Spinal haematoma following epidural anaesthesia in a patient with eclampsia


  • T. S. T. Yuen,

    1. Medical Officer, Department of Anaesthesia, Pamela Youde Nethersole Eastern Hospital, 3 Lok Man Road, Chai Wan, Hong Kong Special Administrative Region, China,
    Search for more papers by this author
  • J. S. W. Kua,

    1. Senior Medical Officer, Department of Anaesthesia, Pamela Youde Nethersole Eastern Hospital, 3 Lok Man Road, Chai Wan, Hong Kong Special Administrative Region, China,
    Search for more papers by this author
  • I. K. S. Tan

    1. Consultant and Director of Intensive Care, Department of Anaesthesia, Pamela Youde Nethersole Eastern Hospital, 3 Lok Man Road, Chai Wan, Hong Kong Special Administrative Region, China
    Search for more papers by this author

J. S. W. Kua Senior Medical Officer, Department of Anaesthesia, Pamela Youde Nethersole Eastern Hospital, 3 Lok Man Road, Chai Wan, Hong Kong Special Administrative Region, China


A patient with a twin pregnancy required a Caesarean section for severe pre-eclampsia. Her platelet count was 71 × 109.l−1. Epidural anaesthesia was performed after platelet transfusion. A spinal epidural haematoma was diagnosed postoperatively. A generalised tonic–clonic seizure sparing the lower limbs enabled early diagnosis to be made. The patient recovered with no permanent neurological damage after laminectomy and clot removal. The risks and benefits of regional techniques require careful consideration, and postoperative monitoring for recovery of neural blockade is essential.

Spinal epidural haematoma after regional anaesthesia is rare. The estimated incidence is 1 in 150 000 following epidural anaesthesia and 1 in 220 000 after spinal anaesthesia [1]. A retrospective study of 505 000 extradural blocks found only one case of spinal haematoma [2]. Dahlgren & Tornebrandt reported three cases of paraplegia caused by spinal haematoma in 9232 patients with deranged haemostasis undergoing epidural anaesthesia [3].

The true incidence of spinal haematoma following central neural blockade in obstetric patients is unknown, and may well be higher than the estimated incidence due to under-reporting. To our knowledge, only one case of haematoma has been reported in a patient with severe pre-eclampsia [4]. We describe a patient who presented with severe pre-eclampsia and developed spinal epidural haematoma after epidural anaesthesia for Caesarean section.

Case history

A 20-year-old Chinese woman, gravida 1 para 0, at 35 weeks' gestation with twin pregnancy was admitted from the antenatal clinic with hypertension (155/95 mmHg) and albuminuria (3+ with reagent strip).

On admission, she complained of a decrease in fetal movement, but no headaches or visual disturbances. There was no history of easy bruising, gum bleeding, previous hypertension, diabetes or neurological disease. Her blood pressure (BP) was 145/102 mmHg. There was gross oedema of her lower limbs but no signs of left ventricular failure. Reflexes were brisk with no clonus. The fundi were normal.

The BP reverted to normal without treatment but albuminuria persisted. On the third day, she complained of headache, nausea, vomiting and mild epigastric pain. Her BP was 177/108 mmHg. She was transferred to the labour ward for further management pending Caesarean section. Blood investigations revealed a haemoglobin of 13.7 g.dl−1, raised urate level of 0.42 mmol.l−1 and normal renal function. The platelet count was 71 × 109.l−1 as compared with 49 × 109.l−1 on the day of admission. No platelet transfusions had been given. The international normalised ratio was less than 1.0, prothrombin time was 9 s and activated partial thromboplastin time was 34.9 s. She had a mildly elevated alanine transaminase of 46 IU.l−1, a raised alkaline phosphotase of 761 IU.l−1, an albumin of 25 g.l−1 and a total bilirubin of 7 mmol.l−1.

The risks and benefits of alternative anaesthetic approaches were discussed with the patient and her spouse, including the risk of spinal haematoma with epidural anaesthesia. The patient chose epidural anaesthesia. Sodium citrate and ranitidine were given as premedication, and six units of platelet concentrate were transfused.

Thirty minutes later, epidural catheter placement was attempted with an 18G Tuohy needle at L2/3 in the midline. The first attempt failed as there was no definite loss of resistance to air and bone was encountered instead. A total of 5 ml of air was injected during the procedure. The Tuohy needle was withdrawn to the subcutaneous layer and redirected. The epidural space was entered this time with loss of resistance to 3 ml of air. Neither attempt was traumatic. After a negative aspiration test, a sensory block to T4 was established with a total of 13 ml 0.5% plain bupivacaine injected in aliquots through the epidural catheter. No haemodynamic disturbances occurred during epidural anaesthesia for the Caesarean section, which lasted 40 min. Both twins were delivered successfully. Peri-operatively, the patient complained of shivering which responded to 10 mg of intravenous pethidine. Haemostasis was adequate at the operative site and the epidural catheter was removed at the end of the operation. She was transferred to the obstetric high-dependency unit (HDU) for close monitoring. A patient-controlled morphine infusion was prescribed for postoperative pain. Her BP remained normal and there was no excessive bleeding from the surgical wound.

The sensory level of the epidural block was monitored and was above the T10 dermatome on arrival in the obstetric HDU. It did not regress below the T10 dermatome during the next hour. One hour later, she developed hypertension and an unusual tonic–clonic seizure that was generalised but spared the lower limbs, which were found to have a motor and sensory (pinprick and temperature) deficit from the L1 dermatome downwards. Upper limb jerks were brisk but knee and ankle jerks were absent. A similar tonic–clonic seizure occurred 1 h later. The neurological deficits of the lower limbs noted earlier persisted.

Computed tomography (CT) scans of the head and lumbar spine were immediately arranged. There were multiple air bubbles at the basal cistern, sulci and parietofrontal region within the skull (Fig. 1). A 2 × 1 cm haematoma at the lumbar region compressing the spinal cord and nerve roots was found along with multiple air bubbles (Fig. 2). Laminectomy and clot evacuation were performed immediately under general anaesthesia avoiding the use of nitrous oxide. At operation, a 4 ml clot extending from T12 to L2 was removed.

Figure 1.

CT scan of brain showing multiple gas bubbles.

Figure 2.

CT scan of lumbar spine showing the haematoma compressing the spinal cord at L1 level. Multiple air bubbles are also evident.

Postoperatively, the patient was managed in the intensive care unit. She was given methylprednisolone, phenytoin, ceftriaxone and magnesium sulphate. The platelet count before laminectomy was 46 × 109.l−1. A total of 20 units of platelet concentrate were given peri-operatively, which increased the count to 98 × 109.l−1. Over the next two days, another nine units of platelet concentrate were transfused before a count of > 100 × 109.l−1 could be achieved. The patient's BP became normal, and lower limb power and sensation gradually improved. She was discharged home nine days later, walking unaided. At follow-up three months later, she had no neurological deficits or sphincter dysfunction.


Spinal haematoma may be spontaneous or idiopathic, or caused by trauma, spinal tumour, vascular anomalies or bleeding disorders [5]. Deranged coagulation as a cause represents a diverse group of patients. These include patients treated with anticoagulants, patients with congenital disorders of haemostasis, conditions such as pre-eclampsia, abruptio placentae, retained dead fetus (> 4 weeks), amniotic fluid embolism and intra-amniotic hypertonic saline instillation [1]. Apart from pre-eclampsia, other causes of thrombocytopaenia include disseminated intravascular coagulation, immune thrombocytopaenic purpura and collagen vascular disorders [6].

The choice of anaesthesia in pre-eclamptic parturients is controversial. The disadvantages of general anaesthesia have been reviewed extensively [7, 8]. Although various pharmacological regimens are available [9] to attenuate the pressor response to intubation, none is ideal. Intubation may be difficult if the airway is oedematous but awake fibreoptic-assisted tracheal intubation can be stressful. A hypertensive crisis during general anaesthesia may cause pulmonary oedema or intracranial haemorrhage. Rapid progression of laryngeal oedema intra-operatively with airway obstruction occurring immediately after extubation has been reported [10]. Other hazards of general anaesthesia include aspiration of gastric contents, weakness due to synergistic effects of magnesium sulphate with neuromuscular blockade and impaired intervillous blood supply [8].

Many anaesthetists prefer to offer epidural analgesia for labour and epidural anaesthesia for operative delivery, provided there are no contraindications. As well as avoiding the problems associated with general anaesthesia, a block extending up to the T4 dermatome blunts the haemodynamic and neuroendocrine stress responses to Caesarean delivery in pre-eclamptic women [11] and may improve intervillous blood flow if hypotension is avoided [12]. However, hypotension induced by sympathetic blockade requires the use of pressor agents or volume load, which can lead to pulmonary oedema.

The lowest acceptable platelet level for safe epidurals is uncertain. Bromage [13] recommended a minimum platelet count of 100 × 109.l−1 without clear and convincing evidence, and this seems to be widely accepted. Recent retrospective reviews of epidurals performed in parturients with platelet counts of less than 100 × 109.l−1 did not reveal any neurological complications [6, 14, 15]. A 1996 survey of anaesthesiologists in the USA suggested that many are increasingly tolerant of low platelet counts [16]. A patient with profound thrombocytopenia (2–4 × 109.l−1) who uneventfully received an epidural anaesthetic for labour has been reported [17]. Platelet dysfunction can occur with pre-eclampsia. However, assessment of platelet function using bleeding time is not useful clinically as a routine test [14, 18, 19]. Thromboelastography showed promising correlation with platelet count and a calculated platelet count of 75 × 109.l−1 should be associated with adequate haemostasis [19]. The technique, although becoming more readily available, is not routinely used. Consensus generally favours obtaining a platelet count as the initial test, with further investigation if the result is abnormal [20].

The problems of pre-eclampsia can extend into the postpartum period and close monitoring after delivery is mandatory. In this case, the sparing of the lower limbs during the seizures could be attributed to a pharmacological paraplegia caused by the residual effects of epidural bupivacaine modifying an eclamptic fit. However, a spinal haematoma could also produce paraplegia, is far more serious and requires urgent investigation and exclusion. The classical presentation of spinal haematoma with backache, hypoalgesia and muscle weakness was not apparent in this case, although this is not uncommon [4]. The CT scan of the lumbar spine showed the haematoma at L1/2 level, which corresponded to the level of the epidural catheter insertion.

A CT scan of the brain was carried out to rule out intracranial pathology. A case of unsuspected intracranial pathology that presented with intrapartum seizure was reported by Cheng & Kwan [21] as causing difficulty in diagnosis. Pneumoencephalus may occur after epidural block [22[23]–24], due to inadvertent injection of air into the subarachnoid or subdural space when air is used for loss of resistance. In keeping with other reports, no free cerebrospinal fluid leakage was seen in our patient. The epidural procedure can be made difficult by the oedema associated with pre-eclampsia, as was reflected by the failed first attempt. Unsuspected dural puncture may have occurred. The signs and symptoms associated with pneumoencephalus may be due to irritation of meninges after cephalic migration of air, and include headache, delayed recovery, confusion, convulsions and hemiparesis [25]. The differential diagnosis of the seizures in this patient could be postpartum eclampsia or pneumoencephalus. Unless worsened by nitrous oxide, there are usually no neurological sequelae with pneumocephalus [26]. We therefore managed the patient intra-operatively without nitrous oxide, and as for eclampsia. The use of saline instead of air for loss of resistance during epidural blocks is likely to reduce the occurrence of pneumoencephalus.

Removal of the epidural catheter can be traumatic and may result in spinal haematoma formation in patients with a defective coagulation profile. The lowest acceptable platelet level for safe removal is also uncertain. Since prior platelet transfusion had been given and the haemostasis of the surgical field appeared normal throughout the operation, the catheter was taken out at the end of the operation. A review suggested waiting until the coagulation status has been corrected [16]. The low platelet count after the Caesarean delivery would have prompted us to postpone removal of the catheter.

A diagnosis of spinal epidural haematoma requires urgent surgical intervention. The earlier the laminectomy, the better the prognosis [27]. Beyond 6–12 h, the chance of complete recovery is remote [4]. Individual cases with complete spontaneous recovery [3, 28] are the exception rather than the rule. In our patient, the neurological deficits of the lower limbs persisted beyond 4 h, so that residual epidural local anaesthetic effect was unlikely to be the cause. In this case, laminectomy was performed 8 h after the epidural anaesthesia and long-term neurological deficit was avoided.

The volume of blood clot removed at surgery was only 4 ml, whereas the amount of autologous blood used for an epidural blood patch is in the range of 20–25 ml. It is interesting that this small amount of blood could be responsible for cord compression and paraplegia. We speculate that a slow bleed in our case confined the haematoma in the epidural space, causing pressure effects on the spinal cord and nerves. In contrast, the immediate spread of injected blood both in the epidural space and intervertebral foramina during an epidural blood patch may relieve the pressure effects.

In summary, isolated reports of uncomplicated epidural catheter placement in patients with thrombocytopenia should be viewed with caution. The risks and benefits of epidural vs. general anaesthesia must be weighed carefully and explained to the patient. No single safe minimal value of platelet count can be recommended for the insertion or removal of an epidural catheter. A post-transfusion platelet count should also be obtained to guide the decision for an epidural block. As much care should be given during catheter removal as during placement. Monitoring the recovery from central neural blockade after an epidural block is essential. The use of saline instead of air for loss of resistance as part of the epidural technique is recommended.


We thank Dr K. H. Fung, Consultant Radiologist of the Pamela Youde Nethersole Eastern Hospital, for the radiological support.