Efficacy of low-dose bupivacaine in spinal anaesthesia for Caesarean delivery: systematic review and meta-analysis

Transcription

1 British Journal of Anaesthesia 107 (3): (11) Advance Access publication 14 July 11. doi: /bja/aer0 REVIEW ARTICLES Efficacy of low-dose bupivacaine in spinal anaesthesia for Caesarean delivery: systematic review and meta-analysis C. Arzola 1 * and P. M. Wieczorek 2 1 Department of Anesthesia and Pain Management, Mount Sinai Hospital and University of Toronto, 600 University Avenue, Room 1514, Toronto, ON, Canada M5G 1X5 2 SMBD-Jewish General Hospital and McGill University, 3755 Côte Ste-Catherine Road, Room A335, Montreal, QC, Canada H3T 1E2 * Corresponding author. carzola@mtsinai.on.ca Editor s key points Spinal anaesthesia for Caesarean section is associated with maternal hypotension. The incidence may be reduced by the use of a lower dose, but this may reduce block efficacy. This meta-analysis of 12 studies with a total of 693 patients used a cut-off dose of.8or 8 mg. Low dose is associated with fewer adverse effects but lower anaesthetic efficacy. Summary. Spinal anaesthesia is the preferred anaesthetic technique for elective Caesarean deliveries. Hypotension is the most common side-effect and has both maternal and neonatal consequences. Different strategies have been attempted to prevent spinal-induced hypotension, including the use of low-dose bupivacaine. We conducted a systematic search for randomized controlled trials comparing the efficacy of spinal bupivacaine in low dose (LD 8 mg) with conventional dose (CD.8 mg) for elective Caesarean delivery. Thirty-five trials were identified for eligibility assessment, 15 were selected for data extraction, and 12 were finally included in the meta-analysis. We investigated sources of heterogeneity, subgroup analysis, and meta-regression for confounding variables (baricity, intrathecal opioids, lateral vs sitting position, uterine exteriorization, and study population). Sensitivity analysis was performed to test the robustness of the results. In the LD group, the need for analgesic supplementation during surgery was significantly higher [risk ratio (RR)¼3.76, 95% confidence interval (95% CI)¼ ] and the number needed to treat for an additional harmful outcome (NNTH) was 4 (95% CI¼2 7). Furthermore, the LD group exhibited a lower risk of hypotension (RR¼0.78, 95% CI¼ ) and nausea/vomiting (RR¼0.71, 95% CI¼ ). Conversion to general anaesthesia occurred only in the LD group (two events). Neonatal outcomes (Apgar score, acid base status) and clinical quality variables (patient satisfaction, surgical conditions) showed non-significant differences between LD and CD. This meta-analysis demonstrates that low-dose bupivacaine in spinal anaesthesia compromises anaesthetic efficacy (risk of analgesic supplementation: high grade of evidence), despite the benefit of lower maternal side-effects (hypotension, nausea/vomiting: moderate grade of evidence). Keywords: anaesthesia, obstetric; anaesthetic techniques, subarachnoid; anaesthetics local, bupivacaine; complications, hypotension; safety, techniques Regional anaesthesia is a major factor in patient safety during Caesarean delivery. 1 Resurgence of spinal anaesthesia as a popular technique was possible due to the development of small-bore needles with pencil-point tips and has become the preferred method of anaesthesia for elective and for many emergency Caesarean deliveries if an epidural catheter is not already in situ. 2 While effective surgical anaesthesia is the primary objective of the spinal technique, it must be accomplished while minimizing maternal and neonatal side-effects. Although various factors influence the appropriate sensory nerve block for surgical anaesthesia, the local anaesthetic dose is the main determinant of its success. 3 Anaesthesia textbooks recommend bupivacaine in a dose of between 12 and 15 mg. 45 However, the use of this dose range has been associated with an incidence of maternal arterial hypotension of 69% to.80%, resulting in maternal and neonatal morbidity. 6 A number of studies have sought an optimal dose of bupivacaine, but produced dissimilar findings with doses ranging from 5 to mg. 78 The use of a lower dose aims to decrease maternal side-effects (hypotension, intraoperative nausea/ vomiting), reduce the time to discharge from the postanaesthesia care unit, and improve maternal satisfaction. 7 However, such a strategy could compromise the adequacy of anaesthesia, and require supplementary analgesia, with possible neonatal consequences and may require conversion to general anaesthesia, a situation known as a risk factor for anaesthesia-related maternal morbidity and mortality. 19 Recent narrative reviews have addressed the controversy of spinal bupivacaine in low dose (LD) Although useful, they are essentially descriptive and conclude with opinionbased recommendations. Systematic reviewing allows the & The Author [11]. Published by Oxford University Press on behalf of the British Journal of Anaesthesia. All rights reserved. For Permissions, please journals.permissions@oup.com

2 Low-dose spinal bupivacaine for Caesarean delivery BJA efficient integration of the evidence due to explicit methods used to summarize the data, limiting bias and improving the accuracy of the conclusions, and thus providing a reliable source and basis that can be used for rational decisionmaking. 12 We therefore conducted a systematic review of the literature and meta-analysis on the efficacy and adverse effects of spinal bupivacaine in LD compared with conventional dose (CD) for elective Caesarean delivery. Methods We conducted a systematic search of the electronic bibliographic databases: Cochrane Central Register of Controlled Trial (CENTRAL), MEDLINE, EMBASE, and LILACS. Both freetext and medical subject headings (MeSH) terms were used regarding type of study, participants, interventions, and outcomes. Reference lists from identified studies and journals which appeared to be associated with the most retrieved citations were then hand-searched. The search was restricted to full reports of randomized controlled trials published in peer-reviewed journals without excluding trials published in languages other than English. No date restriction was applied up to October 08. An update was performed on December 10, using the literature search strategy constructed for MEDLINE. Trials studying ASA I II term parturients for elective or semi-urgent Caesarean delivery under neuraxial spinal anaesthesia [single injection spinal or combined spinal epidural (CSE)], which compared bupivacaine in a dose.8 mg with a dose 8 mg (hyperbaric or isobaric, with or without adjuvants). We selectively chose trials reporting the frequency of intraoperative analgesic supplementation by any route and also conversion to general anaesthesia. Such variables were considered primary outcomes in our review, as surrogate outcomes of efficacy of spinal anaesthesia. Secondary outcomes measured were: maternal adverse effects (hypotension, nausea/vomiting) and neonatal outcomes (Apgar score, acid base status). Patient satisfaction during the intraoperative period and surgical conditions as assessed by the surgeon were also considered as secondary quality outcomes. Eligibility assessment was performed independently in an unblinded, standardized manner by the two review authors using a customized form, while discrepancies were resolved by consensus. With respect to included studies, a formal measure of agreement between the two reviewers was calculated through Kappa statistic. 13 We prepared a flow diagram to summarize the study selection process according to PRISMA. 14 The reviewers extracted data independently using a standardized data collection form managed electronically through the database software Filemaker Pro9 &. Any discrepancy was resolved by discussion and re-inspection of the original trial. The likelihood of the risk of bias was assessed by the review authors independently using two different tools: the Jadad scale 15 and a domain-based evaluation recommended by the Cochrane Handbook for Systematic Reviews of Interventions. 16 Data analysis The main primary outcome was anaesthetic efficacy based on requirement for analgesic supplementation by any route (i.v.; i.m.; inhalation, INH; epidural, EPiD) during the Caesarean delivery. Outcomes were treated as dichotomous variables based mainly on the specific definition that was used in each individual trial and no further attempt was made at standardization, unless stated otherwise in the Results section. We had expected that information extracted from the individual trials might need some transformation (processing) before it was suitable for analysis, situations such as randomization in more than two intervention groups as a multi-arm study receiving similar but non-identical doses of bupivacaine. Therefore, before seeing the data, we decided to follow the approach recommended by the Cochrane Handbook 17 to avoid introducing bias into the analysis. The planned method was to combine within each trial all relevant experimental intervention groups into a single group and to combine all relevant control groups into a single control group. Thus, the number of outcome events and total number per group would reflect such a combined approach. The operational cut-off point was established at 8 mg for the LD (intervention group) and.8 mg for the CD (control group). Even though, evidence from prospective studies assessing maternal haemodynamic stability shows that lowering the spinal dose renders improvement, there are few dose response relationship studies to elucidate and define the LD range in terms of anaesthetic efficacy. Therefore, we have considered what has been summarized by authors in published narrative reviews exploring the literature, and we have established such an arbitrarily operational/ instrumental cut-off point in the review process as recommended by the Cochrane Handbook for Systematic Reviews. Statistical combination of data from two or more separate trials in a meta-analysis was decided based on the evaluation of the clinical and methodological heterogeneity. The inconsistency throughout trials was quantified with the I 2 statistic proposed by Higgins and colleagues, 18 assuming a value more than 50% as a substantial heterogeneity. The summary effect measure chosen was risk ratio (RR) and number needed to treat for an additional harmful outcome (NNTH), along with their corresponding 95% confidence intervals (CI). The meta-analysis was planned to be performed through a random-effects model and Mantel Haenszel (M H) statistical method, anticipating that trials would present multiple dose-schemes and assuming that the actual true effects have a normal distribution. In order to explore publication bias, we performed logarithmic transformation of the RR effect estimates and its standard errors, construction of Begg s funnel plots, and assessment of the degree of symmetry with Egger s test (weighted 309

3 BJA Arzola and Wieczorek regression). Sensitivity analyses were performed to test the robustness of the results. Subgroup analyses and meta-regression were performed on prespecified confounding variables: bupivacaine baricity, position during injection, uterine exteriorization, and original groups vs re-grouping. Short-acting intrathecal opioid use (other than intrathecal morphine) was incorporated as a new confounding variable after data extraction, and also study population. Analyses were conducted using Review Manager 19 (RevMan5, Cochrane Collaboration) and STATA 9.2 (College Station, TX, USA) for Macintosh. This systematic review was carried out using the methods established by the Cochrane Handbook for Systematic Reviews of Interventions and we followed the recommendations and checklist items from the PRISMA Statement for Reporting Systematic Reviews and Meta-analysis. 14 Results The search strategies identified 186 (CENTRAL), 139 (MEDLINE), 366 (EMBASE), and 94 (LILACS) studies and another four from some other sources. After a process of successive screening for eligibility, 35 studies were selected for a more detailed evaluation (Fig. 1). The subsequent eligibility assessment excluded studies; therefore, 15 studies were included for the data extraction process and eventual description of study characteristics involving 1004 participants. Reasons for study exclusion included: (i) mixed CENTRAL 186 identified 19 selected MEDLINE 139 identified 13 selected EMBASE 366 identified 7 selected 35 selected for evaluation interventions without isolation of the efficacy of the intrathecal component (CSE with epidural supplementation, epidural volume extension); (ii) comparison groups not complying with the operational definition of LD and CD; and (iii) lack of reporting primary outcomes data for further analysis. The level of agreement between reviewers in this final assessment yielded a kappa statistic of 0.7, considered to reflect substantial agreement. One of the trials 21 reported two different studies in the same publication, meaning two independent samples of subjects under similar protocol, but independent randomization process and data analysis. Whereas only one of those studies uniformly includes the addition of fentanyl as an adjuvant in groups under comparison, they are presented and counted separately as two studies but one publication reference (Choi and colleagues-a and -b). 21 During the meta-analysis stage, three studies were excluded for presenting clinical co-intervention. Consequently, 12 studies were deemed for the final meta-analysis involving 693 participants (Fig. 1). The 15 studies analysed were published between and 10 and were performed in North America, South America, Europe, 27 Asia, and Africa. 33 We contacted one author for clarification on the primary outcome data. 29 The studies were published in English, Spanish, and Japanese. The mean age of the patients included ranged from 24 to 37 yr. While mean weight reported was between 58 and 62 kg in three studies, between 65 and 70 kg in four LILACS 94 identified 4 selected Other sources 4 selected excluded with reasons 15 included for data extraction (1004 participants) 3 excluded for meta-analysis with reasons 12 included for meta-analysis (693 participants) Fig 1 PRISMA flow chart. 310

4 Low-dose spinal bupivacaine for Caesarean delivery BJA Table 1 Description of the included studies. CSE, combined spinal epidural anaesthesia;, uterine displacement; IB, isobaric bupivacaine; HB, hyperbaric bupivacaine; Fen, fentanyl; Morph, morphine; Ket, ketamine; Lid, lidocaine; Epin, epinephrine; Mepiv, mepivacaine; Epid, epidural; GA, induction of general anaesthesia; LLD, left lateral decubitus; RLD, right lateral decubitus; n/a, information not available; Sx, characteristics of the surgical technique; Ext, Uterine exteriorisation; SuffSx, sufficiency for the surgery; SenB, sensory block; Trend, Trendelenburg position; R-Trend, reverse Trendelenburg; Pain assess, pain assessment; Intraop, intraoperative Study n Comparison groups Neuraxial technique Sx Analg/anaesth supplementation Efficacy criteria Ben-David 7 32 IB 0.5%: 5 mg+fen 25 mg; 10 mg Spinal; sitting; Choi and 60 HB 0.5%: 8 mg; 10 mg; 12 mg Spinal; LLD; colleagues-a 21 Choi and 60 HB 0.5%+Fen 10 mg: 8 mg; colleagues-b mg; 12 mg Spinal; LLD; Kiran and Singal HB 0.5%: 7.5 mg; 8.75 mg; 10 mg Spinal; LLD; Trend Ginosar and 42 HB 0.75%+Fen 10 mg+morph colleagues 23 0 mg: 6 mg; 7 mg; 8 mg; 9 mg; 10 mg; 11 mg; 12 mg CSE; sitting; Nagata and 33 HB 0.5%: 8 mg; 10 mg Spinal; RLD; colleagues 31 R-Trend; Rivero and colleagues HB 0.5%+Fen 15 mg: 7.5 mg; 12.5 mg Carvalho and 48 IB 0.5%+Fen10 mg+morph 0 colleagues 22 mg: 5 mg; 6 mg; 7 mg; 8 mg; 9 mg; 10 mg; 11 mg; 12 mg Guasch and 42 HB 0.5%+Fen mg: 6.5 mg; colleagues mg Kimoto and 47 IB 0.5%+Fen mg: 5 mg; colleagues mg; 10 mg; 12.5 mg Meléndez and 239 HB 0.5%: 7.5 mg+fen 15 mg; colleagues mg+fen 25 mg; 10 mg Bryson and 52 Fen 50 mg+morph 0 mg: IB colleagues % (4.5 mg); HB 0.75% (12 mg) Leo and 60 HB 0.5%+Morph 100 mg: 7 mg; colleagues 29 8 mg; 9 mg Spinal; LLD; CSE; sitting; Ext Fen 1 mg kg 21 (i.v.); GA Pain assess; SuffSx n/a Fen 30 mg (i.v.) Pain assess; SenB T6 n/a Fen 30 mg (i.v.) Pain assess; SenB T6 n/a Ket mg, 10 mg (i.v.) Pain assess; SenB T8 at 4 min Ext n/a Lid 2%+Epin 5 mg ml 21 (5 ml bolus-epid) Analgesic supplementation not needed Pain assess; SenB T6 at 10 min Pain assess; SenB T6 at 10 min n/a Fen (i.v.) Pain Assess; SenB T4 at incision Ext Lid 2%+Epin 5 mg ml 21 (5 ml bolus-epid) Pain assess; SenB T6 at 10 min Spinal; sitting; n/a Fen 100 mg (i.v.); GA Pain assess CSE; LLD n/a Mepiv 2% (10 15 ml bolus-epid) Pain assess; SenB T6 at 10 min Spinal; LLD; Ext Fen 50 mg (i.v.) Pain assess Spinal; sitting; Turhanoglu and IB 0.5%: 4 mg+fen 25 mg; 10 mg CSE; sitting; colleagues 30 Mebazaa and 80 IB 0.5%+Fen 25 mg+morph 100 colleagues 33 mg: 7.5 mg; 10 mg Ext Fen 1.5 mg kg 21 (i.v.); Ket 0.2 mg kg 21 (i.v.) CSE; RLD; n/a Lid 1.5% (alkalinized)+epin 5 mg ml 21 (5 ml bolus Epid) up to ml; Entonox; Fen (iv) Spinal; sitting; Ext No ext Fen 50 mg (i.v.); Lid 1% (7 ml bolus-epid) Fen (i.v.) Pain assess Pain assess; SenB T4 at 15 min; SenB below T6 intraop SenB T6 at 15 min Pain assess studies, and between 70 and 85 kg in eight studies, BMI was mentioned in only two. Sample size ranged from 32 to 239 participants (median¼52) and only two had larger sample sizes of 109 and 239 (Table 1). Power analysis was mentioned in eight of the 15 studies: the variables considered for calculations were hypotension in five studies, intraoperative pain, 25 maximum sensory block, 29 and in one study, interim analysis of their first 10 participants. 27 In seven studies, two groups were compared, three in five studies, and four in one study. 32 A dose-ranging protocol was used in two studies comparing seven and eight groups in each. Bupivacaine dose in the included studies ranged from 4 to 12.5 mg. We attempted to calculate a representative summary dose for each dose-scheme group, considering the studies sample size, thus resembling a weighted mean value. Hence, we obtained a rounded dose of 7 mg for LD and 11 mg for CD. These are somewhat instrumental and intend to keep the meta-analytic approach of the sample size weights of the studies, but we do not expect to consider them as the final nominal summary dose for LD and CD. Intrathecal opioids (other than morphine) were used only in the LD and not in the CD (three studies) and therefore, as a result of treating groups unevenly, this situation was considered an element of co-intervention and source of heterogeneity for the analysis. The anaesthetic technique used in 10 studies was singleinjection spinals while five studies used CSE techniques, in which the intrathecal injection was the main anaesthetic component and the epidural catheter was considered for rescue, analgesic anaesthetic supplementation, or both. Ringer s lactate solution was most commonly used for preloading, in the range of 500 ml in two 311

5 BJA Arzola and Wieczorek studies and ml in the remainder. Four studies also included colloids. Hypotension was managed with ephedrine as a sole drug in 13 studies Ephedrine was a second option to phenylephrine when heart rate was,60 beats min in one study 29 and mephentermine was used in one study. 28 Primary outcomes Anaesthetic efficacy: studies administered intraoperative analgesic/anaesthetic supplementation based on four nonexcluding criteria: (i) a predetermined threshold of pain on a visual analogue scale (VAS), (ii) pain categories of excellent, good, mild, fair, and poor, (iii) any pain during surgery, and/or (iv) failure to achieve a predetermined dermatomal sensory block before start of surgery. The analgesic supplementation during single injection spinals was i.v. fentanyl (ketamine in one study). 28 In CSE studies, supplementation was obtained through the epidural catheter. Conversion to general anaesthesia occurred in only one study 27 in two out of 21 participants in the LD group (0.5% hyperbaric bupivacaine 6.5 mg with fentanyl mg). There were no reported events of conversion to general anaesthesia for CD in any study. Secondary outcomes Maternal adverse effects, hypotension Various criteria were used to define hypotension throughout the studies, considering different cut-off points in arterial pressure or even a combination of two criteria. Some studies recorded hypotension as (i) systolic arterial pressure, mm Hg, (ii) mean arterial pressure,60 mm Hg, 23 or (iii) predetermined percentage decrease in systolic arterial pressure %. Such heterogeneity in variation of clinical definitions for hypotension was anticipated and would have precluded further quantitative analysis. However, pool estimate of combined effect measures was planned to take into consideration such variations across studies. Intraoperative nausea/vomiting Incidence was reported either separately or combined. For the quantitative analysis, it was considered to be whichever was the highest occurrence of nausea, vomiting, or combined nausea/vomiting. Neonatal outcomes (Apgar score, acid base status) While seven studies reported Apgar scores at 1 and 5 min, only two mentioned acid base status. They were all within normal range and showed nonsignificant differences between LD and CD. Patient satisfaction during the intraoperative period Six studies collected data at the post-anaesthesia care unit. Participants were asked to grade their level of overall 312 satisfaction with the anaesthetic (i) on VAS; (ii) in categories as excellent, good, average, or poor; and (iii) participants having had previous Caesarean deliveries were asked if they preferred their present or their previous anaesthetic technique. 24 Scores and categories reported were of high satisfaction without significant differences between LD and CD. Although one study 33 reported a better overall satisfaction in the LD group, a subgroup analysis expressed significant dissatisfaction (average and poor) due to intraoperative pain and to nausea/vomiting when comparing LD with the CD group. Only one study 24 investigated up to 3 days after operation through the quality of recovery (QoR) score 34 and VAS. Follow-up results were similarly high in LD and CD. Surgical conditions assessed by surgeon Surgical conditions assessed by the surgeon were reported in four studies and all showed favourable conditions without differences between LD and CD. The internal validity of included trials was assessed according to methods described by the Jadad scale and Cochrane Collaboration (Table 2). With regard to the Jadad scale (maximum score 5; 3 satisfactory methodological quality), all studies had acceptable levels of methodological quality with low risk of significant bias in the results. We found a source of performance bias based on co-intervention in three studies The included studies presented low risk of bias in their methodology. Quantitative data synthesis We compared LD and CD as the intervention and control groups, respectively, using analgesic supplementation as a surrogate outcome of anaesthetic efficacy. We initially analysed the group of 15 studies included in the data extraction stage, reporting for 1004 patients. In the pooled analysis, LD was associated with a significantly higher risk of analgesic supplementation during Caesarean delivery (RR¼3.30, 95% CI¼ , P¼0.007), but substantial evidence of heterogeneity (I 2 ¼79%, P, ) was observed. On investigating this heterogeneity, a source of clinical co-intervention was detected in three studies The groups being compared were not treated uniformly, and the LD group received intrathecal fentanyl as an adjuvant. After excluding these studies, the final meta-analysis considered the pool estimate from 12 studies reporting 693 participants, which remained significant for a higher risk of analgesic supplementation in LD (RR¼3.76, 95% CI¼ , P, ) once the heterogeneity was resolved (I 2 ¼0%, P¼0.72) (Fig. 2). We calculated an NNTH of 4 (95% CI¼2 7) based on a published risk of analgesic supplementation of 10.9% 35 and 10.1% from the CD group in our analysis. Publication bias was ruled out by construction of Begg s funnel plots and assessment of the degree of symmetry by Egger s test (bias P¼0.274). Sensitivity analyses demonstrated that our findings were not affected by change in the summary effect measure to odds ratio or by analysis through a fixed-effect model.

6 Low-dose spinal bupivacaine for Caesarean delivery BJA Table 2 Risk of bias within studies. Risk of bias: A, low risk; B, high risk; C, unclear/uncertain risk of bias; *co-intervention; Jadad scale 15 Study Sequence generation Allocation concealment Blinding Incomplete outcome data Subgroup analyses and meta-regression did not reveal interaction or effect modification by assessing possible confounding variables (bupivacaine baricity, position during injection, uterine exteriorization, intrathecal opioids other than morphine, weight,70 kg, original groups vs re-grouping, Asian vs non-asian population). Conversion to general anaesthesia occurred in only one study 27 reporting two events in the LD group; therefore, a meta-analysis with a pool estimate of combined effect measures was considered inappropriate. Of the secondary outcomes, the risk of hypotension was lower in the LD group (RR¼0.78, 95% CI¼ , P¼0.005) (Fig. 3). The clinical and methodological heterogeneity of this outcome was confirmed statistically by quantifying a moderate inconsistency across studies (I 2 ¼29%, P¼0.19). Nausea/vomiting as a combined outcome exhibited a lower risk in the LD group (RR¼0.71, 95% CI¼ , P¼0.01) (heterogeneity; I 2 ¼6%, P¼0.39) (Fig. 4) under a fixed-effect model. The other secondary outcomes were not evaluated by meta-analysis due to dissimilarity in the reporting of outcomes. Selective outcome reporting Discussion Free of other bias Jadad scale Randomization Blinding Withdrawals Ben-David and A A A A A B* colleagues 7 Choi and A C A A A A colleagues-a 21 Choi and A C A A A A colleagues-b 21 Kiran and A A A A A A Singal 28 Ginosar and A A A A A A colleagues 23 Nagata and A A A C A A colleagues 31 Rivero and A A A B A A colleagues 26 Carvalho and A A A A A A colleagues 22 Guasch and A C A A A A colleagues 27 Kimoto and A A A A A A colleagues 32 Meléndez and A A A A A B* colleagues 25 Bryson and A A A A A A colleagues 24 Leo and A A A A A A colleagues 29 Turhanoglu and A C A A A A* colleagues 30 Mebazaa and colleagues 33 A C A A A A On the basis of the collected evidence regarding ASA I II term parturients for elective or semi-urgent Caesarean delivery under neuraxial spinal anaesthesia, the risk of intraoperative analgesic supplementation in the LD bupivacainescheme ( 8 mg) is more than three times higher (RR¼3.76, 95% CI¼ , P, ) than the CD scheme (.8 mg). This suggests that one additional patient will require intraoperative analgesic supplementation due to intraoperative pain for every four patients (95% CI¼2 7) receiving an LD spinal rather than a CD spinal. We found no evidence of interaction or effect modification among the possible clinical confounders analysed. The risk of conversion to general anaesthesia could not be estimated based on the included studies (two events in LD). There was a lower risk of maternal side-effects in LD than in CD, with 22% reduction in hypotension and 29% reduction in nausea/ vomiting. Neonatal outcomes were similar in both groups, as were patient satisfaction during the intraoperative period and quality of surgical conditions assessed by the surgeon. 313

7 314 Study or subgroup Choi and colleagues-a 21 Choi and colleagues-b 21 Kiran and Singal 28 Ginosar and colleagues 23 Rivero and colleagues 26 Nagata and colleagues 31 Guasch and colleagues 27 Kimoto and colleagues 32 Carvalho and colleagues 22 Bryson and colleagues 24 Leo and colleagues 29 Mebazaa and colleagues 33 Low dose Events Total Conventional dose Events Total Total (95% CI) 317 Total events 73 Heterogeneity: t 2 =0.00; c 2 =6., df=9 (P=0.72); I 2 =0% Test for overall effect: Z=5.70 (P< ) Weight 16.9% 16.6% 16.5% 2.4% 2.5% 5.4% 17.2% 4.8% 15.8% 2.1% % M-H, Random, 95% CI Year 3.50 (1.16, 10.57) 00 Not estimable (0.65, 6.11) (1.59, 15.00) (0.42, ) 04 Not estimable (0.38, ) (2.72, ) (1.64, 14.67) (0.64, 6.29) (0.64, 6.29) (0.13, 71.51) (2.38, 5.92) M-H, Random, 95% CI Favours LD Favours CD Fig 2 Forest plot for analgesic supplementation comparing LD vs CD: individual trials and meta-analysis. Events, the total numbers with events (primary outcome¼analgesic supplementation) in the intervention (LD) and control (CD) groups; Total, the total numbers of participants in the intervention and control groups; Weight, sample size contribution of the study relative to the pooled sample size of the meta-analysis; M H, Mantel Haenszel methods. BJA Arzola and Wieczorek Downloaded from by guest on October 13, 12

8 315 Study or subgroup Choi and colleagues-a 21 Choi and colleagues-b 21 Kiran and Singal 28 Rivero and colleagues 26 Nagata and colleagues 31 Guasch and colleagues 27 Bryson and colleagues 24 Leo and colleagues 29 Mebazaa and colleagues 33 Low dose Conventional dose Events Total Events Total Weight M-H, Random, 95% CI Total (95% CI) % 0.78 (0.65, 0.93) Total events Heterogeneity: t 2 =0.02; c 2 =11.21, df=8 (P=0.19); I 2 =29% Test for overall effect: Z=2.83 (P=0.005) % 4.6% 3.2% 23.3% 6.6% 2.5% 18.0% 10.8% 23.3% Year 0.86 (0.49, 1.51) (0.35, 1.62) (0.17, 1.14) (0.74, 1.21) (0.26, 0.80) (0., 1.66) (0.71, 1.33) (0.38, 0.96) (0.60, 0.99) 10 M-H, Random, 95% CI Favours LD Favours CD Fig 3 Forest plot for hypotension comparing LD vs CD: individual trials and meta-analysis. Events, the total numbers with events (secondary outcome¼hypotension) in the intervention (LD) and control (CD) groups; Total, the total numbers of participants in the intervention and control groups; Weight, sample size contribution of the study relative to the pooled sample size of the meta-analysis; M H, Mantel Haenszel methods. Low dose Conventional dose Study or subgroup Events Total Events Total Weight Choi and colleagues-a % Choi and colleagues-b % Kiran and Singal % Rivero and colleagues % Ginosar and colleagues % Nagata and colleagues % Carvalho and colleagues % Guasch and colleagues % Bryson and colleagues % Leo and colleagues % Mebazaa and colleagues % M-H, Fixed, 95% CI 0.74 (0.37, 1.45) 0.67 (0.15, 3.01) 1. (0.32, 4.52) 0.23 (0.03, 1.88) 0.15 (0.01, 2.55) 1.00 (0.70, 1.44) 0.92 (0.31, 2.77) 0.71 (0.27, 1.89) 0.83 (0.41, 1.71) 1.75 (0., 7.66) 0.43 (0.22, 0.82) Total (95% CI) % 0.71 (0.55, 0.93) Total events Heterogeneity: c 2 =10.58; df=10 (P=0.39); I 2 =6% Test for overall effect: Z=2.51 (P<0.01) Year M-H, Fixed, 95% CI Favours LD Favours CD Fig 4 Forest plot for nausea/vomiting comparing LD vs CD: individual trials and meta-analysis. Events, the total numbers with events (secondary outcome¼nausea/vomiting) in the intervention (LD) and control (CD) groups; Total, the total numbers of participants in the intervention and control groups; Weight, sample size contribution of the study relative to the pooled sample size of the meta-analysis; M H, Mantel Haenszel methods. Low-dose spinal bupivacaine for Caesarean delivery BJA Downloaded from by guest on October 13, 12

9 BJA Arzola and Wieczorek Table 3 Summary of results (GRADE). N, number of studies; n, number of participants; LD, low dose; (1), one study reporting two events for this outcome; (2), categorical data in four studies (Apgar score 8 at 5 min, score 9 and 10 at 5 min) and continuous data in three studies; (3), two studies collecting data for this outcome; (4), categorical data in three studies and continuous data in three studies; (5), categorical data in three studies and one study reporting no difference Spinal bupivacaine: Low dose vs conventional dose Participants: Pregnant patients ASA I II, elective and semi-urgent Caesarean delivery Intervention: Spinal bupivacaine dose 8 mg Comparison: Spinal bupivacaine dose >8 mg Outcomes (95% CI) N n GRADE Comments Analgesic supplementation 3.76 ( ) High Consistent results in magnitude and direction. Low risk of bias Conversion to general anaesthesia Not estimable (1) 1 42 Low Two events in the LD group Hypotension 0.78 ( ) Moderate Dissimilar criteria definitions Nausea/vomiting 0.71 ( ) Moderate Results from combined events Apgar score Not estimable (2) Moderate Normal score and no differences between groups Acid base status Not estimable (3) 2 85 Low Normal range and non-significant difference Patient satisfaction Not estimable (4) Moderate High satisfaction is not compromised in LD Quality assessed by the surgeon Not estimable (5) Moderate Favourable conditions and no differences The quality of the overall body of evidence for each individual outcome was addressed and summarized through the GRADE system 36 (Table 3). The assessment of the analgesic supplementation as a primary outcome was rated as high quality, in view of low risk of bias in trial design and implementation, precise and consistent results in magnitude and direction, and also identified and explained source of heterogeneity. However, the secondary outcomes were rated as moderate quality, considering relatively different criteria for hypotension; results from combined events for nausea/vomiting; and pool estimates not estimable for neonatal outcomes, patient satisfaction, and quality of surgical conditions. Conversion to general anaesthesia and neonatal acid base status were considered as low quality. Our systematic review has limitations. The search was restricted to full reports of randomized controlled trials published in peer-reviewed journals, excluding other sources of biomedical literature, which could have possibly collected more studies related to the topic. Although the funnel plots and the corresponding statistical tests did not reflect publication bias, the power is low, considering the small number of studies. The two main limitations of this review are related to the operational cut-off point for the bupivacaine dose and outcome definitions across studies. Potential bias as a systematic error can help explain variation in the results of the studies included in a systematic review. 16 Choosing a different cut-off point for the dose-schemes could have changed not only the final meta-analysis but also the early stage of re-grouping the participants within studies for data extraction, and which studies were eligible, that is with a dose scheme from each side of the cut-off point. Sensory blockade to T2 T6 (most commonly T4) is thought to be adequate for Caesarean delivery, but the method used to assess the block is not always standardized or mentioned explicitly A block at T4 to pinprick has been shown to be unreliable at predicting a pain-free Caesarean delivery, 41 but, in contrast, others have shown that a block to T5 to touch, 42 or a block to touch that includes T6 at skin incision, 43 is likely to provide a pain-free procedure. The assumption is made that there is a consistent difference in dermatomal block levels between light touch, pinprick, and cold sensations. 44 Although Russell 43 demonstrated a median difference of two segments between these, it was emphasized that there was no constant relationship either within the group or in individual patients. Brown 45 defined analgesia as loss of sharp pinprick sensation and anaesthesia as loss of touch. Increasing the dose of local anaesthetic used can produce a denser sensory block (anaesthesia) without necessarily causing higher levels of block to pinprick, and with less need for analgesic supplementation. Opioids as adjuvants to neuraxial anaesthesia improve the quality of the block without producing a higher level of analgesia to pinprick. 42 We found that various sensory tests (temperature, pinprick) were used together with various methods of application. Indeed, sometimes the method used to test the block was not explicitly mentioned. The definition of an adequate block ranged from T2 to T6. None of the studies reported touch or light touch for dermatome assessment. Whatever the block level or assessment method, ultimately it is the patient s experience of pain which remains the uppermost outcome of importance. Hence, we used anaesthetic efficacy as assessed by intraoperative analgesic supplementation and conversion to general anaesthesia rather than dermatomal sensory block levels. Studies of prevention and treatment of spinal-induced hypotension show that lowering the dose improves maternal haemodynamic stability, 11 irrespective of what definition is used. Our review shows agreement with this but with moderate quality of evidence. Lastly, ephedrine was used in all studies but one for control of hypotension. 316

10 Low-dose spinal bupivacaine for Caesarean delivery BJA In summary, our review adds to the evidence concerning optimal dosing for spinal anaesthesia. Various prospective studies conclude that an LD scheme is a viable option without compromising the anaesthetic efficacy, but sample sizes have usually been calculated based on variables such as hypotension rather than intraoperative pain. Previous reviews related to this subject have had differing conclusions Studies using dose response curves modelling by logistic regression gave ED 50 and ED 95 values comparable with our weighted mean values for LD and CD, respectively Consequently, lower anaesthetic doses cannot be recommended unless an epidural catheter is in place (CSE) to rescue the block if anaesthesia is inadequate or becomes inadequate during surgery. Low-dose CSE anaesthesia may not be the optimal technique for all patients and institutions. 47 The dilemma of ensuring better anaesthesia while avoiding the higher incidence and severity of hypotension is not yet resolved but we have a better understanding of dose schemes. Acknowledgement The authors thank Dr Jose Carvalho for his valuable mentorship in the research project and development. Conflict of interest None declared. Funding This work was supported by departmental resources. References 1 Ngan Kee WD. Confidential enquiries into maternal deaths: 50 years of closing the loop. Br J Anaesth 05; 94: Gogarten W. Spinal anaesthesia for obstetrics. Best Pract Res Clin Anaesthesiol 03; 17: Hocking G, Wildsmith JA. Intrathecal drug spread. Br J Anaesth 04; 93: Tsen L. Anesthesia for cesarean delivery. In: Chesnutt DH, ed. Obstetric Anesthesia: Principles and Practice. Philadelphia: Elsevier Mosby, 09; Chapter 26, Braveman FR, Scavone BM, Wong CA, Santos AC. Obstetric anesthesia. In: Barash PG, Cullen BF, Stoelting RK, eds. Clinical Anesthesia, 6th Edn. Philadelphia: Lippincott Williams & Wilkins, 06; Lee A, Ngan Kee WD, Gin T. Prophylactic ephedrine prevents hypotension during spinal anesthesia for Cesarean delivery but does not improve neonatal outcome: a quantitative systematic review. Can J Anaesth 02; 49: Ben-David B, Miller G, Gavriel R, Gurevitch A. Low-dose bupivacaine-fentanyl spinal anesthesia for cesarean delivery. Reg Anesth Pain Med 00; 25: Finucane BT. Spinal anesthesia for cesarean delivery. The dosage dilemma. Reg Anesth 1995; : Hawkins JL, Koonin LM, Palmer SK, Gibbs CP. Anesthesia-related deaths during obstetric delivery in the United States, Anesthesiology 1997; 86: Dyer RA, Joubert IA. Low-dose spinal anaesthesia for caesarean section. Curr Opin Anaesthesiol 04; 17: Roofthooft E, Van de Velde M. Low-dose spinal anaesthesia for Caesarean section to prevent spinal-induced hypotension. Curr Opin Anaesthesiol 08; 21: Walker S, Palmer S, Sculpher M. The role of NICE technology appraisal in NHS rationing. Br Med Bull 07; 81 82: Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977; 33: Moher D, Liberati A, Tetzlaff J, Altman DG, PRISMA Group. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. J Clin Epidemiol 09; 62: Jadad AR, Moore RA, Carroll D, et al. Assessing the quality of reports of randomized clinical trials: is blinding necessary? Control Clin Trials 1996; 17: Higgins JPT, Altman DG, Sterne JAC, eds. Chapter 8: assessing risk of bias in included studies. In: Higgins JPT, Green S, eds. Cochrane Handbook for Systematic Reviews of Interventions, Version (updated March 11). The Cochrane Collaboration, 11. Available from 17 Higgins JPT, Deeks JJ, Altman DG, eds. Chapter 16: special topics in statistics. In: Higgins JPT, Green S, eds. Cochrane Handbook for Systematic Reviews of Interventions, Version (updated March 11). The Cochrane Collaboration, 11. Available from 18 Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. Br Med J 03; 327: Review Manager (RevMan) [Computer program], Version Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 08 Higgins JPT, Green S, eds. Cochrane Handbook for Systematic Reviews of Interventions, Version (updated March 11). The Cochrane Collaboration, 11. Available from www. cochrane-handbook.org 21 Choi DH, Ahn HJ, Kim MH. Bupivacaine-sparing effect of fentanyl in spinal anesthesia for cesarean delivery. Reg Anesth Pain Med 00; 25: Carvalho B, Durbin M, Drover DR, Cohen SE, Ginosar Y, Riley ET. The ED50 and ED95 of intrathecal isobaric bupivacaine with opioids for cesarean delivery. Anesthesiology 05; 103: Ginosar Y, Mirikatani E, Drover DR, Cohen SE, Riley ET. ED50 and ED95 of intrathecal hyperbaric bupivacaine coadministered with opioids for cesarean delivery. Anesthesiology 04; 100: Bryson GL, Macneil R, Jeyaraj LM, Rosaeg OP. Small dose spinal bupivacaine for Cesarean delivery does not reduce hypotension but accelerates motor recovery. Can J Anaesth 07; 54: Meléndez H, Gamarra HG, Fernández C, Dulcey R. Eficacia del fentanyl adicionado a bupivacaina en el dolor intraoperatorio en cesárea bajo anestesia subaracnoidea: Ensayo clínico controlado. Rev Col Anest 05; 33: Rivero J, Becerra M, Perea A. Dosis bajas de bupivacaina subaracnoidea reducen la incidencia de hipotensión durante la cesárea? Rev Col Anest 04; 32: Guasch E, Suárez A, Bermejo JM, Gilsanz F. Randomized controlled trial comparing a low dose to a conventional dose of hyperbaric bupivacaine for scheduled cesarean section. Rev Esp Anestesiol Reanim 05; 52: Kiran S, Singal NK. A comparative study of three different doses of 0.5% hyperbaric bupivacaine for spinal anaesthesia in elective caesarean section. Int J Obstet Anesth 02; 11: Leo S, Sng BL, Lim Y, Sia AT. A randomized comparison of low doses of hyperbaric bupivacaine in combined spinal epidural 317

11 BJA Arzola and Wieczorek anesthesia for cesarean delivery. Anesth Analg 09; 109: Turhanoglu S, Kaya S, Erdogan H. Is there an advantage in using low-dose intrathecal bupivacaine for cesarean section?. J Anesth 09; 23: Nagata E, Yoshimine K, Minoda Y, Kawaguchi Y, Sakamoto M, Takehara A. Comparison of 8 mg and 10 mg hyperbaric bupivacaine during spinal anesthesia for cesarean section in Japanese parturients. Masui 04; 53: Kimoto M, Murao K, Shirane A, et al. Appropriate dose of isobaric bupivacaine with fentanyl in spinal anesthesia for cesarean section. Masui 05; 54: Mebazaa MS, Ouerghi S, Ben Meftah R, Ben Cheikh M, Mestiri T, Ben Ammar MS. Reduction of bupivacaine dose in spinal anaesthesia for caesarean section may improve maternal satisfaction by reducing incidence of low blood pressure episodes. Middle East J Anesthesiol 10; : Myles PS, Hunt JO, Nightingale CE, et al. Development and psychometric testing of a quality of recovery score after general anesthesia and surgery in adults. Anesth Analg 1999; 88: Garry M, Davies S. Failure of regional blockade for caesarean section. Int J Obstet Anesth 02; 11: Atkins D, Best D, Briss PA, et al. GRADE Working Group. Grading quality of evidence and strength of recommendations. Br Med J 04; 328: Birnbach DJ, Browne IM. Anesthesia for obstetrics. In: Miller RD, ed. Miller s Anesthesia. Philadelphia: Churchill-Livingston Elsevier, 10; Chapter 69, Gaiser DR. Anesthesia for cesarean delivery. In: Bucklin BA, Gambling DR, Wlody D., eds. A Practical Approach to Obstetric Anesthesia. Philadelphia: Lippincott Williams & Wilkins, 09; Chapter 13, Tsen L. Anesthesia for obstetric care and gynecologic surgery. In: Longnecker DE, ed. Anesthesiology. New York: The McGraw-Hill Companies, 09; Chapter 61, Alahuhta S, Kangas-Saarela T, Hollmén AI, Edström HH. Visceral pain during caesarean section under spinal and epidural anaesthesia with bupivacaine. Acta Anaesthesiol Scand 1990; 34: Pedersen H, Santos AC, Steinberg ES, Schapiro HM, Harmon TW, Finster M. Incidence of visceral pain during cesarean section: the effect of varying doses of spinal bupivacaine. Anesth Analg 1989; 69: Russell IF. Levels of anaesthesia and intraoperative pain at caesarean section under regional block. Int J Obstet Anesth 1995; 4: Russell IF. A comparison of cold, pinprick and touch for assessing the level of spinal block at caesarean section. Int J Obstet Anesth 04; 13: Russell IF. Unreliability of sharp pinprick sensation. Reg Anesth Pain Med 02; 27: Brown D, Wildsmith J, Covino B, Scott D. Effect of baricity on spinal anaesthesia with amethocaine. Br J Anaesth 1980; 52: McNaught AF, Stocks GM. Epidural volume extension and low-dose sequential combined spinal epidural blockade: two ways to reduce spinal dose requirement for caesarean section. Int J Obstet Anesth 07; 16: Benhamou D, Wong C. Neuraxial anesthesia for cesarean delivery: what criteria define the optimal technique? Anesth Analg 09; 109: