We also studied the function of a commercially available Pitot tube sidestream spirometer (Datex CS3 Respiratory Module; Datex-Ohmeda, Helsinki, Finland). Unlike previous reports, we used pure helium because of unavailability, at the time, of helium-oxygen mixture. Materials and Methods Five mechanical ventilators and Pitot tube sidestream spirometer were studied, as described above. The Datex CS3 spirometer was calibrated prior to each experiment using manufacturerrecommended procedures. The ventilatory circuit was checked for the absence of leaks prior to each test. All tests were performed at 700 mm Hg barometric pressure, at room temperature. FIO 2 Measurements Fio 2 was measured by the Datex CS3 spirometer (a rapidresponse paramagnetic oxygen analyzer), and the measurement used in the analysis is termed delivered Fio 2. VT Measurements Vt measurements were performed using a density-independent volume monitor that relies on measurement of water displacement (sidestream spirometry tester, Datex-Ohmeda). At each of the conditions tested, measurements were made visually by two independent investigators simultaneously, who were blinded to Fio 2 and gas mixture. Measurements were made during three breaths cycles at each condition, and the mean of six measurements was calculated and is termed actual delivered Vt for that specific condition. Following each change in ventilatory parameters or Fio 2, 15 min were allowed for equilibration of the system. Inspiratory Vt was recorded from the Datex CS-3 module. Study Protocol The performance of each ventilator was examined separately (Fig 2). Helium was connected to the air inlet of each ventilator, and oxygen was connected to the oxygen inlet. The Siemens 900C ventilator was also studied while helium was administered through the nitrous oxide inlet. The Evita 4 ventilator was studied with the flow sensor and leak compensation turned off, as recommended by the manufacturer. The ventilator was connected using standard tubing to the spirometry transducer of the Datex CS3 monitor (including heat and moisture exchange device), to a standard 90 -angle elbow and through an 8.5-mm internal-diameter endotracheal tube to the water-displacement spirometer. The ventilators were tested in volume-control mode. Preliminary evaluation of the ventilators using pressure-control ventilation and positive end-expiratory pressure (PEEP) showed that unlike during volume-control ventilation, no difference between helium and air was found (data not shown). Each ventilator was studied at three to six preset Vts that were achieved using at least two inspiratory flow rates. The helium and oxygen were mixed so that Fio 2 values between 0.21 and 1.0 were achieved. Mathematical Correction of VT Measurement by the Datex CS3 Spirometer During rapid, turbulent, inspiratory flow such as occurs during artificial ventilation, the relation between the driving pressure and flow velocity and pressure is calculated according to the Bernoulli theorem 8 : P U 2 /2 (1) where density, P driving pressure, and U flow velocity. Rearranging this equation directly calculates flow velocity: Figure 1. Relative density of helium-oxygen and nitrogenoxygen mixtures compared to 100% oxygen. Figure 2. Schematic drawing of the study apparatus. Helium at 140 barometric pressure was connected to the air-supply port of each ventilator through a pressure-reducing valve. The Datex CS3 sidestream spirometer (SSS) was connected distal to the Y-piece, and through an elbow connector to an 8.5-mm innerdiameter endotracheal tube that was attached to the waterdisplacement spirometer. CHEST / 120 / 2/ AUGUST, 2001 583

U 2 P/ 1/ 2 (2) Integration of the flow velocity over time, with knowledge of the cross-section of the resistor, yields Vt. The spirometer is also equipped with a gas analyzer that analyzes oxygen (using a paramagnetic analyzer) and carbon dioxide (infra-red spectroscopy). The algorithm assumes that the balance gas is nitrogen, and the mean density of the gas mixture is calculated and inserted into equation 2. The use of helium instead of nitrogen as the balance gas leads to erroneous flow and Vt calculations because helium has a lower density than nitrogen. Similar errors were described with the use of xenon, which has a greater density than nitrogen 12 : U 2 P/ air 1/ 2 (3) where U erroneous flow calculated during helium ventilation. To calculate the true flow velocity (Utrue), the density of the gas mixture must be known: Utrue 2 P/ He:O 2 1/ 2 (4) Dividing and rearranging equation 3 and equation 4 yields: Utrue U air/ He:O 2 1/ 2 (5) As Vt is calculated by integration of flow velocity over time: Vt Vti air/ He:O 2 1/ 2 (6) Because both gas mixtures (helium-oxygen [He:O 2 ] and nitrogen-oxygen [air]) behave as ideal gases under physiologic conditions, their density is proportional to their mean molecular weight (Fig 1). Therefore, molecular weight (MW) can be substituted for density in equation 6, and a correction factor for Vt measurement by the Datex CS3 monitor can be reached: Data Analysis Vttrue Vti MWair/MWHe:O 2 1/ 2 (7) Data analysis was performed as a function of Fio 2 rather than fraction of inspired helium. The density of the inspired gas was represented by the Fio 2, which is the only measurable independent variable (fraction of inspired helium 1 Fio 2 ). Best-fit curve analysis and regression analysis were performed with software (Excel; Microsoft Corporation; Redmond, WA). Results During helium ventilation, a total of 84 different conditions were studied using the Siemens 900C ventilator, 18 with the Siemens 300 ventilator, 45 with the Evita 4 ventilator, and 25 with the Bear 1000 ventilator. Five measurements were made using the Puritan Bennett 7200 ventilator; however, Vt output of this ventilator was very low, and thus the machine was excluded from further study since it is unsuitable for ventilation with helium. Both delivered Fio 2 and actual Vt were also measured during air ventilation, and all five ventilators plus the Datex CS3 spirometer were in agreement with Fio 2 and Vt settings as well as water-displacement volume measurements during ventilation with air (data not shown). Effect of Helium on Delivered FIO 2 Fio 2 measurements from the Datex CS3 spirometer and from the oxygen analyzers of each of the ventilators equipped with oxygen analyzers were in agreement (data not shown). Measured Fio 2 during helium ventilation with the Siemens 300 and 900C ventilators (helium administered through the air inlet) deviated from the Fio 2 set on the ventilator in a linear fashion (Fig 3, top left, a, and top right, b; Table 1). During the study of the Evita 4 and Bear 1000 ventilators, the relationship between the set and delivered Fio 2 was not linear, but was described by a power function (Fig 3, bottom left, c, and bottom right, d; Table 1). The relation between set and delivered Fio 2 was also nonlinear during helium administration through the N 2 O port of the Siemens 900C (Fig 3, top left, a; Table 1); however, in this instance, the relationship was best described by an exponential equation. Effect of Helium Ventilation on VT Delivery Vt was unaffected by helium during ventilation with the Siemens 300 ventilator (Fig 4, top right, b; Table 1) and actual Vt did not deviate by 5% from set Vt. However, with the Siemens 900C, Evita 4, and Bear 1000 ventilators, actual Vt was affected by the helium concentration in a nonlinear, dose-dependent fashion (Fig 4, top left, a; top right, b; bottom right, d; Table 1). With the Siemens 900C ventilator, alteration in actual Vt was independent of the inlet through which helium was administered and exceeded the set Vt by about 25% when 80% helium was used. However, with the Evita 4 and Bear 1000 ventilators, actual Vt exceeded the set Vt by almost 100% when 80% helium was delivered. Vt was unaffected by the inspiratory flow rate (data not shown). The Performance of a Pitot Tube Spirometer During Helium Ventilation The function of the Datex CS3 respiratory monitoring module was assessed during helium ventilation. At each condition, Fio 2 as well as the inspired and expired Vt measurements were taken directly from the Datex CS3 monitor. The inspired Vt measured by the Datex SC3 monitor was compared to actual Vt. The actual Vt/inspired Vt ratio correlated inversely in a nonlinear way to Fio 2 during ventilation with helium (Fig 5), and reached a value of 1.5 during ventilation with 80% helium. True Vt, which was calculated from inspired Vt according to equation 7, using relative density from Figure 1, correlated to actual Vt (true Vt 1.06 actual Vt 7.6; r 2 0.986; Fig 6). 584 Laboratory and Animal Investigations

Figure 3. Delivered Fio 2 as a function of set Fio 2 during ventilation with helium and oxygen for four mechanical ventilators at all study conditions. Top left, a: Siemens 900C ventilator (triangles helium connected to the N 2 O inlet; squares helium to the air inlet). Top right, b: Siemens 300 ventilator. Bottom left, c: Evita 4 ventilator. Bottom right, d: Bear 1000 ventilator. Mathematical description and r 2 values appear in Table 1. Discussion The present study demonstrates that the function of different respiratory equipment (both ventilators and monitors) is affected by the use of helium. Mathematical correction factors were developed for the equipment studied, and these as well as the graphic results can be used to guide ventilatory therapy with helium when a pure helium source is used. Changes in Gas-Flow Dynamics and Their Effect on the Ventilator Because of its much-reduced density (about one seventh of that of air), helium may alter gas flow dynamics in ventilator tubing and blenders. Gas flow is generally either laminar or turbulent, and gas density is one of the variables that determines flow pattern. Laminar flow predominates when gas with Table 1 Mathematical Relation Between Delivered and Set FIO 2 and VT* Ventilators Delivered Fio 2 r 2 Actual Vt r 2 Siemens 300 1.18 0.21 0.99 Set Vt NA Siemens 900C (air) 1.21 0.27 0.95 Set Vt y 0.13 0.85 Siemens 900C (N 2 O) 0.08e 2.56x 0.99 Set Vt y 0.13 0.85 Evita 4 0.99x 2.67 0.99 Set Vt 1.05 y 0.4 0.97 Bear 1000 1.04x 3.06 0.99 Set Vt 1.024 y 0.37 0.98 *e natural logarithm; x set Fio 2 ;y delivered Fio 2 ;NA not applicable. CHEST / 120 / 2/ AUGUST, 2001 585

Figure 4. True Vt as monitored by water-displacement spirometry (Vtactual), as a function of delivered Fio 2 during ventilation with helium. Each line represents a single Vt setting. The set Vt equaled the actual Vt measured at Fio 2 of 1.0. Top left, a: Siemens 900C ventilator. Top right, b: Siemens 300 ventilator. Bottom left, c: Evita 4 ventilator. Bottom right, d: Bear 1000 ventilator. Mathematical description and r 2 values appear in Table 1. low density flows through a straight, wide, smooth tube, at slow rates. During laminar flow, flow velocity is directly proportional to the pressure gradient, and is independent of gas density as shown by the Hagen-Poiseuille low: P U8 l/ r 4 (8) where P driving pressure, U flow velocity, gas viscosity, and l and r length and radius of the tube, respectively. The transition from laminar to turbulent flow occurs when a critical flow velocity for the specific system is reached, as defined by Reynolds number: Re 2Ur / (9) where Re Reynolds number, and density. For example, gas flow through a smooth, straight tube will be laminar if the Reynolds number is 2,000 and will be turbulent if it is 4,000. 8,11 Once the Reynolds number exceeds the critical limit for the system and turbulent flow ensues, the physical laws governing flow velocity can be derived from the Bernoulli principle (equation 1). The inspiratory-flow regulators and valves, which control the delivery of a preset Fio 2 and Vt, function under turbulent flow conditions because of their physical design. 11 As can be seen from the Bernoulli equation (equation 3), under these conditions flow becomes density dependent. Therefore, the use of helium instead of either air or N 2 O can be expected to erroneously elevate helium flow, thereby reducing Fio 2 and increasing Vt. However, monitoring equipment, which relies on the flow properties of the gas to measure Vt, will read falsely reduced Vt during helium ventilation. Helium differs from air not only in density, but also in viscosity and thermal qualities. The viscosity of helium is about 10% higher than that of air, but the viscosity of gas mixtures is difficult to predict and 586 Laboratory and Animal Investigations

Figure 5. The ratio of Vt monitored by water-displacement spirometry (Vt-actual) and inspired Vt (Vti) monitored by the Datex CS3 spirometer as a function of delivered Fio 2 during helium ventilation (y 0.96x 0.3 ; r 2 0.96). does not always follow a linear pattern. The specific heat and thermal conductivity of helium are much higher than those of nitrogen and oxygen. Therefore, instruments that rely on the thermal properties of the gas to measure flow are liable to produce grossly inaccurate measurements. The flow regulator of the Puritan-Bennett 7200 ventilator is controlled by heated-wire flowmeters and therefore perceives the helium flow to be much higher than actual. This causes the gas flow delivered to the ventilator to be reduced, leading to small Vts. Additionally, Fio 2 is increased. This has been demonstrated previously. 10 Alternatively, during ventilation with helium-oxygen mixtures, volume monitors that rely on heated-wire technology will greatly overestimate Vt. Similar results were shown during xenon ventilation. 12 The Effect of Helium on FIO 2 Delivered Fio 2 deviated from the set Fio 2 for all the ventilators studied. This is in contradiction to the findings in an article by Tassaux et al. 11 However, this difference can be explained by the fact that Tassaux et al 11 used a mixture of 78% helium and 22% oxygen delivered to the ventilator air inlet, Figure 6. Inspired Vt derived from the Datex CS3 spirometer corrected for gas-density difference in relation to actual Vt. See Figure 5 legend for abbreviations. while in the present study 100% helium was delivered through the air inlet. This difference in methodology explains our findings for the change in Fio 2 during the use of Siemens 300 and 900C ventilators (helium administered through the air supply inlet), where Fio 2 roughly equaled 1.2 (set Fio 2 ) 0.2 (Fig 3, top left, a, and top right, b; Table 1). The use of the N 2 O port for the delivery of helium is attractive, because such use would reduce the risk for inadvertent delivery of hypoxic gas mixture. However, when helium was thus administered using the Siemens 900C ventilator, Fio 2 behaved in a nonlinear relation. This was probably due to the high density of N 2 O, which requires special compensation for the varying density of the inspired gas (Fig 3, top left, a; Table 1). During helium ventilation with both the Bear 1000 and Evita 4 ventilators, Fio 2 was nonlinear (Fig 3, bottom left, c, and bottom right, d; Table 1). These results deviate from previous findings, which were based on the use of a gas mixture of 22% oxygen in 78% helium introduced into the air inlet. Such a gas mixture has a density of more than twice that of helium (Fig 1). The difference in gas density flowing through the air-flow regulator may explain why our results show a more extreme nonlinear relationship between set Fio 2 and delivered Fio 2. 11 Effect of Helium Administration on VT The use of helium, as has been shown previously, did not influence the Vt output of the Siemens 300 ventilator. 11 This is probably because the flow-regulating valve has a built-in compensatory mechanism for changing gas density due to change in temperature when gas is supplied from a high-pressure source. 11 The Vt output of the Siemens 900C ventilator was dependent in a nonlinear way on the Fio 2 ; when actual Fio 2 was 20%, Vt was increased by roughly 25%. Tassaux et al 11 showed an increase of 40% in Vt under similar conditions. However, the output of the machine was independent of the gas supply port used for the delivery of helium (air or N 2 O). The Bear 1000 and Evita 4 ventilators also had increased Vt with increasing helium concentration, which behaved nonlinearly. Unlike the Siemens 900 C ventilator, however, both ventilators had almost doubled their output when 80% helium was delivered, which agrees with the findings by Tassaux et al 11 concerning the Evita 4 ventilator. For the Evita 4 ventilator, Tassaux et al 11 showed a linear relationship between set and delivered Vt at Fio 2 50%; however, such gas mixtures would not be clinically useful. During higher helium concentrations, they state that a nonlinear relationship was found, but as no detailed description is available, those results cannot be compared to the present findings. The difference between delivered and set CHEST / 120 / 2/ AUGUST, 2001 587

Vt at various Fio 2 levels can be attributed to the different types of the inspiratory valves, flow regulators, and compensatory mechanisms. 11 The present study addressed volume-control ventilation. Pressure-control ventilation and PEEP were also assessed, but only in a preliminary way (data not shown). During pressure-control ventilation using the current lung model, Vt is dependent on the weight of the displaced water column; therefore, the Vt would not differ between helium and air ventilation. For the same reason, there would be no difference in the effect of PEEP between the two gases. This is in agreement with previous results. 11 Effect of Helium Ventilation on the Performance of a Pitot Tube Spirometer During the study, Vt was monitored through the Datex CS3 sidestream spirometer unit. This spirometer uses the Pitot tube principle. Pressure is monitored continuously before and after a constriction, which creates resistance to flow and induces turbulence. Once turbulence ensues, the Bernoulli principle is used to calculate the Vt. 3,12 However, during slow flow (such as during most of the expiration), flow remains laminar and is therefore calculated according to Poiseuille law, using viscosity rather than density. The Datex CS3 spirometer resorts to tabulated results during laminar flow 3 ; therefore, we decided to analyze the inspired rather than the expired Vt of the Datex CS3 spirometer as has been done by others. 12 Our results show that with simple mathematical correction, the Datex CS3 spirometer can be used to monitor controlled ventilation with helium-oxygen mixtures (Fig 6). Potential Hazards During the Use of Helium for Mechanical Ventilation Although helium ventilation may prove beneficial to patients with small-airway obstruction, its uncontrolled use may be hazardous as ours and previous studies suggest. Helium ventilation leads to severe hypoventilation by some ventilators, while other ventilators may deliver excessively large Vts. The use of helium may lead to erroneous Fio 2 delivery with the risk of delivering hypoxic gas mixtures. Using heliox rather than pure helium instead of air would be safer, as it reduces this risk. However, the most important factor in the use of helium, or otherwise tampering with ventilatory equipment, is vigilant monitoring. It is essential to monitor not only the saturation and arterial blood gas levels of the patient, but also to monitor continuously the Fio 2. While no bedside volume monitor has been shown to function properly with helium, the Datex CS3 sidestream spirometer can be used with appropriate mathematical conversion according to the true density of the delivered gas mixture. Summary At present, no mechanical ventilator is calibrated for the use of helium. Although some mechanical ventilators perform in a predictable fashion during helium ventilation, the uncontrolled use of helium could be hazardous for the patient. The Puritan- Bennett 7200 ventilator does not function during helium ventilation, while the Siemens 300 ventilator is the most accurate machine presently available for ventilation with helium. The Datex CS3 spirometer cannot be used to monitor Vt during helium ventilation, unless mathematical corrections for gas density are applied. However, when the clinical situation calls for ventilation with helium, the ventilators studied can be used according to either the mathematical or graphic corrections. The true Vt delivered to the patient can also be calculated from that measured by the Datex CS3 spirometer. Further studies are required to validate these findings under clinical conditions. ACKNOWLEDGMENT: We thank Rina Ankori and Alex Tuchband for technical assistance. References 1 Barach A. Use of helium as a new therapeutic gas. Proc Soc Biol Med 1934; 32:462 464 2 Barach A. The therapeutic use of helium. JAMA 1936; 107:1273 1280 3 Sondergaard S, Karason S, Lundin S, et al. Evaluation of a Pitot type spirometer in helium/oxygen mixtures. J Clin Monit 1998; 14:425 431 4 Manthous C, Hall J, Caputo M, et al. Heliox improves pulsus paradoxus and peak expiratory flow in nonintubated patients with severe asthma. Am J Respir Crit Care Med 1995; 151:310 314 5 Gluck E, Onorato D, Castriotta R. Helium-oxygen mixtures in intubated patients with status asthmaticus and respiratory acidosis. Chest 1990; 98:693 698 6 Pizov R, Oppenheim A, Eidelman L, et al. Helium versus oxygen for tracheal gas insufflation during mechanical ventilation. Crit Care Med 1998; 26:290 295 7 Schaeffer EM, Pohlman A, Morgan S, et al. Oxygenation in status asthmaticus improves during ventilation with heliumoxygen. Crit Care Med 1999; 27:2666 2670 8 Faber PE. Fluid dynamics for physicists. Cambridge, UK: Cambridge University Press, 1995; 1 77 9 Houck JR, Keamy MF III, McDonough JM. Effect of helium concentration on experimental upper airway obstruction. Ann Otol Rhinol Laryngol 1990; 99:556 561 10 McArthur C, Adams A, Suzuki S. Effects of helium/oxygen mixtures on delivered and died tidal volume during mechanical ventilation [abstract]. Am J Respir Crit Care Med 1996; 153:A370 11 Tassaux D, Jolliet P, Thouret J, et al. Calibration of seven ICU ventilators for mechanical ventilation with helium-oxygen mixtures. Am J Respir Crit Care Med 1999; 160:22 32 12 Goto T, Saito H, Nakata Y, et al. Effects of xenon on the performance of various respiratory flowmeters. Anesthesiology 1999; 90:555 563 588 Laboratory and Animal Investigations