APPARATUS Supplementary oxygenation with the laryngeal mask airway: a comparison of four devices

Transcription

1 APPARATUS Supplementary oxygenation with the laryngeal mask airway: a comparison of four devices P. Peyton, 1 D. Cowie 2 and W. Howard 1 1 Staff Anaesthetist and 2 Anaesthetic Registrar, Department of Anaesthesia, Austin and Repatriation Medical Centre, Studley Road, Heidelberg, VIC 3084, Australia Summary The provision of supplementary oxygen via the laryngeal mask airway used in the recovery room is important for patient safety. Several devices have been described for this purpose, but these studies have not included an accurate measurement of the most clinically important variable, the end-tidal oxygen concentration. We constructed an artificial model of spontaneous ventilation to compare the efficacy and safety of four devices; a circuit filter, a Hudson mask, the `T-bag' device and a T-piece. We combined the use of oximetry with a pneumotachograph to provide a continuous picture of the oxygen delivery characteristics of the devices at flow rates of 2, 4 and 8 l.min 21. The performances of the T-bag and the T-piece were superior to those of the filter and Hudson mask, with end-tidal oxygen concentrations of 46.1%, 45.8% and 35.4%, 34.8%, respectively, at 8 l.min 21. Single point assessments of oxygen delivery, such as peak inspired oxygen concentration, may overestimate the efficacy of test devices. Keywords Equipment: airway; laryngeal mask. Oxygen: delivery systems; measurement.... Correspondence to: Dr P. Peyton Accepted: 14 February 2000 It is common practice to leave the laryngeal mask airway (LMA) in place until the patient has regained consciousness in the recovery room, and this has been supported by recent studies [1±3]. Supplementary oxygen delivery during recovery from anaesthesia is indicated for the following reasons: airway obstruction from reduced airway muscle tone, hypoventilation, pulmonary shunting, cardiovascular depression with reduced O 2 delivery, the potential for shivering resulting in increased O 2 demand, and diffusion hypoxia (following nitrous oxide anaesthesia). Several devices for providing supplementary oxygen through the LMA during emergence have been investigated [4±10]. However, the efficacy of these devices has not been adequately compared because many studies do not include the most clinically important variable, an accurate measurement of end-tidal oxygen concentration. It is this value that will determine the total lung oxygen store and hence the safety margin in the recovery room in the event of an incident which compromises ventilation. Oxygraphy from the nasopharynx has been used to evaluate variable performance oxygen delivery devices [11] and has demonstrated the limitations of gas sampling at a fixed phase of the respiratory cycle. It is of even more importance to examine the dynamic gas concentration waveform in relation to total gas flow, to gain a true picture of gas delivery. Idiosyncracies in the real-time O 2 concentration waveform produced by different O 2 delivery devices make `end-tidal' and `inspired' O 2 concentration readings from some rapid gas analysers unreliable. In this study we combined oxygraphy with the use of a pneumotachograph to provide a continuous picture of the oxygen delivery characteristics of our test devices throughout the respiratory cycle. We hypothesised that we could determine the best device and gas flow to provide supplementary oxygen through the LMA. Methods To achieve a high degree of reproducibility of results, an artificial lung model was used (Fig. 1). This respiration 992 q 2000 Blackwell Science Ltd

2 P. Peyton et al. Supplementary oxygen with the LMA Figure 1 Test lung apparatus ± schematic. simulator models spontaneous ventilation using an MII Vent Aid Test/Training lung consisting of a pair of hinged bellows locked together. Intermittent positive pressure was applied to the driving chamber bellows by an Ulco EV 500 ventilator at a set rate of 16 breath.min 21 and appropriate tidal volume and I : E ratio of 1 : 3. The orifice of the second test lung chamber was connected to a length of tubing of 150 ml volume to simulate anatomical dead space. A Fleisch pneumotachograph (Hans Rudolph, USA), with an internal volume of 88 ml, was incorporated into this dead space volume. Carbon dioxide and nitrogen gases were delivered into the test lung at flow rates calculated to simulate respiratory gas exchange in a spontaneously breathing, anaesthetised patient. This additional volume of gas was allowed to bleed off through a flap valve at the lung inlet. This allowed the inspired and expired tidal volumes to be equal (360 ml), as confirmed with a Wright spirometer. The LMA has tubing of approximately 15 ml volume depending on the size. We used an equivalent volume of polyvinyl chloride tracheal tubing with 15-mm connector at its end. Gas sampling was performed from a sampling catheter advanced through this tubing to a position adjacent to its connection to the test lung, simulating placement just above the laryngeal inlet of an LMA. Oxygen, nitrogen and carbon dioxide concentrations were determined continuously by a Cardiocap II (Datex) gas analyser and recorded at a sampling rate of 150 Hz and displayed in real-time by Labview scientific data display and analysis software running on a Macintosh Power PC (Apple) computer. Simultaneous in-line capnography using a Hewlett-Packard HP 78354A monitor and flow measurement by the pneumotachograph were performed. Flows of individual gases were calculated by multiplication of total gas flow and individual gas concentration, point by point. The application allowed display of gas-flow, oxygen concentration waveform and carbon dioxide concentration waveform, and measurement or calculation of concentrations, partial pressures and flows of gases. Volumes of gas delivered for any interval within the respiratory cycle were calculated by integration of the flow curve vs. time. End-tidal oxygen concentrations were measured by manual identification of the appropriate point of the gas concentration waveform with a cursor. Four principal sources of error apply to this measurement process. A failure to achieve correct phase alignment of flow and concentration waveforms of all measured gases, first of all because of the delay inherent in sidestream sampling systems caused by the time to transport sampled gases from sample point to the gas analyser, and secondly because of the different response times of O 2 (polarographic) and CO 2 (infrared) measurement devices. Thirdly, the finite rise time of these devices may cause underestimation of changes in flow, concentration or both. Finally, appropriate correction must be made for the tidal variation in the mean viscosity of the gas mixture being sampled, when measuring laminar gas flow using a Fleisch pneumotachograph. Correction for the first of these factors was achieved in q 2000 Blackwell Science Ltd 993

3 P. Peyton et al. Supplementary oxygen with the LMA Anaesthesia, 2000, 55, pages 992±999 Figure 2 Test devices. Filter, Hudson mask, T-bag, T-piece. the data processing program by deliberate delay of the faster response waveforms, and then further refined by comparing in-line capnography with the side port sampled capnography. It was noted that the in-line capnograph wave and the side-stream wave would vary in their phase (timing) relationship to each other, presumably as a result of variations in the gas flow sampling rate of the Cardiocap rapid gas analyser. This `wander' was corrected for, on a breath-by-breath basis, by comparing the mid-point of the expiratory upstroke of the capnograph waveforms and coercing the sidestream waveform to overlie the in-line waveform at this point. (Its accompanying oxygraphy wave was moved proportionally.) The in-line capnograph waveform and the pneumotachograph waveform were observed to maintain a constant phase relationship throughout the experiment. A further correction was necessary for the rise time of the rapid gas analysers [12]. The method employed was to assume that the response of the gas analyser in measurement of a step change in gas concentration is exponential in nature. Thus, if the time constant is known, the real concentration change over that time can be predicted from the measured concentration change, on the basis that this would represent 63% of the real change after one time constant. An approximate value for the time constant was calculated from the stated maximum rise time (0±90%) of the analyser. In fact, it was observed that this algorithm overcorrected the concentration change, producing an overshoot in the upstroke or downstroke of the capnograph waveform, before returning the wave to its true plateau value. This is presumably because the stated rise time of the analyser is a maximum value and that performance is often faster than this. Thus, it was necessary to assume a slightly shorter time constant, so that overshooting did not occur. Finally, total flows measured by the pneumotachograph were corrected on a point-by-point basis for changes in average gas viscosity secondary to tidal changes in fractional gas concentrations. Four oxygen delivery devices commonly used in clinical practice were assessed (Fig. 2): 1 The circuit filter (Bact-Trap Mini, Pharma Systems AB, Sweden), an adult hygroscopic heat and moisture exchange filter. Oxygen tubing was attached to the Luerlock port on the non-lung side of the filter. 2 The Hudson mask (Hudson RCI, Mexico) consists of a vented, clear plastic facemask. Oxygen tubing was attached to the orifice in the mask and the mask was placed so that the oxygen inlet at the tip of the nose directly overlay the end of the simulated LMA tube. 3 The `T bag' Oxygen Enhancement Device (Ultimate Medical Pty Ltd. Vic. Australia), an injectionmoulded, polyethylene T-piece and shaped reservoir bag of 300 ml volume. 4 The T-piece, a 15-cm length of corrugated tubing attached to a right angle connector with a side port where oxygen tubing is attached, of total volume 50 ml. As a control, the system was also assessed with no device attached to the LMA tube and no supplementary oxygen (no device, room air). The minute ventilation of the test lung was kept constant between tests by ensuring the end-tidal carbon 994 q 2000 Blackwell Science Ltd

4 P. Peyton et al. Supplementary oxygen with the LMA dioxide remained at 45 mmhg with no device attached. For each test device, data were recorded at oxygen flow rates of 2, 4 and 8 l.min 21. The device was attached to the end of the LMA tubing and the fresh gas flow of oxygen was delivered to the device. Between changes in device or O 2 flow, an equilibration period of 2 min was allowed before measurements began, a time that was observed to be sufficient to allow steady state within the model to be achieved. Nine consecutive breaths were sampled and displayed on the computer. For each test device and flow rate, nine tests were performed representing 81 breaths. The program automatically identified the inspiratory component of each breath. The end-tidal plateau for oxygen concentration was identified manually. The program then averaged the data from all breaths and displayed end-tidal and maximum oxygen and carbon dioxide concentrations, and calculated mean oxygen concentration. The volume (at ambient temperature and pressure dry) of oxygen (Vo 2 ) delivered and carbon dioxide (Vco 2 ) rebreathed during the inspiration were calculated by integration of flow curve for each gas. End points The concentration of oxygen in the end-tidal expired gas was the primary end-point. Secondary end-points were direct measurements of oxygen delivery (mean and maximum inspired oxygen concentration and Vo 2 ) and those concerned with carbon dioxide removal (minimum inspired concentration, end-tidal carbon dioxide concentration and Vco 2 ). Statistical methods The results were separated into groups according to the oxygen delivery device and oxygen flow. Within each group, means and standard deviations (SD) were calculated. The very small SD of each group confirmed the stability and reproducibility of measurements in all situations tested. When appropriate, comparisons between groups were made using anova testing with post hoc Bonferroni testing. Individual comparisons were made using a t-test for paired means. Probability values of less than 0.05 were considered to be significant. Table 1 Baseline respiratory variables (no device, no supplementary oxygen). Results are expressed as mean (SD). Number of independent trials 27 Minute volume; l.min (0.27) End-tidal oxygen concentration; % 16.3 (0.15) Maximum inspired oxygen; % 21.6 (0.50) Mean inspired oxygen concentration; % 20.8 (0.16) Inspired tidal oxygen volume; ml 76.0 (1.0) End-tidal carbon dioxide; mmhg 44.5 (1.3) Inspired carbon dioxide; mmhg 0.02 (0.45) Rebreathed carbon dioxide volume; ml 0.95 (0.13) Note that with this experimental lung there is no difference between inspired and expired volumes, the respiratory quotient is 1.0 and, since there is no water vapour, the ambient gas pressure totals 760 mmhg. concentration was 21.6% with a mean concentration of 20.8%. Oxygen 76 ml was delivered to the lung with each inspiration and an end-tidal oxygen concentration of 16.3% was recorded. The end-tidal carbon dioxide concentration, which was set at 45 mmhg before each test, was measured to average 44.5 mmhg. Performance of devices The end-tidal oxygen concentration achieved varied between 24.5% and 46.8% (Table 2, Fig. 3). The highest concentrations were achieved by the T-bag and the T- piece (46.1% and 45.8% at 8 l.min 21, respectively) compared with the filter and Hudson mask (35.4% and 34.8% at 8 l.min 21, respectively). The ranking of devices from greatest to least efficacy was: T-bag, T-piece, Hudson mask, filter. Because of the controlled nature of the test and the high reproducibility of results within each group, in all cases the differences between each device Results Control The minute volume for the control device was 5.91 (0.27) l.min 21 and this was not significantly different in any of the test groups, confirming that all devices were tested under the same conditions. Baseline variables for the laryngeal mask spontaneous ventilation model are summarised in Table 1. The maximum inspired oxygen Figure 3 Mean end-tidal fractional oxygen concentration. Results are divided by device and oxygen flow rate. q 2000 Blackwell Science Ltd 995

5 P. Peyton et al. Supplementary oxygen with the LMA Anaesthesia, 2000, 55, pages 992±999 Table 2 End-tidal oxygen concentration for each device and oxygen flow rate. Results are expressed as mean (SD). Device O 2 flow rate; l.min 21 Filter Hudson Mask T-bag T-piece No. of independent trials End-tidal oxygen concentration % (0.13) 25.8 (0.18) 39.7 (0.19) 38.6 (0.23) (0.16) 28.9 (0.65) 46.8 (0.21) 43.4 (0.29) (0.21) 34.8 (0.34) 46.1 (0.21) 45.8 (0.18) were highly statistically significant (p, ). Higher oxygen flows resulted in a trend to higher end-tidal oxygen concentrations, although there was no clinically significant difference between 4 l.min 21 and 8 l.min 21 with the T-bag device (46.8% and 46.1%, respectively). The oxygen delivery variables data are summarised in Table 3. The highest maximum inspired oxygen concentration of 68.4% was achieved by the use of 8 l.min 21 with the T-piece, which also generated the best mean oxygen concentration throughout inspiration (56.4%). However, because of variations in gas flow in inspiration, the delivered flow of oxygen varied little between the T- bag at 4 l.min 21, and the T-bag and T-piece at 8 l.min 21 (202, 199 and 202 ml.breath 21, respectively). All oxygen delivery parameters were significantly poorer with the filter and the Hudson mask. There was little change in end-tidal carbon dioxide concentration with any device, the highest recorded being 48.9 mmhg with the filter at an oxygen flow rate of 2 l.min 21 (Table 4). In keeping with this, the volume of inspired carbon dioxide was generally low, ranging from 0.73 to 1.91 ml.breath 21. Discussion The purpose of providing supplementary oxygen is not only to achieve adequate oxygenation in the recovery period but also to increase the margin of safety if ventilation is compromised. A number of previous investigations into the safety and efficacy of devices which supply additional oxygen via the laryngeal mask have relied on the measurement of inspired oxygen concentration as their end-point [4±7, 9]. However it is generally not stated whether the `inspired' oxygen concentration reported is the maximum value obtained, or some other value, such as the timeweighted mean. Different anaesthesia monitors display varying parameters of oxygen delivery, but none reliably accounts for the cyclical changes in oxygen concentration that occurs within each inspiration with these devices. Typically, the F i o 2 displayed is taken at a single point, the oxygen fraction at the lowest carbon dioxide value [11]. The use of a pneumotachograph allows more in-depth understanding of the differences between devices. Measurement and integration of instantaneous flow and gas concentrations provides a clearer picture of the variability of the device performance through the respiratory cycle, and allows comparison of the true flow-weighted mean inspired oxygen concentrations and delivered volume of oxygen between devices. Importantly, the time-weighted mean inspired oxygen concentration does not account for the fact that the determinant of oxygen delivery is the flow of oxygen (the product of flow and oxygen concentration) throughout Table 3 Oxygenation delivery variables for each device and oxygen flow rate. Results are expressed as mean (SD). Device O 2 flow rate; Filter Hudson Mask T-bag T-piece l.min 21 No. of independent trials Maximum inspired oxygen concentration; % (0.20) 34.5 (0.36) 52.6 (0.45) 56.8 (0.35) (0.76) 42.4 (0.67) 61.8 (0.19) 62.9 (0.24) (0.35) 50.1 (0.71) 60.6 (0.13) 68.4 (1.1) Mean inspired oxygen concentration; % (time-weighted) (0.11) 31.5 (0.16) 41.3 (0.62) 38.6 (0.23) (0.34) 36.6 (0.49) 50.1 (0.70) 45.7 (0.22) (0.24) 45.5 (0.29) 53.4 (1.3) 56.4 (0.33) Tidal inspired oxygen volume Vo 2, ml (1.3) 115 (1.2) 167 (6.8) 147 (0.91) (4.1) 132 (2.4) 202 (4.1) 172 (5.8) (1.3) 164 (1.4) 199 (2.8) 202 (1.5) `Effective' inspired oxygen concentration; % (F i o 2 ˆ Vo 2 /V T ) q 2000 Blackwell Science Ltd

6 P. Peyton et al. Supplementary oxygen with the LMA Table 4 Carbon dioxide respiration variables for each device and oxygen flow rate. Results are expressed as mean (SD). Device O 2 flow rate; l.min 21 Filter Hudson Mask T-bag T-piece Minute volume; l.min (0.14) 5.70 (0.23) 5.74 (0.40) 5.61 (0.34) (0.29) 5.87 (0.18) 5.64 (0.35) 5.70 (0.44) (0.35) 5.76 (0.34) 5.51 (0.28) 5.67 (0.19) End-tidal CO 2 ; mmhg (0.31) 46.5 (0.91) 46.2 (0.58) 45.8 (0.67) (0.72) 47.5 (0.74) 44.4 (0.61) 42.7 (0.54) (0.57) 46.5 (0.68) 43.4 (0.53) * Inspired CO 2 partial pressure; mmhg (0.00) 1.23 (0.28) 0.00 (0.00) 0.00 (0.00) (0.00) 0.67 (0.24) 0.04 (0.06) 0.00 (0.00) (0.00) 0.05 (0.60) 0.00 (0.00) * Inspired CO 2 volume; ml (0.18) 1.91 (0.08) 1.66 (0.13) 1.20 (0.06) (0.06) 1.43 (0.07) 1.01 (0.08) 0.73 (0.08) (0.08) 0.91 (0.06) 0.90 (0.15) * *These values were not able to be measured as the higher supplementary O 2 flow rate produced retrograde washout of end-tidal gas at the measurement point of the in-line capnograph sensor during the end-expiratory pause with the T-piece. inspiration. Its integral over time is the inspired oxygen volume (Vo 2 ). We have termed Vo 2 as a fraction of tidal volume, the `effective' inspired oxygen concentration. These data demonstrate the differences which occur between the maximum and time-weighted mean inspired oxygen concentrations and the effective (flow-weighted) inspired concentration. The difference between the timeweighted mean and effective inspired oxygen concentrations is greater for the filter than for the other devices, especially at higher oxygen flow rates. As the oxygen flow increases, the time-weighted mean oxygen concentration is unrealistically elevated compared with the effective oxygen concentration with the filter. It is also of note that the delivery of oxygen was equal from the T-bag at 4 l.min 21 or 8 l.min 21 and the T-piece at 8 l.min 21, despite the fact that the mean and maximum inspired oxygen concentrations were higher from the latter. This is reflected in both end-tidal and effective inspired oxygen concentrations. A distinction should be made between the oxygen concentration at the inlet of a device and that which is achieved at the alveolar level [13]. While an increase in one may be expected to parallel an increase in the other, it is the alveolar concentration which truly reflects the ability of the device to provide oxygen in an effective and safe way. The alveolar concentration is best approximated by the end-tidal oxygen concentration. However, measurement at the device orifice is likely to be made inaccurate by mixing of gases from three sources: alveolar gas, dead-space gas and fresh gas. Typical O 2 concentration waveforms are displayed for each device along with their phase-linked inspiratory oxygen flow curves (Fig. 4). In all cases, the O 2 concentration waveform within each inspiration is multiphasic, unlike that seen with fixed performance systems such as a circle or partial rebreathing anaesthetic circuit. For the filter and Hudson mask, the highest sampled O 2 concentrations occurred at the end of inspiration when inspired flow rates were negligible. In contrast, the corresponding values for the T-bag and T- piece were the lowest concentrations sampled. These values are of least relevance to the calculation of O 2 delivery. Interpretation of these atypical waveforms by gas analysers programmed to calculate inspired or end-tidal O 2 concentrations may be unreliable or incorrect. In our study, the performance of the T-bag and the T- piece in producing increased simulated alveolar oxygenation was superior to that of the filter and Hudson mask. An oxygen flow rate of 8 l.min 21 in either of these devices produced end-tidal oxygen concentrations about three times that of the control device, and roughly one and a half times higher than the filter or Hudson mask. This would represent a significant increase in the margin of patient safety if alveolar ventilation was interrupted during recovery from anaesthesia. In a patient with a typical functional residual capacity of 2 l and O 2 consumption of 250 ml.min 21, an increase in alveolar O 2 concentration of 15% represents an increase of more than 1 min of apnoeic oxygenation. The tidal volume is also an important consideration. In this study, we have used a tidal volume of 360 ml with a respiratory rate of 16 breath.min 21 to represent changes that occur in spontaneous ventilation during recovery from anaesthesia, i.e. more rapid, shallow breaths. It now becomes evident why the T-bag performs so well, even at lower oxygen flow rates. With a reservoir volume of 300 ml, the majority of the inspiratory flow comes from the oxygen-enriched reservoir gas. In a patient taking larger tidal volume breaths, the performance of the T-bag could be expected to decline. A decrease in inspired q 2000 Blackwell Science Ltd 997

7 P. Peyton et al. Supplementary oxygen with the LMA Anaesthesia, 2000, 55, pages 992±999 Figure 4 Oxygen concentration and inspired oxygen flow curves. For each diagram the upper curve is oxygen concentration (%) and the lower oxygen flow (l.min 21 ) for two respiratory cycles. oxygen concentration with a high tidal volume has been demonstrated by Poh and Brimacombe [7]. The ability to achieve a high degree of safety with lower oxygen flow rates is an additional benefit of the T- bag device, but its major advantage over the T-piece lies in an observer's ready ability to judge whether the patient is ventilating, from the tidal movements of the rebreathing bag. All devices and flow rates were found to be safe with regard to their ability to clear alveolar carbon dioxide. Only the Hudson mask group had significantly elevated minimum inspired carbon dioxide concentrations (1.2% at 2 l.min 21 ), and the measured end-tidal concentration was only increased to 46.4 mmhg in this group. This does not represent a clinically important change from the control value of 44.5 mmhg. However, the minimum inspired carbon dioxide concentration is not the most reliable measure of the degree of rebreathing. The device with the highest level of CO 2 rebreathing was the filter at 2 l.min 21, where end-tidal CO 2 was increased to 48.9 mmhg. The minimum inspired carbon dioxide concentration was zero however, owing to the rapid washout of the device and significant entrainment of room air late in inspiration, both of which also contribute to its relatively poor performance in oxygen enrichment. The T-bag has a relatively large rebreathing bag attached, but performed well in terms of minimisation of the degree of rebreathing of alveolar gas, since the bag is designed to fill with expired dead space gas enriched with supplementary oxygen, causing the alveolar gas following to be vented to atmosphere. At a flow rate of 4 l.min 21, the end-tidal CO 2 concentration with the T- bag was not greater than the control value. The advantages of using an artificial lung model to test airway devices are evident: the results are highly reproducible within a group and a smaller number of test runs can be undertaken, avoiding the labour, expense and ethical issues associated with using human subjects. Furthermore, devices whose safety had not been established would be able to be tested or conditions explored outside acceptable limits of safety, without placing patients at risk. Summary The achievement of a high end-tidal oxygen concentration in patients breathing through a laryngeal mask airway in the recovery room is important for patient safety. In this study of devices used to provide supplementary oxygen, the T-bag and the T-piece were found to be significantly superior to the Hudson mask and circuit filter in delivery of oxygen to a lung simulator. Important 998 q 2000 Blackwell Science Ltd

8 P. Peyton et al. Supplementary oxygen with the LMA differences were seen between mean inspired oxygen concentration and effective inspired oxygen concentration because of fluctuations in gas flow and oxygen concentration throughout inspiration. References 1 Brain AIJ. The Intavent Laryngeal Mask Instruction Manual, 2nd edn. Henley-on-Thames, UK: Intavent International SA, Nunez J, Hughes J, Wareham K, Asai T. Timing of removal of the laryngeal mask airway. Anaesthesia 1998; 53: 126±30. 3 Kitching AJ, Walpole AR, Blogg CE. Removal of the laryngeal mask airway in children: anaesthetized compared with awake. British Journal of Anaesthesia 1996; 76: 874±6. 4 Martin J, Brimacombe JR. Oxygen enrichment during emergence with the laryngeal mask: the `T-bag' versus the T- piece. Anaesthesia 1997; 52: 1195±8. 5 Broadway PJ, Royle P. Supplementary oxygen and the laryngeal mask airway. Anaesthesia 1990; 45: 792±3. 6 Goodwin APL. Postoperative oxygen via the laryngeal mask airway. Anaesthesia 1991; 46: Poh J, Brimacombe J. A Comparison of the T-piece, Venturi T-piece and T-bag for emergence with the Laryngeal Mask. Anaesthesia and Intensive Care 1998; 26: 526±8. 8 Kennedy R, Meyer M, Joyce C. Supplemental oxygen using an LMA. Anaesthesia and Intensive Care 1992; 20: Lewis RP, Porter M. Supplementary oxygen and the laryngeal mask airway. Anaesthesia 1991; 46: 70±7. 10 Wilkes AR, Vaughan RS. The use of breathing system filters as oxygen delivery devices. Anaesthesia 1999; 54: 552±8. 11 Waldau T, Larsen VH, Bonde J. Evaluation of five oxygen delivery devices in spontaneously breathing subjects by oxygraphy. Anaesthesia 1998; 53: 256± Noguchi H, Ogushi Y, Yoshiya I, Itakura N, Yamabayashi H. Breath-by-breath VCO 2 and VO 2 require compensation for transport delay and dynamic response. Journal of Applied Physiology 1982; 52: 79± Orme RM, L'E, Williams M. Supplementary oxygen and the laryngeal mask airway ± evaluation of a heat-andmoisture exchanger. Anaesthesia and Intensive Care 1999; 27: 509±11. q 2000 Blackwell Science Ltd 999