Local versus general anaesthesia for carotid endarterectomy

Abstract

Background

Carotid endarterectomy may significantly reduce the risk of stroke in people with recently symptomatic, severe carotid artery stenosis. However, there are significant perioperative risks that may be minimised by performing the operation under local rather than general anaesthetics. This is an update of a Cochrane Review first published in 1996, and previously updated in 2004, 2008, and 2013.

Objectives

To determine whether carotid endarterectomy under local anaesthetic: 1) reduces the risk of perioperative stroke and death compared with general anaesthetic; 2) reduces the complication rate (other than stroke) following carotid endarterectomy; and 3) is acceptable to individuals and surgeons.

Search methods

We searched CENTRAL, MEDLINE, Embase, and two trials registers (to February 2021). We also reviewed reference lists of articles identified.

Selection criteria

Randomised controlled trials (RCTs) comparing the use of local anaesthetics to general anaesthetics for people having carotid endarterectomy were eligible.

Data collection and analysis

Three review authors independently extracted data, assessed risk of bias, and evaluated quality of evidence using the Grading of Recommendations, Assessment, Development, and Evaluations (GRADE) tool. We calculated a pooled Peto odds ratio (OR) and corresponding 95% confidence interval (CI) for the following outcomes that occurred within 30 days of surgery: stroke, death, ipsilateral stroke, stroke or death, myocardial infarction, local haemorrhage, and arteries shunted.

Main results

We included 16 RCTs involving 4839 participants, of which 3526 were obtained from the single largest trial (GALA). The main findings from our meta‐analysis showed that, within 30 days of operation, neither incidence of stroke nor death were significantly different between local and general anaesthesia. Of these, the incidence of stroke in the local and general anaesthesia groups was 3.2% and 3.5%, respectively (Peto odds ratio (OR) 0.91, 95% confidence interval (CI) 0.66 to 1.26; P = 0.58; 13 studies, 4663 participants; low‐quality evidence). The rate of ipsilateral stroke under both types of anaesthesia was 3.1% (Peto OR 1.03, 95% CI 0.71 to 1.48; P = 0.89; 2 studies, 3733 participants; low‐quality evidence). The incidence of stroke or death in the local anaesthesia group was 3.5%, while stroke or death incidence was 4.1% in the general anaesthesia group (Peto OR 0.85, 95% CI 0.62 to 1.16; P = 0.31; 11 studies, 4391 participants; low‐quality evidence). A lower rate of death was observed in the local anaesthetic group but evidence was of low quality (Peto OR 0.61, 95% CI 0.35 to 1.06; P = 0.08; 12 studies, 4421 participants).

Authors' conclusions

The incidence of stroke and death were not convincingly different between local and general anaesthesia for people undergoing carotid endarterectomy. The current evidence supports the choice of either approach. Further high‐quality studies are still needed as the evidence is of limited reliability.

Plain language summary

Is local or general anaesthesia better during surgery to widen the main blood vessel to the brain when it becomes narrowed (carotid endarterectomy)?

Key messages

‐ Current evidence does not show any clear difference between local anaesthesia (where the patient remains awake) and general anaesthesia for the risk of stroke, death, or other unwanted effects for people having surgery to widen a narrowed carotid artery (carotid endarterectomy).

‐ Future studies should recruit more people, analyse and publish information from all of them, and make sure that the researchers assessing the outcomes do not know which type of anaesthetic people had.

What is a carotid endarterectomy?

A stroke happens when blood stops flowing to any part of your brain. The carotid artery is the main vessel supplying blood to the brain. This artery can become narrowed due to fatty deposits that build up over time. Around 1 in 5 strokes is caused by narrowing of the carotid artery. Blood clots can form at the point of narrowing. If a blood clot breaks off into the bloodstream, it can be carried into the brain, where it blocks the blood supply and causes a stroke.

A surgical operation – carotid endarterectomy – removes the inner lining, fatty deposits and any blood clots in the carotid artery and can lower the risk of stroke. However, even with very careful surgery, approximately 1 in 20 people will suffer a stroke caused by the operation itself.

Anaesthetics are medicines that prevent people feeling pain. Surgeons can use either a local anaesthetic, where an area of the body is numbed, or general anaesthetic, where a person is put to sleep. The use of a local anaesthetic rather than a general anaesthetic might lower the risk of a stroke during or after carotid endarterectomy surgery.

What did we want to find out?

We wanted to find out if using local anaesthetic for carotid endarterectomy:

‐ lowers the risk of stroke and death around the time of the operation;
‐ lowers the rate of other unwanted effects; and
‐ is more acceptable to individuals and surgeons

when compared to general anaesthetic.

What did we do?

We searched for studies that compared local and general anaesthetics in people who had a carotid endarterectomy. We compared and summarised their results, and rated our confidence in the evidence, based on factors such as study methods and sizes.

What did we find?

We found 16 studies involving 4839 people. The biggest study included 3526 people and the smallest study had 20 people. The studies were conducted around the world. More men than women were included in the studies, and their average age was 67 years.

Main results

Local anaesthetic makes little to no difference in risk of stroke within 30 days of surgery compared to general anaesthetic.

Local anaesthetic may not reduce risk of death within 30 days of surgery compared to general anaesthetic.

Since neither type of anaesthesia has clear benefits over the other, the choice of which to use can be made on the basis of the clinical situation, and the preferences of the surgeon and patient.

Main limitations of the evidence

We have either little, or moderate, confidence in these results. The quality of the evidence was reduced because, in most studies, it was possible that researchers collecting information about the outcomes of surgery knew which type of anaesthetic people had been given; this could have influenced their assessments. Also, information from some people who were meant to be included in the studies was left out, which also reduces the quality of the evidence.

How up to date is this evidence?

The evidence is current to February 2021.

Summary of findings

Summary of findings 1. Local anaesthesia compared with general anaesthesia for carotid endarterectomy.

Local anaesthesia compared with general anaesthesia for carotid endarterectomy
Patient or population: people with symptomatic or asymptomatic carotid disease Settings: hospital Intervention: local anaesthesia Comparison: general anaesthesia
Outcomes	Relative effect (95% CI)	No of participants (studies)	Quality of the evidence (GRADE)	Comments
Any stroke within 30 days of operation (primary outcome)	OR0.91 (0.66 to 1.26)	4663 (13 trials)	⊕⊕⊕⊝ Moderatea	Local anaesthesia probably results in little to no difference in any stroke within 30 days of operation. (7 trials in the analysis with events; the remainder have no events in both groups)
Death within 30 days of operation	OR 0.61 (0.35 to 1.06)	4421 (12 trials)	⊕⊕⊝⊝ Lowa,b	Local anaesthesia may not reduce death within 30 days of operation. (4 trials in the analysis with events; the remainder have no events in both groups)
Ipsilateral stroke within 30 days of operation	OR 1.03 (0.71 to 1.48)	3733 (2 trials)	⊕⊕⊝⊝ Lowa,b	Local anaesthesia may not reduce ipsilateral stroke within 30 days of operation.
Stroke or death within 30 days of operation	OR 0.85 (0.62 to 1.16)	4391 (11 trials)	⊕⊕⊕⊝ Moderatea	Local anaesthesia probably does not reduce stroke or death within 30 days of operation. (5 trials in the analysis with events; the remainder have no events in both groups)
Myocardial infarction within 30 days of operation	OR 1.53 (0.67 to 3.47)	4567 (12 trials)	⊕⊕⊝⊝ Lowa,b	Local anaesthesia may not reduce myocardial infarction within 30 days of operation. (4 trials in the analysis with events; the remainder have no events in both groups)
Local haemorrhage	OR 0.99 (0.79 to 1.25)	4186 (6 trials)	⊕⊕⊕⊝ Moderatea	Local anaesthesia probably results in little to no difference in local haemorrhage.
Arteries shunted	OR 0.25 (0.22 to 0.29)	4133 (8 trials)	⊕⊕⊝⊝ Lowa,b	Local anaesthesia may result in a large reduction in arteries shunted. (7 trials in the analysis with events; the remainder have no events in both groups)
CI: confidence interval; OR: Peto odds ratio
GRADE Working Group grades of evidence High quality: further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: we are very uncertain about the estimate.

^aDowngraded one level due to very serious risk of bias
^bDowngraded one level due to serious imprecision; wide confidence interval

Background

Description of the condition

Around 20% of people presenting with transient ischaemic attack (TIA) or non‐disabling ischaemic stroke have a significant stenosis with unstable atheromatous plaque at or around the bifurcation of the ipsilateral carotid artery. This plaque gives rise to the embolus. Carotid endarterectomy is an operation to remove this stenosis together with unstable plaque and, therefore, decrease the risk of stroke.

Description of the intervention

Carotid endarterectomy has been shown in large, well‐conducted randomised controlled trials (RCTs) to reduce the risk of stroke in people with recently symptomatic, severe (greater than 70%) internal carotid artery stenosis (ECST 1991; NASCET 1991). In a pooled analysis of data from these RCTs of endarterectomy versus medical treatment, surgery was of marginal benefit in terms of the five‐year risk of ipsilateral ischaemic stroke in those with 50% to 69% stenosis, and was highly beneficial in those with 70% stenosis or greater without near occlusion (Rerkasem 2011; Rothwell 2003). These benefits were seen despite the significant perioperative risks associated with carotid endarterectomy. The risk of stroke or death within 30 days of the operation was between 5% and 7% in the trials. If the risk of perioperative stroke could be reduced, the benefits from carotid endarterectomy would be greater. Thus, it is important to make the operation as safe as possible.

Some perioperative strokes occur during the operative procedure and may relate to reduced blood flow during carotid artery clamping. If the onset of such strokes could be recognised early, it may be possible to reverse the ischaemia by placing a shunt across the clamped artery, thereby increasing blood flow. In people operated on under general anaesthetic, the development of a new stroke is only recognised after recovery from the anaesthetic. In order to minimise the operative risk of stroke, several different approaches to shunting have been adopted when the procedure is performed under general anaesthetic. These are: the placement of a shunt in all individuals (Gumerlock 1988; Javid 1979; Thompson 1979); the placement of a shunt in some individuals thought to be at risk of an operative stroke (Buche 1988; Ricotta 1983; Schweiger 1988; Steiger 1989; Sundt 1986); or avoiding a shunt altogether (Ott 1980; Prioleau 1977; Reddy 1987). Shunts can be avoided because the majority of people tolerate arterial clamping without a shunt. Shunting may be associated with risks such as intimal damage promoting early postoperative thrombosis and late restenosis, which cause stroke. Thus, to surgeons, the selective method appears to be the most appropriate because it implies that only those individuals who are at risk of having a stroke during carotid clamping are exposed to the risks of shunt placement. However, there is little consensus about the best way of identifying those people who are at risk of stroke during the procedure. Several methods have been used to identify people at risk of stroke, including preoperative assessment (e.g. a history of recent stroke or occlusion of the contralateral artery), and a variety of techniques designed to directly or indirectly monitor cerebral blood flow during surgery. Techniques for monitoring blood flow during surgery include electroencephalographic monitoring, somatosensory evoked potential monitoring, transcranial Doppler monitoring, and measurement of the internal carotid artery back pressure (Rerkasem 2010). However, these methods are not reliable for detecting intraoperative stroke (Bass 1989; Gnanadev 1989; Kearse 1992; Kresowik 1991).

How the intervention might work

Performing carotid endarterectomy in people who are awake and under local anaesthetic offers the advantage of accurate assessment of the clinical state of the individual during surgery and the early postoperative period (Benjamin 1993). Any neurological change, either during test clamping or during surgery itself, can be detected early and therefore allow more appropriate use of selective shunting in these individuals. In addition, the cardiac and pulmonary morbidity of general anaesthetic may be avoided (Becquemin 1991; Corson 1987). There is also the suggestion that operation under local anaesthetic may be associated with an overall shorter hospital stay, and lower costs (Godin 1989; Gurer 2003; McCarthy 2001).

However, carotid endarterectomy under local anaesthetic may be associated with certain problems. The operation may be technically more difficult under local anaesthetic, which may increase the risk of a poor result from surgery. People may also undergo undue stress and pain during the operation, which may result in an increased risk of myocardial ischaemia and other complications. Finally, some surgeons may find performing the operation under local anaesthetic stressful. It is also possible that there may be certain advantages to operating under general anaesthetic. For example, there is some evidence that general anaesthetics reduce cerebral metabolic rate and may have a neuroprotective effect in the presence of ischaemia (Markowitz 1984; Michenfelder 1975; Wells 1963).

Why it is important to do this review

Carotid surgery is one of the most common types of vascular surgery. To date, there is no clear evidence that carotid endarterectomy performed under local anaesthesia is associated with reduced mortality. This issue is particularly important in older individuals, who comprise the majority of people who need this type of surgery. The only reliable way to assess the relative risks and benefits of carotid endarterectomy under local anaesthetic versus general anaesthetic is by direct comparison in RCTs. We therefore undertook a systematic review of all such trials. This systematic review is an update of a Cochrane Review first published in 1996 (Tangkanakul 1996), and previously updated in 2004, 2008, and 2013 (Rerkasem 2004; Rerkasem 2008; Vaniyapong 2013).

Objectives

To determine whether carotid endarterectomy under local anaesthetic: 1) reduces the risk of perioperative stroke and death compared with general anaesthetic; 2) reduces the complication rate (other than stroke) following carotid endarterectomy; and 3) is acceptable to individuals and surgeons.

Methods

Criteria for considering studies for this review

Types of studies

All RCTs that compared local with general anaesthetic for carotid endarterectomy and that measured clinically relevant outcomes were eligible for inclusion.

Types of participants

We considered trials that included people undergoing unilateral or bilateral carotid endarterectomy to be eligible for inclusion, whether the initial indication was symptomatic or asymptomatic carotid artery stenosis.

Types of interventions

We sought to identify all trials comparing carotid endarterectomy under local anaesthetic of any type with carotid endarterectomy under general anaesthetic of any type, including both epidural and skin or deep infiltration.

Types of outcome measures

Primary outcomes

The primary outcome was the proportion of participants who had a stroke of any kind (i.e. fatal or non‐fatal, contralateral or ipsilateral or brainstem, haemorrhage or infarction) within 30 days of surgery, and during long‐term follow‐up.

Secondary outcomes

Secondary outcomes included the following.

• Stroke ipsilateral to the operated artery within 30 days of operation and during long‐term follow‐up.

• Deaths from all causes within 30 days of surgery. We tried to classify each death as stroke‐related, related to other vascular disease (cardiac disease, pulmonary embolism, haemorrhage or other vascular disease) or non‐vascular.

• The proportion of participants who had a stroke or died within 30 days of surgery.

• Any myocardial infarction (fatal or non‐fatal) within 30 days of surgery.

• Other significant complications related to surgery (e.g. local haemorrhage (occurred from the artery or neck wound ), pulmonary complications including pneumonia, pulmonary embolism, atelectasis, prolonged intubation, pulmonary oedema, and cranial nerve injuries).

• Blood pressure (the numbers of participants with raised or lowered blood pressure (hypertension or hypotension) during or after surgery).

• Shunted arteries (the percentage of participants in whom a shunt was used during surgery).

• Hospital stay (the total duration of hospital and intensive care unit stay).

• Participants' overall satisfaction with and preference for each type of procedure.

• Surgeons' overall satisfaction with and preference for each type of procedure.

• The feasibility of carrying out carotid endarterectomy under local anaesthetic. We assessed this by calculating the percentage of participants allocated to have the surgery under local anaesthetic but who crossed over to general anaesthetic. We tried to divide this further into participants who had their choice of anaesthetic changed before the procedure was started, and participants who converted from local to general anaesthesia once the procedure had started.

Search methods for identification of studies

See the 'Specialised register' section at Cochrane Stroke's website (apps.ccbs.ed.ac.uk/csrg/entity/searchmethods.pdf). We applied no language restriction in the searches, and we arranged for translation of all possibly relevant publications where required.

Electronic searches

We searched the Cochrane Stroke Group's trials register on 1 February 2021. In addition, we searched the following electronic bibliographic databases from 2013 to 1 February 2021: Cochrane Central Register of Controlled Trials (CENTRAL; 2021, Issue 3 in the Cochrane Library: Appendix 1), MEDLINE (Ovid: Appendix 2), and Embase (Ovid: Appendix 3). We also searched the World Health Organization International Clinical Trials Registry Platform (WHO ICTRP) (www.who.int/clinical-trials-registry-platform/the-ictrp-search-portal) (Appendix 4), and the US National Institutes of Health Ongoing Trials Registry and ClinicalTrials.gov (www.clinicaltrials.gov/) (Appendix 5) on the same date for ongoing trials.

We developed the search strategies with the help of the Cochrane Stroke Group Information Specialist.

Searching other resources

We reviewed the reference lists of all relevant studies.

Data collection and analysis

Three review authors (AR, SN, KR) were involved in this process. One review author (SN) searched for articles through electronic searches, extracted these articles into bibliographic software, and removed duplicates. Two review authors (AR, SN) independently reviewed the titles and abstracts of the remaining articles, excluded any article that was obviously not an RCT, and excluded any article that did not aim to compare the effects between local and general anaesthesia in people having carotid endarterectomy. Any disagreement between the two review authors was resolved by a third review author (KR). We obtained the full texts of the remaining articles at this stage.

Selection of studies

Two review authors (AR, SN) independently considered the retrieved full‐text articles and identified types of study, participant, intervention, and outcome measures. In addition, the two review authors independently classified these articles into either included or excluded studies. We resolved any disagreements between review authors through discussion involving a third review author.

Data extraction and management

From the publications, we extracted details of the method of randomisation, blinding of outcome assessments, losses to follow‐up, cross‐overs and exclusions after randomisation. We also compared participant characteristics (age, sex, vascular risk factors, and indication for surgery) and details of the operation (type of cerebral monitoring, use of carotid patching, use of shunts, use of perioperative antiplatelet therapy) between the treatment groups in each trial. Also, although asymptomatic participants were included in some studies, the data were not available in sufficient detail to allow separate analysis of the outcomes of carotid endarterectomy in symptomatic and asymptomatic participants. However, it is unlikely that the relative effect of local versus general anaesthesia will vary qualitatively with symptom status.

Assessment of risk of bias in included studies

Three review authors (AR, SO, SN) independently assessed the methodological quality of the included trials using the Cochrane risk of bias tool (Higgins 2011). We resolved disagreements in the methodological assessment by reaching consensus through discussion. If an item was assessed as unclear, we contacted trial authors for clarification and to request missing information.

Measures of treatment effect

We estimated treatment effect for the following outcomes within 30 days of surgery: stroke, death, stroke or death, myocardial infarction, local haemorrhage, cranial nerve injuries, blood pressure, shunted arteries, hospital stay, participant and surgeon satisfaction, and feasibility of performing operation under local anaesthetic. We calculated Peto odds ratios (OR) and corresponding 95% confidence intervals (CIs) for discrete outcomes. For continuous outcomes (i.e. hospital stay), we used mean difference (MD) and standard deviation (SD) for analysis.

Unit of analysis issues

An event is the onset of an adverse outcome. We extracted the outcome events reported for each study. Some studies included participants who had bilateral operations, but only reported the number of participants, and not arteries, in each group. However, since bilateral carotid endarterectomy was unusual, we used the number of participants as the number of operations in such studies. Where possible, we used the number of participants, not the number of arteries, in the analysis.

Dealing with missing data

When data were missing, we contacted the corresponding author or a co‐author to request missing information. When missing data could not be obtained, we analysed only the available data.

Assessment of heterogeneity

We assessed heterogeneity between study results using the I² statistic, together with the P value of Cochrane's Q (Higgins 2003). We considered an I² value over 50% as representing substantial heterogeneity. In addition, we considered a P value of Cochrane's Q of less than 0.1 was likely to indicate heterogeneity. We also planned to identify possible sources of the heterogeneity.

Assessment of reporting biases

In an effort to minimise the impact of reporting biases, we sought to identify all relevant trials, including unpublished studies, by searching not only MEDLINE and Embase, but also the Cochrane Stroke Group Trials Register. In addition, we handsearched relevant journals and reviewed the reference lists of all relevant studies. In the previous version of this review, we advertised the review in Vascular News, a newspaper for European vascular specialists. We did not impose any language restriction in the searches and we arranged translation of all relevant non‐English language papers. Given a sufficient number of studies (10 or more studies), we planned to assess publication bias by constructing funnel plots (Page 2021).

Data synthesis

We calculated proportional risk reductions based on a weighted estimate of the odds ratio using the Peto method (APT 1994). We calculated a pooled Peto OR and 95% CI for the following outcomes that occurred within 30 days of surgery: stroke, death, stroke or death, myocardial infarction, local haemorrhage, cranial nerve injuries, and shunted arteries. We performed meta‐analysis using RevMan 5.4 (Review Manager 2014).

Subgroup analysis and investigation of heterogeneity

We intended to investigate substantial heterogeneity (I² over 50% and P < 0.1) using the following subgroup analyses to determine their effects.

• Type of stroke (fatal versus non‐fatal ipsilateral versus contralateral).

• Cause of death (stroke‐related death versus death related to other vascular disease versus death related to non‐vascular cause).

• Type of myocardial infarction (fatal versus non‐fatal).

• Type of blood pressure (hypotension versus hypertension).

• Type of shunted arteries (routine versus selective).

Sensitivity analysis

If I² was higher than 50% and P < 0.1, there would still be uncertainty about whether there was actual heterogeneity. Therefore, we planned to use standard OR with random‐effects model analysis to determine the robustness of outcome effects. In other words, we planned to evaluate the consistency of the results between Peto OR with the fixed‐effect model (our primary analysis model) and standard OR with the random‐effects model.

Summary of findings and assessment of the certainty of the evidence

The summary of findings table contains the effect estimates of the following outcomes of interest between local and general anaesthesia: any stroke within 30 days of operation (primary outcome), death within 30 days of operation, ipsilateral stroke within 30 days of operation, stroke or death within 30 days of operation, myocardial infarction within 30 days of operation, local haemorrhage, and arteries shunted. The table also contains analyses of the quality of evidence, assessed using the Grading of Recommendations, Assessment, Development, and Evaluations (GRADE) tool. The quality of the evidence is assessed based on risk of bias, together with inconsistency, indirectness, and imprecision of the data. The GRADE interpretation is classified into four levels: high, moderate, low, and very low quality evidence.

Results

Description of studies

Results of the search

We obtained a total of 1291 trials from predetermined databases (CENTRAL, MEDLINE, and Embase) and other sources (ClinicalTrials.gov, ICTRP, relevant studies, and reference lists). After we removed duplicates, 1154 trials were recorded for title and abstract screening. Of these, we assessed the full texts of 21 trials. We included 16 RCTs for qualitative assessment, and 14 RCTs for quantitative assessment (Figure 1).

1.

PRISMA study flow diagram

Included studies

We included 16 RCTs, involving 4839 participants, comparing the effects between local anaesthesia and general anaesthesia for carotid endarterectomy. Of these, most studies contained a small number of operated participants, except for the GALA trial, which reported 3526 operations (GALA 2008). Twelve out of 16 included studies identified that, between local and general anaesthesia, male participants were almost comparable (69% for local anaesthetic, 68% for general anaesthetic). Among the included articles, four were unavailable in English publications, and were translated from French (Pluskwa 1989), German (Binder 1999), Serbian (Sindelic 2004), and Czech (Mrozek 2007) into English.

In addition to the eligible articles included in the previous review (Vaniyapong 2013), our exhaustive search identified two new eligible articles (Hussien 2017; Orlický 2019). Of the two new trials, one trial, involving 210 participants, reported stroke incidence (3.3% contributed weight of all stroke events), death incidence (0% contributed weight of all death events), and ipsilateral stroke incidence (4.4% contributed weight of all ipsilateral stroke events) (Orlický 2019). The other trial, involving 30 participants, demonstrated only death incidence (0% contributed weight of all death events) (Hussien 2017). Among all included articles, 13 trials reported a method for local anaesthesia (Binder 1999; Forssell 1989; Hussien 2017; Kasprzak 2006; Luchetti 2008; Mazul‐Sunko 2010; Moritz 2010; Mrozek 2007; Orlický 2019; Pluskwa 1989; Prough 1989; Sbarigia 1999; Sindelic 2004). Of these 13 trials, 12 chose a cervical block for the local anaesthesia group, except for one trial using an epidural block (Pluskwa 1989), while a standard medication was used for the general anaesthesia groups in all trials.

Nine trials reported the indication for shunting (Binder 1999; Forssell 1989; Kasprzak 2006; Luchetti 2008; Mazul‐Sunko 2010; Moritz 2010; Mrozek 2007; Orlický 2019; Sbarigia 1999), with six trials identifying a shunting description for both groups (Binder 1999; Forssell 1989; Kasprzak 2006; Mazul‐Sunko 2010; Moritz 2010; Orlický 2019), and three trials providing details of shunting indication for the local anaesthesia group only (Luchetti 2008; Mrozek 2007; Sbarigia 1999). One trial aimed to follow participants up to one year (GALA 2008). Five trials indicated the follow‐up period: 30 postoperative days (Kasprzak 2006; Orlický 2019; Sbarigia 1999), two postoperative days (Binder 1999), and the time of hospital discharge (Forssell 1989). The other trials did not mention the period of follow‐up, but it is possible that it was to be based on time of hospital discharge.

We describe details of outcome measures in each study in the Characteristics of included studies. Some studies identified any type of stroke. Of these, two studies determined whether the strokes were ipsilateral to the operated artery (GALA 2008; Orlický 2019); however, most strokes appeared to be ipsilateral. Further, the Orlický 2019 trial also identified the proportion of other kinds of stroke (i.e. fatal versus non‐fatal and intraoperative versus postoperative) (Orlický 2019). The GALA trial was the only study illustrating the cause of death and the severity of stroke in terms of disability (GALA 2008). Only one trial formally assessed participant satisfaction (McCarthy 2004). None of the included studies formally assessed surgeon satisfaction.

Excluded studies

We excluded five trials due to unrelated study purpose. Two trials conducted a comparison between two types of local anaesthesia (Ozturk 2018; Ramachandran 2011). The other three trials were comparative studies between general anaesthesia versus combined general and local anaesthesia (Krajnic 2011), general anaesthesia versus general anaesthesia (Martynov 2018), and two different drugs during local anaesthesia (Sindjelic 2010) (see Characteristics of excluded studies).

Risk of bias in included studies

The details of risk of bias are described in Characteristics of included studies and Figure 2.

2.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study

Allocation

Nine studies had a low risk of bias for random sequence generation. These studies used different methods including a block randomisation (Binder 1999), computer‐based randomisation (GALA 2008; Hussien 2017; Kasprzak 2006; Luchetti 2008; Moritz 2010; Pluskwa 1989), zip code (Orlický 2019), and means of causal number (Sbarigia 1999). One study utilised date of birth, which we considered as a high risk of bias (Mrozek 2007). The remaining trials contained unclear risk for random sequence generation (Forssell 1989; Gimenez 2004; Mazul‐Sunko 2010; McCarthy 2004; Prough 1989; Sindelic 2004). For allocation concealment, no trials reported the method for allocation concealment (unclear risk), except for two trials reporting the use of opaque, sealed envelopes, which we deemed to be at low risk of bias (GALA 2008; Hussien 2017).

Blinding

All studies had a high risk of bias for the blinding of participants and personnel. None were able to blind participants and surgical teams to randomised treatment allocation. For the domain of blinding outcome assessment, 13 trials had an unclear risk due to lack of reporting. The remaining three trials reported that outcomes were assessed by neurologists who were blinded to the type of anaesthesia used, and we assessed these as low risk of bias (GALA 2008; Kasprzak 2006; Orlický 2019).

Incomplete outcome data

Seven of the 16 trials had low risk of bias for incomplete outcome data. Among the seven trials, three trials revealed no missing data (Kasprzak 2006; Orlický 2019; Sindelic 2004), while the other four trials described a reason for missing data (GALA 2008; Hussien 2017; Moritz 2010; Sbarigia 1999). Three of the trials with missing data reported that some participants who were randomised for surgery under local anaesthesia actually had surgery under general anaesthesia (GALA 2008; Moritz 2010; Sbarigia 1999). The remaining nine trials did not describe incomplete outcome data, so we assessed these as having unclear risk (Binder 1999; Forssell 1989; Gimenez 2004; Luchetti 2008; Mazul‐Sunko 2010; McCarthy 2004; Mrozek 2007; Pluskwa 1989; Prough 1989).

Selective reporting

We assessed four studies as having high risk of selective reporting because they did not indicate prespecified outcomes (Forssell 1989; Gimenez 2004; McCarthy 2004; Sindelic 2004). We judged the remaining studies as having a low risk of selective reporting bias (Binder 1999; GALA 2008; Hussien 2017; Kasprzak 2006; Luchetti 2008; Mazul‐Sunko 2010; Moritz 2010; Mrozek 2007; Orlický 2019; Pluskwa 1989; Prough 1989; Sbarigia 1999).

Other potential sources of bias

We did not identify other potential sources of bias in any of the studies. We therefore judged that all studies had a low risk of bias for this domain.

Effects of interventions

See: Table 1

See Table 1.

We included 16 RCTs, involving 4839 participants, in this review. However, only 14 RCTs were suitable for meta‐analysis. We performed meta‐analyses for only seven outcomes: any stroke (13 trials), ipsilateral stroke (two trials), death (12 trials), stroke or death (11 trials), myocardial infarction (12 trials), local haemorrhage (six trials), and arteries shunted (eight trials). We only obtained outcomes assessed within 30 days of surgery because none of the included studies reported long‐term results.

Any stroke within 30 days of operation

Of 13 included trials, 154 strokes of any type within 30 days after surgery (74 strokes after local anaesthesia and 80 strokes after general anaesthesia) were reported. The meta‐analysis showed that the incidence of stroke in the local and general anaesthesia groups was comparable (3.2% versus 3.5%) (Peto OR 0.91, 95% CI 0.66 to 1.26; P = 0.58; 13 studies, 4663 participants; moderate‐quality evidence; Analysis 1.1; Figure 3).

1.1. Analysis.

Comparison 1: Local versus general anaesthetic: randomised trials, Outcome 1: Any stroke within 30 days of operation

3.

Funnel plot of comparison 1: local versus general anaesthetic, randomised trials. Outcome 1.1: any stroke within 30 days of operation

Death within 30 days of operation

Within 30 days of operation, a total of 50 deaths were reported in the 12 included trials. There were 19 deaths (0.9%) in the local anaesthesia group compared to 31 deaths in the general anaesthesia group (1.4%). The operative death rate was not substantially lower in the local anaesthetic group compared with the general anaesthetic group (Peto OR 0.61, 95% CI 0.35 to 1.06; P = 0.08; 12 studies, 4421 participants; low‐quality evidence; Analysis 1.2).

1.2. Analysis.

Comparison 1: Local versus general anaesthetic: randomised trials, Outcome 2: Death within 30 days of operation

Ipsilateral stroke within 30 days of operation

Two eligible trials compared the effects between local and general anaesthesia on ipsilateral stroke (GALA 2008; Orlický 2019). The result revealed that the local anaesthesia group and the general anaesthesia group had the same incidence rate of ipsilateral stroke (3.1% in both groups) (Peto OR 1.03, 95% CI 0.71 to 1.48; P = 0.89; 2 studies, 3733 participants; low‐quality evidence; Analysis 1.3).

1.3. Analysis.

Comparison 1: Local versus general anaesthetic: randomised trials, Outcome 3: Ipsilateral stroke within 30 days of operation

Stroke or death within 30 days of operation

Eleven trials reported that the local anaesthesia group had a lower stroke or death rate within 30 days of operation when compared to the general anaesthesia group (3.5% versus 4.1%). However, the meta‐analysis showed there was no different effect in the rate of stroke or death between groups (Peto OR 0.85, 95% CI 0.62 to 1.16; P = 0.31; 11 studies, 4391 participants; moderate‐quality evidence; Analysis 1.4).

1.4. Analysis.

Comparison 1: Local versus general anaesthetic: randomised trials, Outcome 4: Stroke or death within 30 days of operation

Myocardial infarction within 30 days of operation

Twenty‐three participants suffered a myocardial infarction within 30 days of surgery. Fourteen participants (0.6%) in the local anaesthesia groups and nine patients (0.4%) in the general anaesthesia groups had a myocardial infarction. The meta‐analysis showed no different effect in the rate of myocardial infarction between groups (Peto OR 1.53, 95% CI 0.67 to 3.47; P = 0.31; 12 studies, 4567 participants; low‐quality evidence; Analysis 1.5).

1.5. Analysis.

Comparison 1: Local versus general anaesthetic: randomised trials, Outcome 5: Myocardial infarction within 30 days of operation

Other operative complications

Local haemorrhage

Six studies examined the effects of local and general anaesthesia on the rate of local haemorrhage after surgery. In total, 325 participants with local haemorrhages were reported. The local and the general anaesthesia groups had 7.7% and 7.8% of participants with local haemorrhage, respectively, with no different effect between groups (Peto OR 0.99, 95% CI 0.79 to 1.25; P = 0.95; 6 studies, 4186 participants; moderate‐quality evidence; Analysis 1.6). After using a random‐effects model, the effect estimate of the rate of local haemorrhage as computed from the fixed‐effect model was unchanged (OR 0.80, 95% CI 0.33 to 1.94; P = 0.62; Analysis 1.7). No indication of the severity of these bleeds was observed.

1.6. Analysis.

Comparison 1: Local versus general anaesthetic: randomised trials, Outcome 6: Local haemorrhage

1.7. Analysis.

Comparison 1: Local versus general anaesthetic: randomised trials, Outcome 7: Sensitivity analysis: random‐effects model, local haemorrhage

Pulmonary complications

One trial reporting pulmonary complications found no statistically significant difference in the rate of pneumonia under local anaesthesia compared with general anaesthesia (Kasprzak 2006). The GALA trial determined a pulmonary embolism as an outcome and reported that no events in either treatment group were observed (GALA 2008).

Blood pressure

Thirteen trials recorded blood pressure during and after surgery. However, the studies did not consistently report the number of participants with significant hypotension or hypertension or mean arterial pressure during and after surgery. Furthermore, the definitions of hypertension and hypotension varied between trials. We have therefore simply described the results.

Seven studies reported that blood pressure dropped in the general anaesthesia group after induction of anaesthesia (Forssell 1989; GALA 2008; Hussien 2017; McCarthy 2004; Pluskwa 1989; Prough 1989; Sindelic 2004). In one trial, more participants in the general anaesthesia group had significant hypotension during or after surgery compared with the local anaesthesia group (25% versus 7%) (Forssell 1989). Another trial found significantly lower blood pressure in the general anaesthesia group than in the local anaesthesia group (Hussien 2017). However, these findings were not confirmed in Pluskwa 1989. The GALA trial reported on the manipulation of blood pressure (GALA 2008). More general anaesthesia than local anaesthesia participants had their blood pressure manipulated up (43% compared with 17%), and more local anaesthesia participants had their blood pressure manipulated down or not manipulated at all (74% compared with 41%) during or after surgery. The difference in blood pressure manipulation between the two trial arms was statistically significant (GALA 2008).

Five trials showed that blood pressure tended to increase during clamping of the carotid artery in the local anaesthesia group compared with the general anaesthesia group (Forssell 1989; Gimenez 2004; Luchetti 2008; Pluskwa 1989; Prough 1989), but this was not found in another trial (McCarthy 2004). In two trials, there were significantly more participants with hypertension in the local anaesthesia group during surgery than in the general anaesthesia group (Forssell 1989: 36% versus 0%; Pluskwa 1989: 80% versus 20%). Three trials reported that during surgery, the mean arterial pressure in the local anaesthesia group was higher than in the general anaesthesia group (Luchetti 2008; Moritz 2010; Mrozek 2007). Two studies suggested that hypotension was more common in the postoperative period with local anaesthesia than with general anaesthesia (Pluskwa 1989; Prough 1989). Two trials found that participants operated on under general anaesthesia had more postoperative (within day one) hypertension than those operated on under local anaesthesia (Gimenez 2004; Kasprzak 2006).

Arteries shunted

Eight included studies demonstrated that fewer arteries were shunted with the use of local anaesthesia compared to general anaesthesia. Fifteen per cent of participants in the local anaesthesia groups had their arteries shunted compared with 42% of participants in the general anaesthesia groups. As there was significant heterogeneity between studies (I² = 91%), we used a random‐effects model as the main model for meta‐analysis to pool the results (OR 0.24, 95% CI 0.08 to 0.73; P < 0.001; 8 studies, 4133 participants; low‐quality evidence; Analysis 1.8). Sensitivity analysis conducted by fixed‐effect model showed that the effect estimate of the shunting was unchanged (Peto OR 0.25, 95% CI 0.22 to 0.29; P < 0. 001; Analysis 1.9).

1.8. Analysis.

Comparison 1: Local versus general anaesthetic: randomised trials, Outcome 8: Arteries shunted

1.9. Analysis.

Comparison 1: Local versus general anaesthetic: randomised trials, Outcome 9: Sensitivity analysis: random‐effects model, arteries shunted

Hospital stay

The duration of hospital stay was reported in four included trials (Binder 1999; GALA 2008; McCarthy 2004; Orlický 2019). This outcome among the four studies could not be presented in the meta‐analysis due to lack of standard deviation. All included studies only stated that the average time in hospital was not significantly different between the local and general anaesthesia groups.

Participants' overall satisfaction with and preference for each type of procedure

One included study formally assessed participant satisfaction (McCarthy 2004). There was no statistically significant difference in satisfaction between anaesthetic techniques. Of the three participants who had repeat carotid endarterectomies (having had a local anaesthetic for the first operation), none refused repeat randomisation (Forssell 1989). Forssell 1989 reported that one participant in the local anaesthesia group became extremely agitated during the procedure. Another trial evaluated participant satisfaction with a questionnaire (Binder 1999). They found that both types of anaesthesia were equally acceptable but the publication did not describe the questionnaire in detail. All participants preferred the same type of anaesthesia if they needed a second operation, except one participant in the local anaesthesia group (total 27 participants) who wished to have general anaesthesia for any further surgery. Mrozek 2007 asked participants about any unpleasant sensations after surgery and during the postoperative period. A minimum amount of unpleasant sensation was reported for both types of anaesthetic after surgery and during the postoperative period (Mrozek 2007).

Surgeons' overall satisfaction with and preference for each type of procedure

None of the included studies assessed the satisfaction or preferences of surgeons.

Feasibility of performing operation under local anaesthetic

One trial recorded the number of participants randomised to have surgery under local anaesthesia, but who had surgery under general anaesthesia (Forssell 1989). Eight participants crossed over from local to general anaesthesia whilst none switched from general to local anaesthesia. The most common reasons for cross‐over procedure were that the participant changed his or her consent or that the participant had unstable cardiac disease. Seven out of eight participants had their anaesthetic changed before the procedure was started. In another trial, six participants were switched from local to general anaesthesia due to severe agitation (three participants), insufficient anaesthesia under local anaesthesia (two participants), and intravascular injection during application of local anaesthetic agent (one participant) (Kasprzak 2006). Three out of six participants had their anaesthetic changed before the procedure was started. No general anaesthesia cases were switched to local anaesthesia in this study (Kasprzak 2006). In the GALA trial, 167 participants were crossed over before initiation of anaesthesia: 75 participants crossed over from local to general anaesthesia whilst 92 switched from general to local anaesthesia (GALA 2008). Participants allocated to general anaesthesia were more likely to cross over due to a medical decision, whereas participants allocated local anaesthesia were more likely to cross over due to the participant's preference. Sixty‐nine out of 1771 (3.9%) local anaesthesia participants were switched to general anaesthesia after initiation of anaesthesia, 17 before and 52 after the start of surgery. In one trial, two participants switched from local to general anaesthesia (Moritz 2010), and in another trial, no participants switched from local to general anaesthesia (Mrozek 2007).

Publication bias

Publication bias for the outcomes of stroke, death, stroke or death, and myocardial infarction were examined as there was sufficient evidence to do so (10 or more trials) . Of these, the evidence for stroke, death, and stroke or death is likely to have a publication bias  based on the asymmetrical funnel plots (Figure 3; Figure 4; Figure 5), while myocadial infarction is not likely to have publication bias based on the symmetrical funnel plot (Figure 6).

4.

Funnel plot of comparison 1: local versus general anaesthetic, randomised trials. Outcome 1.2: death within 30 days of operation

5.

Funnel plot of comparison 1: local versus general anaesthetic, randomised trials. Outcome 1.4: stroke or death within 30 days of operation

6.

Funnel plot of comparison 1: local versus general anaesthetic, randomised trials. Outcome 1.5: myocardial infarction within 30 days of operation

Discussion

Summary of main results

We included 16 randomised controlled trials (RCTs) investigating the effect of using local versus general anaesthesia on adverse outcomes for people having carotid endarterectomy. Meta‐analysis, involving 4839 participants, found an equivalent effect between the two treatments on the proportion of participants who had strokes, death, stroke or death, myocardial infarction, and other complications, except for arteries shunted within 30 days of operation. In addition to these outcomes, a new finding obtained from our review was an equivalent risk of ipsilateral stroke between the two anaesthesia techniques.

Participants who received local anaesthesia had a lower risk of the arteries being shunted compared to general anaesthesia. Although a superior effect on the occurrence of arteries shunted was revealed under local anaesthesia, we are uncertain about the effect estimate of the arteries shunted due to the low‐quality evidence.

Overall completeness and applicability of evidence

Our review aimed to examine whether local or general anaesthesia produces lower risk of stroke, death, and complications other than stroke, and whether local or general anaesthesia is more acceptable overall to individuals and surgeons. We therefore aimed to include studies directly comparing the two anaesthesia techniques on these outcomes of interests. However, there was insufficient information for some outcome measures, so we were unable to address all of our review objectives. Of the available outcome measures, findings from pooled data demonstrated non‐significant differences in the rate of stroke, death, myocardial infarction, and local haemorrhage between the local and general anaesthesia groups during carotid endarterectomy. Although a large trial was included in this meta‐analysis (GALA 2008), there were no convincing differences between local and general anaesthetic in the rate of stroke, death, and other complications. In our review, fewer arteries were shunted with local anaesthesia, indicating that local anaesthesia may be superior to general anaesthesia for this outcome. This finding showed considerable heterogeneity (I² = 91%) between local and general anaesthesia. One possible reason for this heterogeneity is the different criteria used for shunting. For instance, one eligible study employed shunting for all participants, regardless of necessity (Binder 1999). High heterogeneity is often thought to be a possible result of small‐study effects. However, we could not confirm whether this heterogeneity is a result of small‐study effects because there was an insufficient number of studies to determine the presence of publication bias via a funnel plot.

A new larger RCT included in this review demonstrated that complications associated with carotid endarterectomy were similar to previous studies (Orlický 2019), except for an increase in local haemorrhage with local anaesthetic (8.6% versus 1.9%) (Analysis 1.6). This finding differed from that of previous studies, which tended to favour local anaesthetic (Binder 1999; Forssell 1989; Kasprzak 2006; Sbarigia 1999). Interestingly, the demographic data showed that the number of participants who used anticoagulant preoperatively in the local anaesthetic group was higher than in the general anaesthetic group in the Orlický 2019 study (5.7% versus 1.9%, P < 0.001). Thus, anticoagulant may affect the risk of local haematoma in people given local anaesthetic.

Of the other outcomes available only for narrative review, most included trials (75%) assessed alteration of blood pressure during the operation and postoperatively. It is interesting that two studies suggested that hypotension was more common in the postoperative period with local anaesthesia (Pluskwa 1989; Prough 1989). This may be due to the high rate of blood pressure being manipulated down, but we could not find any hard evidence to support this at the present time.

The choice of anaesthetic technique depends on the clinical situation and the preferences of individuals and their surgeon. In some people, the operation may be technically more difficult under local anaesthesia (e.g. in people with short, wide necks). Some people, perhaps as many as 10%, will refuse to have the operation under local anaesthesia (Forssell 1989), and some surgeons may feel more comfortable performing the operation under general anaesthesia.

Regarding the cost of endarterectomy, one trial was designed to determine whether the type of anaesthesia influenced the cost of endarterectomy (GALA 2008). This trial showed that the expected cost of carotid endarterectomy under local anaesthesia was less than under general anaesthesia (mean difference 178 pounds sterling (GBP)) (Gomes 2010). This difference was mainly due to the longer length of stay in an intensive care unit and the use of consumables such as shunts and patches. A post hoc subgroup analysis (40 participants) from the GALA 2008 study investigated the influence of local versus general anaesthesia on postoperative neurocognitive function. This study showed that local anaesthesia beneficially influenced early postoperative neurocognitive functions. Mazul‐Sunko 2010 found shunting to be the only parameter associated with neurocognitive decline on the first day after carotid endarterectomy. Local anaesthesia was hypothesised to offer an indirect benefit due to the reduced rate of shunting (Mazul‐Sunko 2010). However, given the small size of these studies, early postoperative neurocognitive function requires further investigation (Weber 2009).

It is noteworthy that the meta‐analysis was dominated by the GALA trial because over 70% of the participants included were obtained from this trial (GALA 2008). The GALA trial did receive some criticism that should be of concern, including the very slow rate of recruitment and the lack of a prescriptive study protocol. For example, there was far more blood pressure manipulation in the general anaesthetic group compared to the local anaesthetic group; fewer carotid patches were used in the local anaesthetic group (42% versus 50%) which might have had some influence on outcomes; there was no standardised requirement for indication and management for surgery (i.e. diagnostic methods used, anaesthetic drugs or techniques used, heparin dose, antiplatelet therapy, and endarterectomy technique); and there was a wide variation in shunting practice across different centres in the trial. A further potential criticism of the GALA design is that higher‐risk individuals, who were considered better suited to one or the other type of anaesthetic, may not have been recruited, as the overall stroke risks in the trial were very low for that period. Despite these criticisms, the more pragmatic approach implemented in the GALA trial may have improved the generalisability of the results.

Quality of the evidence

Findings of our meta‐analysis revealed that there were no different effects on incidence of stroke, death, and other complications, except for arteries shunted, between local and general anaesthesia during carotid endarterectomy. Evidence of moderate quality was demonstrated in the effect estimate of stroke, stroke or death, and local haemorrhage in a comparison between the two anaesthesia techniques. The moderate‐quality evidence suggests that local anaesthesia probably does not reduce these outcomes to a greater extent than general anaesthesia. Some evidence judged as moderate quality was downgraded one level due to serious risk of bias. Furthermore, low‐quality evidence was identified in the effect estimates for death, ipsilateral stroke, myocardial infarction, and arteries shunted. These effect estimates suggest that local anaesthesia may not reduce these outcomes to a greater extent than general anaesthesia. We downgraded this evidence one level due to serious risk of bias, and one level due to serious imprecision (wide confidence interval).

One important risk of bias that limited our confidence in the effect estimate was lack of participant and surgeon blinding, which was reported in all included studies. In fact, in this context, it is impossible to blind participants and surgeons to treatment allocation. It should be noted that most studies contained unclear blinding of outcome assessment. Of these, only three studies stated that the outcome assessment was measured by a neurologist blinded to group allocation (GALA 2008; Kasprzak 2006; Orlický 2019). In addition, some randomised participants were removed from the analysis, especially those who crossed over anaesthetic type (GALA 2008; Moritz 2010; Sbarigia 1999). The result would be biased if the participants who were removed differed from participants who remained in the analysis in terms of their characteristics. If the crossed‐over participants who were excluded had in fact received the treatment to which they were originally allocated and were included in the analysis, the treatment effect direction might be changed.

We further explored the fragility index, which was undertaken to determine the robustness of the results. The fragility index result revealed that all outcomes were less robust, with a fragility index of zero, with the exception of arteries shunted, which had a higher fragility index of 488, indicating a more robust result.

Potential biases in the review process

Wherever possible, we have tried to limit the potential for bias throughout the review process. We used pre‐determined criteria to search electronic databases for studies eligible for inclusion in this review. To minimise publication bias, we sought all available data from both published and unpublished sources, and we contacted authors whose publications did not provide sufficient information. In addition, three review authors independently carried out data selection, data analysis, and assessment of risk of bias. Having three review authors work independently reduced the likelihood of bias in publication selection and bias in study assessment.

Nevertheless, there were some possible biases in each article that we could not control. Many studies reported the number of arteries rather than the number of participants. Few trials assessed participants' or surgeons' satisfaction or preference, or the durations of intensive care and overall hospital stay. There was considerable heterogeneity across studies in the use of shunts with both types of anaesthesia. This heterogeneity may be attributed to the different policies in shunting between studies. For example, all participants in the Binder 1999 study were shunted irrespective of treatment allocation. All participants in the local anaesthetic group were shunted despite the fact that surgeons preferred local anaesthetic due to the low rate of shunting. Apart from this trial, the remaining seven RCTs in the pooled analysis used selective shunting. For six RCTs, although the indication of shunting in the local anaesthesia group was not markedly different, the indication for shunting in the general anaesthesia group varied considerably. One study used stump pressure measurement and clinical judgment (Forssell 1989), while another study used a mix of transcranial Doppler, stump pressure measurement, electroencephalography (EEG), and clinical judgment (GALA 2008). Although another three studies used somatosensory evoked potentials as the indication for shunting, details of these indications were different (Kasprzak 2006; Moritz 2010; Orlický 2019). One trial carried out shunting routinely, but the actual rate of shunting was 82% because of expected technical difficulties with shunt insertion in 18% of the cases in the general anaesthetic group (Mazul‐Sunko 2010). The remaining three RCTs did not report the indication for shunting in the general anaesthesia group (Moritz 2010; Mrozek 2007; Sbarigia 1999). All of these differences may explain the considerable heterogeneity in the use of arteries shunted.

In this review, combining the sparse data might be a problem when no events in either or both treatment groups were reported. Zero events in either or both treatment groups was usually not estimated in the analysing process  (Sweeting 2004). For instance, 13 trials involving 4663 participants demonstrated a pooled effect in stroke events. However, only seven trials were actually represented in this outcome effect because the other six remaining trials had no events in both treatment groups. This situation occurred in all outcomes. Thus, our findings should be interpreted with caution that not all included studies represented pooled effects.

Agreements and disagreements with other studies or reviews

This updated Cochrane Review summarises and quantifies the effect of two different anaesthesia approaches (local versus general anaesthesia) in people having carotid endarterectomy. There have been five relevant reviews published previously (Hajibandeh 2018; Rerkasem 2004; Rerkasem 2008; Tangkanakul 1996; Vaniyapong 2013). Three of them included both randomised and non‐randomised controlled trials (Hajibandeh 2018; Rerkasem 2004; Tangkanakul 1996), while the other two only included randomised controlled trials (Rerkasem 2008; Vaniyapong 2013). In general, the findings in this updated review are in line with the previous versions (Rerkasem 2008; Vaniyapong 2013), demonstrating no superior effect in the local anaesthesia group on odds reduction in stroke and death when compared to general anaesthesia. On the other hand, our main findings contrast with those effect estimates from non‐randomised controlled trials showing that local anaesthesia had more significant benefits on the reduction of stroke and death rates than general anaesthesia (Hajibandeh 2018; Rerkasem 2004; Tangkanakul 1996). It should be noted that the better effect under local anaesthesia might be susceptible to bias due to obtaining evidence from trials with a low‐quality research design (non‐randomised controlled trials). With the addition of the GALA 2008 trial, the meta‐analyses showed that these apparent differences were probably due to biases in the non‐randomised comparisons, illustrating the importance of adequately powered randomised controlled trials (Collins 2001).

Authors' conclusions

Implications for practice.

Individuals and surgeons can select an anaesthetic technique for carotid endarterectomy in the knowledge that the current evidence does not show any convincing difference between local and general anaesthesia in the risk of stroke, death, and other complications.

Implications for research.

Most outcome results had reduced reliability due to low‐quality evidence. Thus, future randomised controlled trials with successful blinding of outcome assessment, analysis of all randomised participants, and larger sample sizes are warranted.

What's new

Date	Event	Description
1 February 2021	New search has been performed	We have updated the search to 1 February 2021. A total of 16 trials (4839 operations), with two additional randomised controlled trials included. Of the new trials, one trial reported both stroke (3.3% weight contribution) and death events (0% weight contribution), whereas the other trial only reported the death event within 30 days of operation (0% weight contribution). In addition to the previous version, we could now analyse whether local anaesthesia would lower the incidence of ipsilateral stroke more than general anaesthesia. Only one new trial (4.4% weight contribution) demonstrated an incidence of ipsilateral stroke.
1 February 2021	New citation required but conclusions have not changed	We found that the results for all outcomes were unchanged from the previous version.

History

Review first published: Issue 1, 1996

Date	Event	Description
30 September 2013	New citation required but conclusions have not changed	New first author. Conclusions unchanged
30 September 2013	New search has been performed	The searches have been updated to September 2013. We have identified four new randomised trials. The total number of included trials is now 14 randomised trials of 4596 operations. However, the four new trials did not have any stroke or death events in the perioperative period, so the results for these outcomes have not changed
30 November 2008	New search has been performed	The searches have been updated and completed to November 2008. we have identified three new randomised trial. This most recent study is the biggest trial (3526 operations) in this systematic review. The total number of included trials is now 10 randomised trials of 4335 operations. The non‐randomised studies, which are prone to bias and which were previously included in the review, have now been removed from this version.
6 August 2008	New citation required but conclusions have not changed	There has been a change of authorship.
15 April 2008	Amended	Converted to new review format.

Appendices

Appendix 1. Cochrane Central Register of Controlled Trials (CENTRAL) search strategy

#1 [mh ^“endarterectomy, carotid”]
#2 [mh “carotid arteries”/SU]
#3 [mh “carotid artery diseases”/SU]
#4 [mh “carotid arteries”]
#5 [mh “carotid artery diseases”]
#6 carotid:ti,ab
#7 #4 or #5 or #6
#8 [mh ^endarterectomy]
#9 (endarterectom* or surg*):ti,ab
#10 #8 or #9
#11 #7 and #10
#12 #1 or #2 or #3 or #11
#13 [mh anesthesia]
#14 [mh anesthetics]
#15 (anesthe* or anaesthe*):ti,ab
#16 [mh ^”cervical plexus”]
#17 (cervical NEXT block):ti,ab
#18 (bupivacaine or lidocaine or lignocaine or prilocaine or ropivacaine or mepivacaine or alfentanil or propofol or fentanyl or ketamine or midazolam or sevoflurane or desflurane or etomidate or isoflurane):ti,ab
#19 #13 or #14 or #15 or #16 or #17 or #18
#20 #12 and #19

Appendix 2. MEDLINE searc h strategy (OVID)

1 Endarterectomy, carotid/
2 exp carotid arteries/su
3 exp carotid artery diseases/su
4 exp carotid arteries/
5 exp carotid artery diseases/
6 carotid.tw.
7 4 or 5 or 6
8 endarterectomy/
9 (endarterectom$ or surg$).tw.
10 8 or 9
11 7 and 10
12 1 or 2 or 3 or 11
13 exp anesthesia/
14 exp anesthetics/
15 (anesthe$ or anaesthe$).tw.
16 cervical plexus/
17 cervical block.tw.
18 (bupivacaine or lidocaine or lignocaine or prilocaine or ropivacaine or mepivacaine or alfentanil or propofol or fentanyl or ketamine or midazolam or sevoflurane or desflurane or etomidate or isoflurane).tw.
19 or/13‐18
20 12 and 19
21 exp animals/ not humans.sh
22. 20 not 21.

Appendix 3. EMBASE search strategy (OVID)

1. carotid artery surgery/ or carotid endarterectomy/
2. exp carotid artery/su [Surgery]
3. exp carotid artery disease/su [Surgery]
4. exp carotid artery/
5. exp carotid artery disease/
6. carotid.tw.
7. 4 or 5 or 6
8. endarterectomy/
9. (endarterectom$ or surg$).tw.
10. 8 or 9
11. 7 and 10
12. 1 or 2 or 3 or 11
13. exp anesthesia/
14. exp anesthetic agent/
15. exp local anesthetic agent/
16. (anesthe$ or anaesthe$).tw.
17. cervical plexus/
18. cervical block.tw.
19. (bupivacaine or lidocaine or lignocaine or prilocaine or ropivacaine or mepivacaine or alfentanil or propofol or fentanyl or ketamine or midazolam or sevoflurane or desflurane or etomidate or isoflurane).tw.
20. 13 or 14 or 15 or 16 or 17 or 18 or 19
21. 12 and 20
22. Randomized Controlled Trial/
23. Randomization/
24. Controlled Study/
25. control group/
26. clinical trial/ or phase 1 clinical trial/ or phase 2 clinical trial/ or phase 3 clinical trial/ or phase 4 clinical trial/ or controlled clinical trial/
27. Double Blind Procedure/
28. Single Blind Procedure/ or triple blind procedure/
29. drug comparison/ or drug dose comparison/
30. "types of study"/
31. random$.tw.
32. (controlled adj5 (trial$ or stud$)).tw.
33. (clinical$ adj5 trial$).tw.
34. ((control or treatment or experiment$ or intervention) adj5 (group$ or subject$ or patient$)).tw.
35. (quasi‐random$ or quasi random$ or pseudo‐random$ or pseudo random$).tw.
36. ((control or experiment$ or conservative) adj5 (treatment or therapy or procedure or manage$)).tw.
37. ((singl$ or doubl$ or tripl$ or trebl$) adj5 (blind$ or mask$)).tw.
38. versus.tw.
39. (assign$ or allocat$).tw.
40. controls.tw.
41. trial.ti. or (RCT or RCTs).tw.
42. or/22‐41
43. 21 and 42
44. (exp animals/ or exp invertebrate/ or animal experiment/ or animal model/ or animal tissue/ or animal cell/ or nonhuman/) not (human/ or normal human/ or human cell/)
45. 43 not 44

Appendix 4. WHO ICTRP search strategy

Basic search: anaesthesia AND carotid endarterectomy

Phase are: ALL.

Appendix 5. ClinicalTrial.gov sea rch strategy

anaesthesia AND AREA[StudyType] EXPAND[Term] COVER[FullMatch] "Interventional" AND AREA[ConditionSearch] carotid endarterectomy AND AREA[StudyFirstPostDate] EXPAND[Term] RANGE

Data and analyses

Comparison 1. Local versus general anaesthetic: randomised trials.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Any stroke within 30 days of operation	13	4663	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.91 [0.66, 1.26]
1.2 Death within 30 days of operation	12	4421	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.61 [0.35, 1.06]
1.3 Ipsilateral stroke within 30 days of operation	2	3733	Peto Odds Ratio (Peto, Fixed, 95% CI)	1.03 [0.71, 1.48]
1.4 Stroke or death within 30 days of operation	11	4391	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.85 [0.62, 1.16]
1.5 Myocardial infarction within 30 days of operation	12	4567	Peto Odds Ratio (Peto, Fixed, 95% CI)	1.53 [0.67, 3.47]
1.6 Local haemorrhage	6	4186	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.99 [0.79, 1.25]
1.7 Sensitivity analysis: random‐effects model, local haemorrhage	6	4186	Odds Ratio (IV, Random, 95% CI)	0.80 [0.33, 1.94]
1.8 Arteries shunted	8	4133	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.25 [0.22, 0.29]
1.9 Sensitivity analysis: random‐effects model, arteries shunted	8	4133	Odds Ratio (IV, Random, 95% CI)	0.24 [0.08, 0.73]

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Binder 1999.

Study characteristics
Methods	RCT Block randomisation Blinding: unclear C: unclear Cross‐over: yes, but number excluded during trial was unclear Losses to FU: none
Participants	Austria 1999 46 participants (46 operations) Age mean: 73 years (LA), 68 years (GA) Sex: unclear Comparability: unclear Indications for surgery: TIA, stroke, incidental diagnosis of carotid stenosis
Interventions	LA: superficial and deep block with bupivacaine GA: thiopental, vecuronium, fentanyl Patching: all cases Antiplatelet Rx: unclear Indication for shunting: using intraluminal shunting for all participants
Outcomes	Death, any stroke, TIA, myocardial infarction, time in hospital since surgery, bleeding, mean arterial blood pressure, shunted arteries
Notes	FU: 48 hours Ex: recent neurological deficit < 4 weeks, redo operation, recent myocardial infarction ( < 2 months), ASA score ≥ 4, and any factor precluding randomisation such as pulmonary disease or refusal to participate in the study
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "46 Patienten nach Aufklarung and Unterzeichnung einer Einverstandniserklarung in die Studie aufgenommen und prospektiv randomisiert untersucht"
Allocation concealment (selection bias)	Unclear risk	Not reported
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and surgeons were not blinded to treatment group
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not reported
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Not reported
Selective reporting (reporting bias)	Low risk	All the study's prespecified outcomes of interest were reported
Other bias	Low risk	None suspected

Forssell 1989.

Study characteristics
Methods	RCT C: unclear Blinding: unclear Sequentially numbered envelopes Cross‐overs: 8 LA performed under exclusions during trial: 8 cross‐overs Losses to FU: none
Participants	Sweden 1985 to 1987 100 participants, 111 operations Age (mean): 66 years (LA), 63 years (GA) Male: 71% (LA), 64% (GA) Comparability: groups similar for vascular risk factors Indication for surgery: not reported
Interventions	LA: cervical block and skin infiltration with bupivacaine and mepivacaine or adrenaline GA: thiopental, isoflurane and bupivacaine or adrenaline skin infiltration Patching: not reported Antiplatelet Rx: not reported Indication for shunting: LA: neurological symptoms during/or after 1 minute test clamp; GA: stump pressure < 25 mmHg in TIA, stump pressure < 50 mmHg in vertebrobasilar insufficiency, always if previous stroke
Outcomes	Death, any stroke, myocardial infarction, wound haematoma, blood pressure, shunted arteries
Notes	FU: hospital stay Ex: consent refused, allergy to LA, ongoing heparin infusion, serious chronic cerebral insufficiency, uneasy during previous LA, randomisation miss, anxiety, simultaneous aortic repair, emergent operation
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "The remaining patients were randomised on 111 occasions into two groups, which were comparable (Table 2)"
Allocation concealment (selection bias)	Unclear risk	Not reported
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and surgeons were not blinded to treatment group
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not reported
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Not reported
Selective reporting (reporting bias)	High risk	No prespecified outcomes were reported
Other bias	Low risk	None suspected

GALA 2008.

Study characteristics
Methods	RCT C: central trial office allocation and concealment Blinding: single blinding ‐ independent assessor Block randomisation Cross‐overs: 92/1751 (5.3%) GA participants and 75/1771 (4.2%) LA participants went to theatre, but received the opposite treatment allocation to that allocated at randomisation Losses to FU: 3/3523 (0.09%)
Participants	Multicentre RCT conducted mainly in Europe (95 centres), in 24 countries 2003 to 2008 3526 operations Age (mean): 69 years (LA), 70 years (GA) Male: 71% (LA), 70% (GA) Comparability: groups similar for vascular risk factors Indication for surgery: all participants with symptomatic or asymptomatic carotid stenosis for whom surgery was advised The reasons for using shunts varied in both the LA and GA groups depending on the practice of each trial site. These reasons included: used routinely, drop velocity on TCD, unable to use TCD, contralateral occlusion or near occlusion, low stump pressure, contralateral carotid stenosis, recent stroke, unusual or damaged vein or arteries in head or neck, EEG or evoked potential change, blood pressure drop, falling brain oxygen level, operation converted to vein bypass, and unknown.
Interventions	LA versus GA
Outcomes	Primary outcome: proportion of participants alive, stroke free (including retinal infarction) and without myocardial infarction 30 days post‐surgery Secondary outcomes: proportion alive and stroke free at 1 year and in the longer term, a comparison of health‐related quality of life at 30 days and any surgical adverse events, re‐operation and re‐admission rates, the relative cost of the 2 methods of anaesthesia, length of stay and intensive and high dependency bed occupancy
Notes	FU: perioperative period (30 days after operation) and 1 year follow‐up Ex: a simultaneous bilateral carotid endarterectomy or carotid endarterectomy combined with another operative procedure such as coronary artery bypass surgery
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Central computerised randomisation
Allocation concealment (selection bias)	Low risk	Quote: "the office randomised patients to surgery under either general or local anaesthesia, stratified by centre and with balanced blocks of variable size, ensuring that allocation was completely concealed before the decision to randomise a patient and after baseline data were received"
Blinding of participants and personnel (performance bias) All outcomes	High risk	Quote: "we could not blind patients or the surgical team to randomised treatment allocation"
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "However, the independent stroke physician or neurologist who saw patients 1 month after surgery was unaware of the type of anaesthesia that the patients had received, although this blinding could be broken by the patient or by looking at hospital notes"
Incomplete outcome data (attrition bias) All outcomes	Low risk	Quote: "21 incomplete follow‐up, 1 no follow‐up at all, 1 no post‐surgery form, 19 no physician follow‐up at one month, five of these had patient follow‐up at 1 year"
Selective reporting (reporting bias)	Low risk	All data were analysed and reported as predefined
Other bias	Low risk	Measurement bias: Quote: "a neurologist (CPW), unaware of treatment allocation, then prepared a summary for every patient that, depending on the outcome, was audited by an independent neurologist (PMR) or cardiologist (APB), who were also unaware of treatment allocation" Quote: "Data were analysed by the trial statistician (SCI) and reviewed annually in strict confidence by the Data Monitoring Committee. Everyone else involved in the study was unaware of the treatment allocation until the database was locked" Funding bias: Quote: "The funding source had no role in the study design, data collection, data analysis, data interpretation or writing of the report"

Gimenez 2004.

Study characteristics
Methods	RCT C: unclear Blinding: unclear Cross‐over: unclear Exclusion during trial: unclear Losses to FU: unclear
Participants	Spain 1999 to 2001 93 participants and 93 operations Age: not indicated Male proportion: unclear Comparability: not reported Indication for surgery: not reported
Interventions	LA: not reported GA: not reported Patching and antiplatelet Rx: not reported Indication for shunting: not reported
Outcomes	Blood pressure
Notes	FU: probably hospital discharge Ex: not reported Data were extracted only from abstract and we could not contact the authors of this publication
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "In a prospective randomised study between 1999 and 2001, 93 patients underwent carotid endarterectomy, 47 under GA and 46 under LRA"
Allocation concealment (selection bias)	Unclear risk	Not reported (only abstract available)
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and surgeons were not blinded to treatment group
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not reported (only abstract available)
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Not reported (only abstract available)
Selective reporting (reporting bias)	High risk	No prespecified outcomes were reported
Other bias	Low risk	None suspected

Hussien 2017.

Study characteristics
Methods	RCT C: using opaque, sealed envelopes  Blinding: not reported Cross‐over: not reported Exclusion during trial: 2 allergic reaction, 1 severe vagal response, 1 ventricular tachycardia, 2 severe hypotension Losses to FU: no FU
Participants	Egypt 2017 39 participants, 36 operations, 30 analysed Age (mean): 64.7 years (LA), 65.2 years (GA) Male: 67% (LA), 67% (GA) Comparability: groups similar for vascular risk factors Indication for surgery: ASA physical status II or III
Interventions	LA: superficial and deep cervical plexus block using 0.5% bupivacaine + 2% lidocaine GA: propofol and fentanyl Patching and antiplatelet: not reported Indication for shunting: not reported
Outcomes	Death
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	The sequence generation was administered by a computerised random number generator
Allocation concealment (selection bias)	Low risk	Allocation concealment was conducted using opaque, sealed envelopes
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and surgeon were not blinded to the treatment
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not reported
Incomplete outcome data (attrition bias) All outcomes	Low risk	Missing data were balanced between groups
Selective reporting (reporting bias)	Low risk	All pre‐specified outcomes were reported
Other bias	Low risk	None suspected

Kasprzak 2006.

Study characteristics
Methods	RCT C: unclear Blinding: independent neurologist Cross‐over: 6 participants changed from LA to GA Exclusion during trial: none Losses to FU: none
Participants	Germany 2006 186 participants,186 operations Age (mean): 69 years (LA), 69 years (GA) Male: 67% (LA), 61% (GA) Comparability: groups similar for vascular risk factors Indication for surgery: asymptomatic carotid stenosis > 80% or symptomatic carotid stenosis > 70%
Interventions	LA: superficial and deep cervical plexus block by 0.5% bupivacaine + 1% prilocaine GA: fentanyl, etomidate, vecuronium, isoflurane Patching and antiplatelet: not reported Indication for shunting: LA: motor deficit, aphasia and loss of consciousness during carotid artery clamping; GA: decrease > 30% of amplitude in the baseline somatosensory evoked potential
Outcomes	Death, stroke, myocardial infarction, blood pressure, shunting
Notes	FU: possibly hospital stay Ex: not meeting inclusion criteria, refused to participate, withdrew consent, temporarily not operable, pilot study and other reason
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "randomized by computer random list for one type of anesthesia"
Allocation concealment (selection bias)	Unclear risk	Not reported
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and surgeons were not blinded to treatment group
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "a neurological evaluation by a neurologist and a routine postoperative CT scan were done on day b2 or 3 after surgery. The neurologist was not informed about the type of anesthesia"
Incomplete outcome data (attrition bias) All outcomes	Low risk	Quote: "received allocated treatment n = 95 analysed n = 95"
Selective reporting (reporting bias)	Low risk	All the study's prespecified outcomes of interest were reported
Other bias	Low risk	None suspected

Luchetti 2008.

Study characteristics
Methods	RCT C: unclear Blinding: unclear Cross‐over: unclear Exclusion during trial: unclear Losses to FU: unclear
Participants	Italy 2008 28 participants, 28 operations Age/male: unclear, but publication indicated that demographic data and baseline haemodynamic values were comparable Indication for surgery: unclear
Interventions	LA: superficial cervical plexus block by 0.5% ropivacaine 30 cc GA: superficial cervical plexus block with continuous infusion of remifentanil, propofol with intubation and mechanical ventilation Patching and antiplatelet: not reported Indication for shunting: LA: following carotid clamping, change in mental evaluation defined as agitation, confusion, contralateral weakness, seizure, unresponsiveness
Outcomes	Hemodynamic stability (mean arterial pressure), death, neurological deficit, cardiopulmonary complication
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "They were randomly assigned by means of a computer‐generated random number table to receive 1 of 2 anaesthesia techniques"
Allocation concealment (selection bias)	Unclear risk	Not reported
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and surgeons were not blinded to treatment group
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not reported
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Not reported
Selective reporting (reporting bias)	Low risk	Not all prespecified outcomes were reported ‐ namely, time to recovery of consciousness, perioperative complications such as nausea, vomiting, sweating, and grade on pain perception and comfort. However, these outcomes did not influence the main research question.
Other bias	Low risk	None suspected

Mazul‐Sunko 2010.

Study characteristics
Methods	RCT C: unclear Blinding: unclear Cross‐over: unclear Exclusion during trial: unclear Losses to FU: unclear
Participants	Croatia 2010 57 participants, 57 operations Mean age: 66.2 years (LA); 66 years (GA) Percentage male: 89.6% (LA), 85.7% (GA) Demographic data and baseline data are comparable Indication for surgery: carotid stenosis 70% or more
Interventions	LA: superficial cervical plexus block by levobupivacaine (1.5 mg/kg) and supplemental infiltration by surgeons with 1% lidocaine GA: etomidate in a dosage of 0.2 mg/kg and fentanyl (3 microgram/kg) for indication, vecuronium (0.08 mg/kg) for paralysis, maintain with isoflurane 0.7 to 1.2 MAC in a mixture of oxygen and nitrous oxide 50%:50%. Reversal with neostigmine (2.5 mg) and atropine (1 mg) Indication for shunting LA: following carotid clamping, neurological deficit GA: routine shunting was used except when technical difficulties
Outcomes	Stroke, death, myocardial infarction, shunting
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Elective carotid CEA were prospectively randomised to received either general or regional anaesthesia"
Allocation concealment (selection bias)	Unclear risk	Not reported
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and surgeons were not blinded to treatment group
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not reported
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Not reported
Selective reporting (reporting bias)	Low risk	All prespecified outcomes were reported
Other bias	Low risk	None suspected

McCarthy 2004.

Study characteristics
Methods	RCT C: unclear Blinding: not reported Cross‐over: not reported Exclusion during trial: none Losses to FU: none
Participants	UK 2004 176 participants and 176 operations Age (mean): 71 years (LA), 72 years (GA) Male: 61% (LA), 68% (GA) Comparability: groups similar for vascular risk factors Indication for surgery: not reported
Interventions	LA: not reported GA: not reported Patching: not reported Antiplatelet Rx: not reported Indication for shunting: not reported
Outcomes	Stroke, TIA, myocardial infarction, wound complication
Notes	FU: probably in‐hospital stay Ex: not reported
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote:"The CEA‐EQ questionnaire was administered to 176 CEA patients, prospectively randomised to either GA or LA in two hospitals, the Royal United Hospital Bath and The General Informary, Leeds"
Allocation concealment (selection bias)	Unclear risk	Not reported
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and surgeons were not blinded to treatment group
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not reported
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Not reported
Selective reporting (reporting bias)	High risk	No prespecified outcomes were reported
Other bias	Low risk	None suspected

Moritz 2010.

Study characteristics
Methods	RCT Randomisation: computerised randomisation C: not reported Blinding: not reported Cross‐over: 2 participants crossed over from LA to GA but these 2 participants were excluded from study Exclusion during trial: 6 (2 withdrawal of consent, 4 incomplete data) Losses to FU: not reported
Participants	Germany 2010 96 participants, 96 operations Age (mean): of all participants, 69 years Male: 68.8% (LA), 70.8% (GA) Comparability for vascular risk factors, preoperative symptom, ASA classification Indication for surgery: symptomatic 70% to 99%, asymptomatic 80% to 99%
Interventions	LA: superficial + deep cervical block by 1% prilocaine GA: fentanyl, propofol, rocuronium and anaesthesia maintain by inspired sevoflurane, bolus fentanyl Patching: not reported Antiplatelet Rx: not reported The indication for shunting for LA group was any neurological deterioration like speech abnormality, hemiparesis, or impaired consciousness The indication for shunting in the GA group was N20/P25 amplitude of the somatosensory evoked potential decreased to or below 30% of the baseline value
Outcomes	Stroke, myocardial infarction, cardiopulmonary data (blood pressure, heart rate), comparison neuromonitoring various methods i.e. stump pressure, transcranial Doppler, near‐infrared spectroscopy, somatosensory evoked potentials
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "All patients were randomised to either sevoflurane/fentanyl anaesthesia (GA = general anaesthesia) or regional anaesthesia (RAI) using a computerized system"
Allocation concealment (selection bias)	Unclear risk	Not reported
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and surgeons were not blinded to treatment group
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not reported
Incomplete outcome data (attrition bias) All outcomes	Low risk	Quote: "a total of 106 patients were randomised to the sevoflurane/ fentanyl (n=53) and regional (n=53) anaesthesia groups. Four patients were excluded because of withdrawal of consent (2 in each group), 2 patients because of conversion to general anaesthesia, and 4 patients because of incomplete data acquisition (3 in GA and 1 in RA). Thus, the final analysis was conducted in 96 patients (GA: n=48; RA: n=48)"
Selective reporting (reporting bias)	Low risk	All prespecified outcome were reported
Other bias	Low risk	None suspected

Mrozek 2007.

Study characteristics
Methods	RCT C: unclear Blinding: not reported Cross‐over: no crossovers from LA to GA Exclusion during trial: not reported Losses to FU: not reported
Participants	Olomouc 2007 80 participants, 80 operations Age (mean): 67 years (LA), 67 years (GA) Male: 55% (LA), 87.5% (GA) Comparability for vascular risk factors: not reported Indication for surgery: not reported
Interventions	LA: superficial + deep cervical block by 0.5% bupivacaine under neurostimulator GA: intravenous etomide, thiopental, atracurium, midazolam, fentanyl and atracurium Patching: not reported Antiplatelet Rx: not reported Indication for shunting in LA group was loss of consciousness and loss of motor function following carotid clamping Indication for shunting in GA group: not reported
Outcomes	Hemodynamic parameter (blood pressure, pulse rate) death, stroke, myocardial infarction, participants' subjective feeling
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quote: "the patients were randomised into two groups according to the first six digits of their date of birth (YYMMDD): odds to CB and even to GA"
Allocation concealment (selection bias)	Unclear risk	Not reported
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and surgeons were not blinded to treatment group
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not reported
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Not reported
Selective reporting (reporting bias)	Low risk	All prespecified outcomes were reported
Other bias	Low risk	None suspected

Orlický 2019.

Study characteristics
Methods	RCT Randomisation: zip code C: unclear Blinding: assessor (neurologist) Cross‐overs: none Exclusions during trial: none Losses to FU: none
Participants	Czech Republic 2019 210 participants, 210 operations Age (mean): 68 years (LA), 63 years (GA) Male: 64% (LA), 67% (GA) Comparability: higher age in LA Indication for surgery: ICA stenosis ≥ 50%
Interventions	LA: cervical plexus block with local infiltration GA: received either gaseous volatile anaesthetic agents with muscle relaxant or total intravenous anaesthesia Patching: not reported Indication for shunting (LA): carotid artery clamping Indication for shunting (GA): decreased somatosensory evoked potential amplitude > 50% recorded from the contralateral median nerve or decreased peak systolic velocity in the ipsilateral middle cerebral artery > 70%
Outcomes	Death, any stroke, myocardial infarction, haematoma, blood pressure
Notes	FU: within 30 days of CEA Ex: none
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Using zip code: zip code between 10000 and 49999 were assigned to LA, while zip code between 50000 and 79999 were assigned to GA
Allocation concealment (selection bias)	Unclear risk	Information from study author: "Zip code randomization was used to allocate patients" but we were not sure whether this used a sequentially numbered, concealed envelope or not
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and surgeon were not blinded to the treatment
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Outcome measures were evaluated by a blinded certified neurologist
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing data addressed
Selective reporting (reporting bias)	Low risk	All pre‐specified outcome measures were reported
Other bias	Low risk	None suspected

Pluskwa 1989.

Study characteristics
Methods	RCT C: random number list Blinding: unclear Cross‐overs: none Exclusions during trial: none Losses to FU: none
Participants	France 1989 20 participants, 20 operations Age (mean): 66 years (LA), 63 years (GA) Male: 90% (LA), 70% (GA) Comparability: groups similar for vascular risk factors Indication for surgery: not reported
Interventions	LA: epidural (C7‐T1) by bupivacaine and fentanyl GA: flunitrazepam, fentanyl, vecuronium Patching: not reported Antiplatelet Rx: not reported Indication for shunting: not reported
Outcomes	Death, any stroke, myocardial infarction, blood pressure
Notes	FU: probably hospital discharge Ex: bleeding risk, on anticoagulants
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "La veille de l'intervention. Ces patients ont ete repartis en deux groupes par tirage au sort a partir d'une serie de nombres au hasard"
Allocation concealment (selection bias)	Unclear risk	Not reported
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and surgeons were not blinded to treatment group
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not reported
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Not reported
Selective reporting (reporting bias)	Low risk	Not all prespecified outcomes were reported ‐ namely, heart rate. However, this outcome did not influence the main research question.
Other bias	Low risk	None suspected

Prough 1989.

Study characteristics
Methods	RCT C: unclear Blinding: unclear Cross‐overs: none Exclusions during trial: none Losses to FU: none
Participants	USA 1989 23 participants, 23 operations Age (mean): 67 years (LA), 61 years (GA) Male: 69% (LA), 40% (GA) Comparability: groups similar for preoperative physical status Indication for surgery: not reported
Interventions	LA: superficial cervical block GA: thiopental, pancuronium, isoflurane Patching and antiplatelet Rx: not reported Indication for shunting: not reported
Outcomes	Death, any stroke, myocardial infarction, blood pressure
Notes	FU: probable hospital discharge Ex: 5 participants refused GA so not randomised
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Patients who consented to either form of anaesthesia were randomised to received regional or general anaesthesia"
Allocation concealment (selection bias)	Unclear risk	Not reported
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and surgeons were not blinded to treatment group
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not reported
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Not reported
Selective reporting (reporting bias)	Low risk	Not all prespecified outcomes were reported ‐ namely, intra‐operative and post‐operative intravenous fluid administration and urine output. However, these outcomes did not influence answering the main research question.
Other bias	Low risk	None suspected

Sbarigia 1999.

Study characteristics
Methods	RCT Randomisation: casual numbers C: unclear Blinding: assessor (neurologist) Cross‐overs: 2 exclusions during trials Losses to FU: 18
Participants	Italy 1995 to 1998 107 participants, 107 operations Age (mean): 69.0 years (LA), 70.4 years (GA) Males: 87.3% (LA), 88.5% (GA) Comparability: groups similar for vascular risk factors Indication for surgery: TIA, asymptomatic carotid stenosis > 70%, stroke
Interventions	LA: superficial and deep cervical block with bupivacaine GA: alfentanil + propofol or sodium thiopental + fentanyl + isoflurane or vecuronium + nitrous oxide Patching: LA 36.4%, GA 23.1% Antiplatelet Rx: not reported Indication for shunting: LA: neurological test (toy‐squeaker squeezing test); GA: not reported
Outcomes	Death, any stroke, myocardial infarction, TIA, bleeding, shunted arteries
Notes	FU: 30 days Ex: clinical signs of congestive heart disease, severe valvular heart disease, unstable angina, left bundle branch block (by ECG)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "randomization by means of casual numbers"
Allocation concealment (selection bias)	Unclear risk	Not reported
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and surgeons were not blinded to treatment group
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not reported
Incomplete outcome data (attrition bias) All outcomes	Low risk	2 cross from LA to GA, no dropouts Quote: "two patients were excluded after randomisation and the operations were done under GA"; "in both cases, the anaesthesiologist considered the infiltration of LA to be dangerous"
Selective reporting (reporting bias)	Low risk	All prespecified outcomes were available
Other bias	Low risk	None suspected

Sindelic 2004.

Study characteristics
Methods	RCT Randomisation, concealment and blinding of assessor: unclear Cross‐over: unclear Losses to FU: unclear
Participants	Serbia 50 participants, 50 operations Mean age: 64.4 years (GA), 65.9 years (LA) Comparability: groups similar for vascular risk factors Indication for surgery: unclear Males: 56% (GA), 52% (LA)
Interventions	LA: superficial and deep cervical plexus block: superficial block was done with 15 cc 0.5% bupivacaine and 5 cc 2% lidocaine injection along posterior border of sternocleidomastoid muscle Deep cervical block was performed with 3 injection techniques for blockages of C2, C3 and C4 segment GA: thiopental + fentanyl + rocuronium Patching and antiplatelet Rx and indication for shunting: not reported
Outcomes	Blood pressure
Notes	FU: unclear Ex: unclear
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: " Bolesnici su randomizirani u jednu od dve grupe, shodno anestezioloskim postupeima koji ce se sprovesti u toku operaciji: grupe opste anestezije (OA) i grupe regionalne anesezije (RA)."
Allocation concealment (selection bias)	Unclear risk	Not reported
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and surgeons were not blinded to treatment group
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not reported
Incomplete outcome data (attrition bias) All outcomes	Low risk	No participants lost
Selective reporting (reporting bias)	High risk	Selected to report parameters only at T2 time which is the only one that is significant but not pre‐specified Quote: "Zbog znacajna medjugrupna u grupi OA postojiu vremenu T2 i znacajna razlika u tom vremenu (OA vs RA,p < 0.01)"
Other bias	Low risk	None suspected

ASA: American Society of Anaesthesiologists
C: concealment of allocation
CABG: coronary artery bypass grafting
CEA: carotid endarterectomy
CT: computerised tomography
ECG: electrocardiography
EEG: electroencephalography
Ex: exclusion criteria
FU: follow‐up
GA: general anaesthetic
ICA: internal carotid artery
ICU: intensive care unit
IHD: ischaemic heart disease
LA: local anaesthetic
MCA: middle cerebral artery
RCT: randomised controlled trial
Rx: therapy
TCD: transcranial Doppler
TIA: transient ischaemic attack

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Krajnic 2011	Comparative study between general anaesthesia and combined general anaesthesia and cervical plexus blockade
Martynov 2018	Comparative study between general and general anaesthesia
Ozturk 2018	Comparative study between cervical plexus block (local anaesthesia) and cervical plexus block (local anaesthesia) plus mandibular block
Ramachandran 2011	Comparative study between superficial cervical plexus blocks (local anaesthesia) and deep cervical blocks (local anaesthesia)
Sindjelic 2010	Comparative study between two different drugs during local anaesthesia

Differences between protocol and review

Non‐randomised studies, which are prone to bias and which were previously included in the review, have been removed.

Contributions of authors

Amaraporn Rerkasem, Saritphat Orrapin, Dominic PJ Howard, Sothida Nantakool, Kittipan Rerkasem: designed the conceptual framework; performed searches; drafted, revised, and approved the final manuscript.

Amaraporn Rerkasem, Sothida Nantakool, Kittipan Rerkasem: involved in data collection, data analysis, and data selection.

Amaraporn Rerkasem, Saritphat Orrapin, Sothida Nantakool: judged the quality of evidence.

Sources of support

Internal sources

• Faculty of Medicine, Chiang Mai University, Thailand

• Center for Applied Science, Research Institute of Health Sciences, Chiang Mai University, Chiang Mai, Thailand

External sources

• The Thailand Research Fund, Thailand

• Stroke Prevention Research Unit, Nuffield Department of Clinical Neurosciences, University of Oxford, England, UK

Declarations of interest

AR: none known.
SO: none known.
DPJH: none known.
SN: none known.
KR: none known.

New search for studies and content updated (no change to conclusions)