Remifentanil is an ultra-short acting μ opioid receptor agonist. It is thought to offer some advantages over conventional parenteral opioid analgesia for labour and therefore to have a place in the management of labour pain, despite the greater efficacy of neuraxial analgesia [1, 2]. We developed a local protocol for its use as intravenous patient-controlled analgesia (PCA) in labour based on that described by Hill . Remifentanil was offered to selected women as an alternative to regional analgesia or to intramuscular diamorphine, which is the standard opioid used in our unit. It had, at the time of this reported incident, replaced intravenous morphine PCA as the analgesic option most commonly used by women labouring with a non-viable fetus.
This case report describes the management of a patient, diagnosed with an intrauterine death at 31 weeks’ gestation, who suffered a cardiorespiratory arrest during her induced labour while using a remifentanil PCA. She made a full recovery from resuscitation which included a peri-mortem caesarean section.
A 24-year-old woman with weight 87 kg, height 169 cm, and BMI 29.4 kg.m−2 was diagnosed on ultrasound scan with an intrauterine death at 31 weeks’ gestation. The patient was otherwise well. She had had an uncomplicated emergency caesarean section for a previous delivery. She was not taking any regular medication and had no allergies.
She was given mifopristone 200 μg orally and was booked to attend the labour ward 48 h later for induction of her labour. Induction of labour was commenced as per our local protocol with a misoprostol 100-μg pessary. The patient requested analgesia and oral paracetamol 1 g was given. An hour later (time 0) she took oral codeine phosphate 60 mg, and 60 min later she began to use Entonox (50:50 mixture of nitrous oxide:oxygen).
Two hours after the codeine, the midwife requested an anaesthetic review to discuss further analgesia with the patient as she was complaining of increased pain. At that time the two anaesthetists on labour ward were not able to attend within 30 min, so the midwife gave her diamorphine 10 mg intramuscularly and she continued to use Entonox. Following the diamorphine, the patient was much more comfortable and did not require additional analgesia; 90 min later (time 3 h 30 min), the patient requested further analgesia and the anaesthetist was again asked to review her.
Following a discussion with the anaesthetist, the patient opted for remifentanil PCA. At time 3 h 40 min the anaesthetist set up the PCA with the standard settings, which consisted of bolus 40 μg, lockout period 2 min, no background infusion and a 4-h limit of 5 mg. The PCA was administered via a Graseby infusion pump with locked syringe casing to prevent tampering and a dedicated cannula. The syringe giving set contained an anti-reflux valve and the pump was positioned below the level of the patient as an additional measure to prevent siphoning. The anaesthetist observed the patient directly for the first five PCA boluses over approximately 15 min according to the unit guidelines (time 3 h 55 min). There was no oxygen desaturation on pulse oximetry, although the patient was receiving some supplemental oxygen as she continued to use Entonox as well as the PCA during contractions. When the anaesthetist left the patient she was sitting up in bed.
Fifteen minutes after the anaesthetist left the patient and 35 min after the start of the PCA, (time 4 h 10 min), the patient's mother shouted for help. The midwife was not in the room at the time, having left to collect another pessary. A cardiac arrest call was put out and the emergency buzzer sounded on labour ward. A consultant and a registrar anaesthetist were present on the delivery suite and arrived within a minute of the buzzer sounding. The patient was supine on the bed, unresponsive (Glasgow Coma Score (GCS) 3/15) and deeply cyanosed. A small amount of fluid was present in the airway, which was suctioned. The patient was making no respiratory effort and so manual ventilation was commenced with 100% oxygen using a bag and mask. No peripheral or central pulses were palpable and external cardiac massage was started. The pupils were approximately 5 mm in diameter and equal. The patient's mother reported that the patient had suddenly turned blue, complained that she could not breathe and had then become unresponsive.
The patient had marked masseter spasm, making intubation impossible. Mask ventilation was started while an arrest trolley was obtained which contained airway adjuncts. The chest compliance appeared normal. External cardiac massage with continuous chest compressions was performed at a rate of approximately 100 min−1 along with continuous mask ventilation. The obstetricians were asked to perform an immediate caesarean section in the room. The patient remained supine at this time as a wedge was not immediately available. The cyanosis resolved with manual ventilation, although we were unable to obtain oxygen saturation in the delivery room as the monitor and probe were mislaid in the emergency. As the operation started the carotid pulse became palpable, indicating return of spontaneous cardiac output. Adrenaline had not been given. Delivery of a dead fetus occurred 3 min after the cardiac arrest call was made (time 4 h 13 min). Estimated blood loss was less than 100 ml. Oxytocin was not administered in case it caused deterioration in the cardiovascular state of the patient, but was not required as the uterus contracted fully after delivery.
The patient was immediately transferred to the operating theatre to allow further stabilisation and delivery of care. At this point the patient was making occasional small, ineffective gasps, had a heart rate of 100 beats.min−1 with a palpable radial pulse and GCS 6/15 (eyes 4, movement 1, speech 1). Bag and mask ventilation was continued throughout this time. Due to unavailability of a trained assistant, cricoid pressure had not been applied at any time up to this point but despite this there was no further regurgitation or aspiration of gastric contents. No drugs had been administered. The pupils remained approximately 5 mm in diameter and equal.
A rapid sequence induction was performed with cricoid pressure after administration of thiopental 125 mg and suxamethonium 100 mg. There was a grade-2 view at laryngoscopy and the trachea was uneventfully intubated. Positive pressure ventilation was continued with a FIO2 of 1.0. Normal lung compliance was noted and the cyanosis quickly recovered. Isoflurane was started to maintain anaesthesia and 10 mg of morphine was given intravenously. The pupils remained 5 mm and equal. An arterial catheter was inserted to allow blood pressure monitoring and blood gas analysis. Surgery was re-commenced in a sterile manner to close the abdomen after the caesarean. There was less than 200 ml blood loss during this period and the uterus remained well contracted, so oxytocin was not administered. An inotropic infusion was not required, but boluses of metaraminol 500 μg were given to maintain a mean arterial pressure of 80 mmHg as a preventative measure following brain injury. A quantity of 3 mg was given in total over a 2-h period. A total of 1000 ml colloid and 1000 ml crystalloid had been given. Fundoscopy performed by a medical consultant intensivist at this time was normal.
We were unable to obtain an ECG trace in the labour ward room. The first ECG obtained in theatre on our monitor was sinus rhythm with normal QRS complexes. A 12-lead ECG in theatre showed sinus rhythm with a prolonged QT interval of 466 ms.
Arterial blood gas analysis post-intubation with a FIO2 of 1.0 showed pH 7.28, PaCO2 4.9 kPa, PaO2 66.0 kPa, bicarbonate 16.8 mmol.l−1, base excess −9.2, haemoglobin 11.4 g.dl−1, Na+ 134 mmol.l−1, K+ 3.2 mmol.l−1, glucose 8.1 mmol.l−1 and lactate 6.0 mmol.l−1. In view of her clinical state and rapid improvement, bicarbonate and potassium were not administered . Results 30 min later with FIO2 of 0.8 were as follows: pH 7.40, PaCO2 4.6 kPa, PaO2 48.4 kPa, bicarbonate 20.6 mmol.l−1, base excess −4.3.
The remifentanil PCA pump was checked. Following the first five observed doses, only three further doses (120 μg) had been administered over the next 15 min. The correct volume of drug remained in the syringe and the settings were found to be correct.
A propofol infusion was started to allow transfer to ICU. Magnesium boluses and an infusion were given in case the cardiac arrest was a presenting sign of pre-eclampsia. The patient's cardiovascular and respiratory function remained stable. A reduction in sedation at this point produced eye opening to voice and spontaneous respiratory effort. Sedation was re-commenced to allow safe transfer. A brain CT and CT pulmonary angiogram (CTPA) were performed approximately 1 h post-arrest in order to exclude intracranial pathology or an embolus. Both were normal.
After 12 h the sedation was discontinued and the patient's trachea was extubated uneventfully. She was discharged to the coronary care unit (CCU) the following day. The patient had no neurological deficit on leaving ICU, but no memory of the events that had occurred. The patient remained well on CCU with no further problems. Postoperatively, all ECG recordings were normal with a normal QT interval. A troponin level taken at the time of arrest was not raised. She was discharged home four days post-arrest. An outpatient echocardiogram four weeks later was normal.
The patient was reviewed in the obstetric anaesthetic clinic four weeks after discharge from hospital. She had made a complete physical recovery and had no recall of events after starting the remifentanil PCA and before waking up in ICU.
Safe use of remifentanil PCA for labour analgesia has been well described . More than 50 patients had previously received a remifentanil PCA on our unit without problems. However, this report of cardio-respiratory arrest is more serious than a recent case that involved respiratory arrest alone .
The exact reason for the arrest remains unclear. At the time an embolic episode was considered likely, but a CTPA was normal and the patient did not develop a coagulopathy. No cardiac or other cause was identified and recovery was rapid and complete.
A primary respiratory arrest associated with remifentanil was the presumptive diagnosis. The patient did not have pinpoint pupils at any time during the arrest. The witnessed five remifentanil boluses did not cause a reduction in oxygen saturations or apparent respiratory depression. It seems improbable that three further boluses alone, over 15 min, could lead to the events which occurred; however, it should be noted that the patient had received two other opioids: codeine 60 mg orally 4 h and diamorphine 10 mg intramuscularly 2 h before the event. Diamorphine has a short onset time and half-life when administered by intramuscular injection, but the half-lives of its active metabolites (6 mono-acetyl morphine, morphine, morphine-6 glucuronide) are such that it is likely that our patient had considerable plasma levels of active opioids other than remifentanil at the time of the incident. Remifentanil has a greater potential to cause respiratory depression than other opioids  and remifentanil blood levels may have reached a peak concentration at the time of the arrest to maximise the respiratory depressive effects. The remifentanil ampoules were checked, confirming that the correct dose had been put into the syringe. The syringe driver settings were correct and the correct volume of drug had been administered corresponding to the number of successful demands. No extra drug had been given. The giving set contained an anti-siphon valve and it was administered via a dedicated cannula. There did not therefore appear to be a problem in excess drug administration or equipment failure. We cannot rule out, however, that the patient had positioned her arm resulting in the occlusion of the vein distally to the cannula and so movement caused a subsequent large remifentanil bolus into the central circulation, as has been postulated in other cases . We also do not have records to show if blood pressure was being measured on the same arm as the remifentanil cannula, that might also have the potential to delay drug delivery and lead to a large central bolus.
There was no evidence of a primary cardiac problem causing the cardiac arrest. The QT prolongation found on ECG could have been a result of the low potassium at the time of the arrest . Potassium levels returned to normal 24 h after the arrest and subsequent ECG recordings were normal.
At the time of the arrest, the patient was lying supine without a wedge. Aortocaval compression and/or vasovagal syncope could have contributed to the arrest. A wedge was not immediately available and chest compressions were performed in the supine position. Cardiac output was restored and the baby delivered before a wedge was obtained. We have since updated our maternity resuscitation trolleys so that each one now contains a wedge. Unfortunately, in the heat of the moment, manual displacement of the uterus was not performed either although there were insufficient free hands available to provide this effectively. Had manual displacement been performed, it might have reduced the need for an emergency caesarean. A weak pulse was palpated as the peri-mortem caesarean was started, but to maximise cardiac function and ventilation, surgical delivery was continued.
We have discontinued routine use of remifentanil PCA on our unit other than in exceptional circumstances, as it became clear that our midwives are unable to provide the level of one-to-one care that is required. Indeed, it became clear during the investigation of this incident that our midwives' interpretation of ‘one-to-one’ care and our own as anaesthetists differed significantly, and that midwives felt that considerable periods of absence from a room was permissible during ‘one-to-one’ care. As with the recently published case , the midwife was not present in the room with the patient at the time of the arrest. We have changed our guidelines to emphasise that remifentanil should not be used in opioid-naive women who have recently received other opioids, and that when a woman is using a remifentanil PCA, a trained member of staff must be continuously present in the room. Since making these changes, the use of remifentanil in our unit has declined considerably, but there have been no further adverse incidents.
It has become apparent that our midwives are not adequately trained in the use and interpretation of pulse oximeter and the recording of respiratory rate. Further midwifery training is required in our unit before we are able to provide safe and appropriate patient care with a remifentanil PCA.
One potential advantage of remifentanil PCA over epidural analgesia has been the relative lack of dependence on medical staff to initiate the treatment . If the midwife had been able to start the PCA without a medical review, this case would probably not have occurred: there would not have been a delay in setting up the remifentanil PCA due to lack of availability of the anaesthetist, and so alternative opioids would not have been given.
Published with the written consent of the patient.
No external funding and no competing interests declared.