British Journal of Anaesthesia 103 (5): 62636 (2009)

doi:10.1093/bja/aep272

REVIEW ARTICLES Efcacy and safety of different techniques of paravertebral block for analgesia after thoracotomy: a systematic review and metaregression
1*, A. Scally2 and S. Howell3 A. Kotze
1

Airedale NHS Trust, Steeton, Keighley BD20 6TD, UK. 2School of Health Studies, Bradford University, UK. 3 Academic Unit of Anaesthesia, University of Leeds, UK
*Corresponding author. E-mail: alwynkotze@doctors.net.uk
Various techniques and drug regimes for thoracic paravertebral block (PVB) have been evaluated for post-thoracotomy analgesia, but there is no consensus on which technique or drug regime is best. We have systematically reviewed the efcacy and safety of different techniques for PVB. Our primary aim was to determine whether local anaesthetic (LA) dose inuences the quality of analgesia from PVB. Secondary aims were to determine whether choice of LA agent, continuous infusion, adjuvants, pre-emptive PVB, or addition of patient-controlled opioids improve analgesia. Indirect comparisons between treatment arms of different trials were made using metaregression. Twenty-ve trials suitable for metaregression were identied, with a total of 763 patients. The use of higher doses of bupivacaine (890 990 mg per 24 h compared with 325 472.5 mg per 24 h) was found to predict lower pain scores at all time points up to 48 h after operation (P0.006 at 8 h, P0.001 at 24 h, and P,0.001 at 48 h). The effect-size estimates amount to around a 50% decrease in postoperative pain scores. Higher dose bupivacaine PVB was also predictive of faster recovery of pulmonary function by 72 h (effect-size estimate 20.1% more improvement in FEV1, 95% CI 2.08% 38.07%, P0.029). Continuous infusions of LA predicted lower pain scores compared with intermittent boluses (P0.04 at 8 h, P0.003 at 24 h, and P,0.001 at 48 h). The use of adjuvant clonidine or fentanyl, pre-emptive PVB, and the addition of patient-controlled opioids to PVB did not improve analgesia. Further well-designed trials of different PVB dosage and drug regimes are needed. Br J Anaesth 2009; 103: 62636 Keywords: anaesthetic techniques, regional, thoracic; analgesia, postoperative; pain, acute, regional techniques; surgery, thoracic; thoracic anaesthesia

Thoracotomy frequently causes severe postoperative pain and signicant morbidity.10 27 Atelectasis, pneumonia, pulmonary embolism, and emergency intensive care admission have all been found to be related to poor analgesia and consequent immobility.27 66 Postoperative pain is thought to be the single most important factor leading to ineffective ventilation and impaired secretion clearance after thoracotomy.71 Severe or inadequately treated acute pain after thoracotomy also predicts conversion to chronic post-thoracotomy pain27 66 and long-term post-surgical fatigue.73 Thoracic paravertebral block (PVB) has been shown to provide superior post-thoracotomy analgesia and lung function, compared with systemic opioids or intrapleural local anaesthetics (LA).30 73 Three systematic reviews have compared the efcacy of PVB and thoracic epidural

analgesia (TEA) after thoracotomy.27 30 45 Detterbeck30 found that PVB provided equivalent pain relief to TEA, but did not quantitatively compare complications between the two techniques. A meta-analysis by Davies and colleagues27 showed that PVB provides pain relief as good as TEA, using LA with or without opioid, but with fewer side-effects, technical problems, and failed blocks. Perhaps more importantly, PVB reduced postoperative pulmonary complications by 64% compared with epidural analgesia.27 Recently,45 it was found that PVB provided analgesia after thoracotomy that was comparable with TEA using LA only, but possibly less effective than TEA using LA with opioid. However, PVB reduced the incidence of pulmonary complications compared with systemic analgesia, whereas TEA did not.45 PVB may

# The Author [2009]. Published by Oxford University Press on behalf of the British Journal of Anaesthesia. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournal.org

Efcacy and safety of different techniques of PVB

theoretically be a safer technique than TEA, at least in terms of the chances of serious spinal cord injury from epidural space infection41 or spinal canal haematoma.67 The systematic reviews to date27 30 45 studied PVB as a single generic technique, regardless of drug choice, dose, or administration technique. The optimal drug(s) for PVB have to date not been reviewed. We therefore undertook a systematic review with the aim of determining how the following variables inuence the quality of postthoracotomy analgesia with PVB: (i) LA dose; (ii) administration technique, that is, continuous infusion or intermittent boluses; (iii) choice of LA agent; (iv) the addition of fentanyl or clonidine to LA; (v) pre-emptive PVB; (vi) the addition of patient-controlled opioids to PVB.

trials included for each analysis. Any VAS score obtained between 6 and 10 h after operation was included with those obtained at 8 h, and VAS scores obtained on the rst postoperative day or from 20 to 24 h after operation were grouped at 24 h. Pain scores were assumed to have been obtained at rest unless otherwise stated by the authors. Spirometric measurements were recorded as a percentage of the preoperative values. Complications and sideeffects of the PVB itself were recorded where these were identied by the trial authors. A complication rate of zero was only recorded where it was specically stated that a complication did not occur. The total LA dose in the rst 24 h was calculated as overall indicator of dosage, using the formula: total dose loading dose24(infusion dose h21). Dosages were calculated for a patient weighing 70 kg where weight-related dosages were reported. For dosages reported as a simple range, the midpoint was used.

Methods
This review followed the Quality of Reporting of Meta-analyses (QUORUM) guidelines.57 Data were extracted and analysed in keeping with the methods used in similar meta-analyses27 as far as was appropriate. The MEDLINE and EMBASE databases were searched without language restriction up to May 2008, using the Athens portal of the United Kingdom National Library for Health. The predetermined inclusion criteria were: (i) randomized controlled trials (RCTs) in which at least one trial group received paravertebral LA with or without additives, and (ii) postoperative pain control, pulmonary function, or both reported as outcome measure. Only trials involving adult patients were considered. The keywords paravertebral block, nerve block, paravertebral, extrapleural, subpleural, retropleural, intercostal nerve block, thoracotomy, pneumonectomy, oesophagectomy, and the corresponding Medical Subject Headings (MeSH) were used alone and combined with Boolean operators. The electronic search identied 66 papers for consideration. The abstracts of the identied articles were reviewed to determine if the study met the inclusion criteria of the review. In addition, the reference lists of trials included in previous reviews of analgesia for thoracotomy27 30 45 73 were checked. Data concerning the PVB group of each randomized trial were extracted into Microsoft Excelw and analysed using STATAw (StataCorp LP, College Station, TX, USA). Average pain scores were recorded as if on a 100 mm visual analogue scale (VAS). Where scores were reported on 0 10 long ordinal scales, these were converted to a 100 mm VAS. VAS scores obtained at times close to each other were grouped together to maximize the number of

Statistical analysis
Continuous data are presented as mean values and the standard error (SE) of the mean. Where the standard deviusing the ation (SD) was reported, the SE was calculated p formula for a normal distribution (SESD/ n). When no SD was given, it was imputed with the t-test if the P-value was stated; otherwise the SD was estimated as half of the mean value. When the median and range only were reported for continuous outcomes, the mean and SD were estimated by assuming that the mean was equivalent to the median and that the SD was one-quarter of the range.44 64 For continuous data, indirect comparisons between groups of trials were made by means of metaregression, using univariate analysis, and a random effects model. This was necessary because of great heterogeneity between the original randomized comparisons, described below. Two-tailed P-values were calculated for dichotomous data, using Fishers exact test.

Results
Of the 31 RCTs identied,1 4 7 10 13 15 26 28 29 31 33 36 38 46 six were excluded from meta-analysis. One31 was conducted in patients undergoing minimally invasive coronary artery surgery, rather than thoracotomy. One small trial54 had an unexplained drop-out rate of 30% (three of 10) in the PVB group within 24 h and none in the control group. Another9 compared repeated paravertebral boluses of bupivacaine against systemic analgesia. Both the stated dose (0.1 mg kg21) and bolus size (1 2 ml) of bupivacaine was 10 20-fold lower than every other trial, or the recommended dose. It is therefore probable that the dose quoted in the paper was simply a misprint. No correction was found in the following years issues of the journal, and the author did not reply to communications to conrm the dose used.
52 54 62 63 65 68 70 72 75 80 83 84

627

 et al. Kotze

Papers identified n = 68

Not randomized trials: n = 10

Randomized controlled trials n = 58

No PVB group: n = 27

RCTs with at least one PVB group n = 31

Not published in English: n = 2 No pain measurements: n = 1

RCTs included in review n = 28

Not an RCT in thoracotomy: n = 1 Unexplained patient drop-outs: n = 1 PVB dosage unclear: n = 1

RCTs included in metaanalysis n = 25

Fig 1 QUORUM owchart showing included and excluded studies.

Table 1 Details of included studies. B, bupivacaine Trial n Compared against

The trial was therefore excluded from meta-analysis. Two small trials (with a total of 20 patients in the PVB groups between them) were published in French.1 80 One trial38 investigated specically the haemodynamics of PVB and reported no data on pain or pulmonary function (Fig. 1). The 25 trials included had recruited 763 patients to 31 PVB treatment arms (Table 1). Obtaining original patient data for all the included trials proved impossible. Seven trials conducted comparisons of different paravertebral drug regimes.4 8 15 36 62 71 83 However, these trials evaluated a total of nine different drug and dosage regimes. Four trials were placebo-controlled, with all patients having access to rescue medication and regular simple analgesics.5 29 33 75 Two different PVB regimes were evaluated against control groups who, in turn, received two different analgesic regimens. One of these29 also had a control group receiving only systemic analgesia without a sham PVB. Three trials evaluated the addition

Loading dose

Maintenance

24 h total

PCA

Groups receiving bupivacaine infusions Perttunen and colleagues65 15 Epidural (local only)single-injection IC nerve blocks Epidural (local only) Richardson and colleagues70 46 Bhatnagar and colleagues8 Barron and colleagues4 Sabanathan and colleagues75 Dauphin and colleagues Catala and colleagues15 Richardson and colleagues72 Carabine and colleagues13 Eng and Sabanathan33 Berrisford and Sabanathan6 Luketich and colleagues52 Kaiser and colleagues46 Watson and colleagues83 Deneuville and colleagues29 Wedad and colleagues84 Richardson and colleagues71 Richardson and colleagues69 Richardson and colleagues74
26

8 12 ml 0.25% B 20 ml 0.5% B (pre-incision)20 ml 0.25% B (end of surgery) B 2 mg kg21 (0.125%) B 0.3 ml kg21 0.25%

4 8 ml h21 0.25% B 0.1 ml kg21 h21 0.5% B 0.5 mg kg21 h21 (0.125%) 0.1 ml kg21 h21 0.25% 0.1 ml kg21 h21 0.5% B 0.1 ml kg21 h21 0.5% B 5 ml h21 0.25% B with epinephrine 1:200 000 0.1 ml kg21 h21 0.5% B 5 ml h21 0.25% bupivacaine 0.5 mg kg21 h21 bupivacaine 7 ml h21 0.5% bupivacaine 0.1 ml kg21 h21 0.25% B 0.1 ml kg21 h21 0.5% B 0.1 ml kg21 h21 0.5% B 3 ml h21 0.5% B 6 ml h21 0.25% B 0.1 ml kg21 h21 0.5% B 0.1 ml kg21 h21 of 0.5% B 0.1 ml kg21 h21 of 0.5% B

385 990

Yes Yes

14

22

29 24 15

Two paravertebral groups: plain bupivacaine and bupivacaineclonidine. Plain groups data used Two paravertebral groups: bupivacaine, lidocaine, against placebo. Bupivacaine group used Placebo (both had access to i.m. opioid) Lumbar epidural morphine infusion Two paravertebral regimes: infusion and paravertebral bupivacaine and norepinephrine boluses 6-hourly Interpleural bupivacaine PCA morphine Placebo (both had access to i.m. opioid) Placebo (both had access to i.m. opioid) Epidural (B 0.125% and PF morphine 0.05 mg ml21) Epidural (B 0.25 0.375%fentanyl 2 mg ml21) Two paravertebral groups: bupivacaine 0.5% and lidocaine 1% in identical volumes Placebo and regular i.m. buprenorphine Epidural (local only), interpleural block Pre-emptive PVB vs PVB after surgery

980

No

472.5

No

B 100 mg B with epinephrine 1:200 000 B 0.375% 15 ml with epinephrine 1:200 000 B 150 mg B 25 mg B 100 mg B 100 mg 10 ml 0.5% B 20 ml 0.5%B 10 ml 0.5% B 20 ml 0.5% B 10 ml 0.25% B 20 ml 0.5% B 20 ml 0.5% B

940 950 6.25

No Yes No

22 10 40 25 47 15 23 26 20 56 10 6 465

990 325 940 940 470 940 890 460 385 990 990 990

Yes Yes No No Yes No Yes No No No No No

Interpleural bupivacaine

20 ml 0.5% B

628

Efcacy and safety of different techniques of PVB

of PVB to various systemic analgesics without including a placebo PVB group.9 13 53 Two trials72 74 compared PVB with interpleural analgesia. Nine trials compared PVB with TEA, again using different paravertebral drug and dosage regimes.10 14 16 26 28 46 52 65 70 84 The outcomes reported consistently enough to make meta-analysis reliable were: VAS at 8, 24, and 48 h, maximal expired volume in 1 s (FEV1) at 24, 48, and 72 h after surgery, and LA toxicity (dened as confusion which resolves after cessation of the LA infusion, the occurrence of convulsions, or cardiac arrhythmias).

LA dosage
No trial making a direct comparison between different doses of a given LA was identied. An indirect comparison was possible. The trials that used bupivacaine infusions for PVB were clearly divisible into two groups. In 12 studies patients received bupivacaine 890 990 mg in the rst 24 h after operation (the higher dose LA group).4 6 8 15 25 33 46 70 72 75 83 In seven, patients received 325 472.5 mg per 24 h
Bupivacaine dose vs pain scores at rest: 8 h 100

(the lower dose group).4 13 15 29 52 65 84 Only four of the 19 trials used less than the manufacturers recommended maximum dose of bupivacaine 400 mg82 per 24 h.13 15 65 84 No trial used bupivacaine in the dose range of 473889 mg. Bupivacaine boluses were not considered for dosage meta-analysis, as the time interval between bolus administration and assessment of pain scores could also inuence VAS. Scatter plots of LA dosage vs VAS scores revealed a possible inverse relationship at all time points up to 48 h after surgery (Fig. 2). Metaregression conrmed that this apparent relationship is signicant. Higher dose paravertebral bupivacaine was strongly predictive of lower VAS scores at rest, when compared with lower dose regimes at 8 h after operation (P0.006), 24 h (P0.001), and 48 h (P,0.001). Although there was a trend to improved analgesia on coughing at all time points, the difference did not reach statistical signicance. Far fewer trials reported VAS scores on coughing than at rest: there were only two trials in each dosage group for dynamic analgesia at 8 h,8 15 65 70 and a further one at 24 and 48 h.46
Bupivacaine dose vs pain scores on coughing: 8 h 100
Mean VAS score on coughing

VAS score at rest

80 60 40 20 0 0 1000 200 400 600 800 Bupivacaine dose (mg per 24 h) Bupivacaine dose vs pain scores at rest: 24 h 60 1200

80 60 40 20 0 20 0 200 400 600 800 1000 1200 Bupivacaine dose (mg per 24 h) Bupivacaine dose vs pain scores on coughing: 24 h 100

Mean VAS score on coughing

50
VAS at rest

80 60 40 20 0 0 200 400 600 800 1000 Bupivacaine dose (mg per 24 h) 1200

40 30 20 10 0 10 0 200 400 600 800 1000 1200 Bupivacaine dose (mg per 24 h) Bupivacaine dose vs pain scores at rest: 48 h

Bupivacaine dose vs pain scores on coughing: 48 h

70 60 50 40 30 20 10 0

Mean VAS scores

40 30 20 10 0 0 200 400 600 800 1000 Bupivacaine dose (mg per 24 h) 1200

Mean VAS score on coughing

50

200 400 600 800 1000 Bupivacaine dose (mg per 24 h)

1200

Fig 2 Scatter plots of visual analogue scores vs bupivacaine dosage. Bubble graphs depicting resting and dynamic visual analogue scores at 8, 24, and 48 h after thoracotomy. The diameter of each bubble is proportional to the number of patients in the PVB group(s) of each trial.

629

 et al. Kotze

Table 2 Bupivacaine dosage and pain at rest. Results of a meta-regression showing the effect of higher dose bupivacaine (890 990 mg per 24 h) over lower dose bupivacaine (325 472.5 mg per 24 h). VAS values are presented as mean (SE). 95% CI, 95% condence interval for effect-size estimate; P, probability value for effect size existing by chance VAS lower dose bupivacaine (mm) 8 h after operation 24 h after operation 48 h after operation 40 (8) 34 (5) 27 (3) Effect-size estimate for higher dose bupivacaine (mm) 227 221 217 95% CI (mm) P-value No. of studies

27 to 246 28 to 234 29 to 226

0.006 0.001 ,0.001

10 12 12

Table 3 Bupivacaine dosage and pain on coughing. Results of a meta-regression showing the effect of higher dose bupivacaine (890 990 mg per 24 h) over lower dose bupivacaine (325 472.5 mg per 24 h). VAS values are presented as mean (SE). 95% CI, 95% condence interval for effect-size estimate; P, probability value for effect size existing by chance Pooled VAS low-dose bupivacaine (mm) 8 h after operation 24 h after operation 48 h after operation 29 (2) 32 (2) 26 (2) Effect size estimate for higher dose bupivacaine (mm) 2 3 4 95% CI (mm) P-value No. of studies

215 to 20 211 to 18 210 to 18

0.804 0.625 0.580

4 5 5

Table 4 Bupivacaine dosage and postoperative pulmonary function. Results of a meta-regression showing the effect of higher dose bupivacaine (890 990 mg per 24 h) over lower dose bupivacaine (325 472.5 mg per 24 h). Spirometry values are presented as a percentage of preoperative values, mean (SE). 95% CI, 95% condence interval for effect-size estimate; P, probability value for effect size existing by chance Pooled FEV1 lower dose bupivacaine (%) 24 h after operation 48 h after operation 72 h after operation 54 (5) 60 (7) 46 (7) Effect-size estimate for higher dose bupivacaine (%) 10 11 20 95% CI (%) P-value No. of studies

22 to 23 26 to 28 2 to 38

0.11 0.20 0.02

19 18 6

The effect-size estimate for VAS improvement with higher dose bupivacaine was 26.9 mm (95% CI 7.5 46.3 mm) at 8 h after operation, 21.1 mm (95% CI 8.5 33.6 mm) at 24 h, and 17.4 mm (95% CI 8 26 mm) at 48 h. This represents about a 50% decrease in postoperative pain at rest (Tables 2 and 3). There were no statistically signicant differences in FEV1 at 24 and 48 h between higher and lower dose bupivacaine trials. Pulmonary function recovered faster in the higher dose bupivacaine group; by 72 h after surgery, the difference reached signicance (20.1% better improvement in FEV1, 95% CI 2.08238.07%, P0.029) (Table 4). Three trial arms evaluated PVB with ropivacaine.14 28 53 Only one53 evaluated postoperative pain at all three time points (8, 24, and 48 h), and no trial evaluated postoperative spirometry. No meta-analysis for the effect of ropivacaine dose was therefore performed. However, scatter plots of perceived pain against dosage appear similar to those for bupivacaine (Fig. 3). No meta-analysis for the effect of lidocaine dose was attempted, as the two trial arms of PVB with lidocaine used virtually identical dosage regimes: 1890 mg and 1780 mg per 24 h, respectively.4 83

Ropivacaine dose vs pain scores at rest: 24 h 40 35 30 25 20 15 10 5 0 0 200 400 600 800 Ropivacaine dose (mg per 24 h) 1000 Mean VAS at rest

Fig 3 Scatter plots of visual analogue scores vs ropivacaine dosage. Bubble graphs depicting resting and dynamic visual analogue scores at 24 h after thoracotomy. The diameter of each bubble is proportional to the number of patients in the PVB group(s) of each trial.

Continuous infusion vs intermittent bolus technique

One direct comparison between bolus and infusion technique for maintenance of PVB was found.15 This showed a slight improvement in VAS at 24 h after operation when an infusion regime was used. The difference was 15 mm at rest and 23 mm upon coughing (P0.003 in both cases). The LA dosages used were identical but low. This study did not report on postoperative lung function.

630

Efcacy and safety of different techniques of PVB

An indirect comparison of studies using boluses15 36 and continuous infusions8 13 15 25 29 33 46 47 52 65 70 72 75 83 84 for maintenance revealed that the use of a continuous infusion for maintenance of PVB is associated with an improvement in analgesia at rest at all time points up to 48 h. The effect size was 29.8 mm at 8 h (95% CI 0.9858.7 mm, P0.04), 26.7 mm at 24 h (95% CI 9.2 44.3 mm, P0.003), and 23.3 mm at 48 h (95% CI 13.7 32.9 mm, P,0.001). As was the case for an increase in LA dosage, there was a trend towards improved dynamic analgesia in the few trials that reported on this, but it did not reach statistical signicance (Table 5). Neither of the studies using PVB bolus regimes15 36 evaluated postoperative pulmonary mechanics. There is therefore no evidence available from indirect or direct comparisons about the effect of administration technique on pulmonary mechanics or the incidence of complications.

The use of additives to LA

Clonidine One trial directly evaluated the addition of clonidine (2 mg kg21 h21) to an infusion of bupivacaine for PVB. The authors8 found a modest (9 mm on a 100 mm VAS) improvement in the overall marginal mean postoperative VAS score after thoracotomy. There was no difference in lung function between the groups. When an indirect comparison was made between the results achieved in the clonidine group of the above trial and the trial arms using plain bupivacaine,8 13 15 25 29 33 46 47 52 65 70 72 75 83 84 no difference was found in pain scores (P0.7, Table 6). Fentanyl No direct comparisons between LA-only PVB and PVB with adjuvant opioid were identied. Eighty-nine patients in three trial arms10 62 received LAs with fentanyl during the course of trials comparing PVB with bupivacaine and ropivacaine, and PVB with epidural analgesia, respectively. There was no difference found in analgesia between the trials that added fentanyl to the LA for PVB, and those that did not8 13 15 25 29 33 36 46 47 52 65 70 72 75 83 84 (P0.648, Table 6).

Choice of LA
We identied two direct comparisons between lidocaine and bupivacaine for PVB.4 83 Neither found a difference between bupivacaine and lidocaine in terms of VAS at rest, morphine requirements, or postoperative pulmonary function. Meta-analysis also found no statistically signicant difference (P0.06, Table 6). No studies directly compared bupivacaine and ropivacaine. Regression analysis revealed no difference in analgesic quality between trials that administered bupivacaine8 13 15 25 29 33 46 47 52 65 70 72 75 83 84 or ropivacaine14 28 53 (P0.705, Table 6).

Timing of PVB
Only one study directly compared pre-emptive PVB with an identical regimen established immediately after

Table 5 Effect of continuous infusion techniques. Results of a meta-regression showing the effect of continuous infusion techniques for maintenance, compared with intermittent boluses. VAS values are presented as mean (SE). 95% CI, 95% condence interval for effect-size estimate; P, probability value for effect size existing by chance VAS at rest with bolus regimes (mm) 8 h after operation 24 h after operation 48 h after operation 54 (14) 49 (8) 40 (4) VAS on coughing with bolus regimes (mm) 8 h after operation 48 h after operation 35 (5) 26 (5) Effect size estimate for infusion regimes (mm) 224 223 216 Effect-size estimate for infusion regimes (mm) 214 10 95% CI (mm) P-value No. of studies

21 to 258 216 to 230 213 to 232 95% CI (mm)

0.04 0.003 ,0.001 P-value

16 22 21 No. of studies

26.65 to 35.18 218 to 40

0.18 0.47

7 10

Table 6 Postoperative pain and PVB maintenance regimes. Results of regression analyses showing the effects of different PVB regimes at rest 24 h after operation. VAS values are presented as mean (SE). 95% CI, 95% condence interval for effect-size estimate; P, probability value for effect size existing by chance Pooled VAS at rest, 24 h after operation (mm) Lidocaine vs bupivacaine Ropivacaine vs bupivacaine Clonidine added to local anaesthetic Fentanyl added to local anaesthetic Pre-emptive PVB Morphine via PCA added to PVB 20 (4) 20 (4) 28 (4) 28 (4) 30 (5) 30 (5) Effect-size estimate (mm) 95% CI (mm) P-value No. of studies

222 23 8 5 11 13

246 to 1 221 to 14 234 to 51 219 to 31 29 to 32 27 to 35

0.063 0.705 0.705 0.648 0.294 0.211

14 15 21 21 22 22

631

 et al. Kotze

surgery.71 In an elegant two3 factorial trial, Richardson and colleagues compared the presence or absence of opioid premedication, presence/absence of non-steroidal anti-inammatory premedication, and presence/absence of pre-emptive PVB. Those patients who received all three components of their pre-emptive analgesia regimen had signicantly better analgesia, faster recovery of pulmonary mechanics, and less stress hormone release. Although statistically signicant, the magnitude of the difference in perceived pain with pre-emptive PVB was small: mean VAS scores were improved by 5.1 mm on a 100 mm scale at 24 h after surgery. The observed difference in lung function was of clinical and statistical signicance. The results of the regression analysis do not support the conclusion of the above-mentioned trial. All studies reported on timing of PVB. PVB was established before skin incision in eight.14 52 62 65 70 73 No signicant difference was found between pre-emptive PVB and block established after surgery at rest or on coughing, at 8 or 24 h after operation, although there was a trend towards improved analgesia in those groups who received PVB before skin incision. The respective P-values were 0.187 and 0.294 for analgesia at rest, and P0.07 and 0.270 for analgesia on coughing (Table 6). At 48 h after surgery, pre-emptive PVB seems statistically to be associated with an improvement in analgesia at rest of 13.6 mm on a 100 mm scale (P0.005, 95% CI 4 23 mm), but associated with signicantly higher pain scores on coughing (21.1 mm, 95% CI 13 28 mm, P,0.001).

60 Mean VAS score 50 40 30 20 10 0 0

PCA morphine consumption and pain scores at rest: 24 h

20 40 60 80 Amount of PCA morphine in first 24 h (mg)

100

Fig 4 PCA morphine consumption and pain scores. Bubble graphs depicting resting and dynamic visual analogue scores at 24 h after thoracotomy. The diameter of each bubble is proportional to the number of patients in the PVB group(s) of each trial.

Table 7 Bupivacaine dose and complication incidence Lower dose trials Neurology Cardiac arrhythmias 2/110 0/69 Higher dose trials 4/225 2/173 P-value 1.0 1.0

Patient-controlled analgesia in addition to PVB

This review found no direct comparisons on whether opioid administered via patient-controlled analgesia (PCA) is superior to intermittent administration for postthoracotomy analgesia, when added to PVB. Nine trial regimes included the use of a morphine PCA.26 52 53 65 69 72 74 83 Regression analysis revealed that the use of a morphine PCA did not predict an improvement in postoperative pain (P0.21, Table 6) over the administration of intermittent opioids. There was no detectable relationship between postoperative pain scores and morphine consumption via PCA (P0.93). Higher consumption of PCA morphine did not predict better analgesia, nor did higher pain scores predict consumption of more PCA morphine (Fig. 4).

reported specically whether this complication occurred or not. Neurological effects which may have been due to LA toxicity occurred in four of 225 patients in the higher dose bupivacaine trials, compared with two of 110 patients in the lower dose trials (P1.0). Cardiac arrhythmias occurred in two of 173 patients who received higher dose bupivacaine, and none of the 69 patients who received lower dose bupivacaine (P1.0) (Table 7). No lasting patient harm was reported due to possible LA toxicity.

Discussion
This indirect comparison of data from 25 RCTs found that higher dose LA regimens for PVB and the use of continuous infusions for maintenance are predictive of lower VAS scores up to 48 h after thoracotomy; decreasing postoperative pain by around 50% in each case. Choice of one LA over another, the addition of clonidine or fentanyl, or the use of patient-controlled morphine over intermittent parenteral morphine were not found to be predictive of less pain. Our meta-analysis is different from previous ones on this topic, which were concerned with establishing the superiority of one technique over another. Because of the great heterogeneity between previous direct comparisons, we were unable to conduct a meta-analysis which preserved the randomized nature of the original trials. We therefore include indirect comparisons between the treatment arms of different trials. These are by nature not randomized, and vulnerable to confounding factors and other sources of bias. However, indirect comparisons are common in anaesthesia, for example, the Oxford Pain Relief Units Bandolier league table,85 and meta-analyses of

Complications of PVB
The occurrence of complications of the PVB or of surgery was not as well reported as pain scores and pulmonary function. Possible LA toxicity (manifested by confusion that resolved after LA administration was stopped, convulsions or cardiac dysrhythmias) was the only complication reported in the majority of studies. Only 15 of the 19 studies using bupivacaine8 13 15 25 29 33 46 47 52 65 70 72 75 83 84

632

Efcacy and safety of different techniques of PVB

postoperative nausea and vomiting42 81 and postoperative shivering.43 49 Indirect comparisons are also published in internal medicine.35 37 A systematic review of all medical meta-analyses published over a 5 yr period39 found that 9.5% (31 of 327) contained an indirect comparison. We used metaregression to analyse any continuous data in this review to assess the association between different aspects of PVB technique (exposure) and outcome. Metaregression may partially compensate for the nonrandomized nature of the comparisons where there is heterogeneity between different trials, but the differences within trials and within the groups of trials are small.39 This review deals with just such a situation. Robust indirect comparisons are graded as level II evidence by some authors, alongside direct randomized trials with surrogate outcomes, and ahead of non-randomized direct comparisons.55 56 Post-thoracotomy pain arises as a consequence of pleural and muscular damage, costovertebral joint disruption,70 and intercostal nerve damage.63 Peripheral and central sensitization as a consequence of acute tissue damage and progenitor of severe acute and chronic pain has also been demonstrated in a variety of animal models and experiments in human subjects.23 24 33 There is consensus that a multimodal analgesic regimen, incorporating a form of neural blockade, simple analgesics, and strong opioids, is optimal after thoracotomy.21 27 32 33 48 50 59 67 71 77 78 From our analyses, the strongest predictor of improved analgesia (and hence a lower failure rate) after thoracotomy is a higher dose of LA. Postoperative pain appears to be decreased by around 50% with higher dose regimens, which is both clinically and statistically signicant. Perhaps even more importantly, the improved analgesia from higher dose regimes also translated into better recovery in pulmonary function. Pulmonary complications and mortality data were not reported widely enough for us to examine formally whether higher doses of LA also improve survival. There is good evidence that the addition of PVB improves analgesia, when compared with systemic opioid only.30 45 73 Our meta-analysis found that giving patients free access to opioid via PCA in addition to PVB appears not to improve analgesia greatly. This conrms that some form of neural block may be valuable for postthoracotomy analgesia. There was no clear relationship between opioid requirement and the mean pain scores achieved in the reviewed trials. Higher opioid consumption did not predict better analgesia, and patients did not appear to use more PCA morphine in those studies with higher pain scores. This suggests that patients limited their use of PCA morphine because of other factors than achieving good analgesia; a nding which is supported by literature specic to PCA.11 18 19 34 76 79 Pre-emptive PVB51 58 and the use of an additive to LA were the subject of only one direct comparison each.8 71 In each case, the authors found a modest improvement in perceived pain. Their conclusions are not supported by our ndings. Adjuvant clonidine increased the risk of sedation.8

We were unable to nd any procedure-specic evidence to recommend any one technique of establishing PVB. Of note is that most researchers in this eld have taken steps to ensure that LA spreads longitudinally along the vertebral column and thus covers a number of spinal segments. A single-injection technique has been shown to produce a block that is safe but unpredictable in spread,16 20 22 40 and multiple injections have been shown to produce superior results in patients undergoing breast, thoracic, and upper abdominal surgery.60 61 The improvement in analgesia and pulmonary mechanics with the use of higher doses of paravertebral LA shown above must be balanced against the risk of possible LA toxicity. The absorption of bupivacaine from the paravertebral space has been shown to be rapid, with accumulation to toxic levels shown after prolonged infusion at 0.5 mg kg21 h21, 6 12 17 25 but not when dosage is reduced to 0.25 mg kg21 h21 after 24 h.46 The incidence of possible LA toxicity was not signicantly different between lower and higher dose trials, even though one65 used doses less than the manufacturers recommendation. No patient in any of the trials in this review suffered convulsions. One of the two cases of cardiac arrhythmia (atrial brillation) in the higher dose trials26 was reported by the authors as being related to intrinsic cardiac disease. It was successfully treated without stopping the LA infusion. The other was not commented on.70 A previous systematic review found an incidence of 0.8% for LA toxicity after PVB with bupivacaine. All cases presented as transient confusion only.30 The incidence of serious LA toxicity with higher dose regimes is likely to be low regardless of agent, especially where high doses are administered for 24 h or less. The incidence of such rare events may best be established through a collaborative audit project rather than a research study. Ropivacaine does not appear to accumulate in the same linear manner as bupivacaine, and is seen by some authors as a safer choice for PVB.16 28 53 The trials that evaluated the use of ropivacaine for PVB14 28 53 did not explicitly report on LA toxicity, and this review can therefore present no safety data to compare bupivacaine with other LAs. Ropivacaine appears equipotent with bupivacaine, despite being given in doses far closer to the recommended maximum.2 This review does have weaknesses. As already stated, our comparisons were non-randomized and therefore essentially observational in nature, although the statistical technique of meta-regression partially compensates for this.39 55 The data on postoperative complications should be interpreted cautiously, as the quality of complication reporting in the original studies was variable. There was no consistent reporting of a preoperative risk assessment, and therefore no reliable way of comparing data on mortality or pulmonary morbidity across trials. The exclusion of two small trials not published in English is unlikely materially to affect our results.

633

 et al. Kotze

Table 8 Conclusions Evidence from head-to-head comparisons, supported by regression analysis Continuous infusion techniques are superior to intermittent bolus techniques for maintenance of PVB Bupivacaine and lidocaine are equally effective for PVB The addition of clonidine to LA for PVB increases the risk of undue sedation and hypotensive episodes Higher dose LA produces better analgesia and pulmonary function than lower dose LA Bupivacaine and ropivacaine are equally effective for PVB The addition of fentanyl to a PVB regime does not improve analgesia Pre-emptive PVB produces better results than PVB established at the end of surgery Multiple-injection technique is superior to single-injection technique for establishing PVB Clonidine added to local anaesthetic for PVB improves analgesia

Evidence from indirect comparisons only or evidence from single head-to-head comparisons, not supported by regression analysis

10

11

12

Contradictory evidence from head to head and indirect comparisons

13

14

In conclusion, we present evidence on optimizing PVBs after thoracotomy, from both direct and indirect comparisons (Table 8). Prospective and specic research on the effect of LA dosage on post-thoracotomy pain is needed before denitive conclusions may be drawn on what the optimal dosage may be. Safety data are also urgently required. The question of whether PVB or TEA is best after thoracotomy is yet unanswered. Ultimately, a direct comparison between optimized TEA and PVB regimes is necessary. The denitive answer on how best to manage pain, rehabilitation, and risk in this difcult group of patients is still not clear.

15

16

17

18 19 20

References
1 Alaya M, Auffray J-P , Alouini T, et al. Comparison of extrapleural and interpleural analgesia using bupivacaine after thoracotomy. Ann Fr Anesth 1995; 14: 249 55 2 AstraZeneca UK Limited. Naropin Core Data Sheet. Available from http://www.medicines.org.uk/searchresult.aspx?search=ropivacaine (accessed October 2008) 3 Ballantyne JC, Carr DB, Chalmers TC, Dear KBG, Angelillo IF, Mosteller F. Postoperative patient-controlled analgesia: meta-analysis of initial randomised controlled trials. J Clin Anesth 1993; 5: 182 93 4 Barron DJ, Tolan MJ, Lea RA. A randomized controlled trial of continuous extrapleural analgesia post-thoracotomy: efcacy and choice of local anaesthetic. Eur J Anaesthesiol 1999; 16: 236 45 5 Berrisford RG, Sabanathan S, Mearns AJ, et al. Plasma concentrations of bupivacaine and its enantiomers during continuous extrapleural intercostal nerve block. Br J Anaesth 1993; 70: 201 4 6 Berrisford RG, Sabanathan SS. Direct access to the paravertebral space at thoracotomy. Ann Thorac Surg 1990; 49: 854 7 Berrisford RG, Sabanathan SS, Mearns AJ, Bickford-Smith PJ. Pulmonary complications after lung resection: the effect of

21 22

23

24

25

26

continuous extrapleural intercostal nerve block. Eur J Cardiothorac Surg 1990; 4: 407 11 Bhatnagar S, Mishra S, Madhurima S, Gurjar M, Mondal AS. Clonidine as an analgesic adjuvant to continuous paravertebral bupivacaine for post-thoracotomy pain. Anaesth Intensive Care 2006; 34: 586 91 Bilgin M, Akcali Y, Oguzkaya F. Extrapleural regional versus systemic analgesia for relieving postthoracotomy pain: a clinical study of bupivacaine compared with metamizol. J Thorac Cardiovasc Surg 2003; 126: 1580 3 Bimston DN, McGee JP, Liptay MJ, Fry WA. Continuous paravertebral extrapleural infusion for post-thoracotomy pain management. Surgery 1999; 126: 650 7 Buckley DN. Patient assessment of efcacy of pain management: the fallacy of management by opinion poll. Can J Anaesth 2000; 47: 1164 5 Burlacu CL, Frizelle HP, Moriarty DC, Buggy DJ. Pharmacokinetics of levobupivacaine, fentanyl, and clonidine after administration into the paravertebral space. Reg Anesth Pain Med 2007; 32: 136 41 Carabine UA, Gilliland H, Johnston JR, McGuigan J. Pain relief after thoracotomy: comparison of morphine requirements using an extrapleural infusion of bupivacaine. Reg Anesth 1995; 20: 4127 Casati A, Alessandrini P, Nuzzi M, et al. A prospective, randomized, blinded comparison between continuous thoracic paravertebral and epidural infusion of 0.2% ropivacaine after lung resection surgery. Eur J Anaesthesiol 2006; 23: 999 1004 Catala E, Casa JI, Unzueta MC, et al. Continuous infusion is superior to bolus doses with thoracic paravertebral blocks after thoracotomies. J Cardiothorac Vasc Anesth 1996; 10: 586 8 Cheema S, Richardson J, McGurgan P. Factors affecting the spread of bupivacaine in the adult thoracic paravertebral space. Anaesthesia 2003; 58: 684 711 Cheung SLW, Booker PD, Franks R, Pozzi M. Serum concentration of bupivacaine during prolonged continuous paravertebral infusion in young infants. Br J Anaesth 1997; 79: 9 13 Chumbley GM, Hall GM, Salmon P. Patient-controlled analgesia: an assessment by 200 patients. Anaesthesia 1998; 53: 216 21 Chumbley GM, Hall GM, Salmon P. Why do patients feel positive about patient-controlled analgesia? Anaesthesia 1999; 54: 372 92 Conacher ID, Kokri M. Postoperative paravertebral blocks for thoracic surgery. A radiological appraisal. Br J Anaesth 1987; 59: 155 61 Cook TM, Riley RH. Analgesia following thoracotomy: a survey of Australian practice. Anaesth Intensive Care 1997; 25: 520 4 Cotter JT, Nielsen KC, Guller U, et al. Increased body mass index and ASA physical status IV are risk factors for block failure in ambulatory surgeryan analysis of 9,342 blocks. Can J Anaesth 2004; 51: 810 6 Dahl JB, Rosenberg J, Kehlet H. Effect of thoracic epidural etidocaine 1.5% on somatosensory evoked potentials, cortisol and glucose during cholecystectomy. Acta Anaesthesiol Scand 1992; 36: 378 82 Dahl JB, Rosenberg J, Lund C, Kehlet H. Effect of thoracic epidural bupivacaine 0.75% on somatosensory evoked potentials after dermatomal stimulation. Reg Anesth 1990; 15: 73 5 Dauphin A, Gupta RN, Young AE, Morton WD. Serum bupivicaine concentrations during continuous extrapleural infusion. Can J Anaesth 1997; 44: 367 70 Dauphin A, Lubanska-Hubert E, Young JEM, et al. Comparative study of continuous extrapleural intercostal nerve block and epidural morphine in post-thoracotomy pain. Can J Surg 1997; 40: 431 6

634

Efcacy and safety of different techniques of PVB

27 Davies RG, Myles PS, Graham JM. A comparison of the analgesic efcacy and side-effects of paravertebral vs. epidural blockade for thoracotomya systematic review and meta-analysis of randomized trials. Br J Anaesth 2006; 96: 418 26 28 De Cosmo G, Aceto P, Campanale A, et al. Comparison between epidural and paravertebral intercostals nerve block with ropivacaine after thoracotomy: effects on pain relief, pulmonary function and patient satisfaction. Acta Med Rom 2002; 40: 340 7 29 Deneuville M, Bisserier A, Regnard JF, et al. Continuous intercostals analgesia with 0.5% ropivacaine after thoracotomy: a randomized study. Ann Thorac Surg 1993; 55: 381 5 30 Detterbeck FC. Efcacy of methods of intercostal nerve blockade for pain relief after thoracotomy. Ann Thorac Surg 2005; 80: 1550 9 31 Dhole S, Mehta Y, Saxena H, et al. Comparison of continuous thoracic epidural and paravertebral blocks for postoperative analgesia after minimally invasive direct coronary artery bypass surgery. J Cardiothorac Vasc Anesth 2001; 15: 288 92 32 Eason MJ, Wyatt R. Paravertebral blocka reappraisal. Anaesthesia 1979; 34: 638 64 33 Eng J, Sabanathan S. Continuous extrapleural intercostals nerve block and post-thoracotomy pulmonary complications. Scand J Thorac Cardiovasc Surg 1992; 26: 219 23 34 Everett B, Salamonson Y. Differences in postoperative opioid consumption in patients prescribed patient-controlled analgesia versus intramuscular injection. Pain Manag Nurs 2005; 6: 137 44 35 Felson DT, Anderson JJ, Meenan RF. The comparative efcacy and toxicity of second-line drugs in rheumatoid arthritis. Results of two meta-analyses. Arthritis Rheum 1990; 33: 1449 61 36 Fibla JJ, Molins L, Mier JM, Sierra A, Vidal G. Comparative analysis of analgesic quality in the postoperative of thoracotomy: paravertebral block with bupivacaine 0.5% vs ropivacaine 0.2%. Eur J Cardiothorac Surg 2008; 33: 430 4 37 Garg R, Yusuf S, for the Collaborative Group on ACE Inhibitor Trials. Overview of randomized trials of angiotensin-converting enzyme inhibitors on mortality and morbidity in patients with heart failure. J Am Med Assoc 1995; 273: 1450 6 38 Garutti I, Olmedilla L, Cruz P, et al. Comparison of the hemodynamic effects of a single 5 mg/kg dose of lidocaine with or without epinephrine for thoracic paravertebral block. Reg Anesth Pain Med 2008; 33: 57 63 39 Glenny AM, Altman DG, Song F, et al. Indirect comparisons of competing interventions. Health Technol Assess 2005; 9. Available from http://www.hta.ac.uk/1118 (accessed June 2009) 40 Govenden V, Matthews P. Percutaneous placement of paravertebral catheters during thoracotomy. Anaesthesia 1988; 43: 256 41 Grewal S, Hocking G, Wildsmith JAW. Epidural abscesses. Br J Anaesth 2006; 96: 292 302 r MR. Efcacy, dose response, and 42 Henzi I, Sonderegger J, Trame adverse effects of droperidol for prevention of postoperative nausea and vomiting. Can J Anaesth 2000; 47: 537 51 ` r MR. Metoclopramide in the preven43 Henzi I, Walder B, Trame tion of postoperative nausea and vomiting: a quantitative systematic review of randomized placebo-controlled studies. Br J Anaesth 1999; 83: 761 71 44 Higgins JPT, Green S, eds. Cochrane Handbook for Systematic Reviews of Interventions, Version 5.0.0 (updated February 2008). The Cochrane Collaboration, 2008. Available from www .cochrane-handbook.org 45 Joshi GP , Bonnet F, Shah R, et al. A systematic review of randomized trials evaluating regional techniques for post-thoracotomy analgesia. Anesth Analg 2008; 107: 1026 40

46 Kaiser AM, Zollinger A, De Lorenzi D, et al. Prospective, randomized comparison of extrapleural versus epidural analgesia for postthoracotomy pain. Ann Thorac Surg 1998; 66: 367 72 47 Karmakar MM, Critchley L. Continuous extrapleural intercostal nerve block for post thoracotomy analgesia in children. Anaesth Intensive Care 1998; 26: 115 6 48 Kavanagh BP, Katz J, Sandler AN. Pain control after thoracic surgery. A review of current techniques. Anesthesiology 1994; 81: 737 59 r MR. Single-dose par49 Kranke P, Eberhard LH, Roewer N, Trame enteral pharmacological interventions for the prevention of postoperative shivering: a quantitative systematic review of randomized controlled trials. Anesth Analg 2004; 99: 718 27 50 Kruger M, Sandler AN. Post-thoracotomy pain control. Curr Opin Anaesthesiol 1998; 12: 55 8 nnqvist P-A. Pre-emptive analgesia with thoracic paravertebral 51 Lo blockade? Br J Anaesth 2005; 95: 727 8 52 Luketich JD, Land SR, Sullivan EA, et al. Thoracic epidural versus intercostal nerve catheter plus patient-controlled analgesia: a randomized study. Ann Thorac Surg 2005; 79: 1845 50 53 Marrett E, Bazelly B, Taylor G, et al. Paravertebral block with ropivacaine 0.5% versus systemic analgesia for pain relief after thoracotomy. Ann Thorac Surg 2005; 79: 2109 14 54 Matthews PJ, Govenden V. Comparison of continuous paravertebral and extradural infusions of bupivacaine for pain relief after thoracotomy. Br J Anaesth 1989; 62: 204 5 55 McAlister FA, Laupacis A, Wells GA, Sackett DL. Users guides to the medical literature: XIX. Applying clinical trial results B. Guidelines for determining whether a drug is exerting (more than) a class effect. J Am Med Assoc 1999; 282: 1371 7 56 Moayyedi P. Meta-analysis: can we mix apples and oranges? Am J Gastroenterol 2004; 99: 2297 301 57 Moher D, Cook JC, Eastwood S, Olkin I, Rennie D, Stroup DF. Improving the quality of reports of meta-analyses of randomized controlled trials: the QUORUM statement. Lancet 1999; 354: 1896 900 58 Moiniche S, Kehlet H, Dahl JB. A qualitative and quantitative systematic review of pre-emptive analgesia for postoperative pain relief: the role of timing of analgesia. Anesthesiology 2002; 96: 725 41 59 Myles PS, Bain C. Underutilization of paravertebral block in thoracic surgery. J Cardiothorac Vasc Anesth 2006; 20: 635 8 60 Naja M, Lonnqvist P-A. Somatic paravertebral nerve blockade. Incidence of failed block and complications. Anaesthesia 2001; 56: 1181 201 61 Naja ZM, El-Rajab M, Al-Tannir MA, et al. Thoracic paravertebral block: inuence of the number of injections. Reg Anesth Pain Med 2006; 31: 196 201 62 Navlet MG, Garutti I, Olmedilla L. Paravertebral ropivacaine, 0.3%, and bupivacaine, 0.25%, provide similar pain relief after thoracotomy. J Cardiothorac Vasc Anesth 2006; 20: 644 7 63 Ng A, Swanevelder J. Pain relief after thoracotomy: is epidural analgesia the optimal technique? Br J Anaesth 2007; 98: 159 62 64 ORourke K. Mixed means and medians: a unied approach to deal with disparate outcome summaries. Symposium on Systematic Reviews: Pushing the Boundaries. Oxford, 2002; 49 65 Perttunen K, Nilsson E, Heinonen J, et al. Extradural, paravertebral and intercostal nerve blocks for post-thoracotomy pain. Br J Anaesth 1995; 75: 541 7 66 Pluijms WA, Steegers MAH, Verhagen AFTM, Scheffer GJ, Wilder-Smith OHG. Chronic post-thoracotomy pain: a retrospective study. Acta Anaesthesiol Scand 2006; 50: 804 8

635

 et al. Kotze

67 Richardson J. Paravertebral anesthesia and analgesia. Can J Anaesth 2004; 51: R1 6 68 Richardson J, Jones J, Atkinson R. The effect of thoracic paravertebral blockade on intercostals somatosensory evoked potentials. Anesth Analg 1998; 87: 373 6 69 Richardson J, Sabanathan S, Eng J, et al. Continuous intercostal nerve block versus epidural morphine for postthoracotomy analgesia. Ann Thorac Surg 1993; 55: 377 80 70 Richardson J, Sabanathan S, Jones J, et al. A prospective, randomized comparison of preoperative and continuous epidural or paravertebral bupivacaine on post-thoracotomy pain, pulmonary function and stress responses. Br J Anaesth 1999; 83: 387 92 71 Richardson J, Sabanathan S, Mearns AJ, et al. Efcacy of preemptive analgesia and continuous extrapleural block on postthoracotomy pain and pulmonary mechanics. J Cardiovasc Surg 1994; 35: 219 28 72 Richardson J, Sabanathan S, Mearns AJ, et al. A prospective, randomized comparison of interpleural and paravertebral analgesia in thoracic surgery. Br J Anaesth 1995; 75: 405 8 73 Richardson J, Sabanathan S, Shah R. Post-thoracotomy spirometric lung function: the effect of analgesia. J Cardiovasc Surg 1999; 40: 445 6 74 Richardson J, Sabanathan S, Shah RD, et al. Pleural bupivacaine placement for optimal postthoracotomy pulmonary function: a prospective, randomized study. J Cardiothorac Vasc Anesth 1998; 112: 1669 75 Sabanathan S, Mearns AJ, Bickford-Smith PJ, et al. Efcacy of continuous extrapleural intercostal nerve block on post-thoracotorny pain and pulmonary mechanics. Br J Surg 1990; 77: 2215

76 Salmon P, Hall GM. PCA: patient-controlled analgesia or politically correct analgesia? Br J Anaesth 2001; 87: 815 8 77 Slinger P. New trends in anaesthesia for thoracic surgery including thoracoscopy. Can J Anaesth 1995; 42: R77 84 78 Soto RG, Fu ES. Acute pain management for patients undergoing thoracotomy. Ann Thorac Surg 2003; 75: 1349 57 79 Taylor NM, Hall GM, Salmon P. Patient experiences of PCA. Anaesthesia 1996; 51: 525 8 80 Theissen O, Boileau S, Cornet C, et al. Analgesia after thoracotomy using continuous extrapleural bupivacaine. Ann Fr Anesth 1993; 12: 265 72 81 Tramer MR, Moore RA, McQuay HJ. Prevention of vomiting after paediatric strabismus surgery: a systematic review using the numbers-needed-to treat. Br J Anaesth 1995; 75: 556 61 82 United States Federal Drug Administration. Bupivacaine core data. Available from http://www.drugs.com/pro/bupivacaine.html (accessed July 2008) 83 Watson DS, Panian S, Kendall V, et al. Pain control after thoracotomy: bupivacaine versus lidocaine in continuous extrapleural intercostal nerve blockade. Ann Thorac Surg 1999; 67: 825 9 84 Wedad M, Zaki MK, Haleem M. The effect of addition of wound inltration with local anaesthetics to interpleural block on postthoracotomy pain, pulmonary function and stress response in comparison to thoracic epidural and paravertebral block. Egypt J Anaesth 2004; 20: 67 72 85 www.medicine.ox.ac.uk/bandolier/booth/painpag/acutrev/ analgesics/leagtab.html (accessed January 2009)

636