Effects of temperature on phrenic nerve and diaphragmatic function during cardiac surgery

Transcription

1 British Journal of Anaesthesia 1997; 79: Effects of temperature on phrenic nerve and diaphragmatic function during cardiac surgery G. H. MILLS, Z. P. KHAN, J. MOXHAM, J. DESAI, A. FORSYTH AND J. PONTE Summary We have studied the effects of whole body cooling on phrenic nerve and diaphragmatic function in 26 patients using magnetic stimulation of the phrenic nerves with a pair of Magstim 200 HP stimulator coils during cardiopulmonary bypass. The diaphragmatic electromyogram in response to magnetic pulses was recorded with needle electrodes at two temperatures, approximately 31 C (cold) and approximately 36 C (warm) during the cooling or rewarming phase of hypothermic cardiopulmonary bypass. This 5- C temperature change was associated with clear changes in the evoked electromyographical response of the diaphragm. Median latency between stimulus and electromyographic response was 10.1 (range ) ms during cold and 8.3 ( ) ms during warm stimulation (P 0.001). Median duration of the muscle compound action potential was prolonged and its amplitude reduced in cold compared with warm stimulations (P 0.01). These effects were enhanced by application of ice slush to the heart. We conclude that diaphragmatic function may be affected by mild hypothermia after cardiac surgery. (Br. J. Anaesth. 1997; 79: ). Key words Temperature, effect. Heart, cardiopulmonary bypass. Surgery, cardiovascular. Muscle skeletal, diaphragm. Diaphragmatic dysfunction and phrenic nerve injury are well recognized complications of cardiac surgery that often require prolonged ventilatory support. 1 One possible cause of diaphragmatic dysfunction is exposure of the phrenic nerves to ice slush during cardiopulmonary bypass. Moderate cooling of the phrenic nerves or the diaphragm in animals 2 4 has been shown to slow the velocity of nerve conduction and to prolong the muscle action potential, effects also observed in other skeletal muscle groups. 5 8 Exposure of the phrenic nerve to ice slush has also been studied in animals and shown to cause phrenic nerve conduction block proportional to exposure time, but similar human studies are lacking. The technique of magnetic stimulation of the diaphragm is well suited to study the effects of local cooling on phrenic nerve and diaphragm function during cardiopulmonary bypass, causing minimal interference with the course of surgery. In the absence of gross cardiac or diaphragmatic dysfunction, many centres are aiming to reduce the time of postoperative ventilation in cardiac surgery, usually extubating the trachea of patients within 4 h of surgery. It is believed generally, however, that when patients are hypothermic ( 36 C) on arrival in recovery, mechanical ventilation should be maintained until a core temperature of 37 C is reached, thus imposing a considerable limitation to the fast tracking of these patients. There is, however, no direct evidence that the function of the human diaphragm is impaired by mild hypothermia. In this study our aim was to document the effects of moderate whole body hypothermia (31 32 C) and the use of ice slush around the heart on diaphragmatic function. Patients and methods The study was approved by the King s Healthcare Ethics Committee and informed consent was obtained from 26 consecutive patients (seven females) referred for coronary artery surgery. Patients with neurological conditions or diabetes were excluded. Morphometric data and cardiopulmonary bypass times are shown in table 1. ANAESTHETIC MANAGEMENT Patients were premedicated with temazepam 30 mg, ranitidine 150 mg, papaveretum 11.5 mg and hyoscine 0.3 mg, and anaesthesia was induced with a bolus dose of propofol mg and fentanyl 0.5 mg. Anaesthesia was maintained with an infusion of propofol at a rate of 5 10 mg kg 1 h 1 throughout, and another bolus of fentanyl 0.5 mg before sternotomy. Intubation of the trachea was facilitated by the use of a single dose of either vecuronium 0.08 mg kg 1 (20 patients) or suxamethonium 1 mg kg 1 (six patients) and no GARY H. MILLS, FRCA, JOHN MOXHAM, MD, MRCP (Department of Thoracic Medicine); ZAHID KHAN, FRCA, JOSÉ PONTE, PHD, FRCA (Department of Anaesthetics); JATIN DESAI, FRCS, ANDREW FORSYTH, FRACS (Department of Cardiothoracic Surgery); King s College Hospital, Denmark Hill, London SE5 9RS. Accepted for publication: August 6, Correspondence to J. P.

2 Phrenic nerve and diaphragmatic function and temperature 727 Table 1 Morphometric data for the 26 patients included in the study (mean (SD) or median (range). CPB Cardiopulmonary bypass time; X-clamp aortic cross-clamping time; cold warm time elapsed between recording the evoked EMG in the cold test condition and the warm test condition at the end of bypass Age (yr) 60.0 (39 74) Weight(kg) 80.7 (12.5) Body mass index 27.8 (3.8) CPB (min) 82.4 (23.6) X-clamp (min) 51.7 (14.2) Cold warm (min) 22.6 (10.8) further doses of neuromuscular blocking agents were given throughout surgery. SURGICAL MANAGEMENT All patients had three or four coronary grafts performed. T he two mammary arteries were dissected in all patients. During cardiopulmonary bypass, all patients were cooled actively to a nasopharyngeal temperature of 32 C during suturing of the distal ends of the grafts and then warmed to 38 C before restoring normal circulation. Preservation of the myocardium was achieved with intermittent perfusion of the coronary system via the root of the aorta with blood containing (mmol litre 1 ): KCl 16, MgCl 16 and procaine 1, at C. Sixteen of the 26 patients also had one single application of ice slush around the heart, without insulating pads. Durations of cardiopulmonary bypass and clamping of the aortic root were noted for all patients (table 1). No patient was receiving inotropes during or before the periods of testing diaphragmatic responses to phrenic stimulation. RECORDINGS OF DIAPHRAGMATIC EVOKED EMG SIGNALS After sternotomy, fine needle electrodes were inserted into the thickness of the anterior quadrant of the left hemidiaphragm, approximately 5 cm from the rib insertions, parallel to the muscle fibres. A fine needle temperature sensor was also inserted into the diaphragm in the vicinity of the electrodes. The left hemidiaphragm was chosen because the highest incidence of phrenic dysfunction has been reported on this side The phrenic nerves were stimulated with a 43-mm mean diameter double magnetic coil placed on the anterolateral surface of the neck over the left phrenic nerve. Two double coils were used when both phrenic nerves were stimulated simultaneously Each double coil was powered by a Magstim 200 HP stimulator (Magstim Company, Whitland, Dyfed, Wales). Five consecutive supramaximal stimulations at 15-s intervals, at 100% of maximum stimulator power, were performed at each temperature and the results averaged. The magnetic technique was selected because of the ease with which the phrenic nerve can be stimulated reproducibly. Compared with electrical stimulation, magnetic stimulation is more effective in achieving supramaximal stimulation of the phrenic nerve and is far less sensitive to positioning of the electrodes. 10 The output of the magnetic stimulator was set to maximum in this study, ensuring reliable supramaximal stimulation of the phrenic nerves, as documented in a previous study using the same device. 10 The diaphragmatic evoked electromyograms (EMG) were recorded via sterilized needle electrodes linked to a Magstim Neurosign 2000 amplifier designed for use with patients. The analogue EMG signal was converted to digital form on a National Instruments 12 bit NB-MIO 16 board (National Instruments, Austin, TX, USA) and stored in a MacIntosh 650 Quadra computer (Apple Computer Company, Cupertino, CA, USA). The EMG signal was sampled at 10 khz and processed on-line using a program developed on LabView 2.2 software (National Instruments, Austin, TX, USA). The magnetic pulse stimulation of the phrenic nerves consistently evoked a compound action potential in the EMG tracing recorded from the diaphragm by the needle electrodes (fig. 1). The electromyographical measurements were made in all 26 patients: (a) during cardiopulmonary bypass, before re-warming, while the diaphragmatic temperature was approximately 31 C; and (b) after whole body rewarming when the diaphragmatic temperature was approximately 36 C. In four of the 16 patients who received ice slush on the heart, an additional set of 13 measurements was obtained (c), the first just before applying the ice slush, the subsequent 11 measurements at 5-min intervals for 55 min and the last measurement at the latest possible time before closing the chest. In six patients whose nasopharyngeal temperatures remained at or above 36 C at the time of opening the chest and did not receive ice slush, additional electromyographical measurements were made (d) shortly after sternotomy and (e) during the active cooling period at the beginning of cardiopulmonary bypass. Most of the temperature related data were obtained at the transition from cold to warm occurring at the end of cardiopulmonary bypass, rather than at the transition from warm to cold at the beginning of bypass, for two reasons: first, to allow more time for possible residual effects of the neuromuscular blocking agent (used only to facilitate intubation of the trachea) to dissipate and for stable plasma concentrations of propofol to be achieved; second, most patients had a nasopharyngeal temperature of less than 36 C at induction of anaesthesia and by the time the chest was opened, most had cooled passively to C. In six subjects whose nasopharyngeal temperature had not decreased to less than 36 C when the chest was opened, recordings (d) and (e) were also made. These additional data allowed us to observe if the changes occurring during cooling were similar to those observed during re-warming. OTHER MEASUREMENTS Temperature measurements were made in the nasopharynx with a standard thermocouple sensor and in the diaphragm with a fine myocardial needle thermocouple sensor (Ellab A/S, Roedovre, Denmark) placed close to the EMG electrodes. At

3 728 British Journal of Anaesthesia the time of stimulation, diaphragmatic temperature might differ from nasopharyngeal temperature by C. It was diaphragmatic temperature, however, that was noted for the purpose of the study although nasopharyngeal temperature was used as a guide for perfusion. Arterial blood ph, PCO 2 and PO 2, and plasma concentrations of sodium, potassium and calcium were measured at the time of phrenic nerve stimulation using a self-calibrating Ciba-Corning 288 Blood Gas System, available in the same room. Plasma concentrations of propofol were measured when the electrophysiological tests (a) and (b) were performed. Gastric pressures (Pgas cm H 2 O) were measured after simultaneous stimulation of both phrenic nerves in four patients. A 100-cm balloon catheter (PK Morgan, Rainham, Kent, UK) was passed through the nose into the stomach at the start of operation with the aid of a lubricated size 7 uncuffed and uncut red rubber nasotracheal tube. Recordings were obtained with regularly tested 15 Validyne MP45 pressure transducers and amplifiers (Validyne Corporation, Northridge, CA, USA). The mean values of peak amplitudes of four deflections were calculated at two temperatures. Oesophageal pressure recordings were not made during surgery because the chest was open. Video recordings of the diaphragm during the five phrenic stimulations at each temperature were made in two patients. These were edited subsequently to visually compare the cold and warm twitch responses in quick succession. ANALYSIS OF DATA An example of an evoked compound action potential of the diaphragm obtained with the needle electrodes is shown in figure 1. Three values were derived from the evoked EMG trace, as indicated in figure 1: (a) latency or time elapsed between the magnetic stimulus artefact and the first deflection of the EMG trace (ms); (b) duration or time elapsed between the first deflection and the second peak of the compound action potential (ms); and (c) amplitude or magnitude (mv) of the difference between the positive and negative peak deflections of the compound action potential. Absolute amplitude values (mv) were very variable from patient to patient because of the relative positioning of the recording needle electrodes. Therefore, for the tables and for comparison purposes, we decided to express all amplitude values as relative changes with respect to the first set of measurements obtained during the cold period which were taken as 100. In the six patients in whom the warm to cold transition was also studied, the warm value was set to 100 for that set of measurements only. Latency, duration and amplitude were compared between cold (approximately 31 C, diaphragmatic temperature) and warm (approximately 36 C) conditions using the Wilcoxon signed rank test. A non-parametric test was used to process these data because of the obvious skew in the distribution. These comparisons were obtained independently for the two subsets of patients who received and who did not receive ice slush. Based on Figure 1 Representative example of two evoked compound action potentials recorded directly from the left diaphragm using needle electrodes, in response to pulses of magnetic stimulation to the left phrenic nerve in one patient. Recordings were made at two diaphragmatic temperatures: first, cold (approximately 31 C) and 12 min later, warm (approximately 36 C). S Stimulus artefact; L latency (the bars represent the latencies between stimulus and the first deflection in cold and warm conditions); D durations between the first deflection and the peak on the second wave; and A amplitudes of the warm and cold conditions. On the bottom lefthand corner, horizontal bar 10 ms and vertical bar 0.2 mv. previous studies, the upper limit of normal for latency was taken as 9.5 ms. 16 The influence of possible changes in other factors which might have occurred simultaneously with the change in temperature was also tested. Plasma concentrations of propofol, sodium (Na ), potassium (K ) and calcium (Ca ), and arterial blood-gas tensions were compared between cold and warm periods for each of these factors using the paired t test. Duration of cardiopulmonary bypass and aortic cross-clamping time were recorded. Results TEMPERATURE OF THE DIAPHRAGM Before starting warming at the end of cardiopulmonary bypass, mean diaphragmatic temperature was 31.3 (SD 1.87) C when the cold set of measurements were obtained; nasopharyngeal temperature was at the target value of 32 C. As the warm end-point of 38 C in the nasopharynx was reached, mean diaphragmatic temperature was only 36.5 (1.44) C when the second, warm set of measurements were obtained. This 1.5- C difference between diaphragmatic and nasopharyngeal temperatures was only reduced slightly ( 0.5 C) in the course of the ensuing 5 10 min after the set of warm measurements were obtained and before cardiopulmonary bypass was discontinued. Thus a gradient of approximately 1 C between diaphragmatic and nasopharyngeal temperatures was nearly always present at the end of bypass. PLASMA VARIABLES, CARDIOPULMONARY BYPASS VARIABLES AND BODY TEMPERATURE CHANGES Plasma concentrations of propofol (total range g ml 1 ), Na and Ca, and arterial

4 Phrenic nerve and diaphragmatic function and temperature 729 blood-gas values did not differ (paired t test, except for ph where the Wilcoxon sign rank test was used) between the cold and warm test conditions. Only plasma K concentration was significantly different in the two conditions (P 0.03), being higher during the warm condition by 0.28 mmol litre 1. CHANGES IN LATENCY, DURATION AND AMPLITUDE OF THE EVOKED DIAPHRAGMATIC EMG ASSOCIATED WITH CHANGES IN TEMPERATURE An example of the effect of a change in temperature from 31 to 36 C on the evoked muscle action potential in one patient is shown in figure 1. Data relating diaphragmatic evoked EMG latency, duration and amplitude to diaphragmatic temperature are summarized in table 2 for all patients and for two subgroups, those who did and those who did not receive ice slush. Comparisons between cold and warm conditions for all patients showed significant differences for the three variables; percentage changes in latency, duration and amplitude were 18, 14 and 99%, respectively. Direct observation of the twitch contractions of Table 2 Changes in the diaphragmatic evoked EMG in response to magnetic pulse phrenic stimulation in 26 patients, at two temperature: cold (31.3 C). and warm (36.5 ºC). L Latency, D duration and A amplitude of the evoked compound action potential as a percentage change from the cold condition. Values are medians (range). The Wilcoxon signed rank test was used for comparisons. Data from the 26 patients were first analysed as one population and also as two separate subgroups (those who received ice slush for myocardial preservation and those who did not) Cold Warm P All (n 26) L (ms) 10.1 ( ) 8.3 ( ) D (ms) 11.8 ( ) 9.0 ( ) A ( ) 0.05 Ice (n 16) L (ms) 10.3 ( ) 7.9 ( ) D (ms) 10.6 ( ) 8.6 ( ) A (74 313) 0.05 No ice (n 10) L (ms) 9.8 ( ) 8.7 ( ) 0.01 D (ms) 12.8 ( ) 9.8 ( ) ns A ( ) ns Table 3 Changes in the diaphragmatic evoked EMG in response to magnetic pulse phrenic stimulation in six patients during a temperature transition from warm (36 ºC) to cold (32 ºC). Units of measurement and symbols are as in table 2, except that the baseline value for A was in the warm condition n 6 Warm Cold P L (ms) 8.6 ( ) 10.9 ( ) 0.05 D (ms) 8.2 ( ) 9.4 ( ) ns A (10 34) ns the diaphragm in response to the magnetic pulses showed an obvious difference between the cold and warm testing conditions in all patients. The contraction was more vigorous and shorter in duration under warm conditions. This difference was clear in the edited video recordings in two patients. In the subgroup of six patients also tested during the transition from warm to cold, the mean diaphragmatic warm temperature was 34.8 C (36 C in the nasopharynx) and cold 31.5 C. On cooling, latency increased significantly by 25% and there were no significant changes in duration (12% increase) or amplitude (80% decrease, see table 3). A comparison between the subgroup of six patients who received suxamethonium and the subgroup of 20 patients who received vecuronium for intubation of the trachea showed no differences in the EMG variables latency, duration and amplitude between the cold and warm set of measurements. IMMEDIATE EFFECTS OF ICE SLUSH ON PHRENIC NERVE CONDUCTION The use of ice slush reduced markedly the amplitude of the left evoked diaphragmatic EMG within 5 min in the four patients monitored throughout the ice slush application, and increased markedly latency and duration, as seen in figure 2. In two of the patients the EMG was abolished during the initial 10 min of ice slush use, no diaphragmatic movement being visible in response to phrenic stimulation. Thus data on latency and duration could not be recorded during this period. Within 15 min, the EMG started to recover and after 55 min the mean values for latency, duration and amplitude were, respectively, 134%, 127% and 76% of pre-ice Figure 2 Changes in latency, duration and amplitude, as percentages from baseline pre-ice slush values (P), in the diaphragmatic EMG compound action potential evoked by magnetic pulse stimulation of the left phrenic nerve, in four patients. Values were recorded at 5-min intervals after application of ice slush to the heart and also after systemic warming, just before chest closure (W). In two patients ( and ) the evoked EMG was abolished for 10 min after ice application.

5 730 British Journal of Anaesthesia values. By the time the chest was closed, mean values for latency, duration and amplitude were 118%, 115% and 92% of pre-ice values. Only three of the 26 patients failed to recover the latency of the evoked EMG response to within the normal range (less than 9.5 ms) by the end of surgery; two of these patients had received ice slush, but none had difficulty in weaning from mechanical ventilation after surgery. All patients were breathing spontaneously within 6 h of arrival in the recovery ward. CHANGES IN GASTRIC PRESSURE IN RESPONSE TO PHRENIC STIMULATION The amplitude of the gastric pressure deflections in response to bilateral phrenic stimulation was more variable from patient to patient than the EMG data. Within each patient, however, variability was comparable with that of the EMG data. The amplitude of the deflection was approximately double during warm compared with cold stimulation in the four patients studied. Discussion The main finding in this study was the shortening of latency and increase in amplitude of the evoked diaphragmatic EMG action potential associated with a 5- C increase in temperature. Even the smaller temperature change (3.3 C) in the six patients studied during the warm to cold transition caused a significant increase in latency. These findings are not surprising, but provide the first direct evidence of the effects of small changes in temperature on phrenic nerve and diaphragmatic function in humans. This observation is important because the temperatures tested (31, 34.8 and 36 C) were within the range likely to be found during recovery from major surgery. Another observation was the marked reduction in amplitude of the EMG of the left hemidiaphragm shortly after a single application of ice slush to the heart. It appeared to temporarily abolish the diaphragmatic twitch response to phrenic stimulation within min in approximately 50% of patients (two of four in our series), without subsequent long-term dysfunction. LIMITATIONS OF THIS STUDY An important limitation of this study was the emphasis placed on the electromyographic variables to describe the function of the diaphragm. Ideally, we should have measured changes in twitch tension of the diaphragm in response to impulse stimuli, a notoriously difficult technique, only possible in animal experiments. In human studies, measurement of changes in transdiaphragmatic pressure in response to bilateral twitch stimulation of the diaphragm would have been the technique of choice to assess function, provided the chest cavity was not open. As the opportunity of obtaining measurements during relatively rapid changes in temperature occurred only with the chest open, we had no choice in the method of assessing diaphragmatic function other than the EMG and direct observation of diaphragmatic movement. The video recordings of the twitch responses were useful to confirm the impression gained from direct observation but did not provide quantitative data. Intragastric pressure measurements carried out in four patients during bilateral phrenic stimulation aimed at demonstrating a mechanical correlate of the EMG changes. These results, however, must be interpreted with caution because of many factors (for example, the presence of an open chest distorted by the chest retractor) which may have affected the absolute magnitude of the responses. Nevertheless, despite patient-topatient variation in the amplitude of the baseline pressure deflection, it doubled with the increase in temperature in all four patients tested, which strongly supports the EMG data. The results obtained, however, simply suggest that there was a degree of muscle weakness associated with 3 5 C of hypothermia. They do not allow us to predict if this weakness induces a corresponding degree of hypoventilation in a spontaneously ventilating hypothermic patient. A different type of study design, with emphasis on breathing variables, is required to answer this question. It is generally recognized, however, the difficulty in correlating EMG findings with postoperative diaphragm dysfunction. Ideally, each set of measurements should start with baseline EMG responses at normothermia followed by measurements during hypothermia and then recovery to normothermia. This approach was not possible in the majority of our patients because nasopharyngeal temperature was less than 36 C after induction of anaesthesia or had decreased to C by the time the chest was open. The observed effects on the EMG are unlikely to be related to the anaesthetic drugs used. Plasma propofol concentrations were within the range of light anaesthesia and decreased slightly, but not significantly, between measurements. I.v. induction agents have been shown to reduce the amplitude of the leg EMG response to magnetic cortical stimulation whereas fentanyl has no effects. Data are not available on the direct effects of anaesthetic agents on the motor neurone, skeletal muscle or the neuromuscular junction, except for volatile agents which depress the EMG response evoked by cortical stimulation 19 and the excitability of spinal motor neurones. 20 Vecuronium, used for intubation of the trachea in 20 patients, could have affected the EMG responses as there is evidence that hypothermia prolongs its effect. 21 It is unlikely that vecuronium could account for the changes in EMG variables observed during the short period of systemic warming (approximately 15 min) occurring at least 3 h after the single bolus administration of a moderate dose for intubation. The identical responses observed in the six patients given suxamethonium for intubation also refutes a possible effect of residual block. We do not have a plausible explanation for the slight but significant increase in mean plasma potassium concentration associated with systemic warming. It was, however, too small to have physiological effects

6 Phrenic nerve and diaphragmatic function and temperature 731 and did not correlate with individual changes in electrophysiological variables. CHANGES IN DIAPHRAGMATIC EMG WITH HYPOTHERMIA On the transition from cold to warm in the 10 patients not exposed to ice slush, the only significant change was the shortening of latency in the evoked EMG response. The lack of significance in the changes in duration and amplitude may be explained by the greater variability in the measurements of duration and amplitude, especially amplitude which was sensitive to small changes in electrode position. In the smaller subgroup of six patients in whom measurements were also made during the transition from warm to cold, again, only latency showed a significant change with cooling, despite the smaller sample size and the smaller temperature change (mean 3.3 C). The variability in the measurements in this subgroup was noticeably reduced, perhaps because the patient s systemic cooling took place over a much shorter period of time than warming, so the interval between the measurements at the two temperatures was shorter. EFFECTS OF ICE SLUSH The duration of the conduction block which we observed with ice slush was much shorter than that reported previously in a similar experimental situation in dogs. Exposure of the dog phrenic nerve to ice slush for min caused total phrenic block with a similar, rapid onset (approximately 5 min) but lasting between 6 and 28 days 9 instead of min as in our study. This discrepancy may be because of the pericardial sac preventing direct contact between the slush and the phrenic nerve. Duration of exposure to ice may also be important. Rosou and colleagues, 13 who studied patients in whom a special cooling jacket was placed around the heart for much longer periods of time, observed a 23% incidence of prolonged phrenic paresis. It may be noteworthy that two of the three subjects in our study who, despite normothermia had a prolonged latency at the time of chest closure (but no subsequent clinically obvious postoperative dysfunction) had been exposed to ice slush. We did not, however, expect to be able to correlate measurements obtained during surgery with measurements of diaphragmatic dysfunction observed after surgery in such a small number of patients. Given the average incidence of severe diaphragmatic dysfunction after cardiac surgery, it would require study of at least 400 consecutive patients to be able to include one with this complication. The fact that all patients in this study had an uneventful recovery and were breathing spontaneously 6 h after surgery does not exclude the presence of diaphragmatic dysfunction, especially during the first 3 h of recovery. A different study design is required to examine the effects of postoperative hypothermia on respiratory variables. The clinical implications of our study are mostly apparent when attempts are made to reduce the time to extubation after cardiac anaesthesia (often referred to as fast tracking ). Most patients, if normothermic, are able to maintain acceptable gas exchange when spontaneous ventilation returns shortly after closure of the skin. If hypothermia is present, however, there is diaphragmatic dysfunction and we believe that some type of mechanical assistance to breathing is justified until normothermia is achieved, especially in susceptible patients such as the elderly or the obese. References 1. Diehl JL, Lofaso F, Deleuse P, Similowski T, Lemaire F, Brochard L. Clinically relevant diaphragmatic dysfunction after cardiac operations. Journal of Thoracic and Cardiovascular Surgery 1994; 104: Paintal AS. Block in conduction in mammalian myelinated nerve fibres at low temperatures. Journal of Physiology (London) 1965; 180: Dureuil B, Vires N, Pariente R, Desmonts JM, Aubier M. Effects of phrenic nerve cooling on diaphragmatic function. Journal of Applied Physiology 1987; 63: Doud JR, Walsh JM. Muscle temperature alters the EMG power spectrum of the canine diaphragm. Respiration Physiology 1993; 94: Buchthal F, Engbaek I. Refractory period and conduction velocity of the striated muscle fibre. Acta Physiologica Scandinavica 1963; 59: Jarcho LW, Berman B, Dowben RM, Lilienthal JL. Site of origin and velocity of contraction of fibrillary potentials in denervated skeletal muscle. American Journal of Physiology 1954; 178: Stalberg E. Propagation velocity in human muscle fibres in situ. Acta Physiologica Scandinavica 1966; 70: Iaizzo PA, Poppele RE. Twitch relaxation of the cat soleus muscle at different lengths and temperatures. Muscle and Nerve 1990; 13: Marco JD, Hahn JW, Barner HB. Topical cardiac hypothermia and phrenic nerve injury. Annals of Thoracic Surgery 1977; 22: Mills GH, Kyroussis D, Hamnegard CH, Wragg S, Moxham J, Green M. Unilateral magnetic phrenic stimulation of the phrenic nerve. Thorax 1995; 50: Mills GH, Kyroussis D, Hamnegard C, Polkey M, Green M, Moxham J. Bilateral anterolateral magnetic stimulation of the phrenic nerves. American Journal of Respiratory and Critical Care Medicine 1996; 154: Benjamin JJ, Cascade PN, Rubenfire M, Wajszczuk W, Kerin NZ. Left lower lobe atelectasis and consolidation following cardiac surgery. The effect of topical cooling on the phrenic nerve. Radiology 1982; 142: Rousou JA, Parker T, Engelman RM, Breyer RH. Phrenic nerve paresis associated with the use of iced slush and the cooling jacket for topical hypothermia. Journal of Thoracic and Cardiovascular Surgery 1985; 89: Markand OM, Moorthy SS, Mahomed Y, King RB, Brown JW. Postoperative phrenic nerve palsy in patients with openheart surgery. Annals of Thoracic Surgery 1995; 39: Sixt R, Bake B. A simple pressure calibrator. Scandinavian Journal of Clinical Laboratory Investigation 1976; 36: Laroche CM, Mier AK, Moxham J, Green M. Diaphragm strength in patients with recent hemidiaphragm paralysis. Thorax 1988; 43: Ghaly RF, Stone JL, Levy WJ, Roccoforte P, Brunner EB. The effect of etomidate on motor evoked potentials induced by transcranial magnetic stimulation in the monkey. Neurosurgery 1990; 27: Kalkman CJ, Drummond JC, Ribberink AA, Patel PM, Sano T, Bickford RG. Effects of propofol, etomidate, midazolam and fentanyl on motor evoked responses to transcranial electrical or magnetic stimulation in humans. Anesthesiology 1992; 76:

7 732 British Journal of Anaesthesia 19. Kalkman CJ, Drummond JC, Ribberink AA. Low concentrations of isoflurane abolish motor evoked responses to transcranial electrical stimulation during nitrous oxide opioid anesthesia in humans. Anesthesia and Analgesia 1991; 73: Rampil LT, King BS. Volatile anesthetics depress spinal motor neurons. Anesthesiology 1996; 85: Heier T, Caldwell JE, Sessler DL Miller RD. Mild intraoperative hypothermia increases duration of action and spontaneous recovery of vecuronium blockade during nitrous oxide isoflurane anesthesia in humans. Anesthesiology 1991; 74: