The analgesic efficacy of local infiltration analgesia vs femoral nerve block after total knee arthroplasty: a systematic review and meta-analysis

Transcription

1 British Journal of Anaesthesia, 116 (5): (2016) doi: /bja/aew099 Review Article The analgesic efficacy of local infiltration vs femoral nerve block after total knee arthroplasty: a systematic review and meta-analysis E. Albrecht 1, *, O. Guyen 2, A. Jacot-Guillarmod 3 and K. R. Kirkham 4 1 Department of Anaesthesia, 2 Department of Orthopaedic surgery, 3 Department of Anaesthesia, Lausanne University Hospital, Lausanne, Switzerland, and 4 Department of Anaesthesia, Toronto Western Hospital, University of Toronto, Toronto, Canada *Corresponding author. eric.albrecht@chuv.ch Abstract Many consider femoral nerve block the gold standard in pain management following knee arthroplasty. is an alternate approach that applies the concept of surgical wound infiltration with local anaesthetics. This metaanalysis aims to compare both analgesic treatments for and functional outcomes after total knee arthroplasty. This meta-analysis was performed according to the PRISMA statement guidelines. The primary outcomes were cumulative i.v. consumption, pain scores at rest and on movement on postoperative day one (analogue scale,010). Secondary outcomes included range of motion, quadriceps muscle strength, length of stay and rates of complications (neurologic events, cardiovascular events, falls and knee infections). Fourteen trials, including 1122 adult patients were identified. There was no difference in i.v. consumption (mean difference: 2.0 mg; 95% CI: 4.9, 0.9 mg; I 2 =69%; P=0.19), pain scores at rest (mean difference: 0.1; 95% CI: 0.4, 0.3; I 2 =72%; P=0.80) and pain scores on movement (mean difference: 0.2; 95% CI: 0.5, 0.8; I 2 =80%; P=0.64) on postoperative day one (a negative mean difference favours local infiltration ). The qualities of evidence for our primary outcomes were moderate according to the GRADE system. There were no clinical differences in functional outcomes or rates of complications. Complication rates were captured by three trials or fewer with exception of knee infection, which was sought by eight trials. provides similar postoperative after total knee arthroplasty to femoral nerve block. Although this meta-analysis did not capture any difference in rates of complications, the low number of trials that specifically sought these outcomes dictates caution. Key words: ; nerve block; postoperative pain; regional ; total knee arthroplasty Total knee arthroplasty (TKA) causes moderate to severe postsurgical pain, 1 with femoral nerve block (FNB) considered by many as the gold standard analgesic therapy after this surgery. 25 (LIA) applies the concept of surgical wound infiltration with local anaesthetics 67 to joint surgery. 8 The technique was first reported for knee arthroplasty by Bianconi and colleagues 9 fewer than 15 years ago. Since then, it has gained widespread popularity among orthopaedic surgeons because of its ease of application, cost effectiveness and lack of apparent motor block of the lower limb The initial enthusiasm prompted a number of randomized controlled trials comparing LIA with FNB, which reported conflicting results for analgesic efficacy Several systematic reviews have endeavoured to clarify the magnitude of analgesic effect of both procedures, but their results are limited by the absence of quantitative meta-analysis Accepted: March 1, 2016 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 597

2 598 Albrecht et al. assessment, 1518 the fact that they do not address this question directly, 19 or did not investigate the relative benefit of each intervention on functional recovery. 20 This meta-analysis aims to compare the analgesic efficacy, the functional outcomes and the technique-related complications of FNB and LIA after TKA in adult patients. Methods Literature search and inclusion criteria The authors applied the recommendations of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRIS- MA) statement. 21 The electronic databases MEDLINE (until February 2016), the Cochrane Central Register of Controlled Clinical Trials (until February 2016), and the Excerpta Medica database, EMBASE (until February 2016) were searched with the following terms: Knee joint OR Knee surgery OR Total knee replacement OR Total knee arthroplasty. These search results were associated with OR Periarticular infiltration OR Peri-articular infiltration OR Periarticular injection OR Peri-articular injection OR Intraarticular infiltration OR Intraarticular infiltration OR Intraarticular injection OR Intra-articular injection OR Intraarticular OR Intra-articular. Findings were further restricted by associating with Regional OR Regional anesthesia OR Anaesthetic technique OR Anesthetic technique OR Anaesthesia conduction OR Anesthesia conduction OR Local anaesthetics OR Local anesthetics OR Nerve block OR Peripheral nerve block OR Femoral nerve block OR Adductor canal block OR Saphenous nerve block. The following keywords were also searched: Anaesth*, Anesth*, Nerve*, Replacement*, Arthroplasty*. Search results were limited to randomized controlled trials and humans. No language restriction was placed on the search. Lastly, bibliographies of retrieved articles were scrutinized for any relevant trials not yet identified in the primary search. Population The meta-analysis addresses male or female adults undergoing TKA. Intervention and comparator Only randomized trials comparing LIA to a group of patients having single-shot or continuous femoral nerve, saphenous nerve, or adductor canal blocks were included in the present meta-analysis. Any article that applied the LIA technique described by Kerr and Kohan, 11 in total or in part to a group (infiltration of any layer of the knee joint: posterior part, anterior part, periarticular soft tissue), was included. We excluded trials comparing LIA with a combination of epidural and FNB 2224 or investigating the analgesic efficacy of the combination of LIA and FNB with the combination of sciatic nerve block and FNB at rest and on movement measured at two and 12, and on postoperative day two and three; incidences of postoperative nausea or vomiting, pruritus within the first 24 h postoperatively, and chronic postoperative pain. Additional functional outcomes evaluated were range of motion or knee flexiononpostoperative days one, two and three; quadriceps muscle strength on postoperative days one, two and three; Knee Society score 29 at six weeks, three and 12 months postoperatively; and length of stay. We also aimed to capture any analgesic technique-related complication, such as rates of neurologic events, cardiovascular events, falls, knee joint infections, prosthesis loosening, or revision surgery. Finally, local anaesthetic plasma concentrations were retrieved whenever possible. Trial characteristics Extracted trial characteristics included type (single-shot injection or catheter insertion) and technique of LIA and peripheral nerve block, respectively; type, concentration and volume of local anaesthetics; type of other components used; anaesthetic strategy for surgery, and type and modality of postoperative. Rating of the studies The quality of the research methodology of each randomized trial was assessed following the Cochrane Collaboration s Risk of Bias Tool for randomized controlled trials. 30 Two authors (A.J.G. and K.K.) separately screened, reviewed and rated the items for each trial using this method and extracted data for the analyses. Disagreements with scoring or extracted data were addressed after discussion with a third author (E.A.). Data extraction Means, standard deviation, standard error of means, 95% confidence interval (CI), number of events and total number of participants were extracted from the text, tables or graphs from each source study. The authors of trials that failed to report the sample size or results as a mean and standard deviation, or standard error of the mean, or 95% CI, were contacted twice by to request the missing data or raw data. If no response was obtained, median and interquartile range were used for means and standard deviation approximation, as follows: the mean was estimated as equivalent to the median and the standard deviation was approximated to be the interquartile range divided by All opioids were converted into equi-analgesic doses of i.v. for analysis (i.v. 10 mg=oral 30 mg=iv hydromorphone 1.5 mg=oral hydromorphone 7.5 mg=iv pethidine 75 mg=oral oxycodone 20 mg=iv tramadol 100 mg). Pain scores reported as Visual, Verbal or Numeric Rating Scales were converted to a standardized 010 analogue scale for quantitative evaluations. Finally, we rated the quality of evidence for each outcome following the Grades of Recommendation, Assessment, Development, and Evaluation (GRADE) Working Group system. 34 Outcomes The specific outcomes sought from each article were derived following our approach described in a previous meta-analysis on acute postoperative pain. 28 The primary acute pain-related outcomes were cumulative i.v. consumption, and pain scores at rest and on movement on postoperative day one (24 ). Secondary acute pain-related outcomes sought were cumulative i.v. consumption at two and 12 postoperative h, and on postoperative day two and three; pain scores Statistical analysis Review Manager software (RevMan version 5.3.5; Copenhagen, The Nordic Cochrane Centre, The Cochrane Collaboration 2014) was used to perform meta-analyses. This software estimates the weighted mean differences for continuous data and risk ratio for categorical data between groups. It produces an overall estimate of the pooled effect. As most data sets were heterogeneous, they were analysed using a random effects model, and are presented as the mean difference or relative risk (RR) with

3 vs nerve block after knee replacement %. Our primary outcomes (i.v. consumption, pain scores at rest and on movement on postoperative day one) were analysed in subgroups according to the specialization of the corresponding author (anaesthetist vs orthopaedic surgeon). We conducted a meta-analysis when two or more trials reported similar outcomes. I 2 was used to evaluate heterogeneity. Predetermined thresholds were established for low (2549%), moderate (5074%), and high (>75%) levels. 35 Publication biases were evaluated for our primary outcomes by drawing a funnel plot of standard error of the mean difference (y-axis) as a function of the mean difference (x-axis) and confirmed with Duval and Tweedie s trim and fill test, 36 performed using Comprehensive Meta-analysis Version 2 software (Biostat, Englewood, NJ). A 2-sided P value <0.05 was considered significant. Results One thousand three hundred and forty-seven citations were identified from the literature search strategy, 14 of which met the inclusion criteria, representing a total of 1122 patients (Fig. 1). Table 1 presents the trial characteristics. According to our assessment following the Cochrane Collaboration Risk of Bias tool (Fig. 2), the majority of trials had a low risk of bias. Attempts were made to contact eight authors, and two Identification Regional Regional anesthesia Anaesthetic technique Anesthetic technique Anaesthesia conduction Anesthesia conduction Local anaesthetics Local anesthetics Nerve block Peripheral nerve block Femoral nerve block Adductor canal block Saphenous nerve block Anaesth* as a keyword Anesth* as a keyword Nerve* as a keyword Knee joint Knee surgery Knee replacement Knee arthroplasty Replacement* as a keyword Arthroplast* as a keyword AND 1347 citations Periarticular infiltration Peri-articular infiltration Periarticular injection Peri-articular injection Intraarticular infiltration Intra-articular infiltration Intraarticular injection Intra-articular injection Intraarticular Intra-articular Did not meet inclusion criteria after title review Screening 91 abstracts Did not meet inclusion criteria after abstract review Eligibility 16 full text articles Did not meet inclusion criteria after article review Included 14 full text articles 0 articles retrieved from scanning bibliographies 14 full text articles Fig. 1 PRISMA flow diagram showing literature search results. Fourteen randomized controlled trials were included in the analysis.

4 Table 1 Trial characteristics Reference Group (n) Femoral nerve block Anaesthetic Solution Technique Solution Technique strategy Single-shot injection Ashraf and Ropivacaine 0.2% 150 ml, Infiltration into all layers Ropivacaine 0.2% Ultrasound colleagues, ketorolac 30 mg, ofthekneejoint 30 ml (19), epinephrine 1 mg (total during the surgery Femoral nerve volume 152 ml) (posterior part, block (21) anterior part, periarticular soft tissue) Fan and Ropivacaine 1% 10 ml, Infiltration into all layers Ropivacaine 0.5% Ultrasound General colleagues, sulphate of the knee joint 20 ml combined (79), 10 mg, betamethasone during the surgery with nerve (not detailed) Femoral nerve 5 mg, normal saline (posterior part, stimulation block (78) 38 ml (total volume anterior part, 50 ml) periarticular soft tissue) Moghtadaei Ropivacaine 1% 30 ml, Infiltration into all layers Ropivacaine 1% Nerve and ketorolac 30 mg, of the knee joint 20 ml stimulation anesthesia colleagues, (18), epinephrine 0.5 mg, during the surgery Femoral nerve normal saline ml (posterior part, block (18) (total volume 150 ml) anterior part, periarticular soft tissue) Parvataneni Bupivacaine 0.5% 4080 ml, Infiltration into all layers Not specified Not specified and sulfate 410 of the knee joint anesthesia colleagues, (31), mg, epinephrine 0.3 mg, during the surgery (not detailed) Femoral nerve methylprednisolone (posterior part, block (29) acetate 40 mg, anterior part, cefuroxime 750 mg, periarticular soft normal saline 22 ml tissue) (total volume ml) Uesugi and Ropivacaine 0.75% 20 ml, Infiltration into the Ropivacaine 0.75% Nerve colleagues, 10 mg (male) posterior part of the 20 ml stimulation anesthesia (100), or 5 mg (female), capsule (20 ml) and Femoral nerve epinephrine 0.3 mg, the periarticular soft block (100) dexamethasone 3.3 mg, tissue (2224 ml) normal saline 20 ml (total volume 4244 ml) Continuous infusion and iterative injections Affas and colleagues, (20), Femoral nerve block (20) Ropivacaine 0.2% 150 ml, ketorolac 30 mg, epinephrine 0.5 mg (total volume 156 ml) 30 ml intracutaneously before incision, 80 ml in the posterior part of the capsule and 46 ml infused through an intraarticular catheter placed at the end of the surgery for 24 h Ropivacaine 0.2% 30 ml, followed by 15 ml every 4h for 24 h Nerve stimulation Postoperative Primary outcome Comments undefined nonsteroidal antiinflammatory drugs, i.v. patientcontrolled of, oxycodone Pain scores at 4 - Parecoxib, i.v. patientcontrolled Not specified - of Morphine - ibuprofen, i.v. consumption between 24 and 48 Not specified - celecoxib, ketorolac, oxycodone, i.v. Diclofenac Pain scores in the A sciatic nerve block postoperative was performed period (time in patients period not allocated to specified) femoral nerve block with ropivacaine 0.75% 10 ml. i.v. Pain scores on Patients with a patient-controlled movement femoral nerve of at 24 block received i. v. ketorolac 10 mg every 8h for 24 h (total dose 30 mg). 600 Albrecht et al.

5 Carli and colleagues, Chaumeron and colleagues, Kovalak and colleagues, Kurosaka and colleagues, Kutzner and colleagues, (20), Femoral nerve block (20) (29), Femoral nerve block (30) (29), Femoral nerve block (30) (21), Femoral nerve block (21) (60), Femoral nerve block (60) Ropivacaine 0.2% 100 ml, ketorolac 30 mg, epinephrine 0.5 mg (total volume ml) Ropivacaine 0.2% 150 ml, ropivacaine 1% 10 ml, ketorolac 30 mg, epinephrine 0.5 mg (total volume ml) Levobupivacaine 0.25% 75 ml, epinephrine 0.75 mg (total volume 75 ml) Ropivacaine 0.75% 40 ml, ketoprofen 100 mg, epinephrine 0.5 mg, normal saline 40 ml (total volume 85 ml) Ropivacaine 0.75% 50 ml, sulphate 2 mg, normal saline 148 ml (total volume 350 ml) Infiltration into the posterior part of the capsule (see comments). Second injection through an intra-articular catheter at 24 of an infusion of Ropivacaine 0.5% 50 ml, ketorolac 30 mg, epinephrine 0.25 mg (total volume ml) Infiltration into the posterior part of the capsule and the periarticular soft tissue. Second injection of ropivacaine 1% 15 ml through an intraarticular catheter between 16 and 24 Infiltration into all layers of the knee joint during the surgery (posterior part, anterior part, periarticular soft tissue) Infiltration into all layers of the knee joint during the surgery (posterior part, anterior part, periarticular soft tissue) Continuous infusion through a catheter positioned below the fascia at a rate of 8 ml/h for 44 h (no preliminary infiltration) Ropivacaine 0.2% 8 ml followed by an infusion of ropivacaine 0.2% 8 ml/h for 48h Ropivacaine 0.25% 20 ml, followed by an infusion of ropivacaine 0.2% 8 to 10 ml/h for 48 to 72 h Levobupivacaine 0.25% 10 ml followed by an infusion of levobupivacaine 8 ml/h for 24 h Ropivacaine 0.2% 20 ml followed by an infusion of ropivacaine 0.15% 5 ml/h for 48 h) Continuous infusion of ropivacaine 0.2% at a rate of 8 ml/h for 48 h (no initial bolus injection) Nerve stimulation Nerve stimulation Ultrasound combined with nerve stimulation Ultrasound Not specified (not detailed) General (not detailed) General (not detailed) celecoxib, i.v. patient-controlled of, oxycodone celecoxib, i.v. patient-controlled of for 48 h followed by oral narcotics (hydromorphone, codeine, oxycodone) dexketoprofen, i.v. patient-controlled of tramadol for 24 h followed by oral tramadol Loxoprofen, i.v. patientcontrolled of etoricoxib, oral oxycodone or intramuscular piritramide Cumulative consumption at 48 Cumulative consumption at 48 Not specified Pain scores at 24 Pain scores at 24 All patients received an infiltration of the posterior capsule of the knee with ropivacaine 0.2% 50 ml, ketorolac 30 mg and epinephrine 0.25 mg. Justification was to cover popliteal pain (total volume ml). - Patients with local infiltration had a single injection during the surgery only. Patients with local infiltration had a single injection during the surgery only. - Continued vs nerve block after knee replacement 601

6 Table 1 Continued Reference Group (n) Femoral nerve block Anaesthetic Solution Technique Solution Technique strategy Ng and Ropivacaine 1% 30 ml, Infiltration into all layers Ropivacaine 0.2% Nerve General colleagues, epinephrine 1 mg, of the knee joint 20 ml followed stimulation (16), triamcinolone during the surgery by an infusion of with Femoral nerve acetonide 40 mg, (posterior part, 10 ml/h for 72 h remifentanil block (16) normal saline 70 ml anterior part, (total volume 101,5 ml) periarticular soft tissue) Spangehl Ropivacaine 0.5% 4080 ml, Infiltration into all layers Ropivacaine 0.5% Nerve General and ketorolac 30 mg, of the knee joint 30 ml followed stimulation colleagues, (81), epinephrine mg, during the surgery by an infusion of or (not Femoral nerve sulphate (posterior part, ropivacaine 0.2% ultrasound detailed) block (79) 5 mg, normal saline up anterior part, of 612 ml/h for to a total volume of periarticular soft 48 h 120 ml tissue). Toftdahl and Ropivacaine 0.2% 150 ml, Infiltration into all layers Ropivacaine 1% Nerve colleagues, ketorolac 30 mg, of the knee joint 20 ml, followed stimulation (40), epinephrine 0.5 mg during the surgery by an infusion of Femoral nerve (total volume 152 ml) (posterior part, ropivacaine 0.2% block (37) anterior part, 10 ml/h for 48h periarticular soft with bolus of tissue). Two boluses 20 ml available of 22 ml every 12 h every 8 h for 24 h were injected through the catheter (ropivacaine 1% 20 ml, ketorolac 30 mg and epinephrine 0.5 mg) Postoperative Primary outcome Comments i.v. patient-controlled of diclofenac, gabapentine, i.v. hydromorphone or oral oxycodone ibuprofen, oxycodone, i.v. Cumulative Patients with local infiltration consumption had a at 72 single injection during the surgery only. Pain scores at 24 Patients with local infiltration had a single injection during the surgery only. Oxycodone Patients in group consumption PNB received (time period bupivacaine not specified) 50 mg with 4 mg into the intraarticular drain at the end of surgery (total volume 20 ml). 602 Albrecht et al.

7 vs nerve block after knee replacement 603 Affas et al ref 37 Ashraf et al ref 12 Carli et al ref 13 Chaumeron et al ref 14 Fan et al ref 43 Kovalak et al ref 44 Kurosaka et al ref 45 Kutzner et al ref 46 Moghtadaei et al ref 38 Ng et al ref 39 Parvataneni et al ref 40 Spangehl et al ref 47 Toftdahl et al ref 41 Uesugi et al ref 42 Random sequence generation (selection bias) provided the additional data requested Data were approximated from median and range in one trial, 38 and were not usable in another, 39 as neither standard deviations, confidence interval nor percentiles were reported Five trials compared single-shot injection techniques, whereas nine studied continuous infusion techniques or iterative injections With the exception of four trials where authors used ultrasound guidance and two where it was not specified, all authors performed the FNB with the help of a nerve stimulator only. No trial compared LIA with a saphenous nerve or adductor canal blocks. The total dose of ropivacaine injected for LIA was consistently 300 mg, except in four trials where authors administered bupivacaine 200 to 400 mg, 39 ropivacaine 100 mg 37 or 150 mg, 40 and a continuous infusion of ropivacaine over a period of 44 h. 44 The volumes injected for LIA varied between 150 and 200 ml in five trials, between 100 and 120 ml in three Allocation concealment (selection bias) Blinding of participants and personnel (performance bias) Blinding of outcome assessment (detection bias) Incomplete outcome data (attrition bias) Selective reporting (reporting bias) Other bias Fig. 2 Cochrane collaboration risk of bias summary: evaluation of bias risk items for each included study. Blue circle, low risk of bias; green circle, high risk of bias; pink circle, unclear risk of bias. trials, below 100 ml in five others and 350 ml in one. 44 The solution for LIA contained epinephrine combined with ketoprofen 43 or ketorolac in eight trials, combined with a steroid in four trials, or with in five trials Finally, in one trial the solution contained an antibiotic. 39 Ropivacaine doses for the initial injection of FNB were 60 mg or less in most trials five trials injected ropivacaine 100 mg, mg or 200 mg, while one trial did not specify. 39 There were no differences in i.v. consumption, pain scores at rest or pain scores on movement on postoperative day one (Table 2). Subgroup analysis according to the specialization of the corresponding author (anaesthetist vs orthopaedic surgeon) did not reveal any difference ( consumption: P=0.19; pain scores at rest on postoperative day one: P=0.80; pain scores on movement on postoperative day one: P=0.64). With regards to the funnel plots for our primary outcomes, the Duval and Tweedie s trim and fill test revealed the point estimates for the combined studies to be 0.18 (95% CI: 0.50, 0.13), 0.09 (95% CI: 0.35, 0.16), 0.28 (95% CI: 0.74, 0.18) for i.v. consumption, pain scores at rest, and pain scores on movement on postoperative day one, respectively. These findings suggest an absence of publication bias. The qualities of evidence for our primary outcomes were moderate according to the GRADE working system. Tables 2 and 3 presents secondary acute pain-related outcomes and functional outcomes, respectively. Based on two trials that specifically reported the reduction in i.v. consumption at 12 the difference between groups reached statistical significance, however the clinical impact is questionable. Similarly, range of motion on postoperative day two, Knee Society knee score at six weeks and length of stay were statistically different but direct clinical relevance. There were no significant differences in the other secondary outcomes. No trials aimed to capture chronic postoperative pain. With respect to the occurrence of complications, knee infection was sought by eight trials representing a total of 313 patients in group LIA With the exception of prosthesis loosening or revision surgery that were not captured, individual complications were recorded by three or fewer trials. There were no differences between groups in complication rates (Table 4). No trials reported serum local anaesthetic concentration. Discussion This systematic review and meta-analysis evaluates the postoperative analgesic efficacy and functional outcomes of LIA in comparisontofnb.basedon14randomizedcontrolledtrials and 1122 patients, our results show that both techniques are equivalent in terms of pain and functional outcomes. However, higher doses of local anaesthetic are required within the LIA group. Although the reported complication rate was not different between groups, the small number of trials that specifically sought to capture these outcomes limits the generalizability of this conclusion. The quality of evidence for both the primary and secondary outcomes is moderate to very low because of inconsistency in absolute effects observed for all outcomes and the limited number of trials reporting many outcomes. The desire to optimize care pathways for TKA is a timely issue in many practices. Funding and efficiency concerns encourage the application of evidence-based techniques to improve both early and late outcomes. 48 As an example, the introduction of outpatient TKA requires careful attention to the immediate postoperative issue of with a similar

8 Table 2 Secondary acute pain-related outcomes and opioid-related side-effects. A negative mean difference favours group, and a positive difference, Femoral nerve block group. CI, confidence interval; n, number of events; N, total number of participants; SD, standard deviation Outcomes References Group Mean difference [95% CI] Femoral nerve block or Relative risk [95% CI] Mean or n SD N Mean or n SD N I 2 (%) P value Quality of evidence (GRADE) IV consumption equivalent (mg) Chaumeron [ 6.0, 2.0] 0 < Very low Toftdahl Postoperative day one Affas [ 4.9, 0.9] Moderate Carli Fan Kurosaka Moghtadaei Ng Spangehl Toftdahl Postoperative day two Carli [ 4.3, 2.6] Low Fan Chaumeron Moghtadaei Spangehl Toftdahl Postoperative day three Chaumeron [ 3.9, 5.4] Low Fan Ng Toftdahl Pain scores at rest (analogue scale, 010) two Ashraf [ 2.4, 0.9] Low Chaumeron Moghtadaei Uesugi Moghtadaei [ 0.1, 1.2] Very low Uesugi Postoperative day one Affas [ 0.4, 0.3] Moderate Ashraf Carli Chaumeron Fan Kovalak Kurosaka Kutzner Moghtadaei Ng Spangehl Uesugi Albrecht et al.

9 Postoperative day two Carli [ 0.1, 0.5] Moderate Chaumeron Fan Kovalak Kurosaka Kutzner Ng Spangehl Uesugi Postoperative day three Chaumeron [ 0.1, 0.3] Low Fan Kurosaka Kutzner Ng Pain scores on movement (analogue scale, 010) two Chaumeron [ 3.6, 1.4] Very low Postoperative day one Affas [ 0.5, 0.8] Moderate Carli Chaumeron Fan Kovalak Ng Toftdahl Postoperative day two Carli [ 0.4, 0.3] Low Chaumeron Fan Kovalak Ng Toftdahl Postoperative day three Chaumeron [ 0.1, 0.8] Very low Fan Ng Postoperative nausea and vomiting Fan [0.9, 1.3] Low Kurosaka Moghtadaei Ng Spangehl Toftdahl Uesugi Pruritus Kurosaka [0.5, 1.2] Very low Uesugi Toftdahl vs nerve block after knee replacement 605

10 Table 3 Functional outcomes. A negative mean difference favours Femoral nerve group, and a positive difference, group, except for Length of stay, where the reverse is true. CI, confidence interval; n, number of events; N, total number of participants; SD, standard deviation Outcome Reference Group Mean difference Femoral nerve block [95% CI] Mean or n SD N Mean or n SD N I 2 (%) P value Quality of evidence (GRADE) 606 Albrecht et al. Range of motion (degrees) Postoperative day one Carli [ 11.2, 1.6] Low Chaumeron Kovalak Kutzner Postoperative day two Carli [ 8.8, 2.6] Low Chaumeron Kovalak Kutzner Postoperative day three Carli [ 8.5, 0.2] Low Chaumeron Kutzner Quadriceps muscle strength (pounds) Postoperative day one Ng [ 0.4, 0.1] Very low Postoperative day two Ng [ 0.4, 0.1] Very low Postoperative day three Ng [ 0.2, 0.2] Very low Knee Society score six weeks Carli [ 15.9, 1.6] Very low Kovalak three months Fan [ 1.3, 0.7] Very low Ng months Fan [ 1.2, 0.6] Very low Ng Length of stay (days) Affas [ 0.5, 0.1] Low Carli Chaumeron Fan Kovalak Kutzner Moghtadaei Spangehl

11 vs nerve block after knee replacement 607 Table 4 Complications. CI, confidence interval; n, number of events; N total number of participants Outcome Reference Group Relative risk Femoral [95% CI] nerve block I 2 (%) P value Quality of evidence (GRADE) n N n N Neurologic events Chaumeron [0.1, 4.0] Very low Moghtadaei Uesugi Cardiovascular events Affas [0.4, 33.0] Very low Carli Toftdahl Falls Chaumeron [0.0, 1.8] Low Spangehl Knee infections Carli [0.2, 12.6] Moderate Chaumeron Kovalak Moghtadaei Ng Parvataneni Toftdahl Uesugi attention to quality indicators around functional recovery and complication rates The choice of anaesthetic and analgesic technique may be therefore be critical. However, our results suggest that there is presently no evidence to argue for improved recovery with either LIA or FNB. Apart from the goal of analgesic optimization, peripheral nerve blockade has been questioned because of the potential for motor blockade and suggested potential for falls during the recovery period. Ilfeld and colleagues 51 reported a significant difference in fall rates between placebo infusion vs local anaesthetic (0% vs 7%, P=0.013) during continuous FNB. This risk remains controversial with other trials suggesting no association, 2 including a large administrative database review of 191,570 TKA patients. 52 Although LIA avoids blockade of the motor fibres to the quadriceps muscles, this meta-analysis found no difference in the reported rate of motor related complications, including falls, when compared with FNB. A potential confounder of this comparison was the inclusion of adductor canal studies in the meta-analysis search criteria. However, no trial using this technique met the inclusion criteria for this meta-analysis. Of note, one concern raised with LIA is the large dose of local anaesthetic used during infiltration. Affas and colleagues 53 performed quantitative analysis of plasma ropivacaine concentrations after both LIA and repeated FNB. They identified no difference in maximum plasma concentrations, when similar total doses of ropivacaine were administered, however the median plasma concentration was significantly higher in the LIA group. These authors did not report any signs of clinical local anaesthetic toxicity, a finding consistent with our results. Additional concerns regarding the LIA technique include the wide range of injectate volume between 42 and 350 ml, differences of adjuncts injected, unestablished stability of the mixture and off-label route of administration for certain drugs (e.g., non-steroidal anti-inflammatory drugs). In the absence of a clear rationale behind the mixture administered, animal data and consensus among orthopaedic societies, there is a compelling need for improved standardization of the LIA procedure. Our meta-analysis has several limitations. Although we attempted to explain the observed heterogeneity by grouping results according to the specialization of the corresponding author, heterogeneity remained moderate. This finding may potentially be explained by the large variations in volumes and adjuncts injected for LIA. Except for pain outcomes on postoperative day one, fewer than half of the included trials reported the same acute-pain related outcomes. In addition, functional outcomes were frequently not reported. Consequently, the impact of each intervention on functional outcomes during postoperative recovery remains undetermined. Additional research comparing both techniques with an eye to intermediate and long-term functional recovery, would be a highly valuable addition to the current literature. In conclusion, LIA provides similar postoperative analgesic efficacy to FNB after total knee arthroplasty, but requires a higher dose of local anaesthetic. The reported incidence of complications does not differ between groups although systemic concentrations of local anaesthetic cannot be determined from the included trials. The degree of results heterogeneity suggests that these findings should be interpreted with caution. Similarly, the number of included trials for many outcomes remains small and there is a need for more comprehensive standardized comparative evaluations of functional outcomes after these two techniques. Authors contributions Study design/planning: E.A. Study conduct: E.A., A.J.G., K.R.K. Data analysis: E.A., A.J.G. Writing paper: E.A., K.R.K. Revising paper: all authors Acknowledgments We are grateful to Mrs Isabelle von Kaenel for the assistance in the literature search. Head librarian, Lausanne University Hospital, Lausanne, Switzerland.

12 608 Albrecht et al. Declaration of interest E.A. has received grants from the Swiss Academy for Anaesthesia Research (SACAR), Lausanne, Switzerland (no grant numbers attributed) and from B.Braun Medical AG, Sempach, Switzerland (no grant numbers attributed). Funding This work was supported by departmental funding (Department of Anaesthesia, Lausanne University Hospital, Lausanne, Switzerland). References 1. Gerbershagen HJ, Aduckathil S, van Wijck AJ, Peelen LM, Kalkman CJ, Meissner W. Pain intensity on the first day after surgery: a prospective cohort study comparing 179 surgical procedures. Anesthesiology 2013; 118: Albrecht E, Morfey D, Chan V, et al. Single-injection or continuous femoral nerve block for total knee arthroplasty Clin Orthop Relat R 2014; 472: Paul JE, Arya A, Hurlburt L, et al. Femoral nerve block improves outcomes after total knee arthroplasty: a metaanalysis of randomized controlled trials. Anesthesiology 2010; 113: Fischer HB, Simanski CJ, Sharp C, et al. A procedure-specific systematic review and consensus recommendations for postoperative following total knee arthroplasty. Anaesthesia 2008; 63: Hadzic A, Houle TT, Capdevila X, Ilfeld BM. Femoral nerve block for in patients having knee arthroplasty. Anesthesiology 2010; 113: HannibalK,GalatiusH,HansenA,ObelE,EjlersenE.Preoperative wound infiltration with bupivacaine reduces early and late opioid requirement after hysterectomy. Anesth Analg 1996; 83: Zohar E, Fredman B, Phillipov A, Jedeikin R, Shapiro A. The analgesic efficacy of patient-controlled bupivacaine wound instillation after total abdominal hysterectomy with bilateral salpingo-oophorectomy. Anesth Analg 2001; 93: Horn EP, Schroeder F, Wilhelm S, et al. Wound infiltration and drain lavage with ropivacaine after major shoulder surgery. Anesth Analg 1999; 89: Bianconi M, Ferraro L, Traina GC, et al. Pharmacokinetics and efficacy of ropivacaine continuous wound instillation after joint replacement surgery. Br J Anaesth 2003; 91: Rostlund T, Kehlet H. High-dose local infiltration after hip and knee replacementwhat is it, why does it work, and what are the future challenges Acta Orthop 2007; 78: Kerr DR, Kohan L. : a technique for the control of acute postoperative pain following knee and hip surgery: a case study of 325 patients. Acta Orthop 2008; 79: Ashraf A, Raut VV, Canty SJ, McLauchlan GJ. Pain control after primary total knee replacement. A prospective randomised controlled trial of local infiltration versus single shot femoral nerve block. Knee 2013; 20: Carli F, Clemente A, Asenjo JF, et al. Analgesia and functional outcome after total knee arthroplasty: periarticular infiltration vs continuous femoral nerve block. Br J Anaesth 2010; 105: Chaumeron A, Audy D, Drolet P, Lavigne M, Vendittoli PA. Periarticular injection in knee arthroplasty improves quadriceps function. Clin Orthop Relat R 2013; 471: Andersen LO, Kehlet H. Analgesic efficacy of local infiltration in hip and knee arthroplasty: a systematic review. Br J Anaesth 2014; 113: Fowler SJ, Christelis N. High volume local infiltration compared to peripheral nerve block for hip and knee arthroplasty-what is the evidence Anaesth Intensive Care 2013; 41: Gibbs DM, Green TP, Esler CN. The local infiltration of following total knee replacement: a review of current literature. J Bone Joint Surg Br 2012; 94: Kehlet H, Andersen LO. in joint replacement: The evidence and recommendations for clinical practice. Acta Anaesthesiol Scand 2011; 55: Marques EM, Jones HE, Elvers KT, Pyke M, Blom AW, Beswick AD. Local anaesthetic infiltration for peri-operative pain control in total hip and knee replacement: systematic review and meta-analyses of short- and long-term effectiveness. BMC Musculoskelet Disord 2014; 15: Yun XD, Yin XL, Jiang J, et al. versus femoral nerve block in total knee arthroplasty: A metaanalysis. Orthop Trauma Surg R 2015; 101: Schulz KF, Altman DG, Moher D. CONSORT 2010 statement: updated guidelines for reporting parallel group randomised trials. PLoS Med. 2010; 7: e Meftah M, Wong AC, Nawabi DH, Yun RJ, Ranawat AS, Ranawat CS. Pain management after total knee arthroplasty using a multimodal approach. Orthopedics 2012; 35: e Yadeau JT, Goytizolo EA, Padgett DE, et al. Analgesia after total knee replacement: local infiltration versus epidural combined with a femoral nerve blockade: a prospective, randomised pragmatic trial. Bone Boint J 2013; 95-B: Reinhardt KR, Duggal S, Umunna BP, et al. Intraarticular versus epidural plus femoral nerve block after TKA: a randomized, double-blind trial. Clin Orthop Relat R 2014; 472: Mahadevan D, Walter RP, Minto G, Gale TC, McAllen CJ, Oldman M. Combined femoral and sciatic nerve block vs combined femoral and periarticular infiltration in total knee arthroplasty: a randomized controlled trial. J Arthroplasty 2012; 27: Safa B, Gollish J, Haslam L, McCartney CJ. Comparing the effects of single shot sciatic nerve block versus posterior capsule local anesthetic infiltration on and functional outcome after total knee arthroplasty: a prospective, randomized, double-blinded, controlled trial. J Arthroplasty 2014; 29: Tanikawa H, Sato T, Nagafuchi M, Takeda K, Oshida J, Okuma K. Comparison of local infiltration of and sciatic nerve block in addition to femoral nerve block for total knee arthroplasty. J Arthroplasty 2014; 29: Albrecht E, Kirkham KR, Liu SS, Brull R. Peri-operative intravenous administration of magnesium sulphate and postoperative pain: a meta-analysis. Anaesthesia 2013; 68: Dinjens RN, Senden R, Heyligers IC, Grimm B. Clinimetric quality of the new 2011 Knee Society score: high validity, low completion rate. Knee 2014; 21: Higgins JP, Altman DG, Gotzsche PC, et al. The Cochrane Collaboration s tool for assessing risk of bias in randomised trials. Br Med J 2011; 343: d Collaboration TC. Cochrane Handbook for systematic reviews of interventions version Available from cochrane.org/revman (accessed January 2016)

13 vs nerve block after knee replacement Skaer TL. Practice guidelines for transdermal opioids in malignant pain. Drugs 2004; 64: Silvasti M, Svartling N, Pitkanen M, Rosenberg PH. Comparison of intravenous patient-controlled with tramadol versus after microvascular breast reconstruction. Eur J Anaesthesiol 2000; 17: Balshem H, Helfand M, Schunemann HJ, et al. GRADE guidelines: 3. Rating the quality of evidence. J. Clin. Epidemiol. 2011; 64: Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med 2002; 21: Duval S, Tweedie R. Trim and fill: A simple funnel-plot-based method of testing and adjusting for publication bias in metaanalysis. Biometrics 2000; 56: Fan L, Yu X, Zan P, Liu J, Ji T, Li G. Comparison of local Infiltration with femoral nerve block for total knee arthroplasty: a prospective, randomized clinical trial. J Arthroplasty 2015; 30: Moghtadaei M, Farahini H, Faiz SH, Mokarami F, Safari S. Pain Management for Total Knee Arthroplasty: Single-Injection Femoral Nerve Block versus Local Infiltration Analgesia. Iranian Red Crescent Me 2014; 16: e Parvataneni HK, Shah VP, Howard H, Cole N, Ranawat AS, Ranawat CS. Controlling pain after total hip and knee arthroplasty using a multimodal protocol with local periarticular injections: a prospective randomized study. J Arthroplasty 2007; 22: Uesugi K, Kitano N, Kikuchi T, Sekiguchi M, Konno S. Comparison of peripheral nerve block with periarticular injection after total knee arthroplasty: A randomized, controlled study. Knee 2014; 21: Affas F, Nygards EB, Stiller CO, Wretenberg P, Olofsson C. Pain control after total knee arthroplasty: a randomized trial comparing local infiltration anesthesia and continuous femoral block. Acta Orthop 2011; 82: KovalakE,DoganAT,UzumcugilO,et al. A comparison of continuous femoral nerve block and periarticular local infiltration in the management of early period pain developing after total knee arthroplasty. Acta Orthop Traumato 2015; 49: Kurosaka K, Tsukada S, Seino D, Morooka T, Nakayama H, Yoshiya S. versus continuous femoral nerve block in pain relief after total knee arthroplasty: a randomized controlled trial. J Arthroplasty 2015; doi: /j.arth [Epub ahead of print] 44. Kutzner KP, Paulini C, Hechtner M, Rehbein P, Pfeil J. Postoperative after total knee arthroplasty: Continuous intra-articular catheter vs. continuous femoral nerve block. Orthopade 2015; 44: Ng FY, Ng JK, Chiu KY, Yan CH, Chan CW. Multimodal periarticular injection vs continuous femoral nerve block after total knee arthroplasty: a prospective, crossover, randomized clinical trial. J Arthroplasty 2012; 27: Spangehl MJ, Clarke HD, Hentz JG, Misra L, Blocher JL, Seamans DP. The Chitranjan Ranawat Award: Periarticular injections and femoral & sciatic blocks provide similar pain relief after TKA: a randomized clinical trial. Clin Orthop Relat R 2015; 473: Toftdahl K, Nikolajsen L, Haraldsted V, Madsen F, Tonnesen EK, Soballe K. Comparison of peri- and intraarticular with femoral nerve block after total knee arthroplasty: a randomized clinical trial. Acta Orthop 2007; 78: Kehlet H, Joshi GP. Systematic Reviews and Meta-Analyses of Randomized Controlled Trials on Perioperative Outcomes: An Urgent Need for Critical Reappraisal. Anesth Analg 2015; 121: Cross MB, Berger R. Feasibility and safety of performing outpatient unicompartmental knee arthroplasty. Int Orthop 2014; 38: Lovald S, Ong K, Lau E, Joshi G, Kurtz S, Malkani A. Patient selection in outpatient and short-stay total knee arthroplasty. J Surg Orthop Adv 2014; 23: Ilfeld BM, Duke KB, Donohue MC. The association between lower extremity continuous peripheral nerve blocks and patient falls after knee and hip arthroplasty. Anesth Analg 2010; 111: Memtsoudis SG, Danninger T, Rasul R, et al. Inpatient falls after total knee arthroplasty: the role of anesthesia type and peripheral nerve blocks. Anesthesiology 2014; 120: Affas F, Stiller CO, Nygards EB, Stephanson N, Wretenberg P, Olofsson C. A randomized study comparing plasma concentration of ropivacaine after local infiltration and femoral block in primary total knee arthroplasty. Scand J Pain 2012; 3: 4651 Handling editor: J. G. Hardman