Paravertebral block in paediatric abdominal surgery a systematic review and meta-analysis of randomized trials

Transcription

1 British Journal of Anaesthesia, 118 (2): (2017) doi: /bja/aew387 Review Article REVIEW ARTICLE Paravertebral block in paediatric abdominal surgery a systematic review and meta-analysis of randomized trials E. A. Page and K. L. Taylor* Department of Anesthesia and Pain Medicine, Hospital for Sick Children, 555 University Avenue, Toronto, Ontario M5G 1X8, Canada *Corresponding author: katherine.taylor@sickkids.ca Abstract The increased popularity of paravertebral block (PVB) can be attributed to its relative safety and comparable efficacy when compared with epidural analgesia. It has thus been recommended for open cholecystectomy and other less painful surgeries such as inguinal herniorraphy and appendectomy. We performed a systematic review of PVB in paediatric abdominal conditions to assess its clinical efficacy and side effects compared with other analgesic therapies. A search of Medline, Embase, and Web of Science and hand-searching references from inception date to May 2016 was done. Relevant studies were randomized clinical trials in patients 0 18 years old comparing PVB (single shot or continuous catheter) with any comparator and analgesic medication. Pain scores, rescue analgesia and adverse events were compared. The systematic reviews identified six trials enrolling 358 paediatric patients. PVB medications included bupivacaine, ropivacaine, lidocaine, and fentanyl. Surgical procedures included inguinal herniorraphy, cholecystectomy, and appendectomy. The standardized mean difference in early pain scores favoured PVB: 0.85 [95% confidence interval (CI) ] at 4 6 h and 0.64 (95% CI ) at 24 h. One study reported a reduced length of stay. Parental [odds ratio (OR) 5.12 (95% CI )] and surgeon [OR 6.05 (95% CI )] satisfaction were higher in those receiving a PVB. No major complications occurred with a PVB. PVB resulted in minimally improved pain scores for up to 24 h after surgery, reduced rescue analgesia requirements, and increased surgeon and parental satisfaction. PVB is a good alternative to caudal and ilioinguinal block in paediatric abdominal surgery. Key words: anaesthesia, regional; general surgery; paediatrics; review, systematic Paravertebral block (PVB) is becoming more popular in paediatrics as an alternative to epidural analgesia. 1 The paravertebral space is a triangular-shaped space bordered by the transverse process and superior costotransverse ligament posteriorly, the parietal pleura anteriorly, and the lateral edge of the vertebral body and intervertebral discs medially. The spaces communicate, allowing the local anaesthetic to spread cranially and caudally (limited by the psoas muscle at L1) to cover multiple dermatomes and produce a multilevel block, even from a single injection. 2 The local anaesthetic can also spread laterally to the intercostal space and medially to the epidural space, although the latter is rare. 3 VC The Author Published by Oxford University Press on behalf of the British Journal of Anaesthesia. All rights reserved. For Permissions, please journals.permissions@oup.com 159

2 160 Page and Taylor Editor s key points Paravertebral block (PVB) has an accepted role in adult thoracic surgery. The benefits of PVB include lower risk of sympathetic block and neuraxial injury. This study identifies some beneficial effects of PVB in paediatric practice. Larger randomized trials are needed to confirm or refute these findings. The block is simple to perform, especially with ultrasound guidance, which reduces the time to complete the block. 4 PVB has relatively few complications and comparable efficacy to epidural analgesia in adult thoracic surgery. 1 Advantages of PVB include unilateral blockade and limiting the haemodynamic changes (including hypotension) and stress response resulting from a bilateral sympathetic block. PVB provides a more complete somatosensory block than an epidural without the risk of epidural-related unintentional spinal cord damage. 5 In theory, this allows higher doses of local anaesthetic to be used with less risk. 6 In adults, randomized controlled trials (RCTs) have shown that PVB results in a prolonged period of targeted analgesia and a reduced risk of postoperative nausea, vomiting, and complications. 1 A recent review of bilateral PVB in both adults and children having midline surgery demonstrated that the risk of systemic toxicity is low despite large doses of local anaesthetic. 3 PVB has extremely low failure and complication rates in children. 7 The majority of evidence for PVB points to its efficacy compared with epidural analgesia for thoracic surgery and open cholecystectomy. Increasingly, PVB has been recommended for other less painful surgeries, such as inguinal hernia repair and appendectomy, where an epidural would not normally be considered. 8 For these lower abdominal surgeries, PVB would be considered more invasive than the common alternatives such as caudal, ilioinguinal block or local anaesthesia infiltration. The European Society for Paediatric Anaesthesiology (ESPA) has provided instructions for ultrasound-guided thoracic paravertebral blocks on their website. This includes abdominal surgeries such as inguinal hernia repair and appendectomy. The referenced supporting evidence is based on four systematic reviews and one RCT, but all trials were done in adults. 9 We asked the PICO (patient problem or population, intervention, comparison, and outcomes) question: in paediatric abdominal surgery conditions (P), does paravertebral block (I) compared with all other techniques (C) result in improved clinical outcomes (O) Methods Inclusion criteria were RCTs with paediatric participants 0 18 years of age, with no year or language restrictions. Any level of PVB for a surgical operation in which an incision was made in the abdomen, including laparoscopic, inguinal, and renal surgery, was included. All drug combinations for PVB and all comparators (general anaesthesia, other nerve blocks or placebo) were included. The primary outcomes were pain scores and requirement for rescue analgesia. Secondary outcomes were length of stay (LOS), satisfaction, and clinically relevant adverse effects, such as postoperative nausea and/or vomiting and other complications. A librarian conducted a literature search of the following electronic databases: Web of Science (1945 present), Embase (1947 present), and Medline (1946 present). Search terms included paravertebral near block* or catheter*. The full search strategies for each database are supplied in appendices 1 3. The date of the search was May 20, Two reviewers independently applied inclusion criteria from a review of the titles, abstracts, and keywords. Disagreements were resolved by discussion until consensus was reached. References were then hand-searched by one reviewer. Data were extracted using a data collection form by one reviewer. The following items were recorded: operation, sample size, intervention groups, details of the interventions, sizes of intervention groups, age, outcome definition, outcome units, and outcomes data. Where a pain score was reported at a time point for activity, movement, and rest, the worst score for the time point was used. Authors of original papers were contacted with up to two attempts. The risk of bias of each study was assessed independently by two reviewers according to the guidelines in the Cochrane Reviewer s Handbook. 10 Studies were assessed for random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessors, incomplete outcome data, selective reporting, and any other potential source of bias. Any discrepancies were resolved by discussion. A loss to follow-up rate of 10% was considered the threshold for attrition bias. Meta-analysis was performed using Review Manager (RevMan for Windows, version 5.3; Cochrane Collaboration, Oxford, UK). The summary measure was the standardized mean difference for pain score and the mean difference for time to supplementary analgesia use. The summary measure was the odds ratio (OR) for postoperative vomiting, site tenderness, parental satisfaction, and need for rescue analgesia, and the mean difference for LOS. Subgroup analysis by operation type (inguinal hernia repair) was performed. Variations in study design and the possible risk of bias (methodological heterogeneity) are presented in Table 1. Statistical heterogeneity was calculated using the I 2 statistic to account for the variability in each effect estimate that could be due to a true difference between studies. We considered <30% as low, 30 60% as moderate, and >60% as high. Statistical significance was defined as P<0.05. A random effects model with 95% confidence intervals (CIs) was used due to anticipated heterogeneity and small sample sizes. To synthesize raw data reported in 95% CIs, the method described in the Cochrane Reviewer s Handbook was used to convert these to standard deviations (SDs) via the inverse of the t distribution. 10 Results of the search There were 280 publications identified in the electronic database search and 115 duplicates were removed. Six studies were included in the final, quantitative analysis. The PRISMA 11 flow diagram is reported in Figure 1. Four study authors were contacted by and none provided additional data. No additional studies were found by hand-searching. All included studies were RCTs published in English. The total number of participants in the six studies was 358. All participants were <18 years old. Table 1 provides a full description of the characteristics of the included studies. All studies described single-shot techniques. Naja and colleagues published two studies of inguinal herniorraphy, one comparing

3 Paravertebral block in paediatric abdominal surgery 161 Table 1 Characteristics of included studies.n/n, intervention/total; chole, cholecystectomy; appy, appendectomy; lap, laparoscopic; GA, general anaesthesia; INB, ilioinguinal nerve block; CB, caudal block; LI, local injection at laparoscopic ports; US, ultrasound; NS, nerve stimulator; PVB, paravertebral block. a 20 ml of the mixture contained 6 ml lidocaine, 2%; 6 ml lidocaine, 2%, with 1/200,000 epinephrine; 6 ml bupivacaine, 0.5%; 1 ml fentanyl, 50 lg ml 1 ; and 1 ml clonidine, 75 lg ml 1. b 20 ml of the mixture contained 6.5 ml lidocaine, 2%; 6 ml lidocaine, 2%, with 1/200,000 epinephrine; 6 ml bupivacaine, 0.5%; 1 ml fentanyl, 50 lgml 1 ; and 0.5 ml clonidine, 150 lg ml 1 Study n/n Age, years Surgery Control Guide PVB single shot Method of randomization Lost to follow-up, % Eldeen / Open chole Naja et al / Hernia repair Naja et al / Hernia repair Splinter and Thomson Epidural: 0.25% bupivacaine (0.5 ml kg 1 ), fentanyl (1 lg kg 1 ) US, both 0.25% bupivacaine (0.5 ml kg 1 ), fentanyl (1 lgkg 1 ) GA NS Local anaesthetic mixture a (0.3 ml kg 1 ) INB: Local anaesthetic mixture b (0.3 ml kg 1 ) NS, PVB only Local anaesthetic mixture b (0.3 ml kg 1 ) 18/ Open appy Band-Aid No 0.2% ropivacaine (0.25 ml kg 1 ) with epinephrine 1:200,000 (maximum 5 ml) Tug et al / Hernia repair CB: 0.25% levobupiva-caine (1 ml kg 1 ) Visoiu et al. 41/ Lap chole LI: 0.5% ropivacaine (0.1 ml kg 1 ) (maximum 5 ml) No No 0.25% levobupiva caine (0.2 ml kg 1 ) 0.5% ropivacaine (0.1 ml kg 1 ) (maximum 5 ml) Computer randomization PVB: 0 Epidural: 0 Sealed envelopes PVB: 0 GA: 0 Sealed envelopes PVB: 0 INB: 2.5 Random number table Random number table Computer-generated random number table PVB: 0 Band- Aid: 0 PVB: 0 CB: 0 PVB: 2.38 LI: 0 nerve stimulator guided PVB with general anaesthesia and one comparing nerve stimulator guided PVB with ilioinguinal nerve block. One study compared PVB with caudal block in children also undergoing inguinal hernia repair. 14 One study compared ultrasound-guided PVB with ultrasound-guided epidural in patients with cyanotic congenital heart disease undergoing open cholecystectomy. 15 One study compared PVB with local anaesthetic injection at port sites in children undergoing laparoscopic cholecystectomy. 16 Finally, one study compared PVB (using loss of resistance technique) with a placebo of Band-Aids in paediatric patients undergoing open appendectomy. 17 Risk of bias in included studies Figure 2 summarizes the risk of bias assessment in each domain for the six included studies. Allocation (selection bias) All studies were at low risk of selection bias from random sequence generation, however, five of the six studies did not report adequate information for allocation concealment, therefore the risk of selection bias was unclear. Two of these five studies reported using sealed opaque envelopes but did not state if they were sequentially numbered. Blinding (performance bias and detection bias) Due to the nature of interventions, blinding the anaesthetist, surgeon, and participating child with a placebo block was done in only one of the six studies However, in all five remaining studies the anaesthetist and/or surgeon did not participate in the outcome measurements. Three studies did not report any personnel or participant blinding, therefore it was assumed that this was not done and all received high performance bias ratings. Blinding of outcome assessment was separated by primary outcomes because all pain assessments by hospital staff were blinded but parental assessment of pain and satisfaction was not blinded. Detection bias risk assessment was therefore separated into assessment of pain outcomes, assessment of rescue analgesia outcomes, and assessment of other outcomes (such as satisfaction and complications). One study 15 did not report the blinding status of any of the three outcome assessments and another 14 did not report the blinding status for parents and these were rated as unclear risk. All three outcome groups in two of the studies were rated as low risk for detection bias. Two studies had non-blinded parents assessing satisfaction and pain scores after the child was discharged, introducing bias at this measurement time period. These studies were therefore rated as high risk for detection bias for pain outcomes and other outcomes. Incomplete outcome data (attrition bias) One study excluded an intended patient after allocation for technical reasons and they were not included in the analysis. 16 No study exceeded the attrition threshold for patients lost to follow-up and therefore all six studies were at a low risk of attrition bias.

4 162 Page and Taylor 280 records identified through database searching MEDLINE (n=80) EMBASE (n=135) 115 duplicated records Web of science (n=65) 165 records after duplicates removed 158 records excluded: 43 non-abdominal studies 35 irrelevant studies 26 sudies in adults 20 case reports 14 reviews 9 retrospective studies 6 conference abstracts 165 records screened 5 commentaries or letters 7 full-text articles assessed for eligibility 1 full-text article excluded, done in adults 6 studies for final inclusion in meta-analysis Figure 1 PRISMA flow diagram. Selective reporting (reporting bias) Five of the six studies reported the same outcomes as were prespecified in the methods and one study 14 appeared to be missing outcome data according to the methods, which reported collecting heart rate, blood pressure, and SpO 2 at 5, 10, 15, 30, and 60 minutes and 2, 4, 6, 12, 18, and 24 h postoperatively, although these data were absent from the results. Other potential sources of bias Only one study 16 specified receiving financial support. One study 15 calculated power retrospectively, but it is unclear how this might affect the conclusions drawn. Effects of interventions Table 2 summarizes effect estimates for all of the outcomes included in the meta-analysis. Pain score Five authors reported data on pain scores. Two studies reported pain on a 0 10 visual analogue scale (VAS) One study used a VAS, 16 one used a FLACC (Face, Legs, Activity, Cry, and Consolability) score, 14 and one used a CHEOPS (Children s Hospital of Eastern Ontario Pain Scale) score. 15 The data from Eldeen 15 were not included in the meta-analysis because there was no 95% CI, SD, interquartile range, or other measure of mean distribution provided. Therefore, four studies were included in the pain meta-analysis, which due to the different scales used was reported as a standardized mean difference. Figure 3 represents forest plots for these data. There was a significant standardized mean difference favouring PVB [ 0.85 (95% CI )] for early pain scores (4 6 h), although inconsistency was also high (I 2 ¼88%). The same four studies reported pain scores at 24 h. There was no significant standardized mean difference [-0.38 (95% CI )], which may have been attributed to block duration. The inconsistency was high (I 2 ¼84%). Of note, the

5 Paravertebral block in paediatric abdominal surgery 163 Eldeen 2016 Naja 2005 Splinter 2010 Visoiu 2015 Random sequence generation (selection bias) Allocation concealment (selection bias) Blinding of participants and personnel (performance bias) Blinding of outcome assessment (detection bias): pain outcomes Blinding of outcome assessment (detection bias): rescue analgesia outcomes Blinding of outcome assessment (detection bias): other outcomes Incomplete outcome data (attrition bias) Selective reporting (reporting bias) Other bias Figure 2 The review authors judgements about each risk of bias item for each included study. Table 2 Effect estimates. SMD, standardized mean difference; IV, inverse variance; RE, random effects; MD, mean difference; OR, odds ratio; MH, Mantel Haenszel Outcome Studies Participants Statistical method Effect estimate (95% CI) Pain score at 4 6 h SMD (IV, RE) 0.85 ( 1.58 to 0.12) Pain score at 24 h SMD (IV, RE) 0.38 ( ) Pain score at 4 6 h (inguinal) SMD (IV, RE) 1.19 ( 1.49 to 0.88) Pain score at 24 h (inguinal) SMD (IV, RE) 0.64 ( 1.00 to 0.28) Time to first analgesic requirement, h 2 76 MD (IV, RE) 1.67 ( ) Number of patients needing supplemental analgesia (inguinal) OR (MH, RE) 0.17 ( ) Postoperative vomiting 2 76 OR (MH, RE) 0.32 ( ) Site tenderness OR (MH, RE) 5.15 ( ) Parental satisfaction OR (MH, RE) 5.12 ( ) Surgeon satisfaction OR (MH, RE) 6.05 ( ) mean pain scores from one study 16 were provided as expected scores from a linear mixed-effects model since the raw data were not available. treatment effect, with a mean difference of 1.67 h (95% CI ) and a high level of inconsistency (I 2 ¼95%). Figure 3 shows a forest plot for these data. Hernia repair Subgroup analysis of pain scores after hernia repair included three studies At 4 6 h postoperatively there was a significant standardized mean difference in favour of PVB [ 1.19 (95% CI )] and inconsistency was low (I 2 ¼0%). At 24 h there was also a significant standardized mean difference in favour of PVB [ 0.64 (95% CI )] and inconsistency was moderate (I 2 ¼35%). Appendix 4 shows forest plots for pain scores in this subgroup. Rescue analgesia Tug and colleagues 14 provided data on time to first analgesic requirement for the PVB cohort but not the control. Eldeen 15 and Splinter and Thomson 17 reported complete data on time to first analgesic requirement, which did not show a significant Hernia repair The three inguinal hernia repair studies provided data on the number of patients needing supplemental analgesia postoperatively. This showed a favourable treatment effect of PVB [OR 0.17 (95% CI )] and low inconsistency (I 2 ¼10%). Of note, in one study six patients 14 (two PVB and four caudal patients) had a failed block, defined as needing rescue analgesia after 2 h. The mean duration of analgesia was 1300 min for PVB (SD min). 14 A further 4 patients in the PVB group and 12 patients in the caudal group needed supplemental analgesia. LOS One study 12 showed that the rate of same-day discharge after inguinal hernia surgery was significantly increased in the PVB

6 164 Page and Taylor Pain score at 4-6 hours Std. mean difference Study or subgroup Mean SD Total Mean SD Total Weight IV. random, 95% CI Naja % 1.16 [ 1.76, 0.55] % 1.00 [ 1.47, 0.53] % 1.44 [ 1.97, 0.91] Visoiu % 0.13 [ 0.30, 0.56] Std. mean difference IV, random, 95%CI l00.0% 0.85[ 1.58, 0.12] Heterogeneity; Tau 2 =0.49; Chl 2 =25.34, df=3 (P<0.0001); I 2 = 88% Test for overall effect Z=2.30 (P=0.02) Favours paravertebral Favours control Pain Score at 24 hours Std. Mean Difference Std. mean difference Study or subgroup Mean SD Total Mean SD Total Weight IV. random, 95% CI IV, random, 95%CI Naja % 0.99 [ 1.58, 0.40] % 0.70 [ 1.15, 0.24] % 0.33 [ 0.80, 0.14] Visoiu % 0.44 [0.01, 0.88] [ 0.99, 0.24] Time to first analgesic requirement Mean Difference Study or subgroup Mean SD Total Mean SD Total Weight IV. random, 95% CI Eldeen % 1.12 [ 238, 0.14] Splinter % 4.60 [2.44, 6.76] Not estimable 141 Heterogeneity; Tau 2 =0.33; Chl 2 =19.29, df=3 (P<0.0002); I 2 = 84% Test for overall effect Z=1.21 (P=0.23) 100.0% 2 Favours paravertebral Favours control Mean difference IV, random, 95%CI % 1.67 [ 3.93, 7.27] Heterogeneity; Tau 2 =15.54; Chl 2 =20.07, df=1 (P< ); I 2 = 95% Test for overall effect Z=0.58 (P=0.56) 10 Number of patients needing supplemental analgesia after inguinal surgery Events Total Events Total Weight Study or subgroup Naja % 0.06 [0.01, 0.28] % 0.20 [0.07, 0.56] % 0.25 [0.08, 0.74] Favours control Favours paravertebral % Total events Heterogeneity: Tau 2 =0.04: chi 2 =2.22, df=2 (P=0.33); I 2 =10% Test for overall effect Z=4.91 (P< ) 0.17 [0.08, 0.34] Favours paravertebral Favours control Figure 3 Forest plots comparing PVB with any comparator on pain scores and need for rescue analgesia for all surgeries. group (80% compared with 52% in the general anaesthesia group, P<0.05). No patients with PVB stayed >24 h, compared with 16% in the general anaesthesia group (P<0.05). vascular puncture at a rate of 2.4% in the PVB group and 9.5% in the port site injection group. No study reported time to complete block. Postoperative vomiting and complications PVB did not decrease the incidence of postoperative nausea and vomiting (PONV) [OR 0.32 (95% CI ); I 2 ¼0%] Mild or minimal block site tenderness was significantly increased with PVB [OR 5.15 (95% CI ); I 2 ¼0%]. These data are presented in forest plots in Appendix 5. One study 14 reported the incidence of failed blocks, with a 5.7% failure rate in the PVB group and an 11.4% failure rate in the caudal block group. One study 16 reported incidences of Satisfaction Three studies, all in patients having hernia repairs, provided data on parental satisfaction. This was defined as the number of parents who rated their satisfaction as excellent, and showed a significant treatment effect in favour of PVB [OR 5.12 (95% CI )] and low inconsistency (I 2 ¼0%). Two studies reported surgeon satisfaction, with a pooled estimate in favour of PVB [OR 6.05 (95% CI ); I 2 ¼0%]. Figure 4 shows a forest plot representing these data.

7 Paravertebral block in paediatric abdominal surgery 165 Parental satisfaction Study or subgroup Events Total Events Total Weight Naja % 5.09 [1.45, 17.92] % 6.91 [1.80, 26.54] % 4.33 [1.57, 11.97] % 5.12 [2.59, 10.12] Total events Heterogeneity: Tau 2 =0.00; Chi 2 =0.29, df=2 (P=0.86); I 2 =0% Test for overall effect: Z=4.70 (P< ) Surgeon satisfaction Study or subgroup Events Total Events Total Weight Naja 2005 Total events Heterogeneity: Tau 2 =0.00; Chi 2 =0.04, df=1 (P=0.83); I 2 =0% Test for overall effect: Z=3.57 (P=0.0004) 47.1% 6.77 [1.61, 28.54] 52.9% 5.48 [1.41, 21.33] % 6.05 [2.25, 16.26] Favours control Favours paravertebral Favours control Favours paravertebral Figure 4 Forest plots comparing PVB with any comparator on satisfaction for inguinal surgeries. Discussion This meta-analysis of randomized trials demonstrates that PVB provides a beneficial effect on pain scores at 4 6 h but not at 24 h in children after abdominal surgery. The reduction in pain scores is minimal at both time points and results are based on a meta-analysis of only four trials. PVB did not reduce PONV and seemed to be associated with more site tenderness. In the subgroup analysis of hernia repair, pain scores were reduced by a standardized mean difference of 1.19 at 4 6 h postoperatively. The benefit was maintained at 24 h in this subgroup. There was a small benefit of PVB in reducing rescue analgesia but not for time to first analgesic requirement. Parents and surgeons were satisfied with PVB analgesia. One study reported reduced LOS with the use of PVB. 12 There are a number of limitations to our review based on the quality of the original data. All trials had small sample sizes. Studies used different surgical procedures, comparative variables, and measurement scores. We were able to provide subgroup analysis data for inguinal herniorraphy only. Studies varied in PVB solution and technique for insertion, and only one study 15 used ultrasound guidance. The effect sizes were small. Furthermore, the thresholds used (such as a pain VAS 1 10 score >4 or 5) for needing rescue analgesia were also variable, as were the follow-up periods studied. Given the significant heterogeneity in the methods of the trials, we were unable to attribute a rating to the overall quality of the evidence. In addition, we were unable to perform a meta-analysis on haemodynamic data, as these were reported in only two studies using different measures: absolute values and percentage change from baseline. Importantly, this review showed a higher reported incidence of site tenderness (described as mild or minimal ) with PVB (8.5%) than other studies. A study by Naja and Lönnqvist 6 previously reported a 1.3% rate of site tenderness in adults and a 0% rate in children 7. Ultrasound-guided PVB blocks have a high success rate and appear to be effective in reducing early pain after abdominal surgery with some site tenderness. There are short-term and long-term analgesia benefits in hernia surgery. The improvements in pain scores are minimal for PVB, but importantly the comparators in two of these studies were other regional anaesthetic techniques. If anaesthesia providers are already considering a regional technique for hernia surgery and have the available infrastructure (appropriate monitoring, equipment, training, and ultrasound availability), there are advantages to PVB. The parental and surgeon satisfaction scores are strikingly favourable for PVB compared with caudal and ilioinguinal block. The reasons for this are unclear and not available in the studies included in this review. Surgeons and parents were not blinded for these responses, so they may be biased. From our clinical experience, we hypothesize that parents may appreciate the longer duration of analgesia and the absence of weak legs in their child. The children in these studies were 3 12 years of age and therefore these adverse effects may have been more troublesome than in infants. Surgeon satisfaction is also difficult to explain, as in our experience, a PVB takes longer than the two alternatives and might delay surgery. Of course, parental satisfaction may influence surgeon satisfaction. There are few randomized trials of PVB in paediatrics and only one trial included controversial placebo blocks as a control. Our review of the evidence of PVB compared with standard care allows the clinician to make an informed decision about the risks and benefits of PVBs in paediatric abdominal surgeries and provides further explanation to the indications provided on the ESPA website. From this review, we conclude that PVB is an acceptable alternative for abdominal surgery in children, especially if factors in addition to absolute reduction in pain scores are significant for individual patients.

8 166 Page and Taylor Authors contributions K.L.T. and E.A.P made substantial contributions to the conception and design, literature search, and analysis and interpretation of data. Both were involved in drafting the manuscript. Both authors have read and approved the final article. Acknowledgements The authors thank Alanna Marson for her assistance in the electronic searches for this project. Funding The study was completed without industry sponsorship or financial assistance. Declaration of interests E.A.P has no interest declared. K.L.T. is a member of the Editorial Board for the Journal of Cardiothoracic and Vascular Anaesthesia and a member of the Society of Pediatric Anesthesia (SPA), Congenital Cardiac Anesthesia Society (CCAS) and the Society of Paediatric Anaesthesia in Australia and New Zealand (SPANZA). References 1. Davies RG, Myles PS, Graham JM. A comparison of the analgesia efficacy and side-effects of paravertebral vs epidural blockade for thoracotomy a systematic review and metaanalysis of randomized trials. Br J Anaesth 2006; 96: Albokrinov AA, Fesenko UA. Spread of dye after single thoracolumbar paravertebral injection in infants: a cadaveric study. Eur J Anaesthesiol 2014; 31: Richardson J, Lönnqvist PA, Naja Z. Bilateral thoracic paravertebral block: potential and practice. Br J Anaesth 2011; 106: Boretsky K, Visoiu M, Bigeleisen P. Ultrasound-guided approach to the paravertebral space for catheter insertion in infants and children. Pediatr Anaesth 2013; 23: Eck JB, Cantos-Gustafsson A, Ross AK, Lönnqvist PA. What s new in pediatric paravertebral analgesia. Tech Reg Anesth Pain Manag 2002; 6: Lönnqvist PA, Richardson J. Use of paravertebral blockade in children. Tech Reg Anesth Pain Manag 1999; 3: Naja Z, Lönnqvist PA. Somatic paravertebral nerve blockade. Incidence of failed block and complications. Anaesthesia 2001; 56: Tighe SQM, Greene MD, Rajadurai N. Paraverterbral block. Contin Educ Anaesth Crit Care Pain 2010; 10: Raj N. Thoracic paravertebral block. Available from tions/us-regional-anaesthesia/truncal-blocks/thoracic-para vertebral-block/ (accessed July 1, 2016) 10. Higgins JPT, Green S. Cochrane Handbook for Systematic Reviews of Interventions Version Cochrane Collaboration, Available from (accessed July 25, 2016). 11. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA Statement. Open Med 2009; 3: e Naja ZM, Raf M, El Rajab M, Ziade FM, Al Tannir MA, Lönnqvist PA. Nerve stimulator-guided paravertebral blockade combined with sevoflurane sedation versus general anesthesia with systemic analgesia for postherniorrhaphy pain relief in children: a prospective randomized trial. Anesthesiology 2005; 103: Naja ZM, Raf M, El Rajab M, et al. A comparison of nerve stimulator guided paravertebral block and ilio-inguinal nerve block for analgesia after inguinal herniorrhaphy in children. Anaesthesia 2006; 61: Tug R, Ozcengiz D, Güneş Y. Single level paravertebral versus caudal block in paediatric inguinal surgery. Anaesth Intens Care 2011; 39: Eldeen HMS. Ultrasound-guided thoracic epidural and paravertebral blocks for cholecystectomy in pediatric patients with a cyanotic heart disease: a randomized controlled study. Egypt J Anaesth 2016; 32: Visoiu M, Cassara A, Yang CI. Bilateral paravertebral blockade (T7-10) versus incisional local anesthetic administration for pediatric laparoscopic cholecystectomy: a prospective, randomized clinical study. Anesth Analg 2015; 120: Splinter WM, Thomson ME. Somatic paravertebral block decreases opioid requirements in children undergoing appendectomy. Can J Anaesth 2010; 57: Handling editor: P. S. Myles