British Journal of Anaesthesia 115 (1): 25–32 (2015) doi: 10.1093/bja/aev201 Review Article

Benefits and risks of epidural analgesia in cardiac surgery† 1

Department of Anaesthesia and Intensive Care, IRCCS San Raffaele Scientific Institute, 2Vita-Salute San Raffaele University, Via Olgettina 60, Milan 20132, Italy, 3Anaesthesia and Pain Management Unit, The Royal Melbourne Hospital, and 4Department of Surgery, The University of Melbourne, The Royal Melbourne Hospital, Level 6 Clinical Medical Research Building, Melbourne, VIC 3050, Australia *Corresponding author.E-mail: [email protected]

Abstract Background: Epidurals provide excellent analgesia for cardiac surgery and may reduce complications. However, their use has been tempered because of concern of the rare, but serious complication of epidural haematoma. The aim of this meta-analysis was to assess the effect of epidural on survival and the risk estimate of epidural haematoma. Methods: A systematic review of the literature (Pubmed, Embase, Scopus and the Cochrane Register) and a meta-analysis of the available randomized and case-matched studies were performed to estimate the effect on survival. An international, directed and viral anonymous survey was performed to identify the incidence of haematomas with a corresponding estimate of the number of epidurals performed. Results: Of 66 randomized and case-matched studies, 57 trials including 6383 patients reported the incidence of all-cause mortality at the longest follow up available, with a significant reduction with epidurals (59/3123 [1.9%] vs 108/3260 [3.3%] in the control arm, RR 0.65 [95% CI 0.48–0.86], P=0.003, NNT=70). No epidural haematoma was reported in these 66 trials (3320 epidurals). All other literature revealed nine haematomas in 13 100 patients. Through the anonymous, web-based, viral, international survey, we identified 16 further, non-published, epidural haematomas from 72 400 positioned epidurals. Therefore, a total of 25 haematomas have been identified from an estimate of 88 820 positioned epidurals, producing an estimated risk of 1:3552 (95% CI 1:2552–1:5841). Conclusions: The use of epidural analgesia in cardiac surgery is associated with a reduction in mortality (NNT=70), and with an estimated risk of epidural haematoma of 1:3552. Key words: anaesthesia; epidural; myocardial infarction; spinal epidural hematoma; survival rate; thoracic surgery

Editor’s key points • In this systematic review, the authors demonstrate a clinically significant reduction in post-cardiac-surgery mortality associated with the use of epidurals. • Epidural haematoma, an acknowledged hazard in this context, had a risk of approximately 1:3500.

Epidurals have been widely used in non-cardiac surgery, reporting improved analgesia1 2 and reduced risk of complications,2 3 particularly pulmonary complications.4 A non-randomized report by Wijeysundera and colleagues confirmed a small, but significant reduction in 30-day mortality for thoracic epidurals, in major non-cardiac surgery (1.7 vs 2.0%, relative risk 0.89, P=0.02).3 The incidence of spinal decompression laminectomy

†

This Article is accompanied by Editorial Aev155. Accepted: February 18, 2015 © The Author 2015. Published by Oxford University Press on behalf of the British Journal of Anaesthesia. All rights reserved. For Permissions, please email: [email protected]

25

Downloaded from http://bja.oxfordjournals.org/ at North Dakota State University on July 1, 2015

G. Landoni1,2, *, F. Isella1, M. Greco1, A. Zangrillo1,2 and C. F. Royse3,4

26 |

Landoni et al.

Methods Eligibility criteria We searched all the reports regarding both cardiac surgery setting and an anaesthetic plan involving the use of epidural analgesia, and all reports of symptomatic epidural haematoma, with no language restriction. For the meta-analysis of case-matched and RCTs, we selected all the manuscripts comparing one group of patients receiving epidural analgesia combined with general anaesthesia with one group of patients receiving general anaesthesia only for cardiac surgery. Non-human experimental protocols, studies not reporting data on mortality or myocardial infarction, and duplicate publications were excluded from the meta-analysis.

Information sources An extensive search across different databases (Medline/PubMed, Pubmed Central, Embase, Scopus and the Cochrane Central Register of clinical trials) was performed on December 2013. Corresponding authors of published manuscripts were contacted directly.

Search strategies The Pubmed search included ‘(epidural OR peridural OR neuraxial OR spinal) AND (cardiac OR cardiothoracic OR cardiovascular OR “heart surgery” OR “cardiovascular surgical procedures”) AND (surgery OR surgical)’. The systematic review of the literature aimed to identify: (a) every case report or case series (including, but not limited to, RCTs and case-matched studies) where epidural was used in cardiac surgery; (b) every patient with catheter-related epidural haematoma occurring in cardiac surgery. To increase the sensitivity in identifying epidural haematomas we also used other sensitive search strings, which are listed in Supplementary material S1.

Worldwide survey To determine both the incidence of haematoma (numerator) and the number of epidurals performed (denominator), we conducted a world-wide survey according to the following methods: 1. all survey responses were anonymous to encourage reporting; 2. all authors of any publications revealed in the prior literature search and all known experts in the field were sent survey forms; 3. all people contacted were encouraged to forward on the survey to anyone they knew who had performed epidurals for cardiac surgery; 4. all surveys were crosschecked within each country and against the literature to avoid duplication of the incidence of haematomas or number of epidurals performed. The detailed survey form is available in Supplementary material S2.

Study selection and data extraction Results from database searches were screened at a title/abstract level by trained investigators. A difference in opinion as to the suitability of a reference was resolved by consensus after reviewing the full paper. Corresponding authors were contacted in order to obtain missing data where appropriate. For RCTs and casematched studies, data were extracted for mortality and myocardial infarction. The complete data extraction sheet is shown in Supplementary Table S1. When continuous outcome data were available as median and interquartile range (IQR) only, they were included in pooled analysis after approximation to normal

Downloaded from http://bja.oxfordjournals.org/ at North Dakota State University on July 1, 2015

was 1:5000 and not different from those not receiving epidurals. Taken together, for every patient harmed by an epidural, nine lives could be potentially saved, indicating that benefit exceeded risk in this cohort.3 Conversely, in cardiac surgery, the use of epidurals have been tempered worldwide as a result of the fear that the risk of epidural haematoma might be higher than in other surgical settings, because of the full-dose anticoagulation required for cardiopulmonary bypass (CPB). However, reports of epidural haematomas after placement or removal of the epidural catheter in cardiac surgery have been confined to case reports, with no haematomas described in large audits or randomized controlled trials (RCTs). Estimates of haematoma risk have been attempted over the yrs: using the mathematical model of Hanley and Lippman-Hand for an event that had not yet occurred, Ho AM and colleagues suggested a 95% confidence interval (CI) of 1:1500–1:150 000;5 Bracco and colleagues more recently estimated it to be 1:12 000 (95% CI 1:2100–1:68 000), similar to the incidence of epidural haematoma in the non-obstetric, non-cardiac population (1:10 000, 95% CI 1:6700–1:14 900).6 Further, it should not be forgotten that epidural haematomas can also occur spontaneously, as reported first in 1869 by Jackson.7 Benefits of epidural analgesia have been investigated in cardiac surgery in more than 50 RCTs, most of them reporting positive effects on pain and secondary endpoints such as postoperative complications. However, even if postoperative pain is improved with epidurals,8 benefits other than pain relief have been sought in order to balance the potential risk of haematoma.9 Although most of those RCTs have involved low-risk patients undergoing coronary artery bypass surgery, prior meta-analyses have shown significant benefit with epidurals for a combined outcome of mortality and myocardial infarction,10 for ventilation time, 10 for pulmonary complications, and supraventricular trachyarrhythmias. 11 However, neither meta-analysis was sufficiently large to show significant differences in mortality, despite a clinically important trend (odd ratio 0.7110 and relative risk 0.8111). Therefore, an equipoise still exists for epidural use in cardiac surgery with the potential for a significant reduction in mortality and major complications vs the possibility of a greater haematoma risk than non-cardiac surgery (which does not require full anticoagulation for CPB). The deficiency in the current literature is that the sample size for mortality estimates has been too small to prove the relative risk reduction identified, and similarly continued fear that the epidural haematoma risk may be increased in cardiac surgery, despite repeated risk estimates suggesting that this is not increased. Published reports of epidural use and haematoma occurrence are skewed in both directions: to a lower incidence if only RCTs are analysed, as the incidence of haematoma (the numerator) is zero; to a higher incidence in case reports, since the denominator (the number of epidurals performed to produce the haematoma) is not reported. Our research question was to perform an updated systematic review and meta-analysis, with a sample size that was large enough to show a real benefit or harm in terms of mortality, if present, and to conduct a worldwide survey to identify the risk of epidural haematoma, with identification of both numerator and denominator. The hypotheses based on the previous literature and risk estimates, were that the incidence of catheter-related haematoma was similar to that reported for thoracic epidural use in non-cardiac surgical settings, and that a mortality advantage would exist for epidural use in cardiac surgery.

Epidural in cardiac surgery

| 27

distribution, using a standard transformation formula (IQR=SD 1.35−1, mean=median).

Outcomes For the meta-analysis, the primary outcome was mortality, which was determined at the longest follow up available, whereas the secondary outcomes were myocardial infarction occurrence and length of mechanical ventilation. The incidence of epidural haematomas and the estimated total number of epidurals positioned in cardiac surgery worldwide were used to provide an estimate of risk.

Electronic database search, contact of international expert

11321 citations retrieved from database searches and additional hints

11234 titles/abstracts not eligible for inclusion

Statistical analysis

Results Number of epidurals performed in cardiac surgical procedures worldwide According to the eligibility criteria, we identified 66 published randomized (n=59) or case-matched trials (n=7) including 3320 patients who received epidural analgesia in cardiac surgery (see Fig. 1, Table 1, and, for references, Supplementary material S4 which also shows overlapping publications). In addition we retrieved 130 published case reports, case series and randomized trials not fulfilling the inclusion criteria for a total of further 13 100 patients (references available from authors at request). The anonymous web survey estimated an additional 72 400 epidurals positioned worldwide in the last 20 yrs, resulting in a total of 88 820 epidurals performed.

Meta-analysis The primary outcome of mortality was reported in 57 publications including a total of 6383 patients (3123 who received epidurals and 3260 who received general anaesthesia only). Mortality was significantly lower in the epidural group (59/3123 [1.9%] with epidurals vs 108/3260 [3.3%] in the general anaesthesia arm, risk ratio [RR] 0.65 [95% CI 0.48–0.86], P=0.003, I 2=0%, NNT=70). The forest plot is shown as Fig. 2. The funnel plot (Supplementary Fig. S1) investigated the possibility that small studies

87 abstracts eligible for inclusion and detailed assessment

15 studies not eligible because of explicit exclusion criteria 6 duplicate publications

66 randomized and matched trials finally included

Fig 1 Flow chart of the articles selection process for randomized and matched trials.

with negative effects were unpublished or less accessible than larger studies and clearly showed the absence of small study bias. A sub-analysis was performed on patients included in randomized studies only, which confirmed the direction and magnitude of mortality reduction with epidurals, though the sample size was insufficient to reach significance (27/2054 [1.3%] for epidurals vs 51/2204 [2.3%] for the general anaesthesia group, RR 0.78 [95% CI 0.52–1.18], P=0.24). The secondary outcome of myocardial infarction was reported by 51 publications for a total of 5718 patients. The incidence of perioperative myocardial infarction was significantly lower in the epidural group (67/2785 [2.4%] vs 108/2933 [3.7%] in the control arm, RR 0.68 [95% CI 0.51–0.90], P=0.007, I 2=0%, NNT=78). The forest plot is shown as Supplementary Fig. S4, and the funnel plot, confirming the absence of small study bias, is shown in Supplementary Fig. S5. Length of mechanical ventilation demonstrated a reduction in ventilation time for epidurals when compared with controls (−0.32 h [95% CI −0.50 to −0.14], I 2=82%, P=0.001) and is reported in Supplementary Fig. S6.

Incidence of epidural haematoma We identified 25 epidural haematomas in cardiac surgery (five from published reports,12–16 four from grey literature9 17 and further 16 from the anonymous web survey) from a total of 14 countries: Belgium (1), Brazil (1), France (1), Germany (2), India (2), Italy (1), Japan (2), Korea (1), Malaysia (1), Russia (3), Sweden (2), Turkey (1) United Kingdom (3), and United States (4). The source and location for the haematomas are shown in Table 2. Two neurologic complications were excluded as they were not related to cardiac surgery (one patient with epidural use in refractory angina18 and one patient with stroke, related to the management of the epidural catheter19). No haematoma was reported in any RCT or case-matched study. It is worth noticing that in three out of the

Downloaded from http://bja.oxfordjournals.org/ at North Dakota State University on July 1, 2015

All the randomized and case-matched trials meeting the eligibility criteria were included in the meta-analysis. A sub-analysis was performed including randomized studies only. Calculations were performed with Stata 11 (Stata Statistical Software: release 11, College Station, TX) and RevMan version 5.2 (a freeware available from The Cochrane Collaboration). The detailed meta-analysis methodology is presented in Supplementary material S3. We performed a quality assessment according to the Cochrane Collaboration methods, reported in Supplementary Table S2. Sensitivity pooled analysis were performed including only randomized trials with low risk of bias, and then repeated with studies with low and moderate risk of bias. The risk of epidural haematoma was calculated, dividing the number of haematomas by the total number of epidurals performed, with the Wilson Score Interval used to estimate the 95 and 99% confidence intervals. The incidence of events and risk ratio with 95% CI are presented throughout. The absolute risk reduction is expressed as the number of patients needed to treat (NNT) to prevent an event, calculated as 100 (difference in % event rate between groups)−1.

28 |

Landoni et al.

Table 1 First author, journal, yr of publication, and number of enrolled patients in the 66 randomized or matched trials that compared the use of epidural analgesia on top of general anaesthesia with general anaesthesia alone in cardiac surgical patients (full references are in the Supplementary material S4). *Case-matched trials Journal

Yr

Epidurals

Controls

Amat-Santos IJ35* Bach F36 Bakhtiary F37 Barrington MJ38 Berendes E39 Bichel T40 Brix-Christensen V41 Caputo M4252 Caputo M43 Chae BK44 Crescenzi G45* de Vries AJ46 Dhole S47 El-Baz N48 Fawcett WJ49 Fillinger MP50 Greisen J51 Gurses E52 Hansdottir V53 Heijmans J54 Jakobsen CJ55 Jakobsen CJ56 Jidéus L57 Kendall JB58 Kessler P59* Kilickan L60 Kilickan L61 Kirnö K62 Kirov MY63 Kurtoglu M64 Lagunilla J65 Lenkutis T66 Liang Y67 Liem TH68 Loick HM69 Lundstrom LH70 Mehta Y71 Mehta Y72 Moore CM73 Neškovic V74 Nygard E75 Onan B76 Onan IS77 Palomero Rodriguez MA78 Porizka M79* Priestley MC80 Rajakaruna C81 Rein KA82 Rojas-pé Rez E83 Rosen KR84 Royse C85 Salvi L86* Scott NB87 Sendasgupta C88 Sharma M89 Slinko SK90 Stenger M91* Stenseth R92 Stenseth R93

Heart Acta Anaesthesiol Scand J Thorac Cardiovasc Surg Anesth Analg Arch Surg Paediatric Anaesthesia Acta Anaesth Scand Ann Thorac Surg Anesthesiology Surg Today J Cardiothorac Vasc Anesth J Cardiothorac Vasc Anesth J Cardiothorac Vasc Anesth J Thorac Cardiovasc Surg Anaesthesia J Cardiothorac Vasc Anesth Acta Anaesth Scand Med Sci Monit Anesthesiology J Cardiothorac Vasc Anesth Europ J Anesth (abstract) J Cardiothorac Vasc Anesth Ann Thorac Surg Anaesthesia J Cardiothorac Vasc Anesth J Cardiovasc Surg J Cardiovasc Surg Anesth Analg BMC Anesthesiology Anadolu Kardiyol Derg Acta Anaesthesiol Scand Perfusion J Anesth J Cardiothorac Vasc Anesth Anesth Analg Chest J Cardiothorac Vasc Anesth Ann Card Anaesth Br J Anaesth Vojnosanit Pregl J Cardiothorac Vasc Anesth J Card Surg J Cardiothorac Vasc Anesthesia Minerva Anestesiol J Anesth Anesth Analg Interact Cardiovasc Thorac Surg Acta Anaesth Scand Paediatric Anaesthesia Anesthesiology Ann Thorac Surg J Cardiothorac Vasc Anesth Anesth Analg Ann Card Anaesth Ann Card Anaesth Anesteziol Reanimatol J Cardiothorac Vasc Anesth Acta Anaesth Scand Eur J Cardio-thorac Surg

2012 2002 2007 2005 2003 2000 1998 2009 2011 1998 2009 2002 2001 1987 1997 2002 2013 2013 2006 2007 2007 2012 2001 2004 2005 2005 2008 1994 2011 2009 2006 2009 2012 1992 1999 2005 1998 2010 1995 2013 2004 2013 2011 2008 2011 2002 2013 1989 2003 1989 2007 2007 2001 2009 2010 1999 2013 1994 1996

74 13 66 60 36 12 8 36 109 12 46 30 21 30 8 30 21 32 58 15 10 31 45 10 30 40 30 10 62 34 26 30 32 27 25 30 25 31 9 35 80 20 15 10 15 50 109 8 15 16 37 389 206 15 30 40 508 20 27

61 27 66 60 37 12 8 38 117 12 46 60 20 30 8 30 21 32 55 45 10 31 96 20 30 40 30 10 31 42 26 30 32 27 47 25 117 31 9 46 84 20 15 12 15 50 117 8 15 16 39 389 202 15 30 16 508 10 27 Continued

Downloaded from http://bja.oxfordjournals.org/ at North Dakota State University on July 1, 2015

First author

Epidural in cardiac surgery

| 29

Table 1 Continued First author 94

Svircevic V Tenenbein PK95 Tenling A96 Thorelius J97 Toda A98* Volk T99 Yung MC100

Journal

Yr

Epidurals

Controls

Anesthesiology Can J Anesth Acta Anaesth Scand Eur J Cardio-thorac Surg Asian Cardiovascular & Thoracic Annals Cytokine Zhonghua Yi Xue Za Zhi (Taipei)

2011 2008 2000 1996 2013 2003 1997

327 25 15 14 7 13 20

329 25 15 13 7 13 20

Discussion Our study reports that the use of epidural in cardiac surgery reduces the risk of death, of mechanical ventilation and of myocardial infarction. The serious complication of epidural haematoma, and the number of epidurals performed in cardiac surgical procedures, is widely under-reported in the medical literature. However, the number of patients needed to treat to save a life is 70 and the number treated to harm (haematoma) is 1:3552, which confirms the relative safety of epidural use in cardiac surgery. Even if the present study has several limitations (the metaanalysis included also case-matched non-randomized trials and the viral survey is not as precise as a mandatory register), it also gives important pieces of information to the physicians. Before this study, none of the randomized trials, cohort studies or meta-analyses had sufficient sample size to show a significant difference in mortality rate with epidurals, leading to a high risk of type II errors when concluding that a difference did not exist. Such trials have also been predominantly performed in lower-risk patients undergoing coronary artery surgery. This leads to a reduced event rate compared with all cardiac surgery and, therefore, requires an even larger sample size to show significance. The two most recent meta-analyses,10 11 however showed similar direction and magnitude of mortality reduction to this study. The mechanism for a reduction in mortality with epidural is unknown. Use of high thoracic epidurals improves pain relief,8 blunts the stress response to surgery,20–22 produces coronary artery dilatation23 and improves myocardial oxygen partial pressure,24 may reduce stress-induced immunosuppression,25 and allows a lower dose of general anaesthesia and systemic analgesia,8 which in turn may lead to reduced myocardial injury,10 26 shorter duration of mechanical ventilation,8 27 improved physiotherapy co-operation,20 less confusion development,7 and reduced long term symptoms of depression8 28 and post-traumatic stress disorder.8 All of these are potential mechanism to improve healing and recovery, but we are unable to speculate from our data what the predominant mechanism is. We confirm, however, a reduction in myocardial infarction occurrence using epidural analgesia, which could contribute to reducing mortality in this surgical cohort. According to our data, and in agreement with previous investigations, ventilation time was also significantly reduced with epidurals, which could help patients’ recovery by reducing the incidence of respiratory postoperative complications.

Epidural haematoma and consequent neurological damage including paraplegia is a potential risk for epidural use in all clinical situations including non-cardiac surgery, pain medicine and obstetrics. It is therefore an expected outcome that catheterrelated haematoma will occur with epidural usage in cardiac surgery. The controversy in cardiac surgery9 29 relates to the question of whether the incidence of epidural haematoma will be increased compared with use in non-cardiac surgery, as a result of the full anticoagulation required for CPB. However, epidural haematoma patients were not reported until 2004,12 and the only estimate before that was based on mathematical modelling of an event that had not yet occurred5 and that led to very wide confidence intervals of risk, ranging from 1:1500 to 1:150 000 patients.5 As the reporting of epidurals positioned in cardiac surgery increased over time, and with the publication of haematoma occurrence, the most recent risk assessment in 2007 was an estimated incidence of 1:12 000 patients (95% CI 1:2100–1:68 000), which was comparable with the risk assessment in non-cardiac surgery (1:10 000, 95% CI 1:6700–1:14 900). The largest case cohort study by Wijeysundera and colleagues reported an incidence of spinal laminectomy (which was presumed to be associated with epidural haematoma) of 1:4524 patients (95% CI 1:2695– 1:9090), establishing a benchmark for risk in patients undergoing major non-cardiac surgery.3 They also reported a risk of decompression laminectomy of 1:7246 (95% CI 1:5000–1:10 870) when epidurals were not used, reinforcing that not all epidural haematomas are caused by the insertion of an epidural catheter. We report a sample size nearly double of that by Wijeysundera and colleagues3 and confirm that the risk of epidural haematoma, when using high thoracic epidural analgesia in cardiac surgery, is not increased compared with use in major non-cardiac surgery, where full anticoagulation is not required. It should be acknowledged that, even if the data we collected cannot confirm it (Table 2), epidural hematoma might be triggered by repetitive punctures, bloody taps, impaired anticoagulation or excessive antiplatelet therapy, when the catheter is positioned or removed. Epidural usage has shown superior analgesia with reduction in ventilation time8 27 and pulmonary complications.11 30 It is also associated with a reduced incidence of supraventricular tachycardias10 30 and less confusion.30 Ultimately, it improves overall quality of recovery and promotes earlier hospital discharge.27 Despite these potential advantages, however, in the setting of the risk of haematoma controversy, researchers have attempted to identify whether epidurals reduce mortality, in order to counterbalance the perceived risk of the technique. Overall, when investigating a new technique, it is often assumed by practitioners, that the novel approach is inherently more dangerous than the current established technique. The concept of risk-risk ratio, where the risk of the conventional approach is compared with the risk of the new technique, is difficult to perform in isolation, from the complexity of the surgical procedure

Downloaded from http://bja.oxfordjournals.org/ at North Dakota State University on July 1, 2015

five published epidural hematoma reports patients recovered most of the motor function in both legs. Therefore, the risk of haematoma associated with the use of epidural in cardiac surgery is 1:3552 (95% CI 1:2552–1:5841 and 99% CI 1:2344–1:7326).

30 |

Landoni et al.

Study or Subgroup

8 0 0 1 1 0 0 1 1 1 0 0 0 1 0 1 0 1 0 8 0 0 0 0 1 0 0 0 0 0 0 0 0 1 0 0 0 1 4 1 1 1 1 0 0 5 1 0 14 0 1 3 0 0 0 0 0

74 13 66 60 36 12 8 36 109 46 30 21 30 30 32 58 15 10 31 45 10 30 40 30 62 34 32 27 25 30 25 31 9 35 80 20 15 10 15 50 109 8 15 16 37 389 206 15 508 20 27 327 25 15 14 7 13

3123 Total (95% Cl) 59 Total events Heterogeneity: χ2=19.16, df=29 (P=0.92); I 2=0% Test for overall effect: Z=2.97 (P=0.003)

19 1 0 2 3 0 0 0 0 3 0 0 0 0 0 1 0 0 1 25 0 0 1 1 0 0 0 0 1 0 0 0 0 0 2 0 0 1 3 0 0 0 2 0 1 2 2 0 30 0 0 7 0 0 0 0 0

61 27 66 60 37 12 8 38 117 46 60 20 30 30 32 55 45 10 31 96 20 30 40 30 31 42 32 27 47 25 25 31 9 46 84 20 15 12 15 50 117 8 15 16 39 389 202 15 508 10 27 329 25 15 13 7 13

19.3% 0.9% 1.8% 2.7%

0.5% 0.4% 2.8%

0.5% 0.9% 0.5% 1.4% 14.8%

1.4% 1.4% 0.6%

1.0%

0.4% 2.3%

0.8% 2.8% 0.5% 0.4% 0.5% 1.8% 1.4% 1.8% 1.9% 27.7% 0.5% 6.5%

3260 100.0%

Risk Ratio M-H, Fixed, 95% Cl

0.35 [0.16, 0.74] 0.67 [0.03, 15.34] Not estimable 0.50 [0.05, 5.37] 0.34 [0.04, 3.14] Not estimable Not estimable 3.16 [0.13, 75.20] 3.22 [0.13, 78.17] 0.33 [0.04, 3.09] Not estimable Not estimable Not estimable 3.00 [0.13, 70.83] Not estimable 0.95 [0.06, 14.79] Not estimable 3.00 [0.14, 65.90] 0.33 [0.01, 7.88] 0.68 [0.33, 1.39] Not estimable Not estimable 0.33 [0.01, 7.95] 0.33 [0.01, 7.87] 1.52 [0.06, 36.36] Not estimable Not estimable Not estimable 0.62 [0.03, 14.58] Not estimable Not estimable Not estimable Not estimable 3.92 [0.16, 93.34] 0.21 [0.01, 4.31] Not estimable Not estimable 1.20 [0.09, 16.84] 1.33 [0.36, 4.97] 3.00 [0.13, 71.92] 3.22 [0.13, 78.17] 3.00 [0.14, 64.26] 0.50 [0.05, 4.94] Not estimable 0.35 [0.01, 8.35] 2.50 [0.49, 12.81] 0.49 [0.04, 5.36] Not estimable 0.47 [0.25, 0.87] Not estimable 3.00 [0.13, 70.53] 0.43 [0.11, 1.65] Not estimable Not estimable Not estimable Not estimable Not estimable

Downloaded from http://bja.oxfordjournals.org/ at North Dakota State University on July 1, 2015

Amat-Santos 2012 Bach 2002 Bakhtiary 2007 Barrington 2005 Berendes 2003 Bichel 2000 Brix-Christensen 1998 Caputo 2009 Caputo 2011 Crescenzi 2009 de Vries 2002 Dhole 2001 EI- Baz 1987 Fillinger 2002 Gurses 2013 Hansdottir 2006 Heijmans 2007 Jakobsen 2007 Jakobsen 2012 Jidéus 2001 Kendall 2004 Kessler 2005 Kilickan 2005 Kilickan 2008 Kirov 2011 Kurtoglu 2009 Liang 2012 Liem 1992 Loick 1999 Lundstrom 2005 Mehta 1998 Mehta 2010 Moore 1995 Neškovic 2013 Nygard 2004 Onan B 2013 Onan IS 2011 Palomero Rodriguez 2008 Porizka 2011 Priestley 2002 Rajakaruna 2013 Rein 1989 Rojas-Pérez 2003 Rosen 1989 Royse 2007 Salvi 2007 Scott 2001 Sendasgupta 2009 Stenger 2013 Stenseth 1994 Stenseth 1996 Svircevic 2011 Tenenbein 2008 Tenling 2000 Thorelius 1996 Toda 2013 Volk 2003

Epidural Analgesia Control Risk Ratio Events Total Events Total Weight M-H, Fixed, 95% Cl

0.65 [0.48, 0.86]

108 0.01

0.1 1 10 Favours [experimental] Favours [control]

100

Fig 2 Forest plot for mortality risk of combined epidural-general anaesthesia vs general anaesthesia alone in cardiac surgery.

and from all the factors involved in patients healing and recovery. Risk of mortality is a composite endpoint for all these factors. Our data show that the risk of death is less when using an epidural as part of the anaesthetic technique, and the risk:risk concept favours epidurals for cardiac surgery.

Finally, there are several limitations to our study. Firstly, our data on mortality and myocardial infarction are drawn from many small studies, none of which was adequately powered to measure mortality, and where mortality was measured at different time points. In addition, the RCTs were biased towards a

Epidural in cardiac surgery

| 31

Table 2 References of the manuscripts reporting the nine epidural haematomas that reached publication so far together with the yr of publication and the country where they occurred First author

Yr

Journal

Manuscripts published in peer-reviewed journals Sharma S12 2004 J Cardiothorac Vasc Anesth 2004 Anesth Analg Rosen DA13 Yoshinaga A14 2004 Masui Yoshida S15 2005 Jpn J Thorac Cardiovasc Surg Bang J16 2011 Korean J Anesthesiol

Epidural haematomas (n=5)

Neurological outcome

India US Japan Japan South Korea

1 1 1 1 1

Good recovery at five weeks Complete recovery at six weeks Paraplegia One leg paralysis after one yr Almost complete recovery at six months

1

Paraplegia

3

Paraplegia

Grey literature ( personal communications, non peer-reviewed journals) UK Not available 2004 UK Medical Protection Society: Epidural Emergency, in Education and Publications UK Casebook17 2006 Anesth Analg US Chaney MA9

coronary artery bypass population and lower risk cohorts than those represented in a typical cardiac surgery centre now. It is therefore possible that the effect size is larger than predicted by our analysis. Secondly, we incorporated case-matched studies, which include higher risk patients and on-pump cardiac surgery, other than isolated coronary artery bypass grafting. We assessed the quality of each study to reduce the risk of inclusion bias inherent in non-RCTs, and, when sensitivity analyses were performed, including only studies with low risk of bias and studies with low and intermediate risk of bias, no reduction in mortality was detected. We speculate that the reason is the significant reduction in sample size. Indeed, the most robust research method for both mortality and haematoma assessment would remain a very large scale RCT, but that is not practical and very unlikely to occur, and our study represents the best estimate of events available at the current time. Thirdly, when considering the risk of haematoma formation, it is possible that there are more unreported patients. However, this should also be balanced by more epidural insertions performed. In the absence of a larger prospective registry or very large scale RCT, we believe that our worldwide survey provides the best available approach to identifying the incidence of haematoma (numerator) and the associated number of epidurals performed (denominator). Prior risk estimates were skewed towards a lower risk estimate (if only RCT are included the incidence would be zero), or to a higher risk estimate if case reports were added as they presented the numerator without the associated denominator. Although our approach has limitations, we have captured an incidence of epidurals performed of more than five times the published literature, and likely to represent a substantial proportion of the worldwide experience using this technique. Fourthly, it could be argued that epidural has no importance in cardiac surgery because pain is not an issue in this setting. We therefore would like to underline that the findings of our manuscript support the hypothesis that epidural analgesia is beneficial on clinically relevant endpoints (e.g. mortality reduction) and that epidural in this setting is justified if performed to improve clinically relevant outcomes. Fifthly, the limitations of our survey also should be acknowledged: as it was a viral survey (with information spreading among colleagues) we did not have a ‘response rate’; we asked the participants to estimate the numbers from their countries

rather than to provide recorded data from their centres. Lastly, we only investigated haematoma as the serious complication, and not reports of epidural abscess, osteomyelitis, or direct neural lesions from needle trauma. However, there is no reason to think that peculiarities of cardiac surgery could influence the incidence of catheter-related adverse events other than haematoma in comparison with major non-cardiac interventions. In conclusion, the use of epidural analgesia in cardiac surgery is associated with a reduction in mortality (NNT=70), against an estimated risk of epidural haematoma of 1:3552.

Authors’ contributions All authors developed the concept of this study and the search strategy. All authors contributed to the search of pertinent studies and data extraction. All authors contributed to the statistical analysis, to the first drafting of the manuscript, and to subsequent revisions. All of the authors read and approved the final manuscript.

Supplementary material Supplementary material is available at British Journal of Anaesthesia online.

Declaration of interests None declared.

References 1. Popping DM, Zahn PK, Van Aken HK, Dasch B, Boche R, Pogatzki-Zahn EM. Effectiveness and safety of postoperative pain management: a survey of 18 925 consecutive patients between 1998 and 2006 (2nd revision): a database analysis of prospectively raised data. Br J Anaesth 2008; 101: 832–40 2. Rigg JR, Jamrozik K, Myles PS, et al. Epidural anaesthesia and analgesia and outcome of major surgery: a randomised trial. Lancet 2002; 359: 1276–82 3. Wijeysundera DN, Beattie WS, Austin PC, Hux JE, Laupacis A. Epidural anaesthesia and survival after intermediate-to-high

Downloaded from http://bja.oxfordjournals.org/ at North Dakota State University on July 1, 2015

Country

32 |

4.

5.

6.

7. 8.

10.

11.

12.

13.

14.

15.

16.

17.

18.

risk non-cardiac surgery: a population-based cohort study. Lancet 2008; 372: 562–9 Popping DM, Elia N, Marret E, Remy C, Tramer MR. Protective effects of epidural analgesia on pulmonary complications after abdominal and thoracic surgery: a meta-analysis. Arch Surg 2008; 143: 990–9; discussion 1000 Ho AM, Chung DC, Joynt GM. Neuraxial blockade and hematoma in cardiac surgery: estimating the risk of a rare adverse event that has not (yet) occurred. Chest 2000; 117: 551–5 Bracco D, Hemmerling T. Epidural analgesia in cardiac surgery: an updated risk assessment. Heart Surg Forum 2007; 10: E334–7 Jackson R. Case of spinal apoplexy. Lancet 1869; 2: 5–6 Royse C, Royse A, Soeding P, Blake D, Pang J. Prospective randomized trial of high thoracic epidural analgesia for coronary artery bypass surgery. Ann Thorac Surg 2003; 75: 93–100 Chaney MA. Intrathecal and epidural anesthesia and analgesia for cardiac surgery. Anesth Analg 2006; 102: 45–64 Bignami E, Landoni G, Biondi-Zoccai GG, et al. Epidural analgesia improves outcome in cardiac surgery: a meta-analysis of randomized controlled trials. J Cardiothorac Vasc Anesth 2010; 24: 586–97 Svircevic V, van Dijk D, Nierich AP, et al. Meta-analysis of thoracic epidural anesthesia versus general anesthesia for cardiac surgery. Anesthesiology 2011; 114: 271–82 Sharma S, Kapoor MC, Sharma VK, Dubey AK. Epidural hematoma complicating high thoracic epidural catheter placement intended for cardiac surgery. J Cardiothorac Vasc Anesth 2004; 18: 759–62 Rosen DA, Hawkinberry DW 2nd, Rosen KR, Gustafson RA, Hogg JP, Broadman LM. An epidural hematoma in an adolescent patient after cardiac surgery. Anesth Analg 2004; 98: 966–9, table of contents Yoshinaga A, Sonoda H. Epidural hematoma caused after epidural catheterization for the coronary artery bypass graft: a case report. Masui 2004; 53: 551–4 Yoshida S, Nitta Y, Oda K. Anterior spinal artery syndrome after minimally invasive direct coronary artery bypass grafting under general combined epidural anesthesia. Jpn J Thorac Cardiovasc Surg 2005; 53: 230–3 Bang J, Kim JU, Lee YM, et al. Spinal epidural hematoma related to an epidural catheter in a cardiac surgery patientA case report. Korean Journal of Anesthesiology 2011; 61: 524–7 UK Medical Protection Society: Epidural Emergency, in Education and Publications UK. Casebook, vol. 12. Leeds: UK Medical Protection Society, 2004; 119–120 Richter A, Cederholm I, Fredrikson M, Mucchiano C, Träff S, Janerot-Sjoberg B. Effect of long-term thoracic epidural analgesia on refractory angina pectoris: a 10-year experience. J Cardiothorac Vasc Anesth 2012; 26: 822–8

19. Chaney MA, Labovsky JK. Thoracic epidural anesthesia and cardiac surgery: balancing postoperative risks associated with hematoma formation and thromboembolic phenomenon. J Cardiothorac Vasc Anesth 2005; 19: 768–71 20. Kirno K, Friberg P, Grzegorczyk A, Milocco I, Ricksten S, Lundin S. Thoracic epidural anesthesia during coronary artery bypass surgery: effects on cardiac sympathetic activity, myocardial blood flow and metabolism, and central hemodynamics. Anesth Analg 1994; 79: 1075–81 21. Stenseth R, Bjella L, Berg EM, Christensen O, Levang OW, Gisvold SE. Thoracic epidural analgesia in aortocoronary bypass surgery. II: Effects on the endocrine metabolic response. Acta Anaesthesiol Scand 1994; 38: 834–9 22. Loick H, Schmidt C, Van AH, et al. High thoracic epidural anesthesia, but not clonidine, attenuates the perioperative stress response via sympatholysis and reduces the release of troponin T in patients undergoing coronary artery bypass grafting. Anesth Analg 1999; 88: 701–9 23. Blomberg S, Emanuelsson H, Kvist H, et al. Effects of thoracic epidural anesthesia on coronary arteries and arterioles in patients with coronary artery disease. Anaesthesiology 1990; 73: 840–7 24. Lagunilla J, Garcia-Bengochea JB, Fernandez AL, et al. High thoracic epidural blockade increases myocardial oxygen availability in coronary surgery patients. Acta Anaesthesiol Scand 2006; 50: 780–6 25. Ahlers O, Nachtigall I, Lenze J, et al. Intraoperative thoracic epidural anaesthesia attenuates stress-induced immunosuppression in patients undergoing major abdominal surgery. Br J Anaesth 2008; 101: 781–7 26. Gonca S, Kilickan L, Dalcik C, Dalcik H, Bayindir O. The cardioprotective effects of thoracal epidural anesthesia are induced by the expression of vascular endothelial growth factor and inducible nitric oxide synthase in cardiopulmonary bypass surgery. J Cardiovasc Surg (Torino) 2007; 48: 93–102 27. Caputo M, Alwair H, Rogers CA, et al. Thoracic epidural anesthesia improves early outcomes in patients undergoing off-pump coronary artery bypass surgery: a prospective, randomized, controlled trial. Anesthesiology 2011; 114: 380–90 28. Royse C, Remedios C, Royse A. High thoracic epidural analgesia reduces the risk of long-term depression in patients undergoing coronary artery bypass surgery. Ann Thorac Cardiovasc Surg 2007; 13: 32–5 29. Royse C. Epidurals for cardiac surgery: can we substantially reduce surgical morbidity or should we focus on quality of recovery? Anesthesiology 2011; 114: 232–3 30. Scott NB, Turfrey DJ, Ray DA, et al. A prospective randomized study of the potential benefits of thoracic epidural anesthesia and analgesia in patients undergoing coronary artery bypass grafting. Anesth Analg 2001; 93: 528–35 Handling editor: J. G. Hardman

Downloaded from http://bja.oxfordjournals.org/ at North Dakota State University on July 1, 2015

9.

Landoni et al.