Tracheal tube pressure change during magnetic stimulation of the phrenic nerves as an indicator of diaphragm strength on the intensive care unit

British Journal of Anaesthesia 87 (6): 876±84 (2001) Tracheal tube pressure change during magnetic stimulation of the phrenic nerves as an indicator of diaphragm strength on the intensive care unit G. H. Mills 1*, J. Ponte 2, C. H. Hamnegard 3, D. Kyroussis 1, M. I. Polkey 4, J. Moxham 4 and M. Green 1 1 Respiratory Muscle Laboratory, Department of Thoracic Medicine, Royal Brompton Hospital, London, UK. 2 Department of Anaesthetics, King's College Hospital, London, UK. 3 Department of Pulmonary Medicine and Clinical Physiology, Sahlgrenska University Hospital, Goteborg, Sweden. 4 Respiratory Muscle Laboratory, Department of Thoracic Medicine, King's College Hospital, London, UK *Corresponding author: Department of Surgical and Anaesthetic Sciences, K oor, Royal Hallamshire Hospital, Shef eld S10 2 JF, UK Diaphragm strength can be assessed from twitch gastric (TwPgas), twitch oesophageal (TwPoes), and twitch transdiaphragmatic pressure (TwPdi) in response to phrenic nerve stimulation. This requires the passage of balloon catheters, which may be dif cult. Changes in pressure measured at the mouth during phrenic nerve stimulation avoid the need for balloon catheters. We hypothesized that pressures measured at the tracheal tube during phrenic stimulation, could also re ect oesophageal pressure change as a result of isolated diaphragmatic contraction and, therefore, re ect diaphragm strength. We aimed to establish the relationship between twitch tracheal tube pressure (TwPet), TwPoes, and TwPdi in patients in the supine and sitting positions. The phrenic nerves were stimulated magnetically bilaterally, in 14 ICU patients while supine and on another occasion while sitting up at 45. In the sitting position mean TwPoes was 9.1 cm H 2 O and TwPet 11.3 cm H 2 O (mean(sd) difference ±2.2 (SD 1.5)). In the supine position mean TwPoes was 8.1 cm H 2 O and TwPet 9.9 cm H 2 O (mean difference ±1.8 (2.2)). The difference between TwPoes and TwPet was less at low twitch amplitude; less than 61 cmh 2 O below a mean twitch height of 8 cm H 2 O supine and 10 cm H 2 O sitting. Sitting TwPet was related to TwPoes r 2 =0.93 and TwPdi r 2 =0.65 (P<0.01). Supine TwPet was related to TwPoes r 2 =0.84 and TwPdi r 2 =0.83 (P<0.01). The mean within occasion coef cient of variation while sitting was TwPet=13.3%, TwPoes=13.9%, TwPdi=11.2%, and supine TwPet=11.6%, TwPoes=14.6%, TwPdi=11.8%. We conclude that TwPet re ects TwPoes during diaphragmatic stimulation and is worthy of further study to establish its place as a guide to the presence of respiratory muscle strength and fatigue. Br J Anaesth 2001; 87: 876±84 Keywords: muscle skeletal, diaphragm; equipment, tubes tracheal; intensive care Accepted for publication: July 31, 2001 The assessment of diaphragm strength on the intensive care unit (ICU) is dif cult whether the pressures are recorded during a maximal voluntary effort or a single twitch stimulation of the phrenic nerves. Volitional tests rely on the cooperation of the patient, so measurements may not be possible on the ICU or may underestimate respiratory muscle strength because of the effects of sedation even when using the semivolitional modi cation by Marini, 1 of the maximal inspiratory mouth pressure technique. 2 Conventionally, balloon catheters are passed into the oesophagus and stomach to measure pressure above and below the diaphragm. The difference between these two pressures is transdiaphragmatic pressure, which is an index of diaphragmatic strength during phrenic nerve stimulation. 34 However, the requirement for balloon catheters makes these methods invasive. The oesophageal and gastric balloon are dif cult to position and the measurement is subject to artefact, especially when supine and in the presence of atelectasis. A means of gauging mean pleural or Ó The Board of Management and Trustees of the British Journal of Anaesthesia 2001

Twitch tracheal tube pressure Fig 1 Twitch pressure traces sitting upright, lying at 45 and supine, showing the change in relative size of TwPoes and TwPgas in relation to posture. oesophageal pressure change in response to diaphragmatic contraction, by measuring airway or tracheal tube pressure, would facilitate the assessment of diaphragm strength on the ICU. Pressures measured at the mouth during phrenic nerve stimulation (twitch mouth pressure) have already been used in ambulant patients with electrical and cervical magnetic stimulation, 5 6 although glottic closure can be a problem when mouth pressures are compared with twitch oesophageal pressure (TwPoes). This would not be a problem in the intubated patient, when twitch tracheal tube pressure (TwPet) is measured. Non-volitional tests of diaphragm strength conventionally require electrical stimulation of the phrenic nerves. However, the adequacy and consistency of stimulation may be a problem because of the need to position the stimulating electrodes precisely over the nerves. 7 The technique of bilateral anterolateral magnetic stimulation (BAMPS) over the surface markings of the phrenic nerves in the neck, allows supramaximal bilateral phrenic stimulation in supine patients, without the need for neck exion, which is required with the cervical approach. 8 This produces a pattern of stimulation very similar to bilateral electrical stimulation (BES). 9 10 Therefore, if the BAMPS technique could be combined with TwPet measurements, this would provide a nonvolitional and non-invasive assessment of diaphragm strength on the ICU. These recordings would allow us to assess more readily diaphragmatic strength in the ventilated patient and so help the complex assessment of the balance between the capacity of the respiratory system and the load placed upon it, particularly during weaning from respiratory support. However, even though posture has little effect on TwPdi it does affect the ratio of the decrease in oesophageal pressure to the increase in gastric pressure that make up the change in transdiaphragmatic pressure. 9 11 For example, when the diaphragm is stimulated to contract in the supine position there is a larger decrease in twitch oesophageal pressure (TwPoes) than when sitting, but a relatively smaller contribution from the increase in gastric pressure (TwPgas), so overall twitch transdiaphragmatic pressure (TwPdi) is little affected by posture (Fig. 1). 11 Therefore, it was necessary to assess our new technique in both the supine position, when the oesophageal pressure contribution to changes in TwPdi would be high, and in the sitting poition where the gastric component, which would not be measured by twitch tracheal tube pressure (TwPet), would be greater, to determine if a useful relationship between TwPet and TwPes or TwPet and TwPdi could be established in the two postures. The TwPoes/TwPgas ratio depends on posture, on lung volume, and also varies from individual to individual. However, providing posture and volume are constant TwPoes/TwPgas or TwPoes/TwPdi are constant within individuals. 12 Therefore, within an individual, changes in TwPet re ect TwPoes. A certain percentage change in TwPet would re ect the same change in TwPdi, even without knowing TwPgas. The technique may be useful in longitudinal studies and screening for major weakness, even if absolute values for TwPdi are not possible. This could prove very useful as repeated insertion of balloon catheters often proves too invasive and dif cult. We therefore aimed to measure twitch TwPet proximal to an occluded tracheal tube, during BAMPS in the supine and sitting position, to determine whether this reliably re ected oesophageal pressure change during diaphragmatic contraction and could, therefore, act as an indicator of TwPdi in intubated patients. 13 Methods Study population Fourteen consecutive patients (four female and 10 male) mean age 58 yr (range 45±71), who had undergone coronary artery bypass surgery were studied during the rst 12 h after admission to the ICU. None had any neurological disease or diabetes. Patients with pacemakers or temporary pacing wires were excluded; however, patients with central lines were not. None had pre-existing symptoms or signs of lung disease, although formal lung function testing was not undertaken. The study was approved by the local ethics committee and all patients gave informed written consent. 877

Mills et al. Measurements Pressure measurements were made from oesophageal and gastric balloon catheters 14 (PK Morgan, Rainham, Kent, UK), passed through the nose shortly after induction of anaesthesia via a long uncuffed red rubber nasotracheal tube, which had been passed into the oesophagus and then removed once the balloons were in place. Appropriate positioning of the balloons was initially con rmed using the technique of Ducros. 15 Later, once spontaneous breathing had been resumed, the position of the balloons was once more con rmed by occluding the tracheal tube and asking the patient to make a gentle inspiratory effort; 16 if the balloons were correctly positioned the change in tracheal tube pressure equalled the change in Poes. Pressure measurements were also made from the proximal end of the tracheal tube via a catheter of similar material, length, and diameter as the balloon catheter tubing. Phrenic nerve stimulation The phrenic nerves were stimulated with two 43-mm mean diameter double coils placed anterolaterally on the neck over the phrenic nerves. Each double coil was powered by a Magstim 200HP stimulator (The Magstim Company, Whitland, Dyfed, Wales, UK). The stimulators were charged to 100% of maximum power and triggered simultaneously. This level of power output is known to produce stimulation which is supramaximal or very close to being so. 9 Changes in lung volume alter the position of the inspiratory muscles and so in uence apparent diaphragm strength. For example, at high lung volumes the diaphragm is attened, so reducing its ability to generate negative intrathoracic pressures during a maximal contraction. On the other hand, reductions in lung volume increase the curvature of the diaphragm and increase its ability to generate inspiratory pressures, 17 18 but are associated with atelectasis and worsened compliance. The effect of potentiation may confuse diaphragm strength measurements. A fully potentiated twitch is 50% larger than a twitch in the same subject when fully 19 20 rested. This makes exact comparisons of strength more dif cult to interpret unless patients can be rested for 20 min, avoiding all vigorous respiratory efforts before testing. On the ICU this could be achieved by providing adequate respiratory support. Study design To minimize potentiation, 19 20 all subjects were observed for 20 min to ensure that no vigorous respiratory efforts were made. Potentiation is the phenomenon whereby twitch amplitude depends on the recent contractile history of the muscle. Providing a muscle has not worked so hard that it is fatigued, a recent maximal voluntary contraction will temporarily alter the behaviour of the muscle so that it produces a twitch up to 50% greater than normal amplitude in response to a single twitch stimulation. This effect lasts for 20 min. Manipulations involving the tracheal tube were made as gently as possible to avoid stimulating the airway, which might lead to contractions that could potentiate the diaphragm. Then the tracheal tube was momentarily occluded and the phrenic nerves stimulated. Stimulations were made at end-expiration and before the next inspiration during mechanical ventilation. Those subjects who were spontaneously breathing on the tracheal tube via a T piece, were asked to breath in time with instructions and then relax at the end-expiratory position. The tracheal tube was then occluded and the phrenic nerves stimulated. On each occasion ve stimulations were made. Therefore, two sets of measurements were made on separate occasions in each patient; in a supine position and with the patient sitting. There is evidence to suggest that diaphragm strength can be expected to uctuate in the postoperative period after coronary artery surgery. 21 Therefore, no attempt was made to directly compare measurements made in the supine and sitting up positions. 21 Instead the aim was to determine if TwPet could be a useful indicator of TwPoes and hence TwPdi in either position. In addition, changes in lung volume have been shown to affect TwPoes more than TwPgas and so the relationship between TwPet, TwPoes, and TwPdi could also change over the same time period. 18 However, it was possible to attempt to determine if a linear relationship existed between TwPet, TwPoes, and TwPdi in each of the two different positions. The pressure traces were rejected if the baseline oesophageal or gastric pressure was distorted or if the twitch was imposed on a spontaneous or mechanical breath. The stimulations were 1 min apart. A small ramp of stimulations was made before the recordings to con rm that the stimulations were supramaximal or close to supramaximal. Signal processing Pressures were recorded via identical Validyne MP45 transducers (response range 6150 cm H 2 O) and ampli ed by Validyne carrier ampli ers (Validyne Corporation, Northridge, CA, USA). These were calibrated before each study with a Universal Pressure Meter (Bio-Tek Instruments Inc., USA), which was regularly checked for accuracy with a water manometer. 22 The signals were passed to a 12-bit NB-MIO-16 analogue to digital converter board (National Instruments (NI), Austin, TX, USA), which had been installed within a MacIntosh Centris 650 computer (Apple Computer Inc., Cupertino, CA, USA). Pressure traces were analysed using programs based on LabView 2.2 software (NI). 878

Twitch tracheal tube pressure Data collection Pressure traces were inspected on line during the stimulation procedures to ensure they began on a steady baseline pressure at end-expiration. Twenty-two pressure traces were deemed unsatisfactory on subsequent off-line analysis and so excluded from the study (11 in the sitting position and 11 supine). Therefore, a total of 118 pressure traces out of a possible 140, all demonstrating Pet, Poes, Pgas, and Pdi in a total of 14 patients, were included in the analysis. Fig 2 Pressure responses to BAMPS in a supine patient on the ICU. Irregularities in the oesphageal pressure wave are a result of cardiac artefacts. Statistical analysis The statistical analysis was performed using Statview 4 (Abacus Concepts Inc., Berkley, CA, USA), and included linear regression, Wilcoxon Sign Rank tests and Bland and Altman plots. 23 To allow a comparison of the ratios between TwPet, TwPoes, and TwPdi linear regression was used. The regression equations were also calculated with `no intercepts', that is, made to pass through the zero point on both axes to allow the ratio between the variable on the ordinate and the abscissa to be calculated. Wilcoxon Sign Rank tests were used to compare TwPoes and TwPet. Fig 3 (A) Mean BAMPs TwPoes compared with mean BAMPS TwPet in the supine position and at 45 (cm H 2 O). Line of identity is included. (B) Mean BAMPS TwPoes compared with mean BAMPS TwPdi in the supine position at 45. (C) Mean BAMPS TwPdi compared with mean BAMPS TwPet in the supine position and at 45. Solid regression line indicates TwPet/TwPdi relationship when sitting. Broken regression line indicates TwPet/TwPdi relationship when supine. TwPet generally exceeds TwPoes because TwPoes is adversely affected by local atelectasis and postural effects impeding trasmission of mean pleural pressure. It is likely that the measure TwPoes is an underestimate of mean pleural pressure. Therefore, estimating TwPdi (which has been calculated from the difference between TwPgas and TwPoes) from TwPet values using the regression lines may well tend to produce a slight underestimate of the true TwPdi. 879

Mills et al. Fig 4 Bland and Altman plots comparing the difference between mean BAMPS TwPoes and mean BAMPS Tw Pet against the average of the two techniques in both positions. The differences between the two techniques vary depending on the size of the twitch pressure change. At high twitch pressures, especially when supine and particularly above 9 cm H 2 O, TwPet exceeds TwPoes (2 SD from the mean=2 SD). Table 1 Pressures are measured in cm of H 2 O TwPoes supine TwPet supine TwPoes/TwPet supine TwPoes sitting TwPet sitting TwPoes/TwPet sitting 6.9 10.1 0.68 9.9 13.3 0.74 4.2 4.6 0.92 3.5 4.2 0.83 1.9 2.1 0.92 8.2 11.4 0.73 8.7 11.6 0.74 8.5 11.5 0.74 3.7 4.1 0.92 11.3 13.3 0.86 6.9 8.6 0.81 16.7 19.5 0.85 12.1 19.1 0.63 9.4 10.4 0.93 12.7 15.5 0.82 9.2 12.7 0.73 10.5 10.5 1.00 7.1 8.1 0.87 12.4 12.5 0.99 10.3 12 0.88 9.1 12.9 0.70 11.8 14.4 0.82 9.6 9.3 1.04 3.1 3.8 0.81 4.9 4.0 1.26 6.5 6.3 1.04 10.1 13.7 0.74 12.4 17.9 0.69 Mean 8.1 9.9 0.87 9.1 11.3 0.82 SD 3.5 4.9 0.17 3.5 4.6 0.09 In order to assess the agreement between TwPet and TwPoes we calculated the differences between the means of each of the ve stimulations. The mean of these differences is a measure of accuracy or bias whilst the standard deviation (SD) is a measure of precision as presented in Figure 3. Both bias and precision are necessary to assess agreement. 24 We calculated the limits between which 95% of the differences would lie (`limits of agreement', 2 SD). 23 We also calculated the mean and standard deviation of the ratio of mean TwPet: mean TwPoes to enable us to compare the values of the two methods over the range of twitch pressures observed (Table 1). Bland and Altman plots were constructed to compare the differences between TwPoes and TwPet, against the mean of TwPoes and TwPet (Fig. 3). The ratio of TwPoes to TwPet was calculated to give a measure of the degree of agreement between the two techniques. To assess the spread of data within each set of measurements, the mean coef cient of variation was calculated within each occasion of ve twitches in each subject. An average for each occasion for the 14 subjects was then derived. To assess the spread of data between individuals, the coef cient of variation of TwPoes/TwPet and TwPet/TwPdi were calculated for each position to see if results became more variable in a given posture. TwPgas cannot be measured from mouth or tracheal tube pressures, but if it is a constant fraction of TwPdi in a particular posture, then this would imply that information relating to TwPoes alone may be enough to estimate TwPdi. However, if TwPgas is not a constant fraction of TwPdi, or if this is more of a problem in a particular posture, then estimates of TwPdi will be less reliable. To assess this we calculated the coef cient of variation of mean TwPoes/TwPdi between and within subjects for each posture. 880

Twitch tracheal tube pressure Table 2 Mean SD Coef cient of variation Suppine Tw/Poes/TwPet 0.87 0.17 19.7 TwPoes/Twpdi 0.84 0.11 12.8 TwPet/TwPdi 1.00 0.22 22.1 Sitting TwPoes/TwPet 0.82 0.09 11.4 TwPoes/TwPdi 0.69 0.13 18.9 TwPet/TwPdi 0.86 0.20 23.3 Table 3 Posture Equation r 2 Equation with no intercept r 2 Supine TwPoes=0.873TwPdi±0.3 0.93 TwPoes=0.853TwPdi 0.99 TwPoes=0.653TwPet+1.7 0.84 TwPoes=0.793TwPet 0.97 TwPet=1.173TwPdi±1.3 0.83 TwPet=1.053TwPdi 0.97 TwPdi=0.713TwPet+2.6 0.83 TwPdi=0.923TwPet 0.97 45 TwPoes=0.583TwPdi+1.4 0.66 TwPoes=0.673TwPdi 0.96 TwPoes=0.743TwPet+0.7 0.93 TwPoes=0.803TwPet 0.99 TwPet=0.743TwPdi+1.4 0.65 TwPet=0.843TwPdi 0.95 TwPdi=0.873TwPet+3.5 0.65 TwPdi=1.133TwPet 0.95 Results An example of the pressure traces recorded during magnetic stimulation of the phrenic nerves using BAMPS is shown in Figure 2. Agreement between TwPoes AND TwPet TwPoes plotted against TwPet for the patients in the supine and sitting up positions are shown in Figure 3. The mean TwPoes was 8.1 cm H 2 O and TwPet 9.9 cm H 2 O when supine. The mean TwPoes was 9.1 cm H 2 O and TwPet 11.3 cm H 2 O sitting (Table 1). The mean difference between TwPoes and TwPet measured in the supine position was ±1.8 cm H 2 O (SD 2.2) (Fig. 4). As TwPoes and TwPet increased, the difference between the two methods also increased. The mean difference between TwPoes and TwPet measured in the sitting position was ±2.2 cm H 2 O (SD 1.5) (Fig. 4). As TwPoes and TwPet increased, the difference between the two pressures also increased. In order to standardize for the wide range of measured pressures and to better explore these changes, we expressed them as a ratio of TwPoes/TwPet (Table 1). The 95% con dence intervals for the likely differences (TwPoes ± TwPet) between the two techniques over a range of TwPoes from 2 up to 17 cm H 2 O in the sitting and supine positions was ±5.3 to +1.8 cm H 2 O sitting and ±6.2 to +2.6 cm H 2 O supine. However, the difference between TwPet and TwPoes was less than 61 cm H 2 O at twitch pressures less than 8 cm H 2 O when supine and 10 cm H 2 O when sitting, and then became greater at higher pressures (Fig. 4). TwPet was linearly related to TwPoes in both postures. In the sitting position TwPet was related to TwPoes r 2 =0.93 and TwPdi r 2 =0.65 (P<0.01). In the supine position TwPet was related to TwPoes r 2 =0.84 and TwPdi r 2 =0.83 (P<0.01). TwPoes was less than TwPet. Variability The within occasion coef cient of variation for supine measurements within individual patients was: TwPet 11.6%, TwPoes 14.6%, and TwPdi 11.8%. When sitting the results were similar: TwPet 13.5%, TwPes 13.9%, and TwPdi 11.2%. The variability of between subject comparisons of TwPoes/TwPet, TwPoes/TwPdi, and TwPet/TwPdi are listed in Table 2 for both the supine and sitting postures. The ratio TwPoes/TwPet was slightly closer to 1.0 in the supine position although the between individual variability of this relationship was greater. On the other hand TwPoes made up a greater fraction of TwPdi supine and TwPoes/ TwPdi was more consistent. When TwPet was related to TwPdi variability was relatively high and similar in both positions. The relationship of TwPet, TwPoes, and TwPdi from linear regression TwPet, TwPoes, and TwPdi sitting There were signi cant linear relationships between TwPet, TwPoes, and TwPdi in both the sitting and supine positions (P<0.01) (Fig. 3 and Table 3). Mean TwPet in the sitting position was on average 22% greater than TwPoes (P<0.001, Wilcoxan Sign Rank test). The regression equations without intercept describing the relationship between sitting TwPet and sitting up TwPoes 881

Mills et al. produced an almost identical result: TwPet was typically 25% larger than TwPoes (Table 3). Sitting TwPoes made up 67% of TwPdi; and TwPet was equal to 84% of TwPdi. TwPet, TwPoes and TwPdi supine In the supine position mean values for TwPet were 26% greater than TwPoes (P<0.001, Wilcoxon Sign Rank test). When assessed using linear regression with no intercept, TwPoes made up 85% of TwPdi when supine and TwPet was equal to 105% of TwPdi when supine (Table 3). Therefore, the value for TwPet was a larger fraction of the TwPdi measurement in the supine position compared with the sitting position. Discussion We found that TwPet was linearly related to TwPoes in both positions (Figure 3 and Table 3), although the 95% con dence intervals for the likely differences (TwPoes ± TwPet) between the two techniques were ±5.3 to +1.8 cm H 2 O while sitting and ±6.2 to +2.6 cm H 2 O when supine (Fig. 4). However, the difference between TwPet and TwPoes sitting was less than 61 cm H 2 O at twitch pressures less than 8 cm H 2 O when supine and 10 cm H 2 O when sitting and greater at higher pressures. The between subjects coef cient of variation for TwPes/TwPdi with BAMPS has previously been found to be between 12% 12 and 15%, 9 which is better than the traditional BES. We therefore anticipated the coef cient of variation for TwPet/TwPdi would be similar. However, the between subjects coef cient of variation of TwPet/TwPdi was above 20% in both positions (Table 2) and so, the precise quanti cation of diaphragm strength using a regression equation derived from the results of several individuals is not possible with this level of agreement. However, the technique does retain the potential to identify weak patients, who can then be investigated further. In addition the coef cient of variation of TwPoes/TwPdi within an individual over time is 8.7%. 12 This means that it may be possible to follow changes in strength in an individual with repeated non-invasive studies over a period of time. Factors affecting the transmission of pressure to the oesophagus and airway TwPet was generally higher than TwPoes. Similar results have been seen in previous studies, when twitch mouth pressures were compared with TwPoes. When positioning the balloon catheters in the ICU after cardiac surgery, we could not achieve perfect 1:1 matching of Poes/Pet during a gentle inspiratory manoeuvre against a closed mouthpiece (mean 0.91 (SD 0.09)). Baydur studied normal subjects in laboratory conditions and could not always achieve a 1:1 relationship, particularly in the supine position, where he achieved an average of 0.96, ranging between 0.86 and 1.1. In practice, our mean of 0.91 could explain why our estimate of Poes compared with airway pressure would differ by 9%. This factor could help explain why TwPet is often slightly larger than TwPoes and probably contributes to the greater coef cient of variation of TwPoes/TwPet when supine (19.7%), compared with 11.4% in the sitting position. Indirect evidence from 133 Xe ventilation studies suggested that in supine subjects there might be a gradient of pressure change in the pleura extending from the apex to the base compared to the airway. This may account for some of the difference between Pet and Poes in the supine position. 25 26 Hamnegard found some instances of twitch mouth pressure (TwPmo) exceeding TwPoes in normal subjects. 10 He postulated that the change in TwPmo may more accurately re ect average change in pleural pressure throughout the whole thorax during phrenic stimulation. The same could apply to TwPet measurements. The difference between TwPet and TwPoes does become more prominent as twitch amplitude increases. This may be a problem of transmission of pressure to the oesophageal balloon, which is more dif cult in the supine position. 16 This is also a result of the weight of the heart and mediastinum which compress structures especially the lung, so disturbing the transmission of pressure to the oesophagus. This is even more likely after cardiac surgery when posterobasal atelectasis is extremely common. This makes it dif cult to nd a position where Poes re ects mean pleural pressure. Therefore, it may be that Poes measured with balloon catheters in these subjects when supine tend to underestimate mean pleural pressure change and TwPet may in fact be a closer estimate. On the other hand, in patients with abnormal lungs with areas of slow time constants and with intrinsic positive end expiratory pressure (PEEP) it might be expected that TwPet would be less than TwPoes, because of failure of pressure transmission, rather than the opposite. Cardiac contractions cause 1±2 cm H 2 O reductions in the magnitude of Poes in our experience. This feature is more prominent when supine (mean 3.5 (SD 1.8) cm H 2 O according to Baydur) 16 and can be seen distorting the TwPoes in Figure 1. This will contribute to the relative reduction in TwPoes compared with TwPet on those occasions when the contraction coincides with the stimulation. Clearly glottic closure is not a problem in intubated patients, although activation of some of the laryngeal musculature could occur during magnetic stimulation. This might lead to conformational changes in the airway or alter the position of the tracheal tube or its cuff so affecting TwPet. The importance of these factors remains to be quanti ed. Some of our TwPdi data in this paper suggest the presence of varying levels of diaphragm weakness (mean TwPdi 11 (SD 4.7) cm H 2 O). Our recordings are made after cardiac surgery, which can be associated with phrenic nerve injury or diaphragm weakness. Previous estimates of the incidence of diaphragm dysfunction made slightly later in 882

Twitch tracheal tube pressure the postoperative period vary widely. 27±29 Some of our previous data, 21 looks at the rst 24 h after cardiac surgery, and does appear to show a higher incidence of diaphragmatic weakness (approximately 50% at 12 h), when compared with Polkey's normal data (normal mean TwPdi 26.8 (SD 5.5) cm H 2 O). 30 31 This gradually improved over time. The resulting spread of strength levels is helpful, as it allows us to examine the technique's performance over a range of pressures. The effect of recent wiring of the sternum on lung mechanics is also dif cult to predict. Whether the process of opening and `spreading' the rib cage affects the intercostal nerves is unknown. The impact on the mechanics of the chest wall could alter chest wall compliance. This in turn would alter the impact that contraction of the diaphragm would have on the ratio of TwPoes/TwPgas, and so alter partitioning. However, this is not evident when our partitioning data is compared with previous work. 9 11 Pain might affect the pressures produced during voluntary efforts, but this is less likely during non-volitional stimulations such as used in this study. The effect of the contribution of TwPoes and TwPgas to TwPdi TwPet may be a re ection of the TwPoes component of TwPdi, which forms a larger fraction of TwPdi as we move towards the supine position. Conversely there is a large TwPgas contribution to TwPdi, when sitting, which is not re ected in the TwPoes or TwPet response to diaphragm stimulation. This may explain the lower r 2 value for the association of sitting TwPet with TwPdi and sitting TwPoes with TwPdi. To compound this problem, the fraction of TwPdi made up by TwPoes compared with TwPgas differs greatly between individuals, as seen from the coef cient of variation between individuals of TwPoes/TwPgas: 12.8% supine and 18.9% sitting. Therefore, TwPet may re ect TwPoes, but as TwPoes forms a different fraction of TwPdi in each individual, especially when sitting, and does not re ect the TwPgas component of TwPdi, TwPet while sitting has some limitations in its ability to indicate TwPdi. We have been able to generate linear equations that allow the prediction of TwPdi from TwPet, but the accuracy of prediction will always be hindered by the lack of a TwPgas measurement and the lack of knowledge of partitioning in that individual. However, as the coef cient of variation of partitioning between occasions within an individual from longitudinal studies is less than 10%, serial measurements of TwPet may still be an effective means of recording changes in diaphragm strength over time for that individual. In summary, TwPet combined with BAMPS does re ect TwPoes in response to diaphragmatic stimulation, particularly at low strength levels, allowing an estimate of TwPdi from the TwPet measurement, especially when making supine recordings. This may be a useful means of detecting patients with reduced diaphragm contractility who could then be tested further, by more invasive and potentially more precise techniques. References 1 Marini JJ, Smith TC, Lamb V. Estimation of inspiratory muscle strength in mechanically ventilated patients: the measurement of maximal inspiratory pressure. J Crit Care 1986; 1: 32±8 2 Roman DE, Mills GH. Magnetic stimulation of the phrenic nerves and maximal inspiratory pressure in the assessment of diaphragm strength in the ICU. Eur Respir J 1999; 14 (Suppl. 2): 2359 3 Aubier M, Farkas G, De Troyer A, Mozes R, Roussos C. Detection of diaphragmatic fatigue in man by phrenic stimulation. J Appl Physiol 1981; 50: 538±44 4 Aubier M, Murciano D, Lecocguic Y, Viires N, Pariente R. Bilateral phrenic stimulation: a simple technique to assess diaphragmatic fatigue in humans. J Appl Physiol 1985; 58: 58±64 5 Yan S, Gauthier AP, Similowski T, Macklem PT, Bellemare F. Evaluation of human diaphragm contractility using mouth pressure twitches. Am Rev Respir Dis 1992; 145: 1064±9 6 Hamnegard CH, Wragg S, Kyroussis D, et al. Mouth pressure in response to magnetic stimulation of the phrenic nerves. Thorax 1995; 50: 620±4 7 Mills GH, Kyroussis D, Hamnegard CH, Wragg S, Moxham J, Green M. Unilateral magnetic stimulation of the phrenic nerve. Thorax 1995; 50: 1162±72 8 Similowski T, Fleury B, Launois S, Cathala HP, Bouche P, Derenne JP. Cervical magnetic stimulation: a new painless method for bilateral phrenic nerve stimulation in conscious humans. J Appl Physiol 1989; 67: 1311±8 9 Mills GH, Kyroussis D, Hamnegard CH, Polkey M, Green M, Moxham J. Bilateral magnetic stimulation of the phrenic nerves from an anterolateral approach. Am J Respir Crit Care Med 1996; 154: 1099±105 10 Mills GH, Khan ZP, Hamnegard CH, et al. Effect of temperature on the phrenic nerve and diapleuraquatics during cardiac surgery. Br J Anaesth 1997; 79: 726±32 11 Koulouris N, Mulvey DA, Laroche CM, Goldstone J, Moxham J, Green M. The effect of posture and abdominal binding on respiratory pressures. Eur Respir J 1989; 2: 961±5 12 Mills GH, Turnbull D, Hamnegard CH. The ratio of twitch oesophageal pressure to twitch transdiaphragmatic pressure; how does it vary with time. Eur Respir J 2000; 16 (Suppl. 31): 35 13 Mills GH, Kyroussis D, Hamnegard CH, et al. Twitch endotracheal tube pressure during magnetic stimulation of the phrenic nerves in intubated patients. Am J Respir Crit Care Med 1996; 153: A785 14 Milic-Emili J, Mead J, Turner JM, Glauser EM. Improved technique for estimating pleural pressures from esophageal balloons. J Appl Physiol 1964; 19: 207±11 15 Ducros L, Similowski T, Derenne J-P. Validity of oesophageal pressure measurement for respiratory mechanics studies during ventilation. Eur Respir J 1995; 8: 39S 16 Baydur A, Pangiotis K, Behrakis K, Zin WA, Jaeger M, Emili JAM. A simple method of assessing the validity of the esophageal balloon technique. Am Rev Respir Dis 1982; 126: 788±91 17 Smith J, Bellemare F. Effect of lung volume on in vivo contraction characteristics of human diaphragm. J Appl Physiol 1987; 62: 1893±900 18 Hamnegard CH, Wragg S, Mills GH, et al. The effect of lung volume on transdiaphragmatic pressure. Eur Respir J 1995; 8: 1532±6 19 Wragg SD, Hamnegard CH, Kyroussis D, Road J, Green M, 883

Mills et al. Moxham J. Potentiation of diaphragmatic twitch after its voluntary activation. Thorax 1994; 49: 1234±7 20 Mador JM, Magalang UJ, Kufel TJ. Twitch potentiation following voluntary diaphragmatic contraction. Am J Respir Crit Care Med 1994; 149: 739±43 21 Mills GH, Khan Z, Ponte J, et al. Diaphragm contractility and lung volume after coronary bypass surgery. Am J Respir Crit Care Med 1996; 153: A788 22 Sixt R, Bake B. A simple pressure calibrator. Scand J Clin Lab Invest 1976; 36: 1±2 23 Bland JM, Altman DG. Statistical methods for judging agreement between 2 methods of clinical measurement. Lancet 1986; i: 307±10 24 Altman DG. Analysing data. Statistics in Practise. London: British Medical Association, 1985; 12±14. 25 Sybrecht G, Landau L, Murphy B, Engel L, Martin R, Macklem P. In uence of posture on ow dependence of distribution of inhaled 133Xe boli. J Appl Physiol 1976; 41: 489±96 26 Kaneko K, Milic-Emili J, Dolovich M, Dawson A, Bates D. Ventilation and perfusion as a function of body position. J Appl Physiol 1966; 21: 767±77 27 Markand OM, Moorthy SS, Mahomed Y, King RB, Brown JW. Postoperative phrenic nerve palsy in patients with open-heart surgery. Ann Thoracic Surg 1995; 39: 68±73 28 Estenne M, Yernault J-C, De Smet J-M, De Troyer A. Phrenic and diaphragm function after coronary artery bypass grafting. Thorax 1985; 40: 293±9 29 DeVita MA, Robinson LR, Rehder J, Hattler B, Cohen C. Incidence and natural history of phrenic neuropathy occurring during open heart surgery. Chest 1993; 103: 850±6 30 Polkey MI, Harris ML, Hamnegard CH, et al. Diaphragm strength in elderly and young humans. Am J Respir Crit Care Med 1997; 155: A514 31 Polkey MI, Harris ML, Hughes PD, et al. The contractile properties of the elderly human diaphragm. Am J Respir Crit Care Med 1997; 155: 1560±4 884