Diaphragmatic pacing or 'electrophrenic respiration' is a form of treatment in which the phrenic nerves are stimulated by bursts of radio-frequency energy delivered transcutaneously to implanted electrodes. It is an alternative to positive-pressure ventilation for patients with either chronic hypoventilation or apnoea in the presence of normal phrenic motoneurones and nerve conduction [1,2]. This can arise from disease affecting the respiratory centres, but is also a feature of high tetraplegia from injury of the spinal cord above the outflow of the cervical roots that contribute to the phrenic nerve. The operations for implanting the electrodes for phrenic nerve pacing are undertaken by only a few specialized centres, but patients with high tetraplegia are scattered throughout Europe. With the increased mobility afforded to them by diaphragmatic pacing, they are liable to present to any non-specialist centre and may require anaesthesia for intercurrent emergencies that may or may not be related to their tetraplegia. The requirement for most anaesthetists outside spinal units to anaesthetise a patient with low tetraplegia is quite rare, though the subject still warrants its coverage in recognized monographs . This paper reviews some of the additional complications that could be faced by anaesthetists in dealing with patients with high tetraplegia - the presence of a tracheostomy, the procedural complications in maintaining diaphragmatic pacing, and the ever-present threat of apnoea if it fails. The constraints on the anaesthetic management of patients with high tetraplegia will be illustrated by a review of recent experience in the Spinal Injuries Unit at Southport District General Hospital.
Indications, equipment and surgical considerations
Full accounts of diaphragmatic pacing already exist [4,5] but the salient points will be summarized to place the anaesthetic management in context. When a patient has suffered a cervical cord injury with associated apnoea, some weeks are allowed to elapse to enable any potential for recovery to express itself. If there is no recovery of the myotomes below the second cervical segment, the function of the phrenic nerve itself is evaluated formally by means of transcutaneous phrenic nerve stimulation, with electromyography to determine nerve conduction times and fluoroscopy to display the movement of the appropriate hemidiaphragm. To ensure that an adequate stimulus can be delivered, it may be necessary to undertake this under light general anaesthesia without muscle relaxant. In order to find the phrenic nerve behind the posterior border of the sternomastoid, electrical stimulation is used with square-wave pulses of less than 600 μs duration, at a frequency of 10-20 Hz and a voltage that is increased until the brachial plexus is stimulated or the underlying muscles respond to direct stimulation. The probe is then manipulated until the diaphragm is seen to twitch. In adults, inspiratory descent of the diaphragm on fluoroscopy should be at least 4 cm, and the normal conduction time in adults is 8-10 ms  but is correspondingly less in children  because the phrenic nerve is shorter. The absence of a diaphragmatic response in a patient with good brachial plexus responses to transcutaneous stimulation could in theory be an indication to make a small skin incision and stimulate using an insulated needle, attempting to get as close as possible to the phrenic nerve without actually making a direct and potentially damaging contact with it. We have no experience with this techniqe. In some patients a response has only been obtained at operation to stimulation with pulse widths above 600 μs : this may be attributable to the selective preservation (for whatever reason) of a population of smaller diameter α-motoneurones of about 10 μ diameter, innervating the slow-twitch, fatigue-resistant muscle fibres that make up about 55% of the diaphragm. The larger α-motoneurones of up to 20 μ diameter subserve fast-twitch fibres and usually respond preferentially to electrical stimulation, though it is the slow-twitch fibres that are recruited in physiological breathing. The surgical implantation of the definitive system for electrodiaphragmatic breathing is followed (see later) by a period of intense conditioning with low-frequency stimulation, for which one of the aims is to convert as many as possible of the fast-twitch muscle fibres into fatigue-resistant slow-twitch fibres.
Once functional phrenic nerves have been demonstrated, preparation for diaphragmatic pacing can proceed. The apparatus for definitive pacing consists of external and internal components. The external components consist of (i) a battery-powered, pocket-sized transmitter that delivers bursts of radio-frequency energy via (ii) energy transfer coils that are positioned carefully so as to overlie (iii) internal receivers implanted in a subcutaneous pocket either below the clavicle or on the rib-cage near the costal margin: even slight misapplication of the energy-transfer coils to the skin over the receivers may cause a pacing failure even though the carrier-wave signal may be picked up at some distance on radio. The receivers are connected by internal cables to (iv) stimulating electrodes surgically implanted in close apposition to the phrenic nerves in the neck or thorax. The implanted stimulating electrodes are made of platinum. Either the monopolar (Avery, NY, USA) or quadripolar (Atrostim, Tampere, Finland) electrodes can be sewn in place close to the nerve after delicate dissection to avoid damaging it, but the carousel-type (Medimplant, Wien, Austria) electrodes are stitched to the perineurium itself.
The implantation may be targetted on the phrenic nerve in the neck or in the mediastinum. The potential problems in the neck are that the landmark sternomastoid and scalenus anterior muscles may be grossly atrophic in patients with high tetraplegia, so that identification of the phrenic nerve may be difficult. It is also possible, especially with the Atrostim quadripolar electrode, for cervical phrenic stimulation to recruit branches of the brachial plexus. The movements induced by this and any residual spontaneous contraction of neck muscles can displace the stimulating electrode from optimal placement and result in failure of stimulation. Some people also have an accessory phrenic nerve, which is a post-fixed branch that joins the main trunk intrathoracically. If this were to be the case in a patient with high tetraplegia, it is theoretically possible for a cervically implanted electrode to stimulate only part of the motor supply of the diaphragm and produce only partial diaphragmatic activation. These problems have initiated and strengthened a preference for stimulating the phrenic nerve in the mediastinum.
In mediastinal implantation, the electrodes are stitched in place with care, supported by a flap of pericardium, to ensure that they are at least 5 cm away from the sinoatrial region. This gives a measure of security compared with cervical implantation. Implantation by a limited thoracotomy through the second intercostal space may seem less of a surgical insult than a sternotomy, but even with endobronchial intubation and deflation of the lung, it does not give such good access, especially for the quadripolar electrode. But the sternotomy in those with high tetraplegia does not carry the usual penalty of postoperative pain. It allows easier identification of the phrenic nerves in the chest and is said to carry a lower risk of operative damage to the phrenic nerve : it also requires a shorter operative time which makes it attractive for a number of reasons, including the limitation on heat loss by the patient. There are recognized disadvantages of the sternotomy if the wound becomes infected or re-exploration is required, but despite these, there has been an evolution of the surgical approach at this centre from cervical to limited thoractomy with endobronchial intubation to median sternotomy. When revisions of implantation are necessary, re-implantation tends to be in the alternative site so, whatever the centre's primary preference, it is highly desirable that the surgical and anaesthetic team maintain their expertise with both cervical and intrathoracic implantation.
Whatever site is chosen for implantation, the intra-operative identification of the phrenic nerve is confirmed by eliciting a diaphragmatic response to electrical stimulation, and the final position for implantation is determined as that from which a diaphragmatic twitch can be produced with the smallest possible current. It is therefore obviously necessary to avoid or minimize the use of neuromuscular blocking agents in the anaesthetic technique. Stimulation periods must also be kept short to avoid the onset of fatigue, which can be rapid in a diaphragm that has been deconditioned by disuse. The minimum stimulating currents established at implantation become the standards against which to judge the subsequent performance of the pacing system.
The implantation is followed by a period of caution with such procedures as chest physiotherapy that carry a risk of displacing the stimulating electrodes, and extreme care must always be taken to avoid damage to the implanted receiver and connecting cables. The threshold currents and stimulating currents generally vary little in the established patient. The battery running time is specific to the model and ranges from 2 days to 2 weeks. Some stimulators have limited alarm facilities for battery failure or antenna failure. Nonetheless pacing failures can and do occur, so that continued performance testing is an integral part of further management, as are precautions to re-establish mechanical ventilation in the event of such failure, pending further operations to replace or revise the siting of the implanted components.
Ten years of experience in Southport
The 10 years of the experience with diaphragmatic pacing in the Spinal Injuries Unit at Southport District General Hospital is indicated in Tables 1 and 2. Table 1 lists 10 patients who underwent 18 operations for primary implantation or revision of phrenic pacing systems, and Table 2 lists eight incidental surgical procedures in patients with diaphragmatic pacers. In Table 1, the average age at the time of injury was 18.7 years, the eldest being 46 and the youngest 31/2 years old. All of the Southport patients had sustained spinal cord trauma from road accidents, but in a recent review of 64 patients who required diaphragmatic pacing in 22 institutions across the world, there were 35 children, of whom 14 had Chronic Alveolar Hypoventilation from neurological dysfunction that was more or less localized to the respiratory centres, so that spinal cord function was otherwise relatively normal [Weese-Mayer, personal communication].
The time between the initial injury and the first phrenic nerve implant in the Southport series was between 8 and 19 months with a mean of 13.5 months. All patients had undergone tracheostomy and were dependent on mechanical ventilation. The two patients with injury at C1 also had initially lower cranial nerve involvement which would have been a contraindication to pacing had there not been recovery of bulbar function. The patients with injury at level C2 had variable voluntary respiratory function associated with accessory muscle activity, which had been reinforced by training.
Nine procedures were primary implantations, whilst eight operations in four patients were for revision because of failure of implantable components. Patient four had had his primary implant inserted at another centre and the original anaesthetic records for this patient could not be found. One patient required three operations for receiver and electrode replacements because of broken wires attributed to over-vigorous chest physiotherapy and intermittent function of cervical electrodes associated with changes in posture. Though some patients in the USA have had revision of the subcutaneous receiver implants as day-case procedures under local anaesthesia, these are undertaken in Southport as in-patient procedures which makes for easier post-operative assessment.
The anaesthetist must be clear as to the precise level of spinal cord injury in each patient presenting for anaesthesia, and enquire and examine for symptoms and signs of spinal hyper-reflexia and autonomic dysfunction which may give rise to haemodynamic instability. The proposed implantation sites must be confirmed, and the plan for airway management must be made accordingly. Some patients with heavy tracheal colonization without frank infection will benefit from starting a suitable antibiotic pre-operatively.
Though patients present with a tracheostomy in situ, it is generally not the ideal way to manage the airway during the procedure itself. Even if endobronchial intubation is not contemplated (for collapsing a lung to allow surgical access via thoracotomy), orotracheal intubation is preferred, so as not to impede surgical access, to minimize the risk of bacterial contamination of the adjacent surgical wound and to facilitate intra-operative tracheal suction as indicated. The requirement to avoid or minimize the use of neuromuscular blocking agents in the anaesthetic technique makes it necessary to undertake a careful assessment of the mobility of the cervical spine as an indication of possible additional difficulty during intubation of the larynx, and the likelihood of subglottic stenoses must be considered in those patients who have had long-term tracheostomy.
Patients in Southport received sedative premedication: anticholinergic agents were reserved for intra-operative use unless a patient had copious tracheal secretions associated with bacterial colonisation of the upper trachea. Venous access can be difficult in paralysed spastic limbs because of the reduction in peripheral blood flow compounded by the effects of multiple previous venous cannulation for earlier procedures. Access to the central system via the jugular or subclavian veins may well be precluded because of an existing implant. Although the Southport patients were insensitive to pain in dermatomes below the level of their injury, local anaesthetic was often helpful in preventing strong reflex withdrawal during attempted venepuncture. Veins on the dorsum of the feet were used on eight occasions and were the sole venous access in two patients not undergoing sternotomy. The femoral vein was accessed on three occasions, using a Seldinger technique.
Anaesthesia, haemodynamic responses, temperature and general care
Important features of intra-operative management are the establishment of crystalloid infusion in anticipation of hypotension on induction of anaesthesia, and avoidance of sole reliance on a volatile agent for the same reason. The use of neuromuscular blockers needs to be avoided as far as possible so as to allow optimal placement of the phrenic nerve electrodes, to be judged using the smallest electrical stimulus that will elicit the maximum diaphragmatic response. However, a single judicious dose of a relaxant of intermediate duration is a considerable convenience in establishing the intra-operative airway (see below), and its effects are likely to have abated by the time phrenic stimulation commences.
The disturbances in autonomic function can lead, on the one hand, to a lack of the usual signs of inadequacy of anaesthesia and, on the other, to exagerated haemodynamic responses to nociceptive input such as from thoracotomy or sternotomy. Direct arterial recording has been used more recently in Southport and the tendency to hyper-reflexic hypertension has been adequately controlled with volatile agents . There were episodes of intraoperative hypotension (systolic blood pressure below 85 mmHg) in six cases, usually associated with surgical manipulation around the venae cavae and right atrium. Atropine or glycopyrrolate were used for vagolysis in 10 of 11 procedures and there were no episodes of bradycardia associated with intubation.
Patients with tetraplegia probably lose heat during anaesthesia in much the same way as those who are neurologically intact, but they take much longer to regain normothermia post-operatively, being unable to shiver or vasoconstrict appropriately. Due attention must be paid to the anticipated duration when choosing the surgical procedure (and to the likelihood that sternotomy is nowadays the most brief). Precautions against heat loss should include warming blankets when possible, heat and moisture exchangers, low fresh gas flows in circle systems with CO2 absorption and warming of intravenous (i.v.) fluids. Rectal temperature was measured routinely in the Southport series, and the mean temperatures at start and the finish were, respectively, 36.2°C and 36.1°C.
All patients had a bladder catheter inserted to avoid the possibility of harmful overdistension of the neuropathic bladder whilst maintaining a good diuresis as a prophylactic against urinary tract infection. The patients were lain on a pressure-relieving mattress or pillow packs, with particular care of healed decubitus ulcers. Prophylactic measures were taken routinely against venous thromboembolism and possible implant or chest infection.
Orotracheal intubation was used in the Southport series, except for two adult patients with established subglottic stenosis, in whom the tracheostomy tube was kept in place. After induction of anaesthesia, a single dose of non-depolarizing muscle relaxant of intermediate duration was used to facilitate orotracheal intubation. Because of the insufflation leak around the uncuffed tracheostomy tubes, the anaesthetic room ventilator was unable to maintain adequate ventilation during the onset of relaxation of the vocal cords. It was usually easier to do this with a self-inflating bag and to maintain anaesthesia with i.v. agents. Alternatively, the trachea can be decannulated immediately after induction of anaesthesia, and ventilation controlled by a facemask while an assistant occludes the stoma. A left endobronchial tube was used in the four cases having a thoracotomy incision (with some initial difficulty in one patient with a Mallampati score of 2). After removal of the tracheostomy tube, the stoma was treated with antiseptic, and then covered and sealed to prevent soiling of the adjacent operative site. In Patient 10 (Table 1), a 4-year-old child, the largest size of orotracheal tube which would pass the cricoid ring was 4.0 mm, but it resulted in a large stomal leak. Despite considerable care to apply packing and an airtight dressing to prevent a serious loss of tidal volume during surgery, the dressing was found to have become almost detached by the end of the procedure.
At the end of the procedure, the airway was changed back to a tracheostomy, either cuffed or with a close-fitting uncuffed tube. Although the tracheal stomata in the children were chronic, allowing for easy routine tube changes in the ward, they had constricted somewhat during the period of decannulation required for surgery, and required gentle dilatation with a bevelled orotracheal tube before again accepting the previous size of tracheostomy tube.
Immediate recovery and continuing management
None of the Southport patients had any post-operative pain. All looked well on the day after surgery, except for one patient who developed a haemothorax because of a blocked thoracotomy drain and required bronchoscopy to re-expand the lung. Limited chest vibration and compression is all that is permitted for these patients until the chest wounds have begun healing and the implantation of the electrodes has consolidated. After a few weeks, assisted coughing can be given. This is mostly carried out by abdominal compression, so avoiding chest compression over the vulnerable receiver implant or its cable.
Patients with new electrode systems wait for 2 to 3 weeks for stabilization, after which an intense programme of conditioning is begun to re-accustom the diaphragm to muscular activity (and to increase its content of more fatigue-resistant slow-twitch fibres). It usually takes a further 3 or 4 months until a stable respiratory pattern can be established. The performance of the system is checked at intervals, by comparing the thresholds and minimum pacing currents with those previously recorded, and by spirometry during pacing of one or other hemidiaphragm, or both.
Five of the Southport series have subsequently been able to pace their diaphragms night and day, though one of these has since died from an unrelated cause. Four use mechanical ventilation at night, to rest their diaphragms and maximize their daytime tidal volume, as well as to provide additional humidification. The last patient, a child, is mechanically ventilated night and day because of injury to one of his phrenic nerves. Retention of the tracheostomy tube with a speaking valve is the favoured option, even in patients without subglottic stenosis. Even though patients may be able to breathe normally through their upper airways when awake, there is an undoubted predisposition to obstructive sleep apnoea in tetraplegic patients using diaphragmatic pacing, because of the reduced tone in their genioglossal musculature . Tracheal closure predisposes to inadvertent inhalation of oropharyngeal secretions or food and drink, which predisposes to recurrent bronchorrhoea and chest infection. Finally, patients with high tetraplegia remain indefinitely at risk of needing alternative ventilatory assistance, so that it is in any case prudent to retain an alternative access to their respiratory system.
Though we normally advise patients strongly against tracheal closure for all of the above reasons, we must respect their autonomy. Patient 3 (Table 1) had his tracheostomy closed at his request after his first cervical implantation procedure. He suffered repeated chest infections associated with hypoventilation and required 13 admissions to Southport or to the intensive unit of his local hospital. The hypoventilation was partly attributable to changes in electrode threshold caused by movements of his head and neck. After a re-operation to implant new intrathoracic electrodes, he was managed for 36 h by elective nasotracheal ventilation followed by a strict regimen of nasal intermittent positive breathing (with frequent oropharyngeal suctioning to prevent sputum retention and atelectasis) until the new implants could be used 2 weeks later.
Discharge to non-specialist care
Once the patients are established on electrophrenic respiration, they are candidates for discharge from specialist care. The discharge protocol from the Southport unit includes establishment of liaisons with both community care and the patient's local acute-care hospital. The protocol supplies some basic familiarity with the pacing system and caveats against physical damage and the possible adverse effects of magnetic and radio-frequency fields (such as might be generated not only by diagnostic and therapeutic devices, but also by devices such as mobile phones that are in everyday non-medical use ). It also lists criteria for the provision of further care if the need arises - criteria agreed between the patients, their general practitioners, their local intensive care units and the Southport unit. This allows for a local hospital admission in the event of an acute problem such as a chest infection if it cannot be managed at home, or for re-admission to the Southport spinal unit if that becomes necessary. Criteria for re-admission to specialist care in the case of chest infection would, for instance, include systemic upset with pyrexia, failure to respond to antibiotics, obvious sputum retention, hypoxaemia or the need for frequent tracheal suctioning and intensive chest physiotherapy. In general, any requirement for more prolonged hospital care should prompt efforts to re-admit patients to the specialized spinal units where they have received their primary treatment. Apart from patient 3 described earlier, there has been only one case in the Southport series who could not be managed at home according to the criteria agreed at time of discharge.
Even with the best managed diaphragmatic pacing, lower respiratory tract infections will occur at least as frequently as in normal people. However, the consequences are liable to be more profound. Antibiotics, chest physiotherapy and assisted coughing are, as usual, the mainstay of treatment, but the physiotherapists and nurses must be aware of the position of the implants in the chest wall so they may avoid damaging them. Pulse oximetry, with due attention to the effects of oxygen enrichment, provides adequate monitoring in most cases. If the pulmonary compliance is low, vigilance must be maintained against the possibility of diaphragmatic fatigue and failure, and must be backed up by a readiness to assist ventilation if the work of breathing is unduly increased. Otherwise, it may be reasonable to continue diaphragmatic pacing for short times during such challenging procedures as tracheal suctioning and bed-bathing, provided that the oxygen saturation by pulse oximetry is maintained above 90%.
Table 2 lists seven types of surgical procedures that have been undertaken subsequently in six of the patients in Table 1 for whom implantation or revision was undertaken in Southport. Pre-operative anaesthetic planning covers the adequacy of the patient's pulmonary gas exchange and familiarization with the location, operation, type and supplier of the patient's diaphragmatic pacing system: this may be supplemented, as deemed necessary, by formal checks of minimum stimulus requirements and spirometry, and their dependence on posture. Checks are also needed for the likely dependence on, and availability of, alternative means for supporting breathing and dealing with respiratory secretions.
In Southport, incidental surgery in patients with diaphragmatic pacing has been undertaken where possible under low epidural or spinal blockade to prevent somatic and autonomic reflexes, but light inhalational or i.v. general anaesthesia with continued diaphragmatic pacing is also possible for minor surgery. The procedures in Table 2 that required controlled ventilation were major ones (implantation of sacral root stimulators) that required the patient to be in the prone position. One patient had extracorporeal shock wave lithotripsy solely with diazepam for control of reflexes. Although caution is advised for lithotripsy in patients with phrenic pacers, damage is unlikely as long as the implants are over 5 cm away from the probe.
Travel and accidental medical or surgical care
All patients in the Southport series have been mobilized in wheelchairs, and three have had several holidays involving air travel during which they have used either their diaphragmatic pacers or mechanical ventilation. The option of using electrophrenic respiration during flight is at the discretion of the airline and pilot, because of the potential for radio-frequency interference with the flight controls. A detailed plan for the management of the equipment and respiration needs to be co-ordinated by the medical practitioner in charge of the diaphragmatic pacing programme, who in various parts of the world may be an anaesthetist, a rehabilitation physician, a surgeon or a respiratory physician.
The benefit of increased mobility carries the penalty that these patients may develop problems in places far distant from both their home and tertiary referral centre. This may be as alarming to the patients (who have generally become well informed about the technicalities of their management) and the medical staff of the institution where they present (who are likely to be quite ill-informed on such a specialist topic). The medical staff in question are most likely to be anaesthetists, intensivists or surgeons, and they may be faced with difficult decisions on the logistics of anaesthetic and surgical management, and the choice of anaesthesia for incidental procedures in unfamiliar places may well be profoundly influenced by the patient's previous experience of anaesthesia.
Staff may also need to consider the most appropriate mode of respiratory support in the event of a chest infection or pacer malfunction. A check of the pacer performance may be undertaken by measuring the patient's right and left tidal volumes. This can be carried out by switching the pacemaker to pace one side at a time and measuring the resultant tidal volumes through the tracheostomy - pinching the nose if the tube is cuffless. If the tracheostomy has been closed, this can be carried out using a mouthpiece or facemask. During normal use, many of the patients will have a mild respiratory alkalosis relating to the large tidal volumes necessary for louder talking. The thresholds and stimulation currents can also be checked against the patient's prescribed settings. However, fault finding during a malfunction may turn out to be more complicated. In any event, it is highly desirable, and should usually be possible, to make contact with the tertiary centre for specific advice on the operation of the pacemaker system in question.
The anaesthetist may well be the professional in the best position to ensure that the patient is not put at risk from dysfunction of, or damage to, the pacing system from such devices as magnetic resonance scanners or monopolar diathermy. Of the more clinical aspects of peri-operative management, the interpretation of the cardiovascular measurements and their responses during autonomic hyperreflexia cannot receive enough emphasis. The anaesthetist who successfully meets the challenge of managing a high tetraplegic patient through an unscheduled medical or surgical intervention will undoubtedly emerge from the experience better able to cope with the more mundane problems of everyday practice!
The authors wish to thank those clinicians and other staff who have shared in the care of these tetraplegic patients in their own localities. Their support has facilitated the patients' hospital discharge and has helped restore a quality of life to them.
Atrostim Atrotech OY, PO Box 28, FIN-3372 Tampere, Finland
Avery Mark IV The Dobelle Institute, 61 Mall Drive, Commack, New York 11725-5703, USA
Medimplant Medimplant Inc., Institute fur Biomedizinische Technik und Physik, Allgemeines Krankenhaus Ebene 04/L, 1090 Wien, Wahringer Gurtel 18-20, Austria