Epidural analgesia for cardiac surgery

  • Review
  • Intervention

Authors


Abstract

Background

A combination of general anaesthesia (GA) with thoracic epidural analgesia (TEA) may have a beneficial effect on clinical outcomes by reducing the risk of perioperative complications after cardiac surgery.

Objectives

The objective of this review was to determine the impact of perioperative epidural analgesia in cardiac surgery on perioperative mortality and cardiac, pulmonary or neurological morbidity. We performed a meta-analysis to compare the risk of adverse events and mortality in patients undergoing cardiac surgery under general anaesthesia with and without epidural analgesia.

Search methods

We searched the Cochrane Central Register of Controlled Trials (CENTRAL) (2012, Issue 12) in The Cochrane Library; MEDLINE (PubMed) (1966 to November 2012); EMBASE (1989 to November 2012); CINHAL (1982 to November 2012) and the Science Citation Index (1988 to November 2012).

Selection criteria

We included randomized controlled trials comparing outcomes in adult patients undergoing cardiac surgery with either GA alone or GA in combination with TEA.

Data collection and analysis

All publications found during the search were manually and independently reviewed by the two authors. We identified 5035 titles, of which 4990 studies did not satisfy the selection criteria or were duplicate publications, that were retrieved from the five different databases. We performed a full review on 45 studies, of which 31 publications met all inclusion criteria. These 31 publications reported on a total of 3047 patients, 1578 patients with GA and 1469 patients with GA plus TEA.

Main results

Through our search (November 2012) we have identified 5035 titles, of which 31 publications met our inclusion criteria and reported on a total of 3047 patients. Compared with GA alone, the pooled risk ratio (RR) for patients receiving GA with TEA showed an odds ratio (OR) of 0.84 (95% CI 0.33 to 2.13, 31 studies) for mortality; 0.76 (95% CI 0.49 to 1.19, 17 studies) for myocardial infarction; and 0.50 (95% CI 0.21 to 1.18, 10 studies) for stroke. The relative risks (RR) for respiratory complications and supraventricular arrhythmias were 0.68 (95% CI 0.54 to 0.86, 14 studies) and 0.65 (95% CI 0.50 to 0.86, 15 studies) respectively.

Authors' conclusions

This meta-analysis of studies, identified to 2010, showed that the use of TEA in patients undergoing coronary artery bypass graft surgery may reduce the risk of postoperative supraventricular arrhythmias and respiratory complications. There were no effects of TEA with GA on the risk of mortality, myocardial infarction or neurological complications compared with GA alone.

Resumo

Analgesia Epidural para cirurgia cardíaca

Introdução

Uma combinação de anestesia geral (AG) com analgesia epidural torácica (AET) pode ter um efeito benéfico com resultados clínicos por reduzir o risco de complicações peri operatórias apos a cirurgia cardíaca.

Objetivos

O objetivo dessa revisão foi determinar o impacto peri-operatório da analgesia epidural em cirurgia cardíaca sobre a mortalidade e sobre a morbidade cardíaca, pulmonar ou neurológica. Nós realizamos uma meta-análise para comparar o risco de eventos adversos e mortalidade em pacientes submetidos a cirurgia cardíaca sobre anestesia geral com e sem analgesia epidural.

Métodos de busca

Nós buscamos a Cochrane Central Register of Controlled Trials (CENTRAL) (2012, Volume 12) na Biblioteca Cohcrane, MEDLINE (Pubmed) (1966 até novembro de 2012); EMBASE (1989 até novembro de 2012); CINHAL(1982 até novembro de 2012) e a Science Citation Índex (1988 até novembro de 2012).

Critério de seleção

Nós incluímos ensaios clínicos randomizados comparando resultados em pacientes adultos submetidos a cirurgia cardíaca apenas sob AG ou sob AG combinada com AET.

Coleta dos dados e análises

Todas as publicações encontradas durante as busca foram manualmente e independentemente revisadas pelos dois autores. Nós identificamos 5035 títulos, dos quais 4990 estudos não satisfizeram os critérios de seleção ou eram publicações duplicadas, que foram obtidos das cinco diferentes bases de dados. Nós realizamos uma revisão completa em 45 estudos, dos quais 31 publicações tinham todos os critérios de inclusão. Essas 31 publicações reportaram total de 3047 pacientes, 1578 com AG e 1469 pacientes com AG associado à AET

Principais resultados

Através de nossas buscas (até novembro de 2012) nós havíamos identificado 5035 títulos, dos quais 31 publicações apresentavam nossos critérios de inclusão de reportaram um total de 3047 pacientes. Comparado com AG, o risco relativo para pacientes recebendo AG com AET mostrou uma odds ratio de 0,84 (95%, com intervalo confiança 0,33 a 2,13, 31 estudos) para mortalidade; 0,76(95%, com intervalo de confiança de 0,49 até 1,19, 17 estudos) para infarto do miocárdio e 0,50 (95%, com intervalo e confiança 0,21 a 1,18, 10 estudos) para acidente vascular cerebral. O risco relativo (RR) para complicações respiratórias e arritmias supraventriculares foram 0,68 (95%, com intervalo de confiança de 0,54 a 0,86, 14 estudos) e 0,65 (95%, com intervalo de confiança de 0,50 a 0,86, 15 estudos), respectivamente.

Conclusão dos autores

Esta meta-análise dos estudos identificada até 2010 mostrou que o uso de AET em pacientes submetidos a cirurgia de revascularização do miocárdio pode reduzir o risco de arritmias supraventriculares e complicações respiratórias no pós-operatório. Não existe nenhum efeito da AET associado a AG sobre o risco de mortalidade, infarto do miocárdio ou complicações neurológicas comparados com AG apenas.

Notas de tradução

Tradução da Unidade de Medicina Baseada em Evidências da Unesp, Brazil (Elio Ferreira de Oliveira Junior) Contato: portuguese.ebm.unit@gmail.com Translation notes: CD006715

Plain language summary

Epidural analgesia for heart surgery

The use of thoracic epidural analgesia (TEA) in heart surgery is controversial. TEA produces a high level of pain relief and may reduce some of the adverse effects that are associated with heart surgery. Along with these potential benefits of TEA there is a widespread fear that TEA could cause bleeding into the epidural space (haematoma). We have conducted a review of medical trials to compare the effect of general anaesthesia (GA) alone with GA in combination with epidural analgesia on major complications in patients undergoing elective heart surgery. We identified 5035 titles of which 31 publications met all the inclusion criteria. These 31 publications reported on a total of 3047 patients, 1578 patients with GA and 1469 patients with TEA as well as GA. The risks of death, heart attack and stroke were not statistically significant, however our analysis showed statistically significant reductions in the risk of arrhythmias and pulmonary complications. Despite this, our findings must be viewed with caution. Epidural analgesia in heart surgery remains controversial in the absence of a sufficiently large and statistical significant effect on mortality, stroke, or myocardial infarction and while the risks are unclear.

Resumo para leigos

Analgesia epidural para cirurgia cardíaca

O uso de analgesia epidural torácica (AET) em cirurgia cardíaca é controverso. A AET possui uma alta qualidade para alívio da dor e pode reduzir alguns efeitos adversos que estão associados a cirurgia cardíaca. Em associação com esses potencias benefícios existe o medo amplamente difundido de que a AET possa causar um sangramento para o espaço epidural (hematoma). Nos conduzimos uma revisão de estudos clínicos para comparar o efeito da anestesia geral (AG) sozinha com AG em combinação com epidural sobre as principais complicações em paciente submetidos à cirurgia cardíaca eletiva. Nós identificamos 5035 títulos, dos quais 31 publicações atenderam a todos os critérios de inclusão. Essas 31 publicações reportaram um total de 3047 patientes, sendo 1578 pacientes sob AG e 1469 pacientes com AG associada a AET. Não houve diferença estatística significante nos riscos de morte, ataque cardíaco e acidente vascular cerebral, entretanto a nossa análise mostrou uma redução com significância estatística do risco de arritmias e complicações pulmonares. Em virtude disso, nossos achados devem ser vistos com cuidado. A analgesia epidural em cirurgia cardíaca permanece controversa, com ausência de um efeito grande o suficiente ou com significância estatística sobre a mortalidade, acidente vascular cerebral ou infarto do miocárdio, enquanto que os riscos não são claros.

Notas de tradução

Tradução da Unidade de Medicina Baseada em Evidências da Unesp, Brazil (Elio Ferreira de Oliveira Junior) Contato: portuguese.ebm.unit@gmail.com Translation notes: CD006715

Summary of findings(Explanation)

Summary of findings for the main comparison. Epidural analgesia for cardiac surgery
  1. 1 Narrow confidence intervals, however high variability in effect estimates (I2 = 42%) and the definition of this endpoint was variable among the studies, for this reason the quality of evidence was down graded

    2 There was substantial variability in effect estimates (I2 = 67%)

    3 Large number of participants with narrow confidence intervals and small variability in effect estimates (I2 = 0%)

    4 Large number of participants and small variability in effect estimates, but with wide confidence intervals indicating imprecision in results

    5 The wide confidence intervals indicate imprecision in results

Studies: randomized clinical trials

Patient or population: adult patients undergoing cardiac surgery
Intervention: thoracic epidural anaesthesia
Comparison: general anaesthesia

OutcomesIllustrative comparative risks*Relative
effect
(95% CI)
Number of
participants
(studies)

Quality of the
evidence

(GRADE)

Assumed riskCorresponding risk
GATEA
respiratory complications144 per 1000

100 per 1000

(79 to 141)

0.66 (0.45, 0.98)2011 (14)++OO1
supraventricular arrhythmias372 per 1000

242 per 1000

(186 to 208)

0.65 (0.50, 0.86)2250 (15)++OO2
mortality7 per 1000

6 per 1000

(2 to 16)

0.84 (0.33, 2.13)2877 (31)+++O3
myocardial infarction44 per 1000

33 per 1000

(22 to 52)

0.76 (0.49, 1.19)2127 (17)+OOO4
stroke15 per 1000

8 per 1000

(3 to 18)

0.50 (0.21, 1.18)1791 (10)+OOO5
*The basis for the assumed risk (e.g. the mean control group risk) is provided in footnotes. The corresponding risk is based on the assumed risk in the comparison group and the effect estimate of the intervention.
CI: Confidence interval;
GRADE Working Group grades of evidence
High quality (++++): Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality (+++O): Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality (++OO): 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 (+OOO): We are very uncertain about the estimate.

Background

Description of the condition

Advances in anaesthesiology, surgery, extracorporeal perfusion techniques, and perioperative care medicine have resulted in a reduced risk of perioperative complications and subsequently have improved postoperative outcomes after cardiac surgery (Coriat 2001). The addition of thoracic epidural analgesia to general anaesthesia has been suggested to benefit patients after cardiac surgery in several trials (Royse 2003; Scott 2001). However, the technique is controversial because the insertion of an epidural catheter in patients requiring full heparinization for cardiopulmonary bypass may lead to an epidural haematoma.

Description of the intervention

A combination of general anaesthesia (GA) with thoracic epidural analgesia (TEA) may further improve outcomes after cardiac surgery (Djaiani 2000; Oxelbark 2001; Zarate 2000). TEA is a technique by which continuous infusion of local anaesthetic and opioid is given in the epidural space at high thoracic levels. TEA produces a high level of analgesia and may reduce the adverse effects that are associated with heart surgery.

A possible complication of TEA includes spinal cord compression, caused by a haematoma or abscess, which can result in paraplegia. Systemic anticoagulation is needed during cardiopulmonary bypass and may increase the incidence of epidural haematoma related to the use of an epidural catheter (Bang 2011; Ho 2000; Rosen 2004).

How the intervention might work

TEA may result in enhanced coronary perfusion, improved myocardial oxygen supply and a reduced incidence of dysrhythmias, via intense blockade of the sympathetic nervous system (Royse 2003; Scott 2001). Through the same mechanism, the size of perioperative myocardial infarction may be reduced (Beattie 2003; Scott 2001; Turfrey 1997). TEA may also shorten the duration of tracheal intubation as well as the time spent in an intensive care unit and hospital, which could have major economic benefits (Bowler 2002; Nicholson 2002; Priestley 2002), and may prevent pulmonary complications (Ballentyne 1998; Stenseth 1996).

Why it is important to do this review

A previous meta-analysis of 15 trials by Liu et al (Liu 2004) demonstrated fewer respiratory complications and dysrhythmias in patients assigned to TEA, but failed to show an effect of TEA on mortality or myocardial infarction. However, more trials comparing TEA and conventional anaesthesia for cardiac surgery are now available (Hansdottir 2006; Lundstrom 2005), and these were considered for inclusion in the present review.

The present meta-analysis could elucidate whether or not TEA has a benefit on relevant outcomes such as mortality and myocardial infarction. It does not clarify the clinical costs of TEA, which range from patient discomfort during epidural puncture to paraplegia. When the meta-analysis shows a significant benefit on mortality or other very relevant outcomes, whilst catastrophic complications are not observed in the patient samples that are included, one can argue that TEA should be used; although the safety of the technique has not been completely quantified. On the other hand, if no relevant benefits can be demonstrated, one could argue that TEA should not be applied in cardiac surgery as long as its safety remains unclear.

Objectives

The objective of this review was to determine the impact of perioperative epidural analgesia in cardiac surgery on perioperative mortality and cardiac, pulmonary or neurological morbidity.

Methods

Criteria for considering studies for this review

Types of studies

We included randomized controlled trials (RCT).

Types of participants

We included adult patients undergoing general anaesthesia (GA) for cardiac surgery with and without thoracic epidural analgesia (TEA).

Types of interventions

We included trials that compared cardiac surgery with and without TEA. We excluded studies which compared cardiac surgery with and without spinal anaesthesia.

Types of outcome measures

We did a separate meta-analysis for each of the outcomes listed below. We measured all outcomes at two weeks and six months after surgery.

Primary outcomes
  1. Risk of mortality

  2. Risk of myocardial infarction

We defined myocardial infarction as myocardial specific creatine kinase (CKMB) measurement greater than 50 U/l and peak CKMB/creatine kinase ratio of greater than 10%, or a new Q wave infarction.

Secondary outcomes
  1. Risk of in-hospital pulmonary complications (respiratory insufficiency or pneumonia)

  2. Risk of cardiac events (supraventricular or ventricular dysrhythmias or cardiac arrest)

  3. Risk of neurological complications (epidural haematoma or abscess, transient ischaemic attack or cerebrovascular accident)

Search methods for identification of studies

Electronic searches

We searched the Cochrane Central Register of Controlled Trials (CENTRAL) (2012, Issue 12) in The Cochrane Library, MEDLINE (PubMed) (1966 to November 2012); EMBASE (1989 to November 2012); CINAHL (1982 to November 2012); and the Web of Science (SCI/SSCI) (1988 to November 2012) with relevant search terms relating to heart surgery procedures and neuraxial anaesthetic techniques.

We combined these search terms with a sensitive methodological filter (Appendix 1) for randomized trials, found in Chapter 6 of the Cochrane Handbook for Systematic Reviews of Interventions Version 5.0.2 (Higgins 2011).

Searching other resources

We screened reference lists from retrieved randomized trials, meta-analyses and systematic reviews to identify additional trials. We conducted a search by author for any individual who appeared as an author of two or more relevant articles.

We used PubMed's 'See related articles' function with relevant trial reports to search for additional references. In addition, we included the references of the following reviews on cardiac surgery and epidural analgesia (Ballentyne 1998; Chaney 1997; Liu 2004; Meissner 1997; Rodgers 2000).

Data collection and analysis

Selection of studies

Two authors (VS and MP) independently examined the titles and, where available, abstracts of studies identified using the search strategy, and handsearched reference lists. We retrieved any full studies that met our predefined inclusion criteria. If our judgements on inclusion or exclusion were split we discussed the article until we achieved consensus, with the help of a third review author (DD) where necessary.

Data extraction and management

For each title, abstract or full-text article, the two review authors (VS and MP) used a standardized selection form (Appendix 2) to extract data independently. The review authors were not blinded to the authors' or journals' names. All included trials were thoroughly studied by the two authors using the data collection form. The items that were scored were also listed in a standardized form.

Assessment of risk of bias in included studies

Two review authors (VS and MP) independently assessed the quality of the studies according to the criteria described in the Cochrane Handbook for Systematic Reviews of Interventions Version 5.0.2 (Higgins 2011). We graded the risk of bias for each study in the categories of sequence generation, allocation concealment, blinding, incomplete outcome data and selective reporting. We graded each category as low risk (yes), high risk (no), or unclear risk of bias.

Measures of treatment effect

All outcomes were dichotomous. Both the sample sizes and the numbers of patients with events were summed across groups (Higgins 2011).

Unit of analysis issues

The odds were calculated per treatment group for all outcomes. The Peto odds ratio was used as the pooled measure of effect.

Dealing with missing data

We contacted the first authors of included trials to obtain missing data that were necessary for the meta-analysis.

Assessment of heterogeneity

When significant heterogeneity was suspected from the I2 statistic (I2 = 25% or higher) for heterogeneity, or from visual inspection of the results, we used a random-effects model for pooling. In this model the RR for every trial is weighted by the reciprocal of its variance, whereby studies with a smaller standard error were given more weight than those with a larger standard error. We have explored sources of heterogeneity by looking at possible differences in clinical and methodological factors across the trials.

Assessment of reporting biases

We assessed publication bias for the two primary outcomes by both graphical inspection of the funnel plot and statistical testing of plot asymmetry, using a 95% confidence interval.

Data synthesis

We used RevMan 5.1 for quantitative analysis. As all our outcomes were binary data, we related the numbers reporting an outcome in each group to the numbers at risk to derive a relative risk (RR) and 95% confidence interval, or as an odds ratio (OR) through the Peto method where appropriate. We applied the Peto method to obtain the OR for mortality as there were 15 trials with zero events, and for the myocardial infarction analysis as numbers of events were very small. As a general rule, we used a fixed-effect model for calculations of summary estimates and their 95% confidence intervals (CI).

Subgroup analysis and investigation of heterogeneity

We planned an exploratory subgroup analysis that included trials with two different methods of GA, the fast-track protocol (short-acting opioids) versus regular GA. We also performed a subgroup analysis on the results for mortality at the two-week period, for high and low risk of bias studies. A low risk of bias study was defined as a study where no categories or only the detection bias category (blinding of patient and medical personnel) were inadequate. A high risk of bias study was defined as a study where any of the other bias categories (sequence generation, allocation concealment, incomplete outcome data and selective reporting) were inadequate.

Sensitivity analysis

We conducted a sensitivity analysis in low risk of bias trials to estimate the robustness of the results for mortality at the two-week period.

Results

Description of studies

See Characteristics of included studies and Characteristics of excluded studies.

Results of the search

We identified 5035 titles, of which 4990 studies did not satisfy the selection criteria or were duplicate publications, from the five different databases (Figure 1).

Figure 1.

Study flow diagram.

Included studies

We performed a full review of 45 studies, of which 31 publications met our inclusion criteria (see Characteristics of included studies). These 31 publications reported on a total of 3047 patients, 1578 patients with GA and 1469 patients with GA plus TEA. There were no trials in languages other than English identified.

Excluded studies

We excluded 14 of the 45 assessed studies for the reasons shown in the table Characteristics of excluded studies.

Risk of bias in included studies

Incomplete data were rare and when present this was reported, making attrition bias minimal. Blinding of outcome assessors was rarely mentioned and was treated as a possible risk of detection bias.

Allocation

The risk of selection bias was unclear. We have included only randomized controlled trials in this meta-analysis, however the method of randomization and allocation concealment were not always clearly stated (Figure 2; Figure 3).

Figure 2.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figure 3.

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

Blinding

Performance bias was high in this meta-analysis as the patients and personnel were not blinded in 85% of the trials (Figure 2; Figure 3).

Incomplete outcome data

There was a minimal possibility of attrition bias as the outcome data were not complete in six trials (see Figure 2; Figure 3).

Selective reporting

There was a low risk of reporting bias (Figure 2; Figure 3).

Other potential sources of bias

Other potential sources of bias were not identified.

Effects of interventions

See: Summary of findings for the main comparison Epidural analgesia for cardiac surgery

Mortality

All 31 studies reported mortality (see Characteristics of included studies). Of those studies, 29 (Bach 2002; Bakhtiary 2007; Barrington 2005; Berendes 2003; Brix-Christensen 1998; Caputo 2009;Caputo 2011; de Vries 2002; El-Baz 1987; Fillinger 2002; Hansdottir 2006; Heijmans 2007; Kendall 2004; Kilickan 2006; Kirno 1994; Lagunilla 2006; Liem 1992; Lundstrom 2005; Moore 1995; Nygard 2004; Onan 2011; Rein 1989; Royse 2003; Scott 2001; Sharma 2010; Stenseth 1994; Stenseth 1996; Svircevic 2011; Tenling 1999) reported on mortality within two  weeks after surgery (Analysis 1.1), and two studies (Loick 1999; Priestley 2002) reported mortality after six-months follow-up (Analysis 1.2). Combining data from 29 studies showed no difference in the rate of mortality two weeks after surgery for patients receiving TEA with GA versus GA (OR 0.84; 95% CI 0.33 to 2.13) (Analysis 1.1). Of the 29 included studies, 17 did not provide an estimate resulting in the overall OR being based on 12 studies. The meta-analysis of studies that evaluated mortality six months after surgery included two studies with 170 patients (I2 = 0%) and yielded an RR of 1.35 (95% CI 0.08 to 22.83) (Analysis 1.2).

Subgroup analysis showed that the risk of mortality at the two-week period in both low (Barrington 2005; Berendes 2003; Caputo 2009; Caputo 2011; Hansdottir 2006; Kilickan 2006; Moore 1995; Scott 2001; Svircevic 2011; Tenling 1999) and high (Bach 2002; Bakhtiary 2007; Brix-Christensen 1998; de Vries 2002; El-Baz 1987; Fillinger 2002; Kendall 2004; Kirno 1994; Liem 1992; Lundstrom 2005; Nygard 2004; Onan 2011; Rein 1989; Royse 2003; Sharma 2010; Stenseth 1994; Stenseth 1996) risk of bias groups was not significantly different between the patients receiving TEA compared with patients receiving GA alone (OR 0.83; 95% CI 0.25 to 2.70 and OR 0.85; 95% CI 0.19 to 3.79, respectively). The explanatory subgroup analysis, which included trials with two different methods of general anaesthesia, the fast-track protocol (short-acting opioids) versus regular GA, was not feasible due to lack of data.

Sensitivity analysis (Analysis 6.1) showed that the risk of mortality within two weeks after surgery in low risk of bias studies was not different between the groups (OR 0.83; 95% CI 0.25 to 2.70).

Myocardial infarction

Seventeen studies reported on acute myocardial infarction (AMI). Fifteen studies (Bakhtiary 2007; Barrington 2005; Caputo 2009; Caputo 2011; de Vries 2002; Fillinger 2002; Hansdottir 2006; Heijmans 2007; Kendall 2004; Liem 1992; Onan 2011; Scott 2001; Stenseth 1994; Stenseth 1996; Svircevic 2011), which included 2127 patients, reported AMI after two weeks of follow-up (Analysis 2.1). There was no difference in the rate of AMI between groups of patients receiving TEA compared with patients receiving GA alone (OR 0.76; 95% CI 0.49 to1.19; I2 = 0% ). Three out of the 15 included studies did not provide an estimate, resulting in the overall OR being based on 12 studies. The meta-analysis of studies that evaluated AMI six months after surgery included two studies (Loick 1999; Priestley 2002) with 170 patients (I2 = 0%) and yielded an RR of 0.41 (95% CI 0.06 to 3.03) (Analysis 2.2).

Respiratory events

A total of 14 studies presented data on the number of patients who had respiratory complication within two weeks of surgery. Pooled analysis of data from 2011 patients (Barrington 2005; Berendes 2003; Caputo 2011; de Vries 2002; El-Baz 1987; Fillinger 2002; Hansdottir 2006; Liem 1992; Lundstrom 2005; Royse 2003; Scott 2001; Svircevic 2011; Tenling 1999) showed a lower risk of respiratory events for patient receiving TEA during surgery compared with those receiving GA alone (RR 0.68; 95% CI 0.54 to 0.86). The heterogeneity of these data was high (I2 = 42%) (Analysis 3.1). With the use of a random-effects model the RR for respiratory events within two weeks was RR 0.66 (95% CI 0.45 to 0.98). One study (Priestley 2002) had a six-month follow-up of respiratory events (RR 3.00; 95% CI 0.13 to 71.92); this study was not included in the meta-analysis.

Supraventricular tachyarrhythmias

Thirteen studies (Bakhtiary 2007; Barrington 2005; Caputo 2009; Caputo 2011; de Vries 2002; Fillinger 2002; Hansdottir 2006; Liem 1992; Kilickan 2006; Nygard 2004; Royse 2003; Scott 2001; Svircevic 2011) that included 2250 patients reported on supraventricular tachyarrhythmias (SVT) within the two weeks after surgery. The use of TEA was associated with a lower risk of SVT compared with GA alone (RR 0.65 95% CI 0.50 to 0.86) (Analysis 4.1). However, heterogeneity between the studies was high (I2 = 69%). With the use of a random-effects model the RR for SVT within two weeks was RR 0.65 (95% CI 0.50 to 0.86). After six-months follow-up, the beneficial effect of TEA on SVT was not statistically significant, but was only reported by two trials (Loick 1999; Priestley 2002) (RR 0.76; 95% CI 0.39 to 1.51) (Analysis 4.2).

Neurologic complications

Ten trials reported on stroke. One study (Priestley 2002) had a six-month follow-up (RR of 3.00, 95% CI: 0.13 to 71.92) and nine (Barrington 2005; Caputo 2009; Caputo 2011; Fillinger 2002; Hansdottir 2006; Heijmans 2007; Royse 2003; Scott 2001; Svircevic 2011) had a two-week follow-up. The frequency of  neurological complications was lower in patients receiving TEA compared with those receiving GA alone. Pooled analysis of these 1791 patients, showed that there was not significant difference in the risk of neurological complications (OR 0.50; 95% CI 0.21 to 1.18; I2 = 0%) (Analysis 5.1). Two of the nine included studies did not provide an estimate, resulting in the overall OR being based on seven studies.

There were no reported episodes of epidural haematoma in any of the included studies.

Discussion

We conducted a meta-analysis of clinical trials comparing the effects of cardiac surgery with or without TEA on mortality and cardiac, respiratory and neurological complications. Our meta-analysis showed statistically significant reductions in the risk of supraventricular tachyarrhythmias and respiratory complications. The differences for risk of mortality, myocardial infarction and stroke after TEA were not statistically significant. In order to overcome lack of statistical power due to the small sample sizes of individual clinical studies, we were able to pool a large number of available clinical studies. These studies were performed over a 30-year time span, however, possibly reducing control of variability in the protocols for patient selection, treatment and outcome measurement.

The potential of TEA for decreasing tachyarrhythmias has been reported before (Royse 2003; Scott 2001) and was confirmed in this meta-analysis. However, the included studies were heterogeneous and the confidence intervals were wide. The large study by Scott et al (Scott 2001) contributes the most to this result. In this study, B-blockers were discontinued five days perioperatively, which may explain the large benefit of TEA on supraventricular arrhythmias. Repeating the meta-analysis without the data from the study by Scott results in similar findings. While the Scott study was encouraging, most studies published since then have been unable to repeat its results. A recent, well designed study by Hansdottir (Hansdottir 2006) revealed no reduction in the incidence of adverse events. Recent studies have shown that postoperative supraventricular tachyarrhythmias can also be reduced with clearly less invasive treatments such as B-blockers and amiodarone (Burgess 2006; Kerstein 2004; Nygard 2004). The majority of the studies included in this meta-analysis did not report on whether the patients also used drugs to prevent postoperative arrhythmias. It is, therefore, unclear whether TEA has an additional preventive effect in patients who are also administered prophylactic antiarrhythmic drugs after their operation.

Our meta-analysis showed that TEA results in a statistically significant reduction in postoperative respiratory complications, which is consistent with previous meta-analyses (Ballentyne 1998; Chaney 2006). This may be explained by the superior analgesia after TEA, which facilitates earlier spontaneous respiration in the intensive care unit and faster tracheal extubation. It has been shown that other strategies to allow earlier tracheal extubation can also reduce respiratory complications (Silbert 1998).

There are several limitations in the quality of the included randomized studies that warrant caution in the interpretation of the results of this meta-analysis. First, the time period in which these studies were undertaken spanned over 30 years. The quality of anaesthesiological and intensive care has clearly improved over these years. It is possible that some beneficial effects of TEA, such as earlier extubation, are currently also achieved with modern general anaesthetics. Secondly, the allocation concealment was unclear and blinding information absent in the majority of the trials (Figure 2; Figure 3), and most of the included studies were designed to evaluate the effect of TEA on intermediate or surrogate outcome measures instead of clinical endpoints. The non-standardized coverage of clinical outcomes in most studies carried a high risk of observer bias, in particular when the endpoint adjudication was not blinded. Finally, as in any meta-analysis, our results may be subject to publication bias.

As sympathicolysis and enhanced coronary perfusion can be achieved with TEA, which could result in an improved myocardial oxygen balance (Priestley 2002; Scott 2001), we were interested in a possible difference in risk of myocardial infarction. Our findings are largely comparable to those of the previous two meta-analyses (Ballentyne 1998; Liu 2004). However, it should be noted that the risk of overall bias in this and previous meta-analyses is probably high as the allocation and blinding were unclear in most of the trials. Since we were able to include 27 studies involving 2077 patients, which is substantially more patients than the earlier two previous meta-analyses, the confidence intervals are narrower resulting in more precise effect estimates. Still, for our meta-analysis the statistical power remains too low to reach statistical significance for the beneficial effect estimates for TEA on the risk of mortality, myocardial infarction and stroke. It should be noted that in order to demonstrate statistical significance for a difference in the risk of myocardial infarction from 3.8% after GA to 2.8% after TEA (as found in this meta-analysis), it would require a sample size of at least 10,000 patients.

This meta-analysis shows that the use of TEA in patients undergoing cardiac surgery reduces the risk of postoperative supraventricular arrhythmias and respiratory complications (Summary of findings for the main comparison). The scarcity of events precludes conclusions about mortality, myocardial infarction and stroke, but the estimates suggest a reduced risk after TEA. The risk of side effects of TEA, including epidural haematoma (Bang 2011; Ho 2000; Ho 2006; Rosen 2004), could not be assessed with the current meta-analysis.

Summary of main results

This meta-analysis shows that the use of TEA in patients undergoing cardiac surgery reduces the risk of postoperative supraventricular arrhythmias and respiratory complications (Summary of findings for the main comparison).

Overall completeness and applicability of evidence

The scarcity of events precludes statistically significant conclusions about mortality, myocardial infarction and stroke. The risk of side effects of TEA was not reported in any of the included studies and therefore could not be assessed with the current meta-analysis. In order to find the true risk of side effects, such as neurological complications, an updated review should consider including non-randomized controlled trials in the review.

Quality of the evidence

We identified 11 low risk of bias studies (Barrington 2005; Berendes 2003; Caputo 2009; Caputo 2011; Hansdottir 2006; Kilickan 2006; Moore 1995; Priestley 2002; Scott 2001; Svircevic 2011; Tenling 1999), 18 high risk of bias studies (Bach 2002; Bakhtiary 2007; Brix-Christensen 1998; de Vries 2002; El-Baz 1987; Fillinger 2002; Kendall 2004; Kirno 1994; Liem 1992; Loick 1999; Lundstrom 2005; Nygard 2004; Onan 2011, Rein 1989; Royse 2003; Sharma 2010; Stenseth 1994; Stenseth 1996) and two studies with unclear risk of bias (Heijmans 2007; Lagunilla 2006).

Potential biases in the review process

The risk of overall bias in this and previous meta-analyses is probably high as the allocation and blinding were unclear in most of the trials.

Agreements and disagreements with other studies or reviews

Our findings are largely comparable to those of the previous two meta-analyses (Ballentyne 1998; Liu 2004).

Authors' conclusions

Implications for practice

This meta-analysis showed that the use of TEA in cardiac surgery may reduce the risk of postoperative supraventricular arrhythmias, respiratory complications and neurological complications, but has no effect on the risk of mortality and myocardial infarction.

Implications for research

In view of inconsistent results for different outcomes and the long time period over which the included studies were undertaken, it would be interesting to focus future research on potential benefits of TEA compared to contemporary anaesthesia methods.

Acknowledgements

We would like to thank Mathew Zacharias (content editor), Marialena Trivella (statistical editor), Joanne Guay, Shafi Mussa, R Peter Alston (peer reviewers) and Janet Wale (CARG consumer editor) for their help and editorial advice during the preparation of this review.

We would also like to thank Dr Jane Ballantyne, Dr R Peter Alston, Dr W Scott Beattie Dr Shafi Mussa, Janet Wale, Maryann Napoli and Nete Villebro for their help and editorial advice during the preparation of the protocol for the systematic review.

Data and analyses

Download statistical data

Comparison 1. Thoracic epidural analgesia versus general anaesthesia for mortality
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Mortality two weeks292877Peto Odds Ratio (Peto, Fixed, 95% CI)0.84 [0.33, 2.13]
1.1 All studies292877Peto Odds Ratio (Peto, Fixed, 95% CI)0.84 [0.33, 2.13]
2 Mortality six months2170Peto Odds Ratio (Peto, Fixed, 95% CI)1.35 [0.08, 22.83]
Analysis 1.1.

Comparison 1 Thoracic epidural analgesia versus general anaesthesia for mortality, Outcome 1 Mortality two weeks.

Analysis 1.2.

Comparison 1 Thoracic epidural analgesia versus general anaesthesia for mortality, Outcome 2 Mortality six months.

Comparison 2. Thoracic epidural analgesia versus general anaesthesia for myocardial infarction
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Myocardial infarction two weeks152127Peto Odds Ratio (Peto, Fixed, 95% CI)0.76 [0.49, 1.19]
2 Myocardial infarction six months2170Peto Odds Ratio (Peto, Fixed, 95% CI)0.41 [0.06, 3.03]
Analysis 2.1.

Comparison 2 Thoracic epidural analgesia versus general anaesthesia for myocardial infarction, Outcome 1 Myocardial infarction two weeks.

Analysis 2.2.

Comparison 2 Thoracic epidural analgesia versus general anaesthesia for myocardial infarction, Outcome 2 Myocardial infarction six months.

Comparison 3. Thoracic epidural analgesia versus general anaesthesia for respiratory complications
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Respiratory complications two weeks132011Risk Ratio (M-H, Random, 95% CI)0.66 [0.45, 0.98]
Analysis 3.1.

Comparison 3 Thoracic epidural analgesia versus general anaesthesia for respiratory complications, Outcome 1 Respiratory complications two weeks.

Comparison 4. Thoracic epidural analgesia versus general anaesthesia for supraventricular tachycardias (SVTs)
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 SVTs two weeks132250Risk Ratio (M-H, Random, 95% CI)0.65 [0.50, 0.86]
2 SVTs six months2170Peto Odds Ratio (Peto, Fixed, 95% CI)0.76 [0.39, 1.51]
Analysis 4.1.

Comparison 4 Thoracic epidural analgesia versus general anaesthesia for supraventricular tachycardias (SVTs), Outcome 1 SVTs two weeks.

Analysis 4.2.

Comparison 4 Thoracic epidural analgesia versus general anaesthesia for supraventricular tachycardias (SVTs), Outcome 2 SVTs six months.

Comparison 5. Thoracic epidural analgesia versus general anaesthesia for stroke
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Stroke two weeks91791Peto Odds Ratio (Peto, Fixed, 95% CI)0.50 [0.21, 1.18]
Analysis 5.1.

Comparison 5 Thoracic epidural analgesia versus general anaesthesia for stroke, Outcome 1 Stroke two weeks.

Comparison 6. Thoracic epidural analgesia versus general anaesthesia for mortality - sensitivity analysis
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Low risk of bias studies101794Peto Odds Ratio (Peto, Fixed, 95% CI)0.83 [0.25, 2.70]
Analysis 6.1.

Comparison 6 Thoracic epidural analgesia versus general anaesthesia for mortality - sensitivity analysis, Outcome 1 Low risk of bias studies.

Appendices

Appendix 1. Search strategies

1 CENTRAL
#1. ANALGESIA EPIDURAL explode all trees (MeSH)
#2. ANESTHESIA EPIDURAL explode all trees (MeSH)
#3. ANESTHESIA SPINAL explode all trees (MeSH)
#4. INJECTIONS SPINAL explode all trees (MeSH)
#5. (epidural* or peridural* or spinal* or intraspinal* or intrathecal* or neuraxial*)
#6. (#1 or #2 or #3 or #4 or #5)
#7. CARDIAC SURGICAL PROCEDURES explode all trees (MeSH)
#8. CARDIOPULMONARY BYPASS explode all trees (MeSH)
#9. (#6 and (#7 or #8))
#10. ((coronary next artery next bypass next surgery) or (coronary next artery next surgery) or (coronary next
bypass next graft next surgery) or (coronary next artery next bypass next graft) or (coronary next bypass next
graft) or (coronary next artery next bypass next graft*) or (coronary next bypass next graft*) or cabg or (((off
next pump) or offpump or off-pump) and (coronary next surgery)) or (open next heart next surgery) or (heart
next surgery) or (heart next valve next surgery) or (cardiopulmonary next bypass))
#11. (#7 or #8 or #10)
#12. (#11 and #6)

2 MEDLINE (PubMed)
(("Cardiac Surgical Procedures"[MeSH] OR cardiac surgery[tiab] OR heart surgery[tiab] OR cardiac surgical
procedures[tiab] OR cardiopulmonary bypass[tiab]) NOT Pulmonary Surgical Procedures[MeSH]) AND ("Analgesia, Epidural"[MeSH] OR "Anesthesia, Epidural"[MeSH] OR "Anesthesia, Spinal"[MeSH] OR epidural*[tiab] OR peridural*[tiab] OR extradural*[tiab] OR spinal*[tiab] OR subarachnoid*[tiab] OR intrathecal*[tiab] OR neuraxial*[tiab]) AND ((randomized controlled trial [pt] OR controlled clinical trial [pt] OR randomized controlled trials [mh] OR random alocation [mh] OR double-blind method [mh] OR single-blind method [mh] OR clinical trial [pt] OR clinical trials [mh] OR ("clinical trial" [tw]) OR ((singl* [tw] OR doubl* [tw] OR trebl* [tw] OR tripl* [tw]) AND (mask* [tw] OR blind* [tw])) OR (placebos [mh] OR placebo* [tw] OR random* [tw] OR research design [mh:noexp] OR comparative study [mh] OR evaluation studies [mh] OR follow-up studies [mh] OR prospective studies [mh] OR control* [tw] OR prospective* [tw] OR volunteer* [tw] NOT (animals [mh] NOT human [mh]))

3 Ovid EMBASE
((heart-surgery/) or (cardiopulmonary bypass/)) or ((coronary artery bypass surgery or coronary artery surgery or coronary bypass graft surgery or coronary artery bypass graft or coronary bypass graft or coronary artery bypass graft* or coronary bypass graft* or CABG or ((off pump or offpump or off-pump) and (coronary surgery)) or open heart surgery or heart surgery or heart valve surgery or cardiopulmonary bypass).mp.)) and ((( epidural or peridural or extradural or spinal or subarachnoid or intraspinal or intrathecal or neuraxial).mp.)) or ((spinal anesthesia/ or intraspinal drug administration/ or epidural anesthesia/))) and (((randomized-controlled-trial/ or randomization/ or controlled-study/ or multicenter-study/ or phase-3-clinical-trial/ or phase-4-clinical-trial/ or double-blind-procedure/ or single-blind-procedure/) or (random* or cross?over* or multicenter* or factorial* or placebo* or volunteer*) or ((singl* or doubl* or trebl* or tripl*) adj3 (blind* or mask*)).ti,ab or (latin adj square).mp.) not (animals.sh not (humans.sh and animals.sh)))

4 CINAHL, EBSCOhost
S1 (MM "Heart Surgery+") OR (MM "Cardiopulmonary Bypass+") or (coronary and (surg* or bypass or graft)) or CABG or off?pump or open heart surgery or heart surgery or heart valve surgery or cardiopulmonary bypass
S2 (epidural or peridural or extradural or spinal or subarachnoid or intrathecal or neuraxial) or (TI an?esth*) or (AB an?esth*) or (MM "Anesthesia, Spinal+") OR (MM "Injections, Intraspinal+") OR (MM "Infusions, Intraspinal+") OR (MM "Analgesia, Epidural+") OR (MM "Anesthesia, Epidural+") OR (MM "Epidural Analgesia Administration (Iowa NIC)")
S3 S1 and S2
S4 (MH "Placebos") OR ( (MM "Randomized Controlled Trials") OR (MM "Random Assignment") OR (MH "Prospective Studies") OR (MH "Multicenter Studies") OR (MH "Clinical Trials") OR (MH "Single-Blind Studies") OR (MH "Triple-Blind Studies") OR (MH "Double-Blind Studies") ) OR ( placebo* or multicenter or prospective or ((random* or control*) and trial*) )
S5 S3 and S4

5 Web of Science (SCI/SSCI)
TI=((epidural* OR peridural* OR extradural* OR spinal* OR subarachnoid* OR intrathecal* OR neuraxial*) AND
(anesthes* OR anaesthes* OR analges*) AND (card* surg* OR heart surg* OR CABG OR coronar* arter* bypass* OR coronar* bypass* OR heart* valv* surg*) AND (metaanalysis OR meta-analysis OR review OR consensus OR guideline OR random* OR trial* OR control* OR ((singl* OR doubl* OR trebl* OR tripl*) AND (blind* OR mask*))))

Appendix 2. Processing-form: Epidural versus non-epidural anaesthesia in cardiac surgery

Article nr:

Date:                            //.

Name reviewer:  Svircevic           Passier             van Dijk

First authors name        ..................................................................

Year of publication         ..................................................................

Study Quality

1 Group size

·         Neuraxial                N=

·         Control                   N=

2 Randomized allocation                                    Yes      No        Method unclear

3 Concealed allocation                                       Yes      No       Method unclear

4 Number Crossovers

·         Neuraxial                N=

·         Control                   N= 

5 Maximum number dropouts

·         Neuraxial                N=

·         Control                   N= 

6 Maximum number Loss-to-follow-up

·         Neuraxial                N=

·         Control                   N= 

7 Intention to treat analyses                              Yes      No        unclear

8 Blinded analyses                                           Yes      No        unclear

9 Blinding pre- and post surgery care               Yes      No        unclear

10 Standardized pre- and post surgery care     Yes      No        unclear

11 Blinding end points                                       Yes      No        unclear

12 Standardization endpoints                            Yes      No        unclear

Preoperative data

 13 Age

·         Neuraxial                  Mean=                                               Sd=

·         Control                      Mean=                                               Sd=

 14 Males

·         Neuraxial                  N=

·         Control                      N= 

15 Prior vascular surgery:

·         Neuraxial                  N=                                                    

·         Control                      N=

 16 Diabetic status:        Type                 1                                  2                                  dialysis

·         Neuraxial                              N=                                  N=                                  N=           

·         Control                                 N=                                  N=                                  N=           

 17 Preoperative risk score:         French score                Parsonnet score            Euro score

·         Neuraxial                  Mean=                                               SD=

·         Control                      Mean=                                               SD=

 18 Type(s) of surgery                            ..............................

19 Type of neuraxial anaesthesia:         Intrathecal                       epidural

20 Outcome measures

·         Primary endpoint                           ................................

·         Secondary endpoints                    ................................

 21 Time to follow-up                                ...............................

 22 Main outcomes of this study (give absolute numbers, no percentages)

OutcomeNeuraxial groupControl groupP
Mortality   
MI   
SVT   
Respiratory complications   

 

Other important outcomes:

...............................................................................................................................................................

 ...............................................................................................................................................................

 main conclusion(s) (see last paragraph discussion):

 ...............................................................................................................................................................

 ...............................................................................................................................................................

 Remarks:

 ...............................................................................................................................................................

 ...............................................................................................................................................................

 ...............................................................................................................................................................

 

What's new

DateEventDescription
1 July 2013AmendedGeert J van der Heijden's contact details amended.

History

Protocol first published: Issue 3, 2007
Review first published: Issue 6, 2013

DateEventDescription
21 January 2008AmendedConverted to new review format.

Contributions of authors

Conceiving the review: Vesna Svircevic (VS), Diedrik van Dijk (DD), Cor Kalkman (CK), Arno Nierich (AN), Jan Dieleman (JD)
Co-ordinating the review: VS
Undertaking manual searches: VS, Martijn Passier (MP)
Screening search results: VS, MP
Organizing retrieval of papers: VS, MP
Screening retrieved papers against inclusion criteria: VS, MP
Appraising quality of papers: VS, MP
Abstracting data from papers: VS, MP
Writing to authors of papers for additional information: VS, MP, Geert van der Heijden (GH)
Providing additional data about papers: VS, MP
Obtaining and screening data on unpublished studies: VS, MP, GH
Data management for the review: GH
Entering data into Review Manager (RevMan 5.1): VS
RevMan statistical data: VS, GH
Other statistical analysis not using RevMan: VS, GH
Double entry of data: data entered by person one: VS, MP; data entered by person two: GH, DD
Interpretation of data: VS, MP, GH
Statistical inferences: VS, MP, GH
Writing the review: VS, MP
Performing previous work that was the foundation of the present study: VS, JD, AN, CK, DD
Guarantor for the review (one author): VS
Person responsible for reading and checking review before submission: VS, MP, DD

Declarations of interest

None known

Sources of support

Internal sources

  • University Medical Center Utrecht, Julius Center for Health Sciences and Primary Care, Netherlands.

External sources

  • No sources of support supplied

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Bach 2002

MethodsRandomized controlled trial
Participants40 patients scheduled for elective coronary artery bypass grafting surgery. Exclusion criteria were impaired coagulation, allergies to local anaesthetics, corticoid medication, preoperative signs of infection, renal or liver failure, diabetes mellitus and an impaired left ventricular function (ejection fraction 50%)
Interventions

Continuous epidural infusion of bupivacaine 0.25% versus general anaesthesia group

The 13 patients received an epidural bolus of 10ml of bupivacaine 0.25% through the catheter followed by a continuous epidural infusion of bupivacaine 0.25% (Bupivacain 0.25%, Curasan Pharma AG, Kleinostheim, Germany) adjusted to body height for the whole observation period until 18h after surgery. The control group consists of 13 patients who received an intravenous dopexamine infusion beside general anaesthesia and 14 patients with general anaesthesia alone who received equal volumes of NaCl 0.9% intravenously as placebo in a time-matched fashion.

Outcomes

Inflammatory response

Mortality

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskOne day before surgery and after written informed consent was obtained, patients were randomly assigned to the study groups by drawing lots
Allocation concealment (selection bias)Unclear riskDrawing lots
Blinding (performance bias and detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
High riskOne patient of the control group was excluded during the study. He died 8h after surgery as a result of a non-occlusive ischaemia of the small intestine. All other patients who were initially included terminated the study.
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias

Bakhtiary 2007

MethodsRandomized controlled trial
Participants132 patients with symptomatic coronary artery disease were prospective enrolled in this study. All patients underwent elective OPCAB surgery and were randomized to receive either GA or combined GATEA. Patients with a history of atrial arrhythmias, those undergoing emergency operations, and patients requiring intraoperative inotropic support were excluded from this study.
Interventions

General anaesthesia only or combined general anaesthesia with high thoracic epidural analgesia (66 patients in each group)

TEA: a continuous epidural infusion with ropivacaine 0.16% and sufentanil 1 g/mL at an hourly rate of 2 to 5 mL was started after a bolus dose of 6 mL to provide intraoperative analgesia. GA: propofol (1.5 mg/kg) and remifentanil (1 ug/kg) administered over 120 seconds. After loss of eyelash reflex, 0.1 mg/kg of cisatracurium was administered to facilitate tracheal intubation. Anaesthesia was maintained with continuous infusion of propofol (50-100 ug/kg/min) and remifentanil (0.1-0.3 ug/kg/min). Patients undergoing GA without TEA received intravenous metamizole (Novalgin; Aventis Pharma, Bad Soden, Germany), a peripheral analgesic derived from pyrazolone acid, 15 mg/kg, before skin incision. Intravenous piritramide, a-receptor agonist with a potency of 0.7 compared with morphine, 0.1 mg/kg, was administered after completion of coronary anastomosis and repeated during wound closure.

OutcomesMortality
Myocardial infarction
Supraventricular tachyarrhythmias
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskPatients were randomized to receive either GA or combined GATEA
Allocation concealment (selection bias)Unclear riskNot stated
Blinding (performance bias and detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll pre-specified outcomes reported
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias. Sample size calculation not stated

Barrington 2005

MethodsRandomized controlled trial
Participants

120 patients scheduled for elective coronary artery bypass grafting surgery

All patients scheduled for elective CABG surgery (using cardiopulmonary bypass (CPB)) were eligible. Exclusion criteria were emergency or repeat CABG surgery, combined valve and CABG surgery, aspirin ingestion within 6 days of surgery, a platelet count 150x109/L, an international normalized ratio 1.1, active neurological disease, and cutaneous disorders at the epidural insertion site.

Interventions

General anaesthesia only or combined general anaesthesia with high thoracic epidural analgesia (ropivacaine 1%; 60 patients in each group)

epidural block was established with 5 mL of ropivacaine 1% and fentanyl 50 ug. If required, the block was extended with 2 mL of ropivacaine 1%.

GA was induced with midazolam (0.05 - 0.1 mg/kg), fentanyl (7-15 ug/kg for the GA group and 5-7 ug/kg for the HTEA group), propofol (20 mg increments as required), and rocuronium (0.6 mg/kg). GA was maintained with propofol 3-6 mg/kg/hr. Further doses of rocuronium 10 mg were given only for overt patient movement, with no additional rocuronium given after CPB.

Outcomes

Mortality
Myocardial infarction
Supraventricular tachyarrhythmias
Stroke

Respiratory complications

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskPatients were randomized the day before surgery to two groups. The random allocation sequence was computer-generated in permuted blocks of four and enclosed in sequentially numbered opaque sealed envelopes.
Allocation concealment (selection bias)Low riskOpaque sealed envelopes
Blinding (performance bias and detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll patients were included in the intention to treat analysis; all pre-specified outcomes reported
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias. A formal sample size determination was not possible because primary endpoint data were not available at the time of study inception. The confidence intervals between groups were calculated post hoc to assess the precision of the data and the adequacy of sample size

Berendes 2003

MethodsRandomized controlled trial
Participants

73 patients scheduled for coronary artery bypass who had left ventricular ejection fraction of 50% or more

Eligible patients had coronary artery disease with a left ventricular function of 50% or more and were scheduled for elective CABG. Exclusion criteria were any pre-existing endocrinological diseases, renal insufficiency, coagulation disorders or right and/or left ventricular dysfunction, concomitant disorders of heart valves, having undergone cardiac surgical procedures, acute myocardial infarction and heart failure.

Interventions

General anaesthesia only (37 patients) or combined general anaesthesia with high thoracic epidural analgesia (36 patients) (bupivacaine and sufentanil)

correct position of the epidural was tested by 2 ml of bupivacaine 0.5% with epinephrine. Before induction high TEA was initiated by 6-12 mL of bupivacaine 0.5% and 15-25 ug sufentanil. General anaesthesia was induced in all patients with midazolam 0.1 mg/kg IV and pancuronium 0.1 mg/kg. The GA group received propofol 1.5-3 mg/kg/hr and sufentanil 1-2 ug/kg/hr. In the TEA group they received propofol 1.5-3 mg/kg/hr supported by repetitive injections of sufentanil 0.2-1 ug/kg.

Outcomes

Mortality
Respiratory complications

Left ventricular function

Postoperative complications

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskComputer-generated block randomization
Allocation concealment (selection bias)Low riskAdministered through a sequential opaque envelope technique
Blinding (performance bias and detection bias)
All outcomes
Unclear riskParticipants, staff and research personnel unblinded to the intervention. The investigator collected data, but was not involved in patient care or management
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll pre-specified outcomes reported
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias

Brix-Christensen 1998

MethodsRandomized controlled trial
Participants

16 patients scheduled for elective coronary artery bypass grafting surgery

Patients with diabetes mellitus or cancer were excluded

Interventions

Low dose opioid anaesthesia (epidural: bupivacaine 2 mg/mL and fentanyl 5 ug/mL at 5 ml/hr)) and general inhalational anaesthesia with fentanyl (8 patients in each group)

In the TEA group 8 mL of bupivacaine 0.5 % was injected 30 mins before surgery. Peroperative a continuous epidural infusion was given with bupivacaine 2 mg/ ml and fentanyl 5ug/mL at 5 mL/hr. General anaesthesia was induced with midazolam 0.15 mg/kg and fentanyl 5ug/kg. Muscle relaxation was achieved with pancuronium 0.1mg/kg. Anaesthesia was maintained with enflurane 0.4-0.8%. In the high dose fentanyl group anaesthesia was induced with midazolam 0.15 mg/kg and fentanyl 50 ug/kg. Anaesthesia was maintained with enflurane, midazolam and sufentanil.

Outcomes

Mortality

Cytokine response

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskPatients were randomly allocated into two groups
Allocation concealment (selection bias)Unclear riskNot stated
Blinding (performance bias and detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
Low riskThe peri- and postoperative courses were uneventful for all patients. All data addressed
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias

Caputo 2009

MethodsRandomized controlled trial
Participants74 patients undergoing off-pump coronary artery bypass surgery
Interventions

The anaesthetic technique consisted of premedication with benzodiazepines, and induction with intravenous infusion of propofol at 3 mg/kg/hr combined with fentanyl (10 to 20 g/kg). Neuromuscular blockade was achieved with 0.1 to 0.15 mg/kg pancuronium bromide or vecuronium.

Intervention (36 patients): bilateral neuraxial block was established from T1 to T10 with an initial bolus of 5 mL bupivacaine 0.5% followed by another 5-mL bolus after 10 minutes. After induction of GA and when central haemodynamic status was stable, a continuous infusion of 0.125% bupivacaine and 0.0003% clonidine (150 g in 500 mL) was commenced at an initial rate of 10 mL/h

Control (38 patients): In the GA group, a patient-controlled analgesia intravenous morphine pump was started in the intensive care unit for 48 hours by using a 1 mg bolus dosing with a 5-minute lockout period

Outcomes

Mortality

Stroke

Myocardial infarction

Supraventricular arrhythmias

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskPatients were randomly assigned to receive either general anaesthesia only or general anaesthesia plus epidural. Random treatment allocations were generated by computer in advance of starting the study, using block randomization with varying block sizes. Allocation details were concealed in sequentially numbered opaque sealed envelopes.
Allocation concealment (selection bias)Low riskOpaque sealed envelopes
Blinding (performance bias and detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll patients received the treatment allocated, all pre-specified outcomes reported
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias. Sample size calculation stated

Caputo 2011

MethodsTwo-centre, open, parallel-group, randomized controlled trial
Participants226 adults undergoing primary OPCAB surgery without the use of CPB and cardioplegic arrest
InterventionsIn both groups the induction with propofol at 0.5–1 mg/kg combined with fentanyl (10 –20g/kg). Neuromuscular blockade was achieved with 0.1–0.15 mg/kg pancuronium bromide or vecuronium. Anesthesia was maintained with either isoflurane at 0.8 –1.0 minimal anaesthetic concentration or intravenous propofol 3– 4 mg/kg/hr, at the discretion of the consultant anaesthetist. Patients in the GAE group had a thoracic epidural catheter sited in the operating theatre immediately before surgery at the T2–3 or T3–4 intervertebral space. Bilateral neuraxial block was established from T1 to T10 with an initial bolus of 5 mL bupivacaine, 0.5%, followed by another 5 mL bolus after 10 min. Determination of the spread of block was performed with ethyl chloride spray. After induction of GA and when central hemodynamic status was stable, a continuous infusion of 0.125% bupivacaine and 0.0003% clonidine (150 g in 500 mL) was commenced at an initial rate of 10 ml/hr.
Outcomes

Mortality
Myocardial infarction

Chest infections

Pain scores

Length of hospital stay

Arrhythmias

Stroke

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskRandomized treatment allocations were generated using Stata version 8. They were stratified by consultant team with a 1:1 allocation using blocks of varying sizes.
Allocation concealment (selection bias)Low riskAllocation details were concealed in sequentially numbered, opaque sealed envelopes. These were prepared by the clinical trials and evaluation unit
Blinding (performance bias and detection bias)
All outcomes
High riskNot blinded
Incomplete outcome data (attrition bias)
All outcomes
Low riskThere were 18 protocol violations in patients allocated to GAE who received GA because the epidural could not be inserted
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias. Sample size calculation stated

de Vries 2002

MethodsRandomized controlled trial
Participants90 patients scheduled for elective minimally invasive direct coronary artery bypass surgery
Interventions

30 patients had general anaesthesia and were extubated immediately after surgery (extubated group), 30 patients had a thoracic epidural (bupivacaine and sufentanil) and general anaesthesia and were extubated immediately after surgery (epidural group), and 30 patients had general anaesthesia and were ventilated after surgery (intubated group)

In the extubated group, at induction the patients received midazolam, 0.1 mg/kg, and sufentanil, 0.5 to 1 ug/kg. Anaesthesia was maintained with either isoflurane (0.5% to 0.8%) or propofol (3 to 5 mg/kg/hr) with incremental supplements of sufentanil (25 to 50 g) when the mean arterial pressure (MAP) exceeded baseline values by 20%.

In the epidural group, a thoracic epidural catheter was inserted at the T3-4 level before induction of anaesthesia. Anaesthesia was induced with midazolam, 0.1 mg/kg, and sufentanil, 0.5 to 1 g/kg. Through the epidural catheter, 8 to 10 mL of bupivacaine 0.25% with sufentanil, 25 ug/10 mL, were given 10 minutes before the start of surgery. Anaesthesia was maintained as in the extubated group. In the intubated group, induction of anaesthesia was performed with midazolam and sufentanil as previously reported.15 Briefly, midazolam, 0.1 mg/kg, and sufentanil, 1.5 ug/kg, were slowly infused over 5 minutes. Anaesthesia was maintained with continuous infusions of midazolam and sufentanil at a rate of 2 ug/kg/min for midazolam and 1 ug/kg/hr for sufentanil. In all groups, pancuronium, 0.1 mg/kg, was administered to facilitate endotracheal intubation.

In the extubated group and in the intubated group, postoperative analgesia was given on patient request with piritramide, 0.2 mg/kg intramuscularly. In the epidural group, a continuous infusion of bupivacaine 0.125% and sufentanil, 25 ug/50 mL, was given at 8 to 10 mL/hr. All groups received additional paracetamol suppositories, 1g 4 times daily.

Outcomes

Mortality
Myocardial infarction or ischaemia (ST segment analysis)
Haemodynamics: MAP and heart rate

Pain scores

Length of hospital stay

NotesMIDCAB surgery
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk"90 patients were randomly divided into 3 groups"
Allocation concealment (selection bias)Unclear riskNot stated
Blinding (performance bias and detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
Low riskFive patients were excluded from analysis: three because of surgical reasons and in two patients the epidural technique failed
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias. Sample size calculation stated

El-Baz 1987

MethodsRandomized controlled trial
Participants60 patients, aged 34 to 76 years, after coronary artery bypass surgery (one to four grafts)
Interventions

IV opioids or low dose continuous opioid epidural analgesia (morphine 0.1 mg/mL at 1 mL/hr) with 30 patients in each group.

General anaesthesia was induced with pentothal 3-4 mg/kg followed by succinylcholine 1-2 mg/kg. Anaesthesia was maintained with a nitrous-oxide-oxygen-halothane gas mixture. Pancuronium 0.1-0.2 mg/kg was used for muscle relaxation to facilitate control of ventilation.

The control group received morphine 2 mg/2 hr and as needed. The epidural group received a continuous epidural infusion of 0.1 mg morphine/hr, this was supplemented by 2 mg IV as needed.

Outcomes

Mortality
Respiratory function

Cardiovascular parameters

Pain relief

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk"Patients were randomly divided into two equal groups of 30 patients"
Allocation concealment (selection bias)Unclear riskNot stated
Blinding (performance bias and detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
Low riskAppears have no incomplete outcome data
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias

Fillinger 2002

MethodsRandomized controlled trial
Participants

60 patients scheduled for elective coronary artery bypass

Patients scheduled for elective CABG surgery were considered eligible for inclusion in the study in the absence of any specific contraindication to the use of EAA (heparin or warfarin (Coumadin (Endo Laboratories Inc, Wilmington, DE)) anticoagulation, pre-existing coagulopathy, infection at insertion site, or septicaemia).

Interventions

General anaesthesia with IV morphine or combined general anaesthesia with high thoracic epidural analgesia (bupivacaine 0.5% at 4-10 mL/hr) with 30 patients in each group.

On arrival in the operating room, intravenous sedation was initiated with incremental doses of fentanyl ≤2 ug/kg, and midazolam, 0≤.05 mg/kg. GA was induced in all patients with intravenous fentanyl (5 to 20 g/kg total), midazolam (0.1 mg/kg), thiopental (1 to 2 mg/kg), and pancuronium or vecuronium for muscle relaxation and tracheal intubation. Inhaled isoflurane was used for anaesthesia maintenance. Muscle relaxation was reversed at the end of the operation with neostigmine and glycopyrrolate.

For participants randomized to receive TEAA, a thoracic epidural catheter was inserted after vascular catheter placements and before induction of GA. After a negative epidural test dose of 3 mL of 1.5% lidocaine with 1:200,000 epinephrine, the epidural catheter was injected with preservative-free morphine, 20 ug/kg, and 0.5% bupivacaine, in 5 mg increments, to a total loading dose of 25 to 35 mg of bupivacaine. Then the epidural catheters were continuously infused with 0.5% bupivacaine with morphine, 25 ug/mL, at 4 to 10 mL/hr beginning after the induction of GA. Clinical signs of inadequate epidural analgesia (haemodynamic response to surgical stimulation) during surgery were managed with a 3 mL bolus of the infusion solution followed by an increase in the infusion rate of 1 mL/hr (to a maximum of 10 mL/hr).

After surgery: patients in the GA group received intravenous morphine for postoperative analgesia according to a standard CTICU protocol, whereas patients in the TEAA group received an epidural infusion of 0.125% bupivacaine with morphine, 25 ug/mL, at 4 to 10 mL/h.

Outcomes

Mortality
Myocardial infarction
Supraventricular tachyarrhythmias
Stroke
Respiratory (time to tracheal extubation)

Pain control

Duration of hospital stay

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskComputerized randomization
Allocation concealment (selection bias)Unclear riskNot stated
Blinding (performance bias and detection bias)
All outcomes
High riskParticipants, staff and research personnel unblinded to the intervention
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll pre-specified outcomes reported
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias

Hansdottir 2006

MethodsRandomized controlled trial
Participants113 patients undergoing elective cardiac surgery (CABG, cardiac valve procedures, combined CABG and valve procedures or the Maze procedure, with or without CABG), absence of contraindications to epidural anaesthesia, abnormal coagulation tests (i.e., partial thromboplastin time 45 sec or prothrombin time (international normalized ratio) 1.5 or a platelet count 80,000), or recent (1 week) treatment with thrombolytic or potent antiplatelet drugs (streptokinase, alteplase, clopidogrel, abciximab, tirofiban, integrelin). Aspirin treatment was not considered a contraindication to the placement of a thoracic epidural catheter.
Interventions

General anaesthesia with postsurgical PCA with morphine or combined general anaesthesia with high thoracic epidural analgesia (bupivacaine 5 mg/mL at 0.05 mL/kg/hr)

All patients (58 patients in epidural group and 55 in the control group) received infusions of propofol and remifentanil to a target anaesthetic depth of 15-25 AAI using an auditory evoked response monitor. Tracheal intubation was facilitated with 0.5 mg/kg atracurium. After induction of anaesthesia and insertion of monitoring devices, patients in the PCTEA group received an epidural 0.1 mL/kg bolus dose followed by a continuous 0.05 mL/kg/hr infusion with bupivacaine (5 mg/mL). In the PCTEA group, postoperative pain treatment was achieved by epidural bolus doses of 2 mL of the mixture 1 mg/mL bupivacaine plus 2 g/mL fentanyl plus 2 g/mL adrenaline, a lockout interval of 20 min, and a background epidural infusion of 0.1 mL/kg/hr. An epidural 0.1 ml/kg loading dose of this mixture was given at the end of surgery. In the PCA group, postoperative pain treatment was achieved by intravenous PCA morphine with bolus doses of 0.01 mg/kg and a lockout interval of 6 min with no background infusion. A loading dose of 0.1 mg/kg morphine was given in the operating room when remifentanil infusion was stopped.

Outcomes

Myocardial infarction
Supraventricular tachyarrhythmias
Stroke
Respiratory (lung volumes)

Pain and sedation scores

Length of hospital stay

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskPatients were randomly assigned the day before surgery to one of two regimens
Allocation concealment (selection bias)Low riskSealed envelopes
Blinding (performance bias and detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
Low risk113 patients were randomized, 110 patients received allocated treatment, and 97 patients were eventually analysed
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias. Sample size calculation stated

Heijmans 2007

MethodsRandomized controlled trial
Participants

60 patients undergoing elective cardiac surgery

Exclusion criteria included left ventricular ejection fraction of less than 25%, hypothermic circulatory arrest, recent myocardial infarction, preoperative inotropic or intra-aortic balloon pump metabolic, or neurologic diseases. All patients were receiving chronic B-adrenoceptor blocking drugs

Interventions

Alfentanil or low dose remifentanil or high dose remifentanil or epidural (15 patients in the epidural and 45 patients in the control group)

A test dose of 2 mL of lidocaine 2% was given to test for the correct position of the catheter. A loading dose of 10 mL of bupivacaine 0.25% with 2.5 mg of morphine was infused over 1 hour.

In the present study, 4 groups of patients were compared as follows: group 1 (AG): a loading dose of alfentanil, 50 ug/kg, was infused over 4 minutes, and, thereafter, alfentanil was infused at a maintenance rate of 1 ug/kg/min throughout surgery; group 2 (HDRG): a loading dose of remifentanil, 2.5 ug/kg, was infused over 4 minutes, and, thereafter, remifentanil was infused at a maintenance rate of 0.5 ug/kg/min throughout surgery; group 3 (LDRG): a loading dose of remifentanil, 2.5 ug/kg, was infused over 4 minutes, and, thereafter, remifentanil was infused at a maintenance rate of 0.25 ug/kg/min; and group 4 (TEG): a loading dose of remifentanil, 2.5 ug/kg, was infused over 4 minutes, and, thereafter, remifentanil was infused at a maintenance rate of 0.125 ug/kg/min and via thoracic epidural infusion bupivacaine 0.375% plus morphine 0.2 mg/mL were administered at a rate of 1.5 mL/hr throughout surgery. The initial infusion setting for propofol on the Diprifusor was a plasma concentration of 2 ug/mL to be reached in 4 minutes.

At arrival in the ICU, a sedative-analgesic infusion of propofol, 0.5 mg/kg/hr, together with alfentanil, 0.1 ug/kg/min, in group 1 and remifentanil, 0.025 g/kg/ min, in groups 2, 3, and 4, were started for 4 hours. If necessary, propofol was increased to achieve the desired level of sedation (Ramsay sedation score 3, 4, or 5). Additionally, acetaminophen, 1 g, 4 times daily, was started as a basic analgesic. In the TEG group, the catheter was left in position 48 hours postoperatively and bupivacaine 0.125% and morphine, 0.2 mg/mL, were infused at a rate of 1.5 mL/hr. Fifteen minutes before cessation of the sedative analgesic infusion, pirinitramide, a synthetic morphine derivative (analgesic potency in comparison with morphine [1] is 0.7) 0.15 mg/kg, intravenously, was administered in the remifentanil groups 2 and 3, and placebo was administered to the AG in a blinded fashion.

Outcomes

Infectious parameters (CRP, IL-6)

Mortality
Supraventricular tachyarrhythmias
Stroke

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk"60 patients scheduled to undergo coronary artery bypass surgery were randomized"
Allocation concealment (selection bias)Unclear riskNot stated
Blinding (performance bias and detection bias)
All outcomes
Low riskThe study was blinded for the opioid infusion, except in the thoracic epidural group
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot stated
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias. Sample size calculation not stated

Kendall 2004

MethodsRandomized controlled trial
Participants

26 patients undergoing off pump coronary artery bypass grafting

Patients undergoing emergency surgery and those with unstable angina were excluded from the study. Patients with plasma creatinine values greater than 160 mmol/L were also excluded from the study. Patients taking anticoagulant therapy and those with any other contraindication to the insertion of a thoracic epidural were also excluded.

Interventions

General anaesthesia: propofol or isoflurane or epidural: isoflurane with bupivacaine

In the propofol group (9 patients), anaesthesia was induced with fentanyl 10-15 ug/kg and propofol delivered by a DiprifusorTM pump (AstraZeneca, London, England), aiming for a target blood concentration in the range 4-8 ug/mL

In the isoflurane group (9 patients), anaesthesia was induced with etomidate 0.2 mg/kg and fentanyl 10-15 ug/kg. Anaesthesia was maintained with isoflurane in oxygen and air at an end-tidal concentration of 1%. This was discontinued at the end of the procedure, and a propofol infusion was started to provide sedation if required until the criteria for tracheal extubation were met

In the epidural group (8 patients), anaesthesia was as for the isoflurane group, with the exception that fentanyl was given in a dose of 1.5 ug/kg epidural: bupivacaine 0.1% with fentanyl
5 ug/ml was given, followed by an infusion of 0.1 ml/kg/hr, adjusted according to haemodynamic values

Outcomes

Mortality
Myocardial infarction (troponin levels, new Q waves, loss of R progression)

Prolonged intensive care stay (GT; 24 hr)

Duration of tracheal intubation

Return to operating room

Intraoperative haemodynamic parameters

NotesSurgery with the octopus
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskPatients were randomly allocated to one of three groups using a shuffled, sealed envelope technique
Allocation concealment (selection bias)Low riskSealed envelopes
Blinding (performance bias and detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
Unclear risk"Three patients were excluded from the study and further analysis. Their treatment was re-randomized and reallocated,
providing 30 complete data sets for analysis"
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias. Sample size calculation not stated

Kilickan 2006

MethodsRandomized controlled trial
Participants

80 patients scheduled for elective coronary artery bypass grafting

Exclusion criteria: contraindications for the epidural technique, contraindications to any of the intended drugs in the treatment protocol, alcohol abuse, cognitive impairment

Interventions

General anaesthesia alone or general anaesthesia in combination with high thoracic epidural analgesia (40 patients in each group).

Anaesthesia was induced in all groups by 0.2 mg/kg midazolam, 10-15 ug/kg fentanyl and 0.1 mg/kg vecuronium. Anaesthesia was maintained with propofol 2-6 mg/kg/hr and 2 ug/kg fentanyl intermittently. In the TEA group, 1 hr before surgery a bolus of 20 mg of bupivacaine was administered and a continuous infusion was started (bupivacaine 0.25 % at 8 ml/hr).

Postoperatively, in the TEA group received an epidural infusion of bupivacaine 0.125%, 4-10 ml/hr. The control group received a standard analgesia protocol (morphine PCA)

Outcomes

Mortality
Bcl-2 immunoreactivity (rate of apoptosis)

Haemodynamic parameters (CI)

Reperfusion ventricular fibrillation

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk80 patients scheduled for elective CABG were randomized in two groups
Allocation concealment (selection bias)Low riskSealed envelopes
Blinding (performance bias and detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll pre-specified outcomes reported
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias. Sample size calculation not stated

Kirno 1994

MethodsRandomized controlled trial
Participants

20 patients undergoing coronary artery bypass grafting

All patients had a history of stable ischaemic heart disease with 2 or 3 vessel coronary artery disease and an ejection fraction of > 50%. Patients with co-existing valvular anomaly, arrhythmias or diabetes mellitus were not included.

Interventions

General anaesthesia (fentanyl nitrous oxide) and general anaesthesia and epidural analgesia with 10 patients in each group.

Anaesthesia was induced with thiopental 3-5 mg/kg, followed by pancuronium 0.1 mg/kg IV. Fentanyl was given in incremental doses up to a total amount of 10-15 ug/kg before sternotomy. The patients were ventilated with 70% nitrous oxide in oxygen. In the TEA group 3-3.5 mL mepivacaine was given epidurally after induction.

Outcomes

Mortality

MAP

Regional myocardial oxygen consumption

Myocardial ischaemia

Haemodynamic parameters

Noradrenaline spillover (sympathetic nervous system activation)

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskThe patients were randomized into two groups
Allocation concealment (selection bias)Unclear riskNot stated
Blinding (performance bias and detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll pre-specified outcomes reported
Selective reporting (reporting bias)Unclear riskNot stated

Lagunilla 2006

MethodsRandomized controlled trial
Participants

50 patients submitted for scheduled coronary revascularization of, at least, the left anterior descending (LAD) coronary artery were enrolled in the study.

The exclusion criteria were as follows: patient refusal, urgent or emergency procedure, unstable haemodynamic status (myocardial infarction less than 2 weeks prior to surgery, requirement for inotropic drugs or for intra-aortic balloon contrapulsation), absence of normal sinus rhythm, neurologic or neuromuscular disorders, previous thoracic or cervical spine surgery or trauma, significant left main coronary artery stenosis, anticoagulation, haematologic disorders and infection at the puncture site.

Interventions

General anaesthesia with a thoracic epidural with either ropivacaine or a saline solution with 25 patients in each group.

Thereafter, general anaesthesia was induced with midazolam 1 mg, remifentanil at 0.7 mg/kg/min, cisatracurium in a single bolus of 0.2 mg/kg and 2-3% inhaled sevoflurane in oxygen. All patients received the corresponding volume of plain saline or 0.3% ropivacaine, in a blind manner, accordingly to the group to which they had been randomized. Then, an epidural infusion of 0.3% ropivacaine at 5-7 ml/hr (reduced to 2-3 mL/h during grafting) was started in all patients. In the postoperative period, 0.2% ropivacaine with 5 mg/mL fentanyl was used for analgesia in all patients, employing a patient controlled system.

Outcomes

Mortality

Haemodynamic parameters

Intramyocardial oxygen partial pressure

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskPatients enrolled in the study were randomly assigned, randomization was performed using a computer-generated random sequence
Allocation concealment (selection bias)Unclear riskComputer-generated random sequence
Blinding (performance bias and detection bias)
All outcomes
Low riskPatients enrolled in the study were randomly assigned to receive either epidural saline or epidural 0.3% ropivacaine
Incomplete outcome data (attrition bias)
All outcomes
Low riskOne patient was excluded in the ropivacaine group due to a wet tap (no catheter placement). One patient in the saline group became unstable after anaesthesia induction and required inotropic support; this patient was subsequently withdrawn from the study.
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias. Sample size calculation stated

Liem 1992

MethodsRandomized controlled trial
Participants

54 patients scheduled for elective coronary artery bypass surgery

To be admitted to the study they were required to have normal or only moderately impaired left ventricular (LV) function (ejection fraction greater than 40%) as assessed by preoperative LV cineangiography and a LV end-diastolic pressure of less than 18 mmHg. Patients who had a myocardial infarction in the 7 days preceding surgery. Pre-existing haemorrhagic diathesis, or valvular heart disease were excluded from the study.

Interventions

High thoracic epidural with bupivacaine 0.375% plus sufentanil 5 ug/mL in combination with general anaesthesia midazolam/N2O and general anaesthesia with midazolam/fentanyl (27 patients in each group).

IV anaesthesia in the GA group was induced over a period of 10 minutes with midazolam, 0.1 mg/kg, sufentanil, 5 ug/kg, and pancuronium, 0.1 mg/kg. Anaesthesia was maintained with an infusion of midazolam, 0.1 mg/kg/hr, and sufentanil, 2.5 ug/kg/hr. Pancuronium, 0.025 mg/kg, was administered every hour.

Epidural analgesia in the TEA group was induced, over a period of 10 minutes, with a mixture of 0.375% bupivacaine plus sufentanil, 1:200,000 (i.e., 5 ug/mL) in a dose of 0.05 mL/cm body length. Continuous epidural analgesia was then started with 0.125% bupivacaine plus sufentanil, 1:1,000,000 (i.e., 1 ug/mL) at an infusion rate of 0.05 mL/cm body length/hour. Just before induction of GA, measurements were taken (point 2). IV anaesthesia was induced with etomidate, 0.2 mg/kg, midazolam, 0.1 mg/kg, and pancuronium, 0.1 mg/kg. IV anaesthesia was maintained with an infusion of midazolam, 0.1 mg/kg/hr, and after the establishment of cardiopulmonary bypass (CPB), a bolus of midazolam, 0.05 mg/kg, was administered IV; pancuronium, 0.025 mg/kg, was administered every hour.

Outcomes

Mortality
Myocardial infarction
Haemodynamic parameters
Respiratory complications

VAS scores

Time to awakening

Arrythmias (tachycardia)

Adrenergic responses

NotesAll three articles
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskOn the day before surgery, the patients were assigned randomly to either a TEA or a GA group
Allocation concealment (selection bias)Unclear riskNot stated
Blinding (performance bias and detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
Low riskRe-sternotomy was necessary in two patients (1 in each group). In one TEA group patient, the epidural catheter was dislocated. These patients were excluded from the study
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias

Loick 1999

MethodsRandomized controlled trial
Participants

70 patients scheduled for elective coronary artery bypass grafting

Seventy-two patients scheduled for elective CABG fulfilled the study criteria and were selected for the investigation. Disorders of the intestine and liver, gastritis, ulcera ventriculi and duodeni, autonomic neuropathy, and diabetes mellitus (patients receiving insulin or oral hypoglycaemic drugs) were exclusion criteria

Interventions

General anaesthesia with clonidine IV or general anaesthesia with thoracic epidural analgesia with 25 patients in epidural group and 45 in the control group (with and without clonidine).

Anaesthesia was induced with sufentanil (1-2 mg/kg) and propofol (1-2 mg/kg). Pancuronium (0.1 mg/kg) was used to facilitate tracheal intubation. Anaesthesia was maintained by a continuous infusion of sufentanil (1-2 ug/kg/hr) and propofol (1-3 mg/kg/hr) throughout the surgical procedure. On the day of surgery, after insertion of the invasive catheters, the patients received 8-12 mL of bupivacaine 0.375% and 16-24 ug of sufentanil into the epidural space. In the clonidine group, patients received a standardized opioid-based general anaesthesia supplemented with IV clonidine. In the control group, patients received a standardized opioid-based general anaesthesia without any supplement. Postoperative pain management consisted of an IV application of 1 g of paracetamol four times daily. Additionally, patients in the clonidine and control groups were allowed to administer a bolus of 2 mg of piritramide IV on demand by a patient-controlled analgesia device with a lockout time of 20 min. Patients in the TEA group received a continuous infusion of 2-3 mL of bupivacaine 0.75% via the epidural catheter. If the patient was < 65 yr of age, the bupivacaine solution contained 1 ug/mL sufentanil, but this drug was not given to older patients. If pain relief was not sufficient, patients administered an additional 2 mL of bupivacaine with a lockout interval of 20 min. IV piritramide was supplemented for additional pain relief.

Outcomes

Mortality
Myocardial infarction and ischaemia

Haemodynamics, lactate, cortisol plasma levels, and cardiac enzymes

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk"The patients were randomly allocated to one of the following three study groups"
Allocation concealment (selection bias)Unclear riskNot stated
Blinding (performance bias and detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot stated
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias

Lundstrom 2005

MethodsRandomized controlled trial
Participants

A total of 50 patients undergoing elective coronary artery bypass grafting

The criteria for inclusion in the study were age greater than 18 years, sinus rhythm on the ECG preoperatively, and written and oral informed consent. The criteria for exclusion from the study were oral anticoagulation and coagulopathy

Interventions

Conventional IV anaesthesia (24 patients) or general anaesthesia combined with thoracic epidural anaesthesia followed by postoperative epidural analgesia (26 patients) with bupivacaine

Epidural analgesia was induced with 8 to 10 mL of bupivacaine, 5 mg/mL. Anaesthesia was induced with IV midazolam, 3 to 5 mg, fentanyl, 0.3 mg, and a dose of pancuronium, 0.1 mg/kg. Anaesthesia was maintained with isoflurane and a continuous epidural infusion of bupivacaine, 1.25 mg/mL, with morphine, 25 ug/mL and 5 mL/h. Top-up bolus doses of 4 mL of bupivacaine, 5 mg/mL, were administered hourly during the operation. The administration of fentanyl was restricted to the dose given at the induction of general anaesthesia. Postoperatively, the epidural infusion continued, and bolus doses of 4 mL of 0.25% bupivacaine were administered as needed Anaesthesia was induced with midazolam, 3 to 5 mg, fentanyl, 15 to 30 ug/kg, and pancuronium, 0.1 mg/kg, which were administered at induction only and not repeated. Anaesthesia was maintained with isoflurane. Analgesia in the CON group was provided with intermittent IV morphine boluses (2.5 to 5 mg) until the morning of the first postoperative day when morphine, 5 to 10 mg, orally were given as needed. Postoperatively, all patients received oral paracetamol, 1 g every 6 h.

OutcomesMortality
Respiratory complications: episodic hypoxaemia
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk"Patients were randomly assigned to receive either conventional IV anaesthesia (CON) or general anaesthesia combined with TEA. The randomization list was generated from a table of random numbers."
Allocation concealment (selection bias)Unclear riskNot stated
Blinding (performance bias and detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll pre-specified outcomes reported
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias. Sample size calculation stated

Moore 1995

MethodsRandomized controlled trial
Participants

18 patients undergoing elective coronary artery bypass grafting

They had no history of metabolic or endocrine disease and had a normal bleeding time

Interventions

General anaesthesia alone or combined with thoracic epidural analgesia with 9 patients in each group

Bupivacaine 0.5% was given in 2 mL increments to achieve a adequate block, followed by an infusion of 0.375% bupivacaine at 5-8 mL/hr. General anaesthesia was induced by all patients with sufentanil 10 ug/kg, followed by a sleep-dose of thiopentone. Pancuronium 0.1 mg/kg was given and ventilation was with nitrous oxide in 50% oxygen. Post-OK analgesia was with 0.25 % bupivacaine at 5-8 mL/hr, increments of 2.5 mg papaveretum IV was given in the control group.

Outcomes

Mortality

Plasma catacholamines

Plasma cortisol

Serum insulin and growth hormone

Haemodynamic parameters

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskPatients were allocated by selection of a sealed envelope
Allocation concealment (selection bias)Low riskSealed envelopes
Blinding (performance bias and detection bias)
All outcomes
Unclear riskNot stated
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskOne patient had a severe haemorrhage and data from this patient were not presented
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias. Sample size calculation not stated

Nygard 2004

MethodsRandomized controlled trial
Participants

163 patients scheduled for elective coronary artery bypass grafting

Patients scheduled for elective CABG were screened for participation in the study. Inclusion criterion was sinus rhythm. Exclusion criteria were off-pump surgery; implanted pacemaker; use of amiodarone within 4 months of enrolment; a history of amiodarone toxicity; known thyroid disease, liver disease, or uncontrolled heart failure; a resting heart rate of less than 50 beats/min in the absence of medical therapy known to slow the heart rate; anticoagulant medication with warfarin; coagulopathy; pregnancy; and use of antiarrhythmic drugs other than alfa1-receptor antagonists, calcium channel antagonists, and digoxin.

Interventions

Group E (44 patients) had perioperative TEA, group E and A had TEA and amiodarone (35 patients), group A had amiodarone (36 patients), and group C served as control (48 patients).

On arrival in the operating room, epidural analgesia was induced with 8 to 10 mL of bupivacaine, 5 mg/mL. Anaesthesia was induced with intravenous (IV) midazolam, 3 to 5 mg, fentanyl, 0.3 mg, and pancuronium, 0.1 mg/kg. Anaesthesia was maintained with isoflurane in oxygen and a continuous epidural infusion of bupivacaine, 1.25 mg/mL, with morphine, 25 ug/mL, 5 mL/h. Additional bolus doses of 4 mL of bupivacaine, 5 mg/mL, were given hourly during the operation. Fentanyl was restricted to the dose given at induction of anaesthesia. Postoperatively, the epidural infusion was continued for 4 days. In groups A and C, anaesthesia was induced with midazolam, 3 to 5 mg, fentanyl, 15 to 30 ug/kg, and pancuronium, 0.1 mg/kg. Anaesthesia was maintained with isoflurane in oxygen. Postoperative analgesia in the non-TEA groups was provided with intermittent IV morphine boluses (2.5 to 5 mg) until the morning of the first postoperative day when 5 to 10 mg of oral morphine was given as needed.

Outcomes

Mortality

Supraventricular tachyarrhythmias

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskPatients were randomly assigned to four groups; randomization was 1:1:1:1
Allocation concealment (selection bias)Unclear riskThe randomization list was generated from a computerized table of random numbers
Blinding (performance bias and detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
Low riskOf the 196 patients included, 163 were evaluated: 18 patients had surgery cancelled, and 4 patients had a change in surgical procedure. One withdrew consent preoperatively, and 6 withdrew consent postoperatively. One patient had a stroke before surgery, and in 1 patient placement of the epidural catheter was unsuccessful. Two patients were excluded because of protocol violations.
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias. Sample size calculation stated

Onan 2011

MethodsA prospective randomized study
Participants30 patients with documented 3-vessel coronary artery disease who were scheduled for elective coronary artery bypass graft surgery
InterventionsIn both groups, anesthesia was induced with midazolam (0.1-0.2 mg/kg), fentanyl (7-10 g/kg), and rocuronium (0.6 mg/kg). The patients in the GA TEA group received a 20-mg bolus of 0.25% bupivacaine through the epidural catheters 1 hour before surgery. During the intraoperative period, 0.25% bupivacaine was infused at a rate of 20 mg/hr. The patients received a continuous epidural infusion of 0.125% bupivacaine, 4 to 10 mL/h, after surgery to attain sensory blockade.
Outcomes

Immunoreactivity

(Mortality)
(Myocardial infarction)

Graft blood flow

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk"The patients were assigned randomly into 2 groups"
Allocation concealment (selection bias)Unclear riskNot clear
Blinding (performance bias and detection bias)
All outcomes
High riskNot blinded
Incomplete outcome data (attrition bias)
All outcomes
Low riskData appear to be completed
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias. Sample size calculation stated

Priestley 2002

MethodsRandomized controlled trial
Participants

100 patients scheduled for elective coronary artery bypass grafting

All patients presenting for elective CABG were eligible for inclusion unless they had one or more of the following exclusion criteria: contraindications to the epidural technique (e.g., pre-existing coagulopathy, anticoagulation (i.e., full therapeutic doses of standard or low-molecular-weight heparin, warfarin, thrombolytic drugs, or potent antiplatelet drugs), or systemic or local infection); arthritis of the thoracic or cervical spine with a history of associated neurologic deficit; coexisting surgery (e.g., valvular, carotid, or aortic surgery); contraindications to any of the intended drugs in the treatment protocol; significant alcohol or other substance abuse; cognitive impairment; or other reason for inability to comply with treatment as assessed by the investigators.

Interventions

High (T1 to T4) thoracic epidural anaesthesia (TEA) with ropivacaine 1% (4 mL bolus, 3-5 mL/h infusion), with fentanyl (100 ug bolus, 15-25 ug/hr infusion) and a propofol infusion (6 mg/kg/hr) in 50 patients. Another 50 patients (the general anaesthesia group) received fentanyl 15 ug/kg and propofol (5 mg/kg/hr), followed by IV morphine patient-controlled analgesia.

On the day of surgery, after monitoring devices were inserted, an epidural bolus of 4 mL ropivacaine 1% and fentanyl 100 ug was administered, with supplemental ropivacaine 1% given as necessary to obtain blockade. After the bolus, an epidural infusion was commenced with ropivacaine 1% and fentanyl 5 ug/mL, and this continued for 48 h at 3-5 mL/hr. General anaesthesia (GA) in the TEA group consisted of IV fentanyl 2 ug/kg plus propofol 15 mg/kg/hr until consciousness was lost, followed by a propofol infusion at 6 mg/kg/hr, which continued until wiring of the sternum was completed. The GA group received fentanyl 15 ug/kg in divided doses from the induction of anaesthesia to sternotomy, in addition to propofol 15 mg/kg/hr until loss of consciousness, followed by a propofol infusion of 5 mg/kg/h (the larger fentanyl dose allowed a slightly smaller propofol dose than in the TEA group) until sternal closure. During cardiopulmonary bypass (CPB), a morphine infusion was commenced at 40 ug/kg/hr.

All patients received pancuronium 0.1 mg/kg for muscle relaxation

Outcomes

Mortality
VAS pain score
Stroke
Respiratory parameters: spirometry results

Time to extubation

Length of hospital stay

Mobilization goals

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskBlock randomization using sealed envelopes was used; patients at high risk were randomized separately
Allocation concealment (selection bias)Low riskSealed envelopes
Blinding (performance bias and detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot stated
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias. Sample size calculation stated

Rein 1989

MethodsRandomized controlled trial
Participants16 male patients requiring aorta-coronary bypass with extra-corporal circulation
Interventions

General anaesthesia alone or general anaesthesia combined with thoracic epidural analgesia (8 patients in each group)

All patients received morphine-scopolamine as premedication. General anaesthesia was obtained by the administration of thiopentone, pancuronium, nitrous oxide, diazepam and oxygen. The control group received fentanyl 54 ug/kg , the study group received fentanyl 14 ug/kg. TEA at T4-T5, initial dose of 50 mg bupivacaine, followed by 20 mg/hr during surgery. Postoperatively the control patients received morphine, the TEA patients were treated with epidural bupivacaine at 20 mg/hr.

Outcomes

Transcapillary fluid balance

Postoperative haemodynamics

(Mortality)

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk"16 male patients were allocated at random to two groups"
Allocation concealment (selection bias)Unclear riskNot stated
Blinding (performance bias and detection bias)
All outcomes
Unclear riskNot stated
Incomplete outcome data (attrition bias)
All outcomes
Low riskOne patient died 9 hr postoperatively and was excluded from the final analyses
Selective reporting (reporting bias)Unclear riskNot stated

Royse 2003

MethodsRandomized controlled trial
Participants76 patients receiving elective coronary artery bypass grafting with cardiopulmonary bypass
Interventions

High thoracic epidural analgesia with ropivacaine and fentanyl (37 patients) or IV morphine analgesia (39 patients)

Acetaminophen, indomethacin, and tramadol were allowed as supplemental analgesia in both groups

Epidural at T1-2 or T2-3. Eight mL of 0.5% ropivacaine with 20 ug of fentanyl was administered prior to induction of anaesthesia. Thereafter, ropivacaine 0.2% with fentanyl 2 ug/mL was infused at a rate of 5 to 14 mL per hour, adjusted to attain a sensory blockade of T1 to T10, and was ceased at 6:00 am on postoperative day 3. Anaesthesia consisted of midazolam (3 to 5 mg), fentanyl (200 ug), and a target-controlled infusion of propofol. For the control group anaesthesia consisted of midazolam (3 to 5 mg), propofol (2 to 4 g/mL), and a 2-stage target controlled alfentanil infusion (2 ug/mL, reduced to 0.05 ug/mL after cardiopulmonary bypass, and ceased after sternal wiring. Nurses were permitted to administer boluses of morphine in the intensive care unit until the patient was awake. This was followed by demand patient controlled intravenous morphine (1 mg bolus with 5 minute lockout period), which was continued until 6:00 am on postoperative day 3.

Outcomes

Pain

Physiotherapy cooperation

Depression and post-traumatic stress

Somatosensory sensitization

Lung function

Intraoperative haemodynamics

(Mortality)
(Myocardial infarction)

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskEighty patients undergoing elective CABG were randomized
Allocation concealment (selection bias)Unclear riskNot described
Blinding (performance bias and detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
Low risk4 patients were withdrawn: 1 patient withdrew itself from the study after randomization, deciding not to participate in research, and 2 patients failed epidurals and 1 patient from the control group requested the epidural
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias. Sample size calculation not stated

Scott 2001

MethodsRandomized controlled trial
Participants408 patients undergoing elective coronary artery bypass grafting with a normal coagulation screen and an ejection fraction of > 35%
Interventions

General anaesthesia alone (202 patients) or general anaesthesia combined with thoracic epidural analgesia (bupivacaine and clonidine) in 206 patients

All patients received target-controlled infusions (TCIs) of propofol and alfentanil for anaesthesia and analgesia, respectively. In Group GA, TCI of alfentanil continued for 24 hr and was then converted to a patient-controlled analgesia (PCA) IV morphine pump for another 48 h by using 1 mg bolus dosing with a 3 min lockout period. Patients in Group TEA had a thoracic epidural catheter sited in the operating theatre immediately before surgery at the T2-3 or T3-4 interspace with an initial bolus of 5 mL bupivacaine 0.5% followed by another 5 mL bolus after 10 min. After induction of GA and when central haemodynamic status was stable, a continuous infusion of 0.125% bupivacaine and 0.0006% clonidine (300 ug in 500 mL) commenced at an initial rate of 10 mL/hr. In Group TEA, the epidural infusion continued for 96 hr. “Top-up” bolus doses up to a maximum of 4 mL of 0.25% bupivacaine were administered either when the patient complained of pain.

Outcomes

(Mortality)
Myocardial infarction
Supraventricular arrhythmias
Stroke or cerebrovascular accident
Pulmonary complications

Acute confusion

Significant bleeding

Renal failure

Incidence of major organ complications

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk"Patients were randomized to one of two regimens...by using cards drawn from a sealed envelope"
Allocation concealment (selection bias)Low riskSealed envelope
Blinding (performance bias and detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
Low risk12 subjects had insufficient data
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias. Sample size calculation stated. Interim analysis performed and presented after 120 patients

Sharma 2010

MethodsRandomized controlled trial
ParticipantsSixty patients scheduled for elective OPCAB
InterventionsPatients in both the groups received general anaesthesia as per hospital protocol. Intravenous propofol (1-1.5 mg/kg) for induction of anesthesia along with fentanyl citrate (2-3 µgm/kg), midazolam (0.04 mg/kg) and vecuronium bromide (0.1 mg/kg) while isoflurane (1- 2 MAC) in air and oxygen mixture was used for maintenance of anesthesia. An epidural catheter was placed at C7-T1 /T1 -T2 level. Intrathecal placement was ruled out by using 3 mL of 2% lignocaine as a test dose. Once on the operation table, the patient was administered a bolus dose of 8-10 mL of bupivacaine (0.25%), inducing sensory block till at least T4. After confirming the block by loss of sensation to cold and pin prick, general anesthesia was administered. Bupivacaine infusion (0.125%) with 1µg/mL fentanyl citrate) at the rate of 5 mL/hr was commenced and continued till 3rd postoperative day for providing intra- and postoperative analgesia.
Outcomes

Postoperative spirometric values

Postoperative PaO2/FiO2 ratio and PaCO2 values

Painscores

(Mortality)

NotesOnly obese patients (BMI > 30 kg/m2) were included
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk"Patients were randomized into two groups of 30 each."
Allocation concealment (selection bias)Unclear riskNot clear
Blinding (performance bias and detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
Low riskData appear to be completed
Selective reporting (reporting bias)Unclear riskAppears to be free of other sources of bias. Sample size calculation not stated

Stenseth 1994

MethodsRandomized controlled trial
Participants28 male patients < 65 years and with an ejection fraction of > 50% undergoing cardiac surgery. They were all ASA III with double or triple vessel disease
Interventions

General anaesthesia with high dose fentanyl (10 patients), general anaesthesia with low dose fentanyl and thoracic epidural analgesia (18 patients).

Epidural at T4-5 or T5-6. 10 mL 0.5% bupivacaine, top up doses of 4 mL 0.5% bupivacaine during surgery. GA with thiopentone, pancuronium. High dose fentanyl was 20-30 ug/kg, low dose fentanyl was 5 ug/kg. HF received morphine IV on demand, both epidural groups received a continuous epidural infusion of bupivacaine 0.5 % at 3 mL/hr with additional top ups of 4 mL every hour.

Outcomes

(Mortality)
(Myocardial infarction)

Hypertension

Tachycardia

Other haemodynamic parameters

Use of vasoactive drugs

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk30 male patients gave informed consent and were randomized into one of three groups
Allocation concealment (selection bias)Unclear riskNot stated
Blinding (performance bias and detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
Low riskTwo patients were excluded from the final analysis due to surgical problems
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias. Sample size calculation not stated

Stenseth 1996

MethodsRandomized controlled trial
Participants52 male patients age < 65 years and with an ejection fraction of > 50%. They were all in NYHA class III and presented with double- or triple-vessel disease
Interventions

High dose fentanyl general anaesthesia and an epidural group receiving low dose fentanyl general anaesthesia + thoracic epidural analgesia with bupivacaine 0.5% (26 patients in each group)

Patients allocated to the epidural group had an epidural catheter inserted at level T4- T5 or T5-T6. Epidural group patients received an epidural injection of 10 mL of bupivacaine 5 mg/mL. Top up bolus doses of 4 mL bupivacaine 5 mg/mL were given hourly during the operation. General anaesthesia consisted of thiopentone, diazepam, nitrous oxide and pancuronium (with no difference between the groups) in addition to fentanyl. The control group patients received a total of 55 ug/kg of fentanyl, the epidural group a total of 15 ug/kg. Control group patients received morphine IV on demand for pain relief during the first 20 hr after surgery. During the next 2 days daily paracetamol 3.2 g and codeine 240 mg (Paralgin forte) were given rectally as basal analgesia, which was supplemented with morphine IV as needed. Epidural group patients received a continuous epiduraI infusion of bupivacaine 5 mg/ml, 3 ml/h with additional top up doses of 4 mL bupivacaine every 4 hr and supplementation with IV morphine when needed. They received morphine epidurally 4-6 mg 3-4 times a day for the next 2 days, supplemented with bupivacaine 5 mg/mL and morphine IV when needed. For both groups, supplementation was most often needed in association with physiotherapy and mobilization.
From the 3rd postoperative day all patients received only Paralgin forte on request

Outcomes

(Mortality)
(Myocardial infarction)

Extubation time

Spirometry results after surgery

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk54 male patients were randomized into two groups
Allocation concealment (selection bias)Unclear riskNot stated
Blinding (performance bias and detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
Low riskTwo patients were excluded from the analysis because of the suffered complications
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias. Sample size calculation not stated

Svircevic 2011

MethodsTwo centre, randomized controlled trial
Participants656 patients scheduled for elective cardiac surgery, including off-pump procedures
InterventionsNot known
Outcomes

Mortality
Myocardial infarction

Pulmonary complications

Length of hospital stay

Arrhythmias

Stroke

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskThe random allocation sequence was concealed and computer-generated in permuted unequal blocks, accessible through an Internet site
Allocation concealment (selection bias)Low riskThe random allocation sequence was concealed
Blinding (performance bias and detection bias)
All outcomes
High riskNot blinded
Incomplete outcome data (attrition bias)
All outcomes
Low riskOne patient was excluded because his surgery was cancelled, and one patient withdrew his consent after randomization
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias. Sample size calculation stated

Tenling 1999

MethodsRandomized controlled trial
Participants28 patients scheduled for coronary artery bypass grafting. Inclusion criteria were stable angina pectoris with a left ventricular ejection fraction of more than 40%
Interventions

Perioperative and postoperative TEA ( bupivacaine) was added to general anaesthesia (GA) in 14 patients, and 14 patients receiving GA alone served as controls

Epidural at T3-4 or T4-5. Epidural analgesia was induced with 8 to 12 mL of bupivacaine, 5 mg/mL. The block was maintained with an infusion of bupivacaine, 5 mg/mL, 4 to 8 mL/hr, until the patient arrived at the intensive care unit. Anaesthesia was induced with fentanyl, 5 to 10 ug/kg; thiopental, 1.5 to 2.5 mg/kg; and pancuronium, 0.1 mg/kg. In the TEA group, fentanyl was restricted to the dose administered at induction. From then on, analgesia was achieved with TEA. The GA group was administered small doses of fentanyl as needed (1 to 2 mg total). In both groups, anaesthesia was maintained with inhaled isoflurane, 0.5 to 1.0 minimal alveolar concentration.
In the GA group, postoperative analgesia was achieved with titrated doses of ketobemidone, 1 to 3 mg
intravenously, according to standard procedures. The TEA group
received a continuous infusion of bupivacaine, 2 mg/mL, and sufentanil, 1 ug/mL epidurally (3 to 7 mL/hr), from arrival at the ICU until the end of the study on the day after surgery.

OutcomesMortality
Respiratory: ventilation/perfusion mismatch, atelectasis, time to extubation
Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk"After giving their informed consent, 30 patients scheduled for CABG were randomized to receive either general anaesthesia (GA group) or GA with TEA (TEA group). Randomization was achieved with sealed envelopes"
Allocation concealment (selection bias)Low riskSealed envelopes
Blinding (performance bias and detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
Low riskTwo patients were excluded from the analyses
Selective reporting (reporting bias)Low riskAppears to be free of other sources of bias. Sample size calculation stated

Characteristics of excluded studies [ordered by study ID]

StudyReason for exclusion
Chae 1998No adequate sequence generation
Dohle 2001No adequate sequence generation and no allocation concealment
Fawcett 1997No adequate sequence generation
Greisen 2012Not randomized
Jakobsen 2012No relevant outcomes reported
Jideus 2001Not randomized
Kessler 2005No adequate sequence generation
Liang 2012Comparison between epidural anesthesia perioperatively and postoperatively
Liem 1992aNo relevant outcomes reported
Liem 1992bNo relevant outcomes reported
Mehta 1998No adequate sequence generation and no allocation concealment
Olivier 2005No comparison to general anaesthesia
Stenseth 1995No relevant outcomes reported
Thorelius 1996No adequate sequence generation

Ancillary