We describe the management of a 62-year-old man who developed severe pain, cramps, paraplegia and pulmonary oedema after the accidental administration of potassium chloride into the subarachnoid space. In addition to supportive treatment, we performed cerebrospinal fluid lavage with saline 0.9%. The patient recovered well without any permanent injury.
A 62-year-old man weighing 75 kg was scheduled for right total hip replacement. His medical history included hypertension, although he was not being treated for this; he had undergone spinal surgery (L5 laminectomy with posterolateral L4-S1 fusion) one year previously. We decided to use spinal anaesthesia. Lumbar puncture was performed with the patient in the left lateral position, at the L3-L4 space with a 25-G Quincke-type spinal needle, without complications. When the anaesthetist performing the block needed to fill the syringe with local anaesthetic, his assistant mistakenly passed him an ampoule containing 10 ml potassium chloride 7.45% (B Braun Medical Ltd, Queluz de Baixo, Portugal) instead of a 10 ml vial of isobaric 0.5% bupivacaine (B Braun Medical Ltd). Two millilitres of the contents of the ampoule were aspirated into the syringe and then injected into the subarachnoid space, with 20 μg fentanyl. Ten minutes after injection, there was only patchy blockade, the loss of sensation being inadequate for the planned surgery. In view of the patient's previous spinal surgery, this did not surprise the anaesthetist, who then repeated the lumbar puncture and administered the same drug dosage once more. Immediately after the second injection, the patient started to complain of severe pain, pruritus, cramps, progressive weakness in both lower limbs (left more than right) and marked sweating. There was complete anaesthesia below the T10 level. His blood pressure had increased to 190/100 mmHg and his heart rate to 120 beats.min−1. As the block distribution suggested a hyperbaric rather than an isobaric solution, and the clinical picture was so very different from what we expected, we reviewed all the medications administered. This revealed the error. We started antihypertensive treatment with labetalol, but the patient remained hypertensive (160/80 mmHg), tachycardic (100 beats.min−1), agitated and sweaty. Ten minutes after the second injection, the sensory level had risen to T4, there was flaccid paraplegia with loss of deep tendon reflexes, and pain in both lower limbs. We then began cerebrospinal fluid lavage with the patient in the left lateral position and twenty degrees head up tilt, using a 22-G Quincke-type spinal needle. Over a period of five minutes, we aspirated a total of 50 ml cerebrospinal fluid, which we replaced, over a further five-minute period, with the same volume of saline 0.9%. Immediately after the lavage, the patient developed signs and symptoms of pulmonary oedema; this required sedation, tracheal intubation and mechanical ventilation. However, before sedation, we tested the patient once more and found persistent paraplegia, but regression of the sensory level to T10. One hour later, we repeated the lavage, this time withdrawing 20 ml cerebrospinal fluid over one minute and injecting the same amount of saline 0.9% over the same period of time. The patient was transferred to the intensive care unit. Eight hours after the lavage, the patient was haemodynamically stable, with normal arterial blood gas values, and starting to move his legs, so we extubated his trachea. At this stage, we found that cutaneous sensation had returned to normal. There were no further neurological deficits, and the signs of autonomic hyperactivity had disappeared. A magnetic resonance scan of both thoracolumbar spine and brain was normal. The patient underwent the planned hip replacement operation two months later, under spinal anaesthesia, without complications.
Medication errors are common in anaesthetic practice, in regional as well as in general anaesthesia . Recently reported examples have involved administration of the wrong drug  and the wrong arterial line flush solution  and the accidental flushing of residual drugs through intravenous cannulae (a particular problem in children) . Other substances that have accidentally entered the neuraxis are chlorhexidine  and insulin .
In this case, the use of non-sterile plastic ampoules of isobaric bupivacaine 0.5% obliged the anaesthesiologist to draw up the drug held by the assistant to maintain aseptic technique. Although pre-operative assessment is a vital part of peri-operative care , in this case, the knowledge gained about the patient's spinal surgery  provided a plausible but misleading explanation for the patchy initial block, leading to the administration of a second dose of spinal potassium chloride. Nor would other commonly advocated safeguards have been useful in this case. For instance, although much has been made in the UK of the move to non-Luer connectors for neuroaxial procedures, in this particular case the use of a non-Luer syringe would not have prevented the drawing-up of the wrong drug from the vial . Furthermore, designating some specific serious untoward incidents as ‘never events’ may also not be effective. In the UK, the mis-administration of potassium chloride intravenously is one such ‘never event’ , but mis-administration intrathecally is not mentioned, although it is clearly also dangerous.
In the past, potassium chloride has often been confused with saline 0.9% because the two solutions were presented in similar ampoules distinguished only by different coloured writing [11, 12]. These reports and the data collected on clinical errors have prompted recommendations for safe practice from organisations in the UK , USA and Canada. In these countries, potassium chloride has been removed from ward stock and kept in a separate area.
In Portugal, our supplier produces isobaric bupivacaine and potassium chloride in similar ampoules (Fig 1). However, hospitals have not developed local policies regarding the storage of potassium chloride. We hope that publishing this case report will help prevent recurrence of this type of incident . Furthermore, although there is currently no national incident reporting scheme in Portugal , we did, however, report our incident within the hospital and to B Braun Medical Ltd.
Accidental epidural and subarachnoid accidental injections of potassium chloride, leading to temporary or permanent deficits and death have previously been reported [11, 12, 16-20]. Epidural injection may be more common because isobaric local anaesthetics are more frequently used in the epidural space rather than in the subarachnoid space. One study has correlated different epidural potassium chloride concentrations with temporary or permanent sensory and motor deficits , but not when potassium chloride has been injected into the subarachnoid space. Meel et al. reported the only case in the literature of subarachnoid potassium chloride administration, which resulted in the patient's death 2.5 h after the injection . Our patient received a total of 4 mmol potassium chloride intrathecally, which caused pain and pruritus secondary to the hyperosmolarity of the solution's irritating the cell membrane [20, 21]. Also, the neurological damage might be caused by the high extracellular concentration of potassium in the spinal canal . The autonomic dysfunction (tachycardia and hypertension) is thought to be related to the cephalad diffusion of potassium in the cerebrospinal fluid causing sympathetic stimulation, because potassium potentiates the depolarisation of sympathetic neurons [11, 12, 19, 20]. The pulmonary oedema was probably caused by histamine release [11, 22].
Although Lin et al.  reported that the use of saline 0.9% in a case of accidental epidural injection of potassium raised the level of sensory blockade, other authors reported that saline injection into the epidural space accelerated the systemic absorption of pancuronium and vecuronium (‘volume effect’) [23, 24]. In theory, cerebrospinal fluid lavage dilutes and removes the drug, thus limiting neural damage. This is well demonstrated with antineoplasic drugs and with ionic contrast medium [25-28]. For example, with accidental intrathecal vincristine administration, cerebrospinal fluid drainage and early irrigation has been related to a good outcome (prolonged survival with no encephalopathy) [25-28]. Also, cerebrospinal fluid lavage appears to be an effective treatment in cases of accidental intrathecal injection of local anaesthetics secondary to epidural catheter misplacements [29-31].
In our patient, the more severe neurological signs were found on the left side. This probably reflected greater potassium chloride concentration on the dependent side. We felt that we could aspirate a higher amount of potassium chloride using the left lateral position during cerebrospinal fluid removal. In cases of unintended intrathecal injection of ionic contrast media, most authors emphasise the importance of minimising the amount of agent reaching the brain by maintaining the patient in a head-up position . We thought this position would not only prevent the cephalad spread of potassium chloride but also contain the higher concentration of potassium chloride in the lower spinal segments. We still do not know what the maximum and/or safe volumes are for aspiration of cerebrospinal fluid and saline replacement. Despite the fact that in the lumbosacral region the cerebrospinal fluid volume is approximately 30 ml , we withdrew 50 ml cerebrospinal fluid (as reported by Kaiser et al. ). This was because the patient had symptoms explained by the rostral spread of potassium (tachycardia, hypertension and high sensory block). We needed to perform two spinal lavages because after the first one, the patient still had symptoms attributable to intrathecal potassium chloride. We used a 22-gauge needle for the lavage because it has less internal resistance, allowing easier cerebrospinal fluid aspiration than a thinner needle .
We substituted the same volume of saline 0.9% for cerebrospinal fluid to try to avoid ‘spinal headache’. We withdrew and replaced 20 ml after an hour, as we felt that this would be safe, given that the normal rate of cerebrospinal fluid production is approximately 20 ml.h−1 . Saline 0.9%, lactated Ringer's solution and Plasma-lyte ™ (Baxter Médico-Farmacêutica, Sintra, Portugal) have all been used for cerebrospinal fluid lavage and endoneurosurgery [27, 31, 36]. We used saline 0.9% because it does not contain potassium, unlike the others.
In retrospect, it would have been informative to send a cerebrospinal fluid sample for potassium assay to enable correlation between concentration and symptoms/signs, and relate this to the other reported cases .
To our knowledge, there are no previous reports of the successful treatment of accidental spinal potassium administration in the literature. The reversal of adverse effects in our patient raises the question whether the recovery in our patient resulted from the actual removal of potassium chloride or from the simple dilution of potassium chloride and/or saline instillation. This point has been discussed in relation to the treatment of intrathecal overdose of morphine or bupivacaine [31, 35]. We feel that the 70 ml cerebrospinal fluid lavage removed a large portion of the intrathecal potassium chloride, enabling full recovery, unlike the patient reported by Meel et al., who did not undergo spinal lavage .
We cannot draw definitive conclusions based on one patient, but the success of this case suggests that, in the case of accidental intrathecal administration of potassium, cerebrospinal fluid lavage may be beneficial. Several factors may be implicated in drug errors: similarity between ampoules; lack of attention; staff training; working conditions e.g. fatigue; and storage of similar ampoules in the same place . Thus, it is likely that measures are needed to improve care across a number of aspects. These are likely to involve the non-technical skills of anaesthetic practice [37, 38]: communication ; shared awareness of work and its problems amongst all members of the anaesthetic team ; and checking [41-43]. Specific design features such as: standardised vial and/or label colours; different ampoule design; and the separation of intravenous from regional anaesthetic drugs are also likely to help.
The authors thank Dr Paulo Costa, Dr Seabra Lopes and to the intensive care unit staff (Centro Hospitalar do Porto) for their support. The authors also thank Marie Rogers for her assistance with the English language used.
Published with the written consent of the patient.
No external funding and no competing interests declared.