BJA Advance Access published online on January 12, 2009
British Journal of Anaesthesia, doi:10.1093/bja/aen360
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Major complications of central neuraxial block: report on the Third National Audit Project of the Royal College of Anaesthetists
1 Department of Anaesthesia, Royal United Hospital, Combe Park, Bath, UK
2 Wrexham Maelor Hospital, Wrexham, UK
3 University of Dundee, Dundee, UK
* Corresponding author. E-mail: timcook007{at}googlemail.com
Accepted for publication November 19, 2008.
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
Background: Serious complications of central neuraxial block (CNB) are rare. Limited information on their incidence and impact impedes clinical decision-making and patient consent. The Royal College of Anaesthetists Third National Audit Project was designed to inform this situation.
Methods: A 2 week national census estimated the number of CNB procedures performed annually in the UK National Health Service. All major complications of CNBs performed over 1 yr (vertebral canal abscess or haematoma, meningitis, nerve injury, spinal cord ischaemia, fatal cardiovascular collapse, and wrong route errors) were reported. Each case was reviewed by an expert panel to assess causation, severity, and outcome. ‘Permanent’ injury was defined as symptoms persisting for more than 6 months. Efforts were made to validate denominator (procedures performed) and numerator (complications) data through national databases.
Results: The census phase produced a denominator of 707 455 CNB. Eighty-four major complications were reported, of which 52 met the inclusion criteria at the time they were reported. Data were interpreted ‘pessimistically’ and ‘optimistically’. ‘Pessimistically’ there were 30 permanent injuries and ‘optimistically’ 14. The incidence of permanent injury due to CNB (expressed per 100 000 cases) was ‘pessimistically’ 4.2 (95% confidence interval 2.9–6.1) and ‘optimistically’ 2.0 (1.1–3.3). ‘Pessimistically’ there were 13 deaths or paraplegias, ‘optimistically’ five. The incidence of paraplegia or death was ‘pessimistically’ 1.8 per 100 000 (1.0–3.1) and ‘optimistically’ 0.7 (0–1.6). Two-thirds of initially disabling injuries resolved fully.
Conclusions: The data are reassuring and suggest that CNB has a low incidence of major complications, many of which resolve within 6 months.
Keywords: anaesthetic techniques, epidural; anaesthetic techniques, subarachnoid; anaesthetic techniques, regional, caudal; audit; complications, death; complications, neurological
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
Central neuraxial block (CNB) techniques can produce highly effective pain relief for a wide variety of indications and may decrease patient morbidity after major surgery, although the extent of the latter benefit is not agreed universally.1 2 In recent years, both large randomized controlled trials (RCTs)3 and meta-analysis2 have led to conflicting conclusions and interpretations regarding the outcome benefit of CNB techniques. However, the risk–benefit analysis must also take into account both the rate of failure of the techniques, which may be higher than some accept,4 and the incidence of complications.5–7 In the past, these complications have been serious enough to turn the speciality away from the techniques almost entirely, particularly when reports of paraplegia after spinal anaesthesia in both the USA8 and the UK9 led to the near abandonment of CNB in the UK for more than two decades after the Second World War. With the techniques now used widely again, there are reports of, and commentaries on, major complications from both the UK and elsewhere.10–17 Most recently, Christie and McCabe18 reported a series from one hospital that, with a very high incidence of major sequelae, achieved some prominence.
Knowledge of the incidence of such complications should be an essential component of the clinical decision-making and consent processes, but there are few good data which can be quoted to support such discussions, leaving both patient and clinician in a quandary. Figures (ranging from 1:1000 to 1:100 000) are quoted, but their doubtful validity questions the ability to obtain genuinely informed consent from patients offered these procedures. Recognizing this, and that neither RCT nor meta-analysis is an appropriate method for identifying rare events, the Council of the Royal College of Anaesthetists devoted its third National Audit Project to this topic. The aim was a prospective attempt to identify both numerator (number of major complications) and denominator (number of CNB) information for a 12 month period by a review across the breadth of anaesthetic and pain management practice in the UK National Health Service (NHS). Follow-up (as far as an anonymous reporting system would allow) would extend to 6 months so that final outcome, and incidence, could be assessed and give some indication of the prognosis of such events.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
A two-part project was devised: first, an assessment of the number of CNBs performed annually in the UK NHS (for denominator information); and second, an audit of the major complications of these procedures performed during a 12 month period (for numerator information). Discussions with the Centre of Research Ethics Committees (now National Research Ethics Service) indicated that ethical approval was not required, and the processes involved were agreed with the Patient Information Advisory Group of the Department of Health. The project was advertised widely throughout 2006 and 2007 through direct contact with the relevant organizations in anaesthesia, pain management, neurology, spinal surgery, radiology, and neuroradiology (Appendix 1). The aims and processes of the project were explained and the information was cascaded down to the members of those organizations at regular intervals.
Denominator data
A detailed description of the first part, the ‘snapshot’ survey (census) to determine denominator information, has been published already,19 but a brief summary is appropriate here. Between March and September 2006, the anaesthetic department of each NHS hospital believed to be performing surgery was contacted, asked to participate, and to nominate a ‘local reporter’ (LR) to co-ordinate the project locally. Each LR was asked to collect information on the number of CNBs performed over a 2 week period at the end of September 2006 or an equivalent period at about that time. The blocks were classified as epidurals, spinals, combined spinal–epidurals (CSEs), and caudals for each of the five indications: adult perioperative, obstetric (both labour analgesia and operative delivery), chronic pain, paediatric perioperative, and administered by a non-anaesthetist. We did not request data on CNB that were attempted and failed as we considered it unlikely that all cases would be recorded reliably. For each category, the reporters indicated whether their data were ‘accurate’, a ‘close estimate’, or an ‘approximate estimate’. The mechanism of data collection was not specified and reminders to return information were sent at regular intervals by post, e-mail, and telephone as necessary. Data were summed to give cumulative totals for a nominal 2 week period and, based on the annual results of one large district general hospital (Royal United Hospital, Bath), these figures were then multiplied by 25 to give an approximation of annual activity.
Event reporting (numerator data)
The same LR system was used to identify complications of CNB, but direct reports from any clinician in all relevant specialities were promoted with the aim of ensuring complete capture of all possible cases. We accepted reports even if the attempted CNB was abandoned: as such, there is a potential to slightly overestimate the incidence of complications because we did not include these attempts in the denominator. The formal audit period was September 1, 2006, to August 31, 2007, inclusive, but reporting was actively encouraged until March 31, 2008, for the same reason. Information was sought on all major complications of CNB with the potential for serious patient harm, including infection, haematoma, nerve damage, and cardiovascular collapse (detailed in Table 1). In addition, because of current concern about wrong route errors (i.e. a drug intended for the epidural or subarachnoid space inadvertently administered i.v., or vice versa),20 reports on these events were encouraged even when no injury occurred.
|
Primary notification of an event was by e-mail, with reports accepted from any source. The project team was able to exclude obviously irrelevant cases at this stage, but otherwise the LR for the relevant hospital was asked to obtain the details and upload them to a secure, password-protected website (the National Confidential Acute Pain Critical Incident Audit, NCAPCIA, www.ncapcia.org.uk). The information requested depended on the type of incident, but the questions were designed to gain a full picture of the procedure and the presentation, severity, and consequences of the complication. The NCAPCIA administrator (D.C.) was able to access these reports and request updates as required, being the only person who knew their source: this was essential to allow requests for clarification and updates of information while maintaining confidentiality. Each case was reviewed in detail by a panel representing all the specialities involved in the project (Appendix 2), and the following details were confirmed:
- type of block and indication for its performance (as described above). Procedures performed for the control of non-operative acute pain (e.g. fractured ribs and pancreatitis) were included in the perioperative group;
- category of complication (Table 1);
- correctness of diagnosis;
- date of CNB within the audit period;
- CNB was performed in an NHS hospital;
- severity of patient outcome (see below), initially and at 6 months (or later where such information was available);
- causation: whether the CNB was the cause of the patient injury: certain, likely, possible, unlikely, and no link.
Severity of complications
Severity of initial and final harm was recorded in a variety of ways. First, it was categorized according to the National Patient Safety Agency (NPSA) severity of outcome scale for patient safety incidents (Table 2).21 Patient harm was graded as ‘temporary’ if the incident met the NPSA criteria for moderate injury, and ‘permanent’ if the outcome was worse than this (severe injury or death). Secondly, where injury was permanent, or assumed to be so, the features were classified as follows:
- sensory only;
- motor: motor weakness of whatever severity, with or without sensory symptoms;
- paraplegia: paraplegia or tetraplegia with or without additional motor or sensory symptoms;
- death: this was classified as ‘direct’ (e.g. a cervical abscess leading to tetraplegia, respiratory failure, and death) or ‘indirect’ when the CNB was followed by a series of other events leading to death (e.g. an abscess requiring decompression with good neurological recovery, but complicated by a fatal pulmonary embolism).
|
Permanent lessening of bodily functions, sensory, motor, physiologic, or intellectualInterpretation of reports
In a proportion of cases, LRs were not able to provide full details of cases and patient progress, and some information was incomplete in spite of follow-up requests. Therefore, the reports required some ‘interpretation’ by the review panel, which assumed the worst, unless there was evidence to refute it:
- Diagnosis: where this was uncertain, cases were included: only those with clear evidence of incorrect diagnosis were excluded.
- Causation and outcome: these were particularly difficult to judge in a number of cases, and this led to a decision to quote rates of complications in two ways, that is, in terms of both ‘worst’ and ‘best’ case scenarios, defined in the results as ‘pessimistic’ and ‘optimistic’ incidences. When causation was judged certain, likely, possible, or unlikely, cases were included in the ‘pessimistic’ analysis, but those judged as unlikely were excluded from the ‘optimistic’ analysis. Similarly, efforts were made to determine the patient outcome at 6 months after the CNB. Where outcome at 6 months (or later) was available, this was used in the final judgement, but if such outcome information was only available from an earlier date, that outcome was assumed to have persisted—the ‘pessimistic’ outcome.
- Thus, the results are presented both cautiously (the ‘pessimistic’ figures) and pragmatically (the ‘optimistic’ figures).
Validation of data
Requests were made to several organizations for information which might validate (i.e. confirm the completeness of) both denominator and numerator data. For the denominator, this included the National Joint Registry, the National Obstetric Anaesthesia Database, and the Department of Health Hospital Episodes Statistics. For the numerator, we sought evidence of relevant cases from the NHS Litigation Authority (NHSLA) and National Reporting and Learning Service (NRLS) of the NPSA, the Medical Protection Society, and the Medical Defence Union. Medical journals were checked for reports of relevant cases and authors contacted as necessary. The internet search engine ‘Google’ was used to search for news items published on the internet with the words (epidural, spinal, death, abscess, haematoma, and infection).
Incidence calculations
Cases were included in the numerator where a complication of CNB led to permanent patient harm and the CNB had been performed within the audit period and in an NHS hospital.
The data were entered into a Microsoft Excel 2007 spreadsheet (Microsoft Corporation, USA) and incidences were calculated (by dividing the numerator for a given group by the relevant denominator). Confidence intervals (CIs) were derived using binomial probability tests with the stat-conf programme (Handbook of Biological Statistics 2008, http://udel.edu/~mcdonald/statconf.html). The primary end-points of the study were the incidences (both ‘pessimistic’ and ‘optimistic’) of permanent harm due to complications of the various types of CNB performed within the 1 yr audit period in an NHS hospital. The incidence of decompressive laminectomy in adult patients undergoing a perioperative epidural block was also calculated.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
This report focuses primarily on the quantitative aspects of the project. A full report, with expanded clinical details and analysis to identify clinical learning points will be published simultaneously by the Royal College of Anaesthetists (www.rcoa.ac.uk).
By September 2006, all 309 hospitals which had been contacted had agreed to participate and had appointed an LR.
Denominator data (snapshot returns)
The original publication of ‘snapshot’ data was based on reports from 97% of hospitals,19 but since then returns have been received from the final 3%. Thus, the denominator data (Table 3) used in the calculation of incidences of complications are based on returns from all the hospitals surveyed, 92% of them grading their figures as ‘accurate’. Extrapolating to annual activity by using a multiplier of 25 (see comment above) suggests that a total of just more than 700 000 CNB procedures [
325 000 (46% of total) subarachnoid blocks, 293 000 (41%) epidurals, 42 000 (6%) CSEs, and 47 000 (7%) caudals] are performed annually in the UK. The majority of CNBs were performed for obstetric (45%) or perioperative care (44%) indications.
|
None of the databases consulted in an attempt to validate these data provided information that could be used for that purpose.
Numerator data (complications reported)
Event returns and validation of completeness
In total, 108 cases were reported directly to the project team or through NCAPCIA, with 84 of these being considered appropriate for panel review. The 24 cases eliminated by the project team before panel review were all minor complications of no relevance to the problems under consideration: when there was the slightest doubt, the cases were included for review.
The NHSLA and NRLS databases were screened by the NPSA for all reports relating to CNB performed in the audit period. This identified 1700 cases reported to the NRLS (of which 13 were reported to have a serious or fatal outcome) and five cases notified to the NHSLA. The audit lead (T.M.C.) reviewed an unselected subset of 200 of the NRLS cases, all NRLS cases with a serious or fatal outcome, and all NHSLA cases. The NRLS review identified only one case meeting the criteria of the current project (in the 13 serious cases): this had already been reported. Two NHSLA cases were potentially relevant. One (a wrong route injection error) clearly met NAP3's inclusion criteria, but did not match the details of any cases reported at that time. A second case (of nerve injury) possibly met the inclusion criteria, but it was not clear whether it had been reported or not. Both hospitals were contacted by the NPSA and asked to report the case if it met inclusion criteria and had not been reported already. The wrong route injection case was subsequently reported to NCAPCIA and is included with those reviewed in detail.
Review of the literature identified three potential cases for inclusion, but discussion with the authors of the papers indicated that they did not meet the criteria. Internet-based news ‘alerts’ identified the wrong route injection case also identified by NHSLA screening. Other sources of validation did not identify any further cases.
Sources and timing of reports
Although the methodology of the system meant that anonymous reporting was possible, the majority (67) of cases were from identified individuals: 56 anaesthetists, nine neurologists, and two acute pain nurses. Similarly, other details cannot be described in full, but reports were received from all areas of the UK. Four hospitals reported more than one event, but two of these had neurosurgical units and were reporting complications of CNBs which had been performed elsewhere. It was not possible to obtain detailed information about the dual reports from the other two hospitals.
Events were notified throughout the audit period, but only one was reported after December 2007 and that was in August 2008, 5 months after the formal closure date. However, review indicated that it should be included in the analysis, even at a late stage.
Review panel assessments
Eighty-four cases were reviewed. Thirty-two cases were either performed outside the period of the audit, not performed in the NHS, or the complication did not meet the diagnostic criteria of the audit (wrong diagnosis, no link between CNB and notified complication, or full recovery of the complication at the time it was notified). Fifty-two cases therefore met all of the audit's inclusion criteria and efforts were made to follow-up these cases for a minimum of 6 months (Table 4). All 84 were reviewed for learning points for the clinical report to be published elsewhere (www.rcoa.ac.uk), but the remaining 52 are the focus of this analysis of the determination of permanent injury after CNB. Of these 52 patients, 22 made a fully documented complete recovery from their serious complication (NPSA classification ‘moderate’, Table 2): seven epidural abscesses, seven nerve or spinal cord injuries, three cardiovascular collapses [requiring cardiopulmonary resuscitation or admission to intensive care (ICU)], three cases of meningitis, one vertebral canal haematoma, and one other (intrathecal opioid overdose leading to respiratory arrest). These cases are not considered further. The remaining 30 events were used in the calculation of the ‘pessimistic’ incidences of permanent harm after CNB techniques. Detailed review indicated that in 16 cases, the patients were either likely to make a good recovery or the attribution of the permanent harm to the block was tenuous. This left 14 events for the calculation of the ‘optimistic’ incidences.
|
Patient characteristics
Events were distributed across both genders and the range of ASA status, with the majority of events occurring after elective surgical procedures and about half the CNBs having been performed by consultants and half by other grades (Table 5). There were no children in the 52 patients in the audit, and the majority of cases occurred in patients aged more than 50 yr. In the 30 patients with permanent harm (judged ‘pessimistically’), the complications were spread across all types of CNB: 18 (60%) followed epidural block and seven (23%) spinal anaesthesia. As far as clinical indication was concerned, 25 (83%) were in the perioperative group (Table 6).
|
|
Incidence of permanent harm
Considering the overall totals first, the incidence of any permanent injury (NPSA classifications serious and fatal, Table 2) after all CNBs in this survey is 4.2 per 100 000 (95% CI 2.9–6.1; equivalent to 1 in 23 500) using the ‘pessimistic’ assessment of outcome, and 2.0 per 100 000 (95% CI 1.1–3.3; 1 in 50 500) with the ‘optimistic’ assessment. However, there was a considerable variation between the incidences after different types of block. In both ‘pessimistic’ and ‘optimistic’ assessments, epidural and CSE were associated with higher incidences than both spinal and caudal blocks. Looking at clinical indication also revealed similar variation.
By using the subgroups we used in the census phase (Table 3), it is possible to calculate incidences for each of the subgroups. We report these for completeness (Tables 7–10), but as discussed below caution against their over-interpretation. The incidence of complications was highest after perioperative use and considerably lower in other groups (Tables 7 and 8). The incidence of permanent injury after adult perioperative epidural anaesthesia or analgesia was ‘pessimistically’ 17.4 per 100 000 (95% CI 7.2–27.8; 1 in 5800) and ‘optimistically’ 8.2 per 100 000 (95% CI 3.5–16.1; 1 in 12 200). Twelve patients in this category underwent decompressive laminectomy (seven for abscess, four for vertebral canal haematoma, and one as a result of nerve injury in association with spinal stenosis), an incidence of 12.3 per 100 000 cases (95% CI 6.3–21.4, 1 in 8100). One patient declined laminectomy.
|
|
|
Paraplegia and death are the worst possible outcomes so figures for these (13 ‘pessimistic’ and five ‘optimistic’) were extracted and analysed in the same way. The overall incidence of these two complications in this series is ‘pessimistically’ 1.8 per 100 000 (95% CI 1.0–3.1; 1 in 54 500) and ‘optimistically’ 0.7 in 100 000 (95% CI 0–1.6; 1 in 141 500) (Tables 9 and 10). The patterns revealed are similar to those seen in the analysis of all permanent complications.
|
Six patient deaths were reported (two abscesses, three cardiovascular collapses, and one wrong route error). All were included in the ‘pessimistic’ assessment, giving a rate of <1 in 100 000 (0.8 per 100 000: 95% CI 0–1.8), and three in the ‘optimistic’ group, a rate of <1 in 200 000 (0.4 per 100 000: 95% CI 0–1.2). Four of the deaths were considered to be directly associated with CNB and two indirectly.
Consideration of the cases with a fatal outcome (Table 11) may clarify how determinations of ‘pessimistic’ and ‘optimistic’ decisions were made, and illustrate the need to present the outcome data in both ways.
|
We followed the progress of those patients reported to the project with an initially serious neurological injury in whom we were able to determine a final outcome (Table 12). Patients were included even if they did not meet inclusion criteria (e.g. incidents occurring outside the audit dates or in private hospitals).
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
The results of this large prospective project are largely reassuring with the incidence of permanent injury being lower than in other equivalent or related studies.18 22–24 Assessed ‘pessimistically’ the incidence of permanent injury after CNB was 4.2 per 100 000, and of paraplegia/death was 1.8 per 100 000. ‘Optimistically’ the incidence of permanent injury was 2.0 per 100 000 and of paraplegia/death 0.7 per 100 000. The incidence of complications of epidural and CSE were at least twice those of spinals and caudals.
Previous studies have focused on the neurological complications of CNB, but this project took a broader approach and included all major complications of CNB, whether leading to neurological or other major sequelae. As a result, several deaths and major complications from wrong route errors or cardiovascular collapse were identified.
An internal NPSA paper describes epidural anaesthesia and its multiple potential complications well: ‘a complex amalgam of clinical judgement, technical skills, materials and equipment, drug delivery systems, patient supervision and care pathways. In addition to inherent complications in the procedure, each of these facets has the potential to generate patient harm through a combination of patient characteristics, human error or shortfalls in performance, equipment dysfunction and broader system failures. As a consequence, an enormous number of injuries can result’.25 This description is applicable to all forms of CNB and encapsulates the complexity of these seemingly simple procedures. The results of this national project reflect the complexities of both CNB and the interpretation of its sequelae.
Data interpretation
The data contain both clinical uncertainty and statistical uncertainty. We have presented the results in both ‘pessimistic’ and ‘optimistic’ terms to acknowledge the clinical uncertainty. As the case descriptions of the patients who died illustrate, in many cases, the interpretation of clinical descriptions was difficult because causation may be uncertain within a complex train of events. In other cases, the degree to which CNB led to final outcome may be uncertain. As an example we do not know whether spinal cord ischaemia after general anaesthesia in elderly frail patients who also have an epidural in place is caused by the CNB or simply coincidental: there were four such cases. Further, the final outcome was not always clear. One option would have been to be more decisive and simply present one ‘best guess’ result, but this would be an inappropriately simplistic response to the reality of complex clinical data. In 11 of 84 cases, interpretation was hampered by incomplete information: gaps were interpreted pessimistically even though this may mean that some patients were included inappropriately.
The statistical uncertainty is accommodated by the use of 95% confidence intervals (CIs) for all calculated incidences. In many cases, CIs are large, an inevitable consequence of the low or zero numerators of some groups. The data with the narrowest CIs are those with larger numerators and large denominators. Data with low or zero numerators are difficult to interpret.26 27 For zero numerators, we used the recommended ‘rule of 3’ (which states that for n observations with a zero numerator, the upper 95% confidence limit is 3/n) to calculate the upper confidence limit.26 The importance of this is that the main results have quite narrow CIs (e.g. pessimistic incidence of permanent injury from any CNB; 4.2 per 100 000 cases, 95% CI 2.9–6.1). In contrast, some of the sub-classifications of the data have very wide CIs (e.g. optimistic incidence of death or paraplegia after spinal anaesthesia in children 0 per 100 000 cases, 95% CI 0–921.8). This makes such data, particularly those with zero numerators, very difficult to interpret, and we would advise extreme caution in doing so.
The nature of this project means that whatever incidence is calculated from our data, it can only be a minimum incidence: unreported or wrongly excluded cases would increase the rates. Each additional case would increase the pessimistic incidence by 3%.
Data reliability and validation
The first and most obvious question is, ‘are the results robust?’ We consider the denominator(s) to be robust because they are based on a census of activity of the entire relevant population not a sample. All UK NHS hospitals committed to the project and all returned census data with more than 92% of these data being reported as ‘accurate’. Therefore, any error in the denominator is small.
Within the numerator data, there are both ‘known unknowns’ and ‘unknown unknowns’.28 The known unknowns are those cases which were reported, but where detail was inadequate for robust decisions on the nature or outcome of the event. In 11 cases (13%), insufficient information prevented determination of long-term outcome: in each, no recovery was assumed. Therefore, several cases have been classified ‘pessimistically’ as suffering permanent injury when full recovery may have occurred: this will have increased the resulting incidence of such complications. The unknown unknowns are those cases which may exist, but were not notified and therefore have not been included in incidence calculations. Inevitably, it is impossible to determine their number and futile to speculate on numbers, but every effort was made to disseminate widely information about the project, both within and outside the anaesthetic speciality. That 100% of hospitals volunteered an LR to the project, 100% returned census data, and more than 10% of cases were notified by non-anaesthetists attests to wide awareness and enthusiasm for the project.
Several sources were searched in an effort to validate the denominator (the number of procedures performed annually) and numerator (the number of relevant complications). These sources were either incomplete, from different populations, not validated themselves, or were impossible to correlate with the data presented here. None of the sources searched provided any information which conflicted with this project's data. Validation attempts showed that most cases of significant injury after CNB had not been notified to other national databases of clinical incident (e.g. NRLS). This raises concerns over the current under reporting of serious clinical incidents to the NRLS. It is, however, recognized that a number of data sources are required to fully capture and characterize clinical incidents.29 In contrast, validation attempts only identified one case that had, at that time, definitely not been reported to us and we subsequently learned of this case by other means also.
In spite of the inability to validate data externally, comparisons are possible with other data published recently. A UK audit of more than 10 000 paediatric epidurals reported a similarly low number of major complications, no deaths, and a permanent neurological injury incidence of 1 in 10 663,30 and thus is consistent with this survey. A survey of UK hospitals by Meikle and colleagues,31 identified knowledge of 40 vertebral canal haematomas occurring in a 6 yr period. Their annual rate of seven cases per year is very similar to that of this project: eight cases of vertebral canal haematoma were reported in 1 yr, with five meeting full inclusion criteria. In a Canadian series, the rate of decompressive laminectomy was 21 per 100 000 cases.32 In our equivalent subgroup (adult, non-obstetric perioperative epidurals), the incidence of decompressive laminectomy was 12.3 per 100 000: within the confidence limits of the Canadian data. It should be noted that Canadian and UK practice in selecting patients for laminectomy may well differ and our cohort contains nine cases which might have undergone laminectomy if the threshold for it was lower.
Comparison with other studies
The burden of neurological complications from CNB compared with other causes such as general anaesthesia and surgery is not well reported. A recent review of 54 cases from a UK medical defence organization found 72% were ‘surgical’ and 28% ‘non-surgical’.33 This, somewhat limited report indicates that neurological injury associated with regional anaesthesia is much less frequent than that related to surgery. Further, the incidence of such injury may differ little between regional and general anaesthesia.34
The best information available previously on major complications after regional anaesthesia comes from surveys in two Scandinavian countries, Finland and Sweden, both having ‘no fault’ compensation schemes and populations small enough to allow for central reporting systems. In Finland, a survey of 720 000 procedures performed between 1987 and 1993 found that the incidence of major complications was one in 22 000 after spinal anaesthesia and one in 19 000 after epidural block.22 In Sweden, a survey of 1.7 million procedures performed between 1990 and 1999 found an incidence of severe neurological complications of one in 20 000–30 000 after spinal anaesthesia, one in 25 000 after obstetric epidural, and one in 3600 after non-obstetric epidural.23 Both reviews were retrospective.
In the UK, Christie and McCabe18 retrospectively recorded 12 major complications after 8100 perioperative epidurals (1 in 675) in one hospital. This approximates to 148 per 100 000 epidurals. As nine patients made a full recovery, permanent injury was three in 8100 (37 per 100 000, 95% CI 7.6–108). Our point estimates for permanent injury after adult perioperative epidural are: pessimistic 17.4 per 100 000 (95% CI 7.2–27.8) and optimistic 8.2 per 100 000 (95% CI 3.5–16). Although the CIs from these data are narrower than those of Christie and McCabe, there is significant overlap. The figures reported here come from a population some 12 times larger so that the point estimates and CIs are likely to be more robust.
Cameron and colleagues35 reported a similar, retrospective, single hospital series, from Australia. Two vertebral canal haematomas and six epidural abscesses followed 8210 ‘acute pain’ epidurals. One laminectomy was required and there were no cases of permanent neurological injury. The incidences of vertebral canal haematoma (24 per 100 000, 95% CI 3–88), abscess (73 per 100 000, 95% CI 27–159), laminectomy (12 per 100 000, 95% CI 1–68), and permanent neurological harm (0 in 100 000, 95% CI 0–45) are again broadly consistent with those reported here.
Clinical implications
In the current series, as in the Swedish study, most complications of CNB occurred when epidural block was used in the perioperative period. Whether this was because it was used in higher risk patients is not something that this project can identify, but a higher (or lower) incidence of complications in one subgroup does not necessarily equate to the procedure being less (or more) appropriate for them. There are both statistical and clinical reasons for this. First, Moen and colleagues’23 figure of one in 1800 major complications in women having epidural anaesthesia for knee arthroplasty is often quoted, but the absence of any complications in men having the same procedure for hip arthroplasty or spinal anaesthetic for knee arthroplasty is rarely mentioned. Denominators for these groups were as low as 7000 and are too small for robust point estimates of incidences.
Secondly, the clinical perspective of the appropriateness or safety of a CNB procedure must recognize the potential benefits of that procedure (compared with other techniques) and risks other than the major ones reported here. Such risk–benefit analyses will differ between subgroups of patients and procedures so, for both statistical and clinical reasons, comparisons between subgroups should be made with considerable caution.
The patient characteristics are also relevant. More complications were reported in females than in males, but permanent injury was of equal incidence. Although many patients experiencing complications were aged >70 yr, a significant proportion was <50 yr of age (Table 5). More than half of the patients were fit and well (estimated ASA grades I–II), and most patients were undergoing major, elective surgery with CNB being performed by consultants. However, denominator data for these observations were not collected, so the extent (if any), to which these factors are associated with, or causal of, adverse outcomes cannot be determined. Notwithstanding this, patients who developed spinal cord ischaemia, vertebral canal haematoma, and epidural abscess were usually elderly, many were infirm and most undergoing major surgery. In contrast, patients suffering (non-ischaemic) nerve injury were more frequently young and healthy. These differences reinforce that comparisons between subgroups may not be valid.
Accepting these cautions, several clinical findings are of note. More complications occurred with perioperative epidural than in any other subgroup, although the four perioperative deaths all occurred in association with spinal or CSE block. Obstetric, chronic pain, and paediatric groups had a low incidence of major complications. This series includes one of the largest cohorts of each subgroup and, as such, those results are reassuring.
Concerns have been raised previously about the safety of CSE,36–38 and in this series, it had a relatively high incidence of complications. It represented only 5.9% of all CNBs performed, but led to 13–14% of permanent injuries and 15–40% of cases of paraplegia/death. Two of the deaths followed its use.
Of perhaps greater concern is the continuing problem with ‘wrong route’ injection errors: nine cases are reported here, six in obstetric practice. There was one death, but no other patient harm. A further similar death occurred in an obstetric unit shortly before this audit started:16 judged by the coroner to be an ‘unlawful killing’.39 Subsequently, an NPSA-published safety alert20 and multi-professional best practice guidance40 have highlighted the problem and identified measures to reduce its occurrence. That one in four respondents to a recent survey of 206 UK obstetric units reported knowledge of such an event indicates that this may be a major problem.41 Several alternatives, to remedy these potentially fatal mix-ups, have been advocated, but until a robust solution is universally in place, these events are likely to continue. This might be termed a national ‘systems error’. It is beyond the remit of this review to evaluate solutions, but clearly one must be found.
Prognosis of neurological complications
Most reviews of serious complications of CNB do not report their prognosis. All major complications are important, but the incidence of permanent harm is the most critical outcome. In Christie's series, three-quarters of identified patients made a full recovery. In this project, it was possible to monitor the progress of 41 initially major neurological complications of CNB (Table 12), and in 25 (61%) complete, or almost complete, recovery was documented. Neurological injury associated with spinal cord ischaemia or vertebral canal haematoma had a notably poor prognosis, whereas all patients with meningitis and the majority of patients experiencing nerve injury and abscess recovered fully. As we did not set out to identify all mild or moderate complications of CNB, unreported minor cases will have occurred and some may have resulted in permanent harm.
Overview
This project attempted to identify the incidence of major complications resulting in permanent harm after CNB in NHS hospitals in the UK. The number of such procedures was estimated in a 2 week census, and the complications of all CNBs performed over 1 yr in the NHS were identified, followed up, and analysed in detail. Analysis of the data suggests a lower incidence than reported previously in other series, usually of smaller numbers of patients, but there can be no certainty that all relevant cases were identified. There would need to be a considerable number of additional cases for the results of this project to be changed significantly, but if anyone is aware of such an unreported case meeting the inclusion requirements (see Methods section), the review panel would welcome further reports (in confidence to Professor Wildsmith at jaww{at}doctors.org.uk). If a substantial number of reports is made, the results will be updated in the future.
|
Funding |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
The project was funded by The Royal College of Anaesthetists. It would not have been possible without time and support given freely by many individuals and organizations.
Appendix 1: supporting organizations
The project was endorsed by the following organizations and specialist societies which played an important role in the promotion and dissemination of information about the project: Association of Anaesthetists of Great Britain and Ireland, Association of British Neurologists, Association of Paediatric Anaesthetists, British Association of Spinal Surgeons, British Pain Society, British Society of Neuroradiologists, European Society of Regional Anaesthesia (Great Britain and Ireland Section), Medical Defence Union, Medical Protection Society, National Confidential Acute Pain Critical Incident Audit, National Patient Safety Agency, Obstetric Anaesthetists Association, Royal College of Radiologists, and Society of British Neurological Surgeons. The project was also endorsed by the Chief Medical Officers of England (Sir Liam Donaldson), Northern Ireland (Dr Elizabeth Mitchell), Scotland (Dr Harry Burns), and Wales (Dr David Salter).
Appendix 2: the review panel
The review panel was composed of experts in CNB and its complications, as follows (in alphabetical order with their nominating groups): Dr David Bogod (Obstetric Anaesthetists Association), Dr Iain Christie (Association of Anaesthetists of Great Britain and Ireland), Dr David Counsell (National Confidential Acute Pain Critical Incident Audit), Dr Max Damian (Association of British Neurologists), Dr Barrie Fischer (European Society of Regional Anaesthesia: Great Britain and Ireland Section), Dr Richard Howard (Association of Paediatric Anaesthetists), Professor Ravi Mahajan (Royal College of Anaesthetists), Dr Angelique Mastihi (Medical Protection Society), Mrs Anne Murray (Patient Liaison Group, Royal College of Anaesthetists), Mrs Joan Russell (National Patient Safety Agency), Dr Nick Scott (European Society of Regional Anaesthesia: Great Britain and Ireland Section), Dr Andrew Vickers (British Pain Society), Professor Tony Wildsmith (Royal College of Anaesthetists). Several panel members also brought medico-legal expertise to the review panel. The panel was chaired by Dr Tim Cook (Project Lead, Royal College of Anaesthetists) who obtained additional expertise, as required, from individuals nominated by specialist microbiological or radiological organizations.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
First, our thanks go to the network of local reporters who collected the data for this project and supplied the detailed clinical reports, as well as the other individuals who notified us of cases. The following organizations were represented at a preliminary meeting and their unanimous support contributed to the development and success of the project: Association of Anaesthetists of Great Britain and Ireland (Professor Mike Harmer), British Pain Society (Dr Beverley Collett and Dr Andrew Vickers), European Society of Regional Anaesthesia, Great Britain and Ireland Section (Dr Barrie Fischer), National Confidential Acute Pain Critical Incident Audit (Dr David Counsell), Patient Liaison Group of the Royal College of Anaesthetists (Mrs Anne Murray), Acute Pain Nurses (Ms Sharon Kitkatt), National Patient Safety Agency (Mrs Joan Russell), Council of the Royal College of Anaesthetists (Dr Anne May, Professor Tony Wildsmith). We are also indebted to the President, Council, and the Head of Professional Standards (Mr Charlie McLaughlan) at the Royal College of Anaesthetists. We would also like to acknowledge the advice of Mrs Karen Thomson, Patient Information Advisory Group at the Department of Health and Ms Alexandra Cronberg, statistician at the National Patient Safety Agency. Finally, special thanks go to the chief administrator for the project at the Royal College of Anaesthetists, Ms Shirani Nadarajah.
|
Footnotes |
|---|
This article is accompanied by Editorial I. 
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
1 Ballantyne JC, Carr DB, deFerranti S, et al. The comparative effects of postoperative analgesic therapies on pulmonary outcome: cumulative meta-analysis of randomised controlled trials. Anesth Analg (1998) 86:598–612.[Abstract]
2 Rodgers A, Walker N, Schug S, et al. Reduction of postoperative mortality and morbidity with epidural or spinal anaesthesia: results of overview of randomised trials. Br Med J (2000) 321:1493–7.
3 Rigg JAR, Jamrozik K, Myles PS, et al, for the MASTER Anaesthesia Trial Study Group. Epidural anaesthesia and analgesia and outcome of major surgery: a randomised trial. Lancet (2002) 359:1276–82.[CrossRef][Web of Science][Medline]
4 McLeod GA, Dell K, Smith C, Wildsmith JAW. Measuring the quality of continuous epidural block for abdominal surgery. Br J Anaesth (2006) 96:633–9.
5 Buggy DJ, Smith G. Epidural anaesthesia and analgesia: better outcome after major surgery? Growing evidence suggests so. Br Med J (1999) 319:530–1.
6 Wildsmith JAW. State of the art: regional anaesthesia. Anaesthesia (2003) 58:1200–3.[CrossRef][Web of Science][Medline]
7 Wildsmith JAW. No sceptic me, but the long day's task is not yet done: the 2002 Gaston Labat lecture. Reg Anesth Pain Manag (2002) 27:503–8.
8 Kennedy F, Effron AS, Perry G. The grave spinal cord paralyses caused by spinal anesthesia. Surg Gynecol Obstet (1950) 91:385–98.[Web of Science][Medline]
9 Cope RW. The Woolley and Roe case. Anaesthesia (1995) 50:162–73.[CrossRef][Web of Science][Medline]
10 American Society of Regional Anesthesia. Consensus statements on central nerve block and anticoagulation. Reg Anesth Pain Med (1998) 23:S2.[CrossRef]
11 Checketts MR, Wildsmith JAW. Editorial: central nerve block and thromboprophylaxis—is there a problem? Br J Anaesth (1999) 82:164–7.
12 Burke D, Wildsmith JAW. Editorial: meningitis after spinal anaesthesia. Br J Anaesth (1997) 78:635–6.
13 Checketts MR, Wildsmith JAW. Editorial: accidental i.v. injection of local anaesthetics: an avoidable event? Br J Anaesth (1998) 80:710–1.
14 Fettes PDW, Wildsmith JAW. Editorial: somebody else's nervous system. Br J Anaesth (2002) 88:760–3.
15 Grewal S, Hocking G, Wildsmith JAW. Epidural abscess. Br J Anaesth (2006) 96:292–302.
16 Available from http://news.bbc.co.uk/1/hi/england/wiltshire/7226836.stm (accessed September 30, 2008).
17 Available from http://timesonline.co.uk/tol/life_and_style/health/article3201003.ece (accessed September 30, 2008).
18 Christie IW, McCabe S. Major complications of epidural analgesia after surgery: results of a six-year survey. Anaesthesia (2007) 62:335–41.[CrossRef][Web of Science][Medline]
19 Cook TM, Mihai R, Wildsmith JAW. A census of UK neuraxial blockage: results of the snapshot phase of the 3rd National Anaesthesia Project. Anaesthesia (2008) 63:143–6.[Web of Science][Medline]
20 Available from http://www.npsa.nhs.uk/nrls/alerts-and-directives/alerts/ epidural-injections-and-infusions/ (accessed September 30, 2008).
21 Seven Steps to Patient Safety: A Guide for NHS Staff (2004) National Patient Safety Agency.
22 Aromaa U, Lahdensuu M, Cozantis DA. Severe complications associated with epidural and spinal anaesthesias in Finland 1987–1993. A study based on patient insurance claims. Acta Anaesthesiol Scand (1997) 41:445–52.[Web of Science][Medline]
23 Moen V, Dahlgren N, Irestedt L. Severe neurological complications after central neuraxial blockades in Sweden 1990–1999. Anesthesiology (2004) 101:950–9.[CrossRef][Web of Science][Medline]
24 Phillips JMG, Stedeford JC, Hartsilver E, Roberts C. Epidural abscess complicating insertion of epidural catheters. Br J Anaesth (2002) 89:778–82.
25 Catchpole K, Bell D, Johnson S, Boult M. Reviewing the evidence of patient safety incidents in anaesthetics. In: Internal Report. (2006) The National Patient Safety Agency.
26 Hanley JA, Lippman-Hand A. If nothing goes wrong, is everything alright? J Am Med Assoc (1983) 259:1743–5.
27 Newman B. If almost nothing goes wrong, is almost everything all right? Interpreting small numerators. J Am Med Assoc (1995) 274:1013.[CrossRef][Web of Science][Medline]
28 News Briefing—Secretary Rumsfeld and Gen. Myers. February 2002. Transcript. Available from http://www.defenselink.mil/transcripts/transcript.aspx?transcriptid=2636 (accessed September 30, 2008).
29 Runciman WB. Lessons from the Australian Patient Safety Foundation: setting up a national patient safety surveillance system—is this the right model? Qual Saf Health Care (2002) 11:246–51.
30 Llewellyn N, Moriarty A. The national pediatric epidural audit. Paediatr Anaesth (2007) 17:520–33.[CrossRef][Medline]
31 Meikle J, Bird S, Nightingale JJ, White N. Detection and management of epidural haematomas related to anaesthesia in the UK: a national survey of current practice. Br J Anaesth (2008) 101:400–4.
32 Wijeysundera DN, Beattie WS, Austin PC, Hux JE, Laupacis A. Epidural anaesthesia and survival after intermediate-to-high risk non-cardiac surgery: a population-based cohort study. Lancet (2008) published online August 11, doi:10.1016/S0140-6736(08)61121-6.
33 Anthony S, Lalanda M. Damaged nerves. Medical Protection Society United Kingdom Casebook (2008) 16:12–3.
34 Jenkins K, Barker AB. Consent and anaesthetic risk. Anaesthesia (2003) 58:962–84.[CrossRef][Web of Science][Medline]
35 Cameron CM, Scott DA, McDonald WM, Davies MJ. A review of neuraxial epidural morbidity: experience of more than 8,000 cases at a single teaching hospital. Anesthesiology (2007) 106:997–1002.[CrossRef][Web of Science][Medline]
36 Reynolds F. Damage to the conus medullaris following spinal anaesthesia. Anaesthesia (2001) 56:238–47.[CrossRef][Web of Science][Medline]
37 Cook TM. 201 combined spinal–epidurals for anaesthesia using a separate needle technique. Eur J Anaesthesiol (2004) 21:679–83.[Web of Science][Medline]
38 Furguson S, Brighouse D, Valentine S. An unusual complication following combined spinal–epidural anaesthesia for caesarean section. Int J Obstet Anesth (1997) 6:190–3.[CrossRef][Web of Science][Medline]
39 Available from http://news.bbc.co.uk/1/hi/england/wiltshire/7226836.stm (accessed September 30, 2008).
40 Good Practice in the Management of Continuous Epidural Analgesia in the Hospital Setting, November 2004. Royal College of Anaesthetists. Available from http://www.library.nhs.uk/guidelinesFinder/viewResource.aspx?resID=121622 (accessed September 30, 2008).
41 Jones R, Swales HA, Lyons GR. A national survey of safe practice with epidural analgesia in obstetric units. Anaesthesia (2008) 63:516–9.[CrossRef][Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  A. R. Bodenham and S. J. Howell General anaesthesia vs local anaesthesia: an ongoing story Br. J. Anaesth., December 1, 2009; 103(6): 785 - 789. [Full Text] [PDF]
|
|
|
|
|
|
M. Scott, J. Stones, N. Payne, T. M. Cook, B. Fischer, D. Bogod, J. A. Wildsmith, D. Counsell, I. Christie, and M. Damien Antiseptic solutions for central neuraxial blockade: which concentration of chlorhexidine in alcohol should we use? Br. J. Anaesth., September 1, 2009; 103(3): 456 - 457. [Full Text] [PDF]
|
|
|
|
 |
|
 S. Choi and R. Brull Neuraxial Techniques in Obstetric and Non-Obstetric Patients with Common Bleeding Diatheses Anesth. Analg., August 1, 2009; 109(2): 648 - 660. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
V. Moen, L. Irestedt, N. Dahlgren, T. M. Cook, D. Counsell, and J. A. Wildsmith Major complications of central neuraxial block: the Third National Audit Project: some comments and questions Br. J. Anaesth., July 1, 2009; 103(1): 130 - 132. [Full Text] [PDF]
|
|
|
|
|
|
S. Boye and J. Schumacher Diagnosis of vertebral canal haematoma by myelography and spiral computer tomography in a patient with an implantable cardioverter-defibrillator contraindicating magnetic resonance imaging Br. J. Anaesth., July 1, 2009; 103(1): 137 - 138. [Full Text] [PDF]
|
|
|
|
|
|
S. J. Fowler, T. M. Cook, D. Counsell, and J. A. W. Wildsmith Incidence of severe complications after central neuraxial block Br. J. Anaesth., June 1, 2009; 102(6): 882 - 883. [Full Text] [PDF]
|
|
|
|
|
|
R. M. Grounds, T. M. Cook, D. Counsell, and J. A. W. Wildsmith Is the outcome for central neuraxial blockade really reassuring? Br. J. Anaesth., May 1, 2009; 102(5): 714 - 716. [Full Text] [PDF]
|
|
|
|
|
|
D. J. Buggy Central neuraxial block: defining risk more clearly Br. J. Anaesth., February 1, 2009; 102(2): 151 - 153. [Full Text] [PDF]
|
|
E-letters:
Read all E-letters
- Incidence of severe complications after centroneuraxial block
- Steven J Fowler
- British Journal of Anaesthesia, 2 Feb 2009 [Full text]
- Is the outcome for central neuraxial blockade really reassuring?
- R Michael Grounds
- British Journal of Anaesthesia, 20 Feb 2009 [Full text]
- Re: Incidence of severe complications after centroneuraxial block
- Tim Cook, et al.
- British Journal of Anaesthesia, 20 Feb 2009 [Full text]
- Concerns regarding the infectious complications of neuraxial block
- Prasad K Narasimha
- British Journal of Anaesthesia, 25 Feb 2009 [Full text]
- Re: Concerns regarding the infectious complications of neuraxial block
- Tim M Cook
- British Journal of Anaesthesia, 2 Mar 2009 [Full text]
- Re: Is the outcome for central neuraxial blockade really reassuring?
- Tim M Cook
- British Journal of Anaesthesia, 2 Mar 2009 [Full text]
- Antiseptic solutions for central neuraxial blockade – which concentration of chlorhexidine in alcohol should we use?
- Michael J Scott, et al.
- British Journal of Anaesthesia, 11 May 2009 [Full text]
- Re: Antiseptic solutions for central neuraxial blockade – which concentration of chlorhexidine in alcohol should we use?
- Tim M Cook, et al.
- British Journal of Anaesthesia, 16 Jun 2009 [Full text]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||