Death by regional block: can the analgesic benefits ever outweigh the risks?


  • M. Fredrickson,

  • W. Harrop-Griffiths

This issue of Anaesthesia contains a case report by Yanovski et al. of a patient’s death after local anaesthetic injection through a catheter that was intended to have been placed adjacent to the interscalene brachial plexus [1]. Postmortem computerised tomography showed the catheter to be in an intrathecal position. The patient was a healthy 20-year-old undergoing surgery that was neither urgent nor life-threatening. Yanovski et al. deserve praise for reporting this case, which in turn deserves careful consideration in an attempt to answer the question of whether continuous interscalene analgesia should be abandoned because of the addition of the worst possible adverse outcome to its list of known complications.

Many assert that the introduction of the interscalene block by Alon P Winnie in 1970 revolutionised pain management after shoulder surgery, particularly for its most painful manifestation – rotator cuff repair [2, 3]. The subsequent development of brachial plexus catheterisation allowed further prolongation of the beneficial effects of this block: randomised studies have confirmed that continuous interscalene block significantly improves pain control after major shoulder procedures for up to three days [3]. Before the introduction of this technique, opioid administration for several days was not uncommon after shoulder surgery, and with that opioid requirement came the inevitable side effects of sedation, nausea, vomiting and pruritus, and the consequent requirement for ongoing hospitalisation. Large, prospective studies now totalling >2000 patients have confirmed that in experienced hands, the technique has an excellent safety profile [4–6]. However, just like general anaesthesia and opioid analgesia [7], if performed incorrectly, both single-injection and continuous interscalene block have the potential to pose a significant risk to patient safety.

Only a year after Winnie’s description, bilateral epidural block was reported [8], and two years after that, total spinal anaesthesia [9] – all fortunately without lasting adverse effects. In 1980, Barratell [10] reported the first permanent upper limb neurological deficit, presumably related to injection directly into a nerve, nerve root [11, 12] or the spinal cord, and since then there have been at least four further cases of permanent neurological deficit probably related to this mechanism [13, 14]. One of the reports describing four cases of permanent spinal cord damage [13] led in part to a recommendation by the American Society of Regional Anesthesia (ASRA) that interscalene block be placed in a separate category of peripheral nerve blocks, and decreed that these blocks: “should not be performed in anesthetized or heavily sedated adult or pediatric patients”. The assumption was that the awake or lightly sedated patient would report the pain and paraesthesia that may accompany needle entry and injection into neural tissue [15]. The statement that interscalene blocks are potentially more dangerous than other peripheral nerve blocks is supported by Auroy et al.’s data on the incidence of complications from >250 000 nerve blocks studied prospectively: the overall incidence of nerve damage was around 1:5000; for interscalene blocks it appears to be about 1:3500 [16, 17]. Is this block truly more dangerous than all others, and if so, why?

All blocks carry with them the generic risks of regional anaesthesia: failure; nerve damage; local anaesthetic toxicity; bleeding; infection; and damage to surrounding structures. The anatomical proximity of the brachial plexus roots to critical structures makes accidental injection into the neuraxis, a vessel or pleural puncture very possible, and many anaesthetists rightly regard the anterior triangle of the neck as ‘tiger country’. There has also been speculation that a relatively high incidence of abnormal nerve function after interscalene block (up to 15% within the first 10 days in one study [18]) results from the high ratio of neural to connective tissue in the cervical nerve roots [19], and their relative immobility as they emerge from the spine. Nerve root immobility increases the likelihood of impalement by the needle, while the relative lack of nerve root connective tissue may increase the risk of neurological dysfunction if intraneural injection occurs. Admittedly, this relatively high incidence may in part be related to the fact that patients are more likely to identify abnormal sensation in the hands than the feet. Of greatest concern is the potential threat to the neuraxis. Alon Winnie’s original description of the block recommended a needle orientation aimed directly towards the neuraxis: “perpendicular to the skin in all planes, that is, mesiad, dorsad, and slightly caudad” [2]. Further, he advised that if it was not possible to elicit paraesthesia, the needle should be advanced until a transverse process is contacted, and then ‘walked’ along the process until paraesthesia is evoked [2]. Research using magnetic resonance imaging and cadavers has suggested that the needle angulation originally described by Winnie is such that if the needle is inserted too far, entry through an intevertebral foramen and into the spinal canal becomes possible [20, 21]. How far is too far?

Of the 20 published reports (in 24 patients) of total spinal, epidural or permanent neurological deficit following single-injection interscalene block (see Appendix S1; online supporting information), excessive needle depth can be implicated in all on the basis of needle contact with the transverse process, a needle depth ≥2.5 cm, contralateral paraesthesia, or failure to report needle depth. Studies have confirmed that in the vast majority of patients, an insertion depth >2.0 cm is unnecessary when using the Winnie approach, which gives the needle the trajectory that has the shortest distance between the skin and the brachial plexus. Another key safety factor for neuraxial complications is the needle insertion angle. In 1997, Meier et al. [22] described a variation of the Winnie technique specifically to facilitate interscalene catheter placement. The approach has been popularised by Borgeat et al. [4], who termed it the ‘modified lateral’ approach, while others have used the term ‘anterolateral’ [23]. A key safety feature of this approach is not only a needle orientation more in line with the direction of the plexus, but also – and importantly – not orientated medially, i.e. the needle trajectory is more parallel to the neuraxis and may even be directed away from it. A more caudally directed needle means that a needle so inserted is much less likely to head towards the neuraxis, and even if encroaching on the vertebral column, is less likely to pass through an intervertebral foramen into the spinal canal [20, 21].

As noted above, placement of interscalene catheters to provide continuous regional analgesia has become popular in many centres. However, plexus catheterisation is not without its dangers, particularly if performed incorrectly. Of the five published reports of neuraxial complications resulting from catheterisation [24–28], four involved catheter placement using a Winnie-type medial needle angulation (or the approach was not specifically described), while one report involved catheter advancement >5 cm beyond the needle tip. The optimal distance to which a catheter should be passed beyond the needle’s tip remains a controversial issue. However, concerns exist that excessive catheter advancement might increase the likelihood that the catheter may pass into the epidural space, through a dural cuff into the subarachnoid space, or at the very least towards a less desirable anatomical location. Consequently, most experienced clinicians insert catheters no more than 3–5 cm beyond the tip of the introducing needle. Catheter migration is a concept that is commonly invoked when a catheter is found in a location that was not intended. We do not believe in the concept of catheter migration in this context and would contend that there are no reports of adverse outcomes that can definitively be ascribed to movement of a catheter once placed in a patient and secured. We believe catheter migration to be synonymous with incorrect catheter placement.

When inserting interscalene catheters, many anaesthetists inject local anaesthetic through the needle, rather than through the catheter, to maximise the chances of primary block success. Maximising primary block success may be relevant when the block is to be used for surgical anaesthesia, and some also argue for this approach on the basis that it is most likely to ensure effective analgesia during the first 12–18 h after surgery when pain is at its most intense. However, another common approach when surgery is performed under general anaesthesia is to inject the first local anaesthetic bolus through the catheter rather than the needle. This provides valuable confirmation of satisfactory catheter placement when the patient awakes in recovery pain-free. It is also common and intuitively correct practice to inject the first local anaesthetic bolus through the catheter when the patient is in a closely monitored environment, so that accidental intravascular or neuraxial injection can be readily identified and rapidly treated.

Ultrasound guidance for block performance and catheter placement is becoming popular in developed countries. One of the main advantages of this technology when compared with neurostimulation-assisted landmark techniques is that it provides useful information regarding appropriate needle depth [29, 30]. However, like the paraesthesia technique and neurostimulation, it too is limited by false negatives: observation of the nerve root and needle tip is not always possible, so the same principles for needle depth and orientation described above also apply for ultrasound-guided techniques. Similarly, observing the catheter (or catheter injectate) can be technically challenging. However, it is encouraging that there has as yet been no neuraxial complication reported that has resulted from an interscalene block performed under ultrasound guidance, but this should not lead to complacency.

We now turn to the details of Yanovski et al.’s tragic case. We note the following:

  •  The approach was not ‘lateral’ but medial, as described by Winnie [2].
  •  Seven centimetres of catheter were threaded beyond this medially-orientated needle.
  •  The first local anaesthetic bolus was injected down the needle before catheter insertion, which resulted in the patient’s emerging from general anaesthesia pain-free and with an ipsilateral and unilateral dense brachial plexus block, confirming that the needle itself was extrathecal at the time of injection.
  •  A catheter infusion was not started, and the first catheter bolus injection (10 ml bupivacaine 0.25%) was performed at night on an orthopaedic ward; the resident who injected the local anaesthetic left ‘after a short stay’.

Janovski et al. responded appropriately to these events by reviewing their practices and introducing protocols aimed at maximising the safe use of peripheral nerve catheters for postoperative analgesia. We would also add these further comments.

The interscalene brachial plexus block is, by its very nature, a block performed close to the neuraxis, and therefore a high index of suspicion should be maintained about the possibility of injection into the neuraxis. The local anaesthetic drugs injected through the needle or catheter can on occasion gain access to the epidural or subarachnoid spaces, and both single-shot blocks and the first injection down the catheter (or the start of an infusion) should therefore only be performed in an environment that allows the ready identification and rapid management of any resulting complications.

The title of this editorial poses a question that we have not yet answered. It will come as no surprise to those who know us to be regional analgesia enthusiasts to hear that our answer is a unanimous and perhaps ever-so-slightly raucous “yes!”, and that we would readily point to the fact that no analgesic regimen with any drug or drug combination by any route is without significant complications. In particular, we would argue that the leading competitor to continuous regional analgesia – intravenous patient-controlled opioid analgesia – is likely to have been responsible for scores or perhaps even hundreds of deaths since its introduction [7]. Further, we believe that scrupulous attention to detail in the performance of the block and catheterisation, and careful postoperative management of the catheter infusion, will allow the provision of excellent analgesia at an acceptably small, but never zero, incidence of serious complications.


The authors thank Professor Alain Borgeat for his translation and review of the cited French and German language case reports. No external funding or competing interests declared.