The impact of peri-operative intravenous lidocaine on postoperative outcome after elective colorectal surgery A meta-analysis of randomised controlled trials

s excluded (n = 385) Not colorectal surgery (n = 104) Not ReT (n = 62) Animal studies (n = 76) Systematic review or meta-analysis (n = 49) letter, editorial, review (n = 33) No relevant clinical outcomes (n = 37) Paediatric cases (n = 24) Infusion commenced postoperatively (n = 2) No control group (n = 2) Included emergency cases (n = 3) No relevant clinical outcomes (n = 5) Full texts excluded (n = 12) Records screened (n = 22) Full text assessed for eligibility (n = 22) Studies included in quantitative evidence synthesis (n = 10) Additional records identified through other sources (n = 39) In cl u d ed S cr ee n in g E lig ib ili ty Id en ti fi ca ti o n PRISMA diagram of studies considered for inclusion from the initial literature search. Eur J Anaesthesiol 2020; 37:1–12 CE: Swati; EJA/EJA-D-19-00766; Total nos of Pages: 12;


Introduction
There has been significant interest in the potential for perioperative intravenous lidocaine (IVL) infusion in patients undergoing abdominal surgery, 1 with increasing evidence in a range of surgical specialties, including colorectal surgery, 2 hepatobiliary surgery, 3 obstetrics 4 and gynaecology. 5 Studies have suggested that IVL conveys a postoperative benefit in terms of its analgesic, anti-inflammatory and opioid-sparing effects, resulting in reduced postoperative pain, reduced time to return of gastrointestinal function, and reduced nausea and vomiting. 6 In addition, there is some evidence for a beneficial effect on intra-operative parameters, including a reduction in anaesthetic requirement 7,8 and improved recovery from general anaesthesia. 9 A Cochrane review 1 compared IVL infusion with placebo or thoracic epidural analgesia for postoperative pain and recovery in a total of 68 randomised controlled trials (RCTs) across a range of surgical specialties, including general, spinal, endocrine and cardiothoracic surgery, and gynaecology.This demonstrated an unclear effect of IVL vs. placebo on pain scores, recovery of gastrointestinal function, postoperative nausea and overall opioid requirement, and highlighted poor quality evidence. This review 1 did not seek to differentiate the benefits in individual surgical specialities such as colorectal alone, although a comparison was made between open and laparoscopic abdominal surgery. This included a range of nongastrointestinal procedures so is not directly comparable to the aims of the current meta-analysis.
Specific to the field of colorectal surgery, a recently published systematic review 10 examined the role of IVL in the setting of elective colorectal surgery and concluded that this provided limited benefit in the reduction of early postoperative pain and morphine requirement when compared with placebo, and a variable degree of improvement when compared with epidural analgesia. This has been followed by a meta-analysis within colorectal surgery, which found improved time to recovery of gastrointestinal function as well as reduced hospital length of stay. 11 The recently published Enhanced Recovery After Surgery (ERAS) Society Recommendations for peri-operative care in elective colorectal surgery 12 have concluded that although 'the use of lidocaine infusions to reduce opioid use and nausea in colorectal surgery is now well established', the benefit in terms of reduction of postoperative ileus is unclear. Despite this, no meta-analysis has been conducted to date to assess the role of combined intra-operative and postoperative IVL in colorectal surgery specifically. In addition, substantial additional evidence has been published 2,13,14 since the previous systematic review. 10 The aims of this meta-analysis were to examine the effect of peri-operative IVL on postoperative outcome in patients undergoing elective colorectal surgery, including postoperative pain, morphine consumption, time to return of gastrointestinal function, hospital length of stay and complications; to study the role of peri-operative lidocaine infusion in laparoscopic vs. open elective colorectal surgical procedures; to identify the optimal dosing and infusion regimen as well as duration of infusion; and to determine the incidence of local anaesthetic toxicity.

Materials and methods
Search strategy A search of PubMed, Scopus and the Cochrane Library databases was conducted up to 5 November 2018 in order to identify full-text studies examining the impact of perioperative IVL in elective colorectal surgery. The electronic search terms adopted were (intravenous OR infusion) AND (lidocaine OR lignocaine) AND (colon OR rectal OR colorectal OR proctectomy OR colonic), with no limitation placed on data or language for inclusion. The bibliographies of all studies that met the inclusion criteria were hand-searched for any additional suitable articles and relevant conference abstracts to ensure study inclusion was as complete as possible, and this accounts for the additional 39 manuscripts identified from other sources. The meta-analysis was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. 15 Selection of articles Following the exclusion of initial studies on the basis of article title and abstract by two independent researchers (KR and HJE), the remaining full-text articles were screened in detail for inclusion. Studies were included if they were RCTs that evaluated the role of peri-operative intravenous IVL infusion vs. placebo in adult patients undergoing elective colorectal surgery. Exclusion criteria comprised paediatric patients, noncolorectal or emergency procedures, non-RCT methodology or lack of any relevant clinical outcome measures. Studies that included more than two study arms, but had IVL and placebo groups, were included and only those groups pertinent to this metaanalysis were considered. No consideration was given to how long the lidocaine infusion was continued after surgery, but to be eligible for inclusion, the infusion had to commence before the surgical incision.

Data extraction
Study data were extracted from the included RCTs by one author (KER) and checked by another (HJE). The primary outcome measure was postoperative pain scores up to 48 h [at rest and movement scored on the visual analogue scale (VAS)]. In line with the previous Cochrane meta-analysis 1 and reflecting the International Association for the Study of Pain (IASP) recommendations, we have considered a 1-cm difference in VAS (on a 0 to 10 cm scale) as a clinically, rather than just statistically, significant difference. Secondary outcome measures included time to return of gastrointestinal function, both in terms of flatus and defecation, postoperative morphine requirement, incidence of prolonged ileus, surgical site infection, anastomotic leak, signs of local anaesthetic toxicity and overall hospital length of stay. In addition, data were collated on patient baseline demographics[(age, sex, American Society of Anesthesiologists' (ASA) physical status], operative variables (operating time, estimated blood loss, nature of colorectal resection and indication for surgery) and details of the lidocaine infusion (dose, starting point, duration postoperatively and any bolus dose administered) as well as the details of the placebo administered. The studies included were stratified according to whether the patients underwent open or laparoscopic resection. If the data necessary for meta-analysis of continuous variables were not available, the corresponding author was approached to provide the raw data, and if a response was not received, the technique described by Hozo et al. 16 was employed to estimate the mean and standard deviation from the median and interquartile range [IQR].
Where results were available only in graphical format and the authors did not respond to the request for raw data, data were extracted in either direct or indirect form using plotdigitizer (www.plotdigitizer.sourceforge.net). Where opioid drugs other than morphine were provided by the study, previously described conversion methods were used to standardise all opiates to an equivalent morphine dose. 17 Risk of bias was assessed using the Cochrane Collaboration tool, 18 which focuses upon random sequence generation (selection bias), allocation concealment (selection bias), blinding of participants and personnel (performance bias), blinding of outcome assessment (detection bias), incomplete outcome data (attrition bias) and selective reporting (reporting bias).

Statistical analysis
Statistical analysis was performed using RevMan 5.3 18 (Cochrane, London, UK). Dichotomous variables were analysed using the Mantel-Haenszel random effects model and quoted as a risk ratio with 95% CIs. Continuous variables were analysed using the inverse-variance random effects model and quoted as a mean difference with 95% CI. Data were used to construct forest plots, with a P value less than 0.05 on two-tailed testing indicating a statistically significant difference. Study heterogeneity and inconsistency were assessed using the I 2 statistic, 19 with 25% or less representing low heterogeneity, 25 to 50% as moderate and more than 50% high heterogeneity.

Protocol registration
The protocol for this meta-analysis was registered with the PROSPERO database (www.crd.york.ac.uk/prospero) -registration no. CRD42018115916.

Results
The initial literature search identified a total of 489 potentially eligible full-text articles, of which a total of 10 studies were included in the meta-analysis 2,13,20-27 ( Fig. 1). Two studies initially identified as potentially eligible were excluded subsequently because the Peri-operative lidocaine in elective colorectal surgery 3 lidocaine infusion was not commenced until after the completion of surgery. 14,28 Overall, the risk of bias within the included studies was moderate and the quality of the studies was similarly moderate (Fig. 2). In the assessment of publication bias, markers of imprecision and inconsistency (as estimated from the sample size and CI of the effect sizes), and the quality of evidence was low to moderate, in keeping with the recent Cochrane metaanalysis 1 on a similar topic.
VAS score on coughing at 48 h postoperatively was considered in six studies, 2,23-27 three in laparoscopic surgery 2,24,26 and three in open procedures. 23,25,27 At 48 h postoperatively, IVL did not significantly affect VASrated pain on coughing in any of the groups (Fig. 4).
The time to passage of flatus was considered in eight studies 2,13,21,23-27 including a total of 398 patients. Overall, IVL infusion was not associated with a significant difference in the time to passage of flatus (mean difference À5.33 h, 95% CI À11.53 to 0.88, P ¼ 0.09, I 2 ¼ 90%; Fig. 5). When the four studies in laparoscopic surgery were considered, an IVL infusion (n ¼ 201) was not associated with any difference in the time to passage of flatus (mean difference À3.78 h, 95% CI À12.88 to 5.32, P ¼ 0.42, I 2 ¼ 87%). However, in patients undergoing open surgery (four studies; n ¼ 197), IVL was associated with a significant reduction in the time to passage of flatus of 8.4 h (95% CI À13.7 to À3.1, P ¼ 0.002, I 2 ¼ 31%).
Time to defecation was examined in seven studies including a total of 378 patients. Overall, the use of IVL was associated with a significant reduction in time to defecation (mean difference À12.06 h, 95% CI À17.83 to À6.29, P ¼ 0.0001, I 2 ¼ 93%; Fig. 5). When the five studies involving laparoscopic surgery were considered (n ¼ 261), IVL was associated with a significant reduction in the time to defecation (mean difference À12.33 h, 95% CI À18.63 to À6.03, P ¼ 0.0001, I 2 ¼ 96%). However, no difference was seen in patients undergoing open surgery (mean difference À11.04 h, 95% CI À23.56 to 1.48, P ¼ 0.08, I 2 ¼ 0%), although this was based upon data from two studies alone including a total of 107 patients.
The incidence of prolonged postoperative ileus was considered in four studies, 2,21,22,25 of which three were conducted in patients undergoing laparoscopic surgery 2,21,22 and one in open surgery. 25 Overall, the use of IVL was not associated with any difference in the incidence of prolonged postoperative ileus in either group (Fig. 5).
A total of five studies considered the time to tolerance of oral intake, 2,20-22,25 of which four studies were conducted in laparoscopic surgery (253 patients) and one in open procedures (60 patients) (Fig. 6). Overall, the use of IVL was not associated with a significant difference in time to tolerance of enteral intake (mean difference À3.02 h, 95% CI À6.06 to 0.01, P ¼ 0.05), and in just those undergoing laparoscopy (mean difference À1.96 h, 95% CI À4.97 to 1.04, P ¼ 0. 20). No meta-analysis was conducted on those undergoing open surgery, as this only considered data from a single study.
Only two studies 21,22 included data on the incidence of surgical site infection (n ¼ 128). IVL was not associated with any difference in the incidence of surgical site            25 Both overall and in those undergoing laparoscopic surgery, the use of IVL was not associated with any difference in the incidence of anastomotic leak. Due to a lack of data, no meta-analysis was performed in those undergoing open surgery.
Six studies examined the incidence of local anaesthetic toxicity in IVL vs. placebo including 300 patients, 2,13,20,22,24,27 with four studies 2,20,22,24 conducted on patients undergoing laparoscopic surgery and two in those undergoing open procedures. 13,27 Only one study 2 had any incidence of local anaesthetic toxicity, with all other studies reporting no events, hence a meta-analysis was not conducted on this outcome. There was only one case of local anaesthetic toxicity, which was mild in nature, with a metallic taste and tinnitus.

Discussion
This meta-analysis of 10 RCTs examining the impact of IVL vs. placebo in elective colorectal surgery has demonstrated that IVL is associated with a statistically significant reduction in VAS pain scores at rest at early time points (4, 12 and 24 h) and pain scores on coughing at 12 h only. However, when these are interpreted in light of the IASP threshold for a clinically significant difference in VAS-rated pain scales, these cannot be interpreted as clinically significant results. In addition, there was a significant reduction in time to defecation and hospital LOS in those who underwent IVL infusion vs. placebo.
No difference was seen in the time to passage of flatus, pain at rest at 48 h postoperatively, pain on coughing at 4, 24 and 48 h postoperatively, overall morphine consumption or complication rates (Supplementary Digital Content Table 3, http://links.lww.com/EJA/A277). However, the results should be interpreted with some caution in respect of the time to first flatus and hospital length of stay, as these differences are almost entirely the result of one or two studies (Kuo et al. 27 for the time to first flatus and Kaba et al. 26 and Tikuisis et al. 22 for the hospital LOS). This renders the conclusion potentially weaker and further evidence is needed to provide a more definitive answer. When studies in laparoscopic surgery were analysed, the benefits of peri-operative IVL were more pronounced. IVL was associated with a significant reduction in time to defecation, pain at rest, and at 12 and 24 h, pain on coughing at 12 h and overall hospital LOS, but no difference in pain scores at any other time point, time to return of flatus, morphine requirement or surgical complications. However, again, the differences in pain score cannot be considered clinically significant. When open surgery studies were considered, IVL was associated with a significant reduction in time to flatus but not faeces, pain at rest at early time points (4, 12 and 24 h) and on coughing at 12 h only, but there was no difference in the hospital LOS or morphine requirement.
The details of the lidocaine infusion differed greatly between studies included within this meta-analysis. In terms of commencement of the IVL infusion, one study commenced 30 min before the start of surgery, 27 five studies commenced before induction of anaesthesia, 20,22,23,25,26 one started at the time of induction 2 and three commenced after induction or at the time of skin incision. 13,21,24 In addition, there was a degree of variability in the dosage of lidocaine. In terms of the lidocaine bolus, all but one study 24 administered a bolus before commencing the infusion, with the most common dosage being 1.5 mg kg À1 , which was used in seven studies, 2,13,20,22,23,25,26 with one study 27 administering 2 mg kg À1 , and one 21 1 mg kg À1 . This was also reflected in differing doses for the IVL infusion, with the most common rate of 2 mg kg À1 h À1 being administered in four studies, 20,22,23,25 followed by 1 mg kg À1 h À1 in two studies, 13,21 3 mg kg À1 h À1 in one study 27 and 1.5 mg kg À1 h À1 in one study. 2 One study administered a weight-dependent rate of infusion, 24 and one study reduced the infusion rate between intra-operative and postoperative stages. 26 There was also variability in the infusion duration; three studies administered the infusion only intra-operatively, 20,23,27 and three commenced intra-operatively and continued for 24 h. 21,22,26 Two studies discontinued the infusion at 4 h postoperatively, 2,25 one after 48 h 13 and one 24 on return of gastrointestinal function or at day 5, whichever was sooner. The level of heterogeneity between different infusions is a potentially significant source of bias. There are insufficient data for the individual regimens to separately analyse these and draw meaningful conclusions. There were significant levels of heterogeneity observed in the meta-analyses. Overall, 10 of the 14 analyses demonstrated high levels of heterogeneity, with a similar level in the analyses concerning laparoscopic surgery (n ¼ 10/15). This was lower when studies of open surgery alone were analysed (n ¼ 5/12), but this remains a significant confounder.
One previous systematic review 10 has examined the impact of IVL in elective colorectal surgery, which included five RCTs comparing IVL and placebo. That demonstrated that IVL provided limited benefit in the reduction of early postoperative pain and morphine consumption. No meta-analysis of the available data was conducted, and since its publication, two further RCTs 2,13 have been published that are included in the current meta-analysis. In addition, a meta-analysis of nine RCTs 11 has recently been published focusing on the time to return of gastrointestinal function following IVL in patients undergoing colorectal surgery. That meta-analysis did not include two RCTs, 2,20 which were potentially eligible, and also included one RCT which administered the lidocaine in the postoperative setting only, 28 which the current meta-analysis excluded due to the potential heterogeneity that this causes. The findings of the current meta-analysis mirror these previous studies in their observation of a statistically significant improvement in early postoperative pain and time to return of gastrointestinal function in those receiving IVL. However, the current meta-analysis did not find any difference in morphine consumption rates, contrary to the previous systematic review. In addition, no conclusions were drawn regarding the relative effectiveness of IVL in open vs. laparoscopic surgery by the previous systematic review.
A recently updated meta-analysis 1,29 examined the role of IVL in all branches of surgery which included a total of 68 RCTs, finding IVL to be associated with a statistically significant benefit in terms of pain scores. However, this did not meet the threshold for a clinically relevant difference. A significant difference was also seen in gastrointestinal recovery, hospital length of stay and opioid requirement. This Cochrane review 1 and a recently published commentary article 30 make note of the significant level of heterogeneity introduced by including studies conducted in a range of surgical specialties, which is avoided in the current meta-analysis. There is also significant variability in the details of the IVL infusion, which are also relevant to the current metaanalysis and are due to a lack of standardisation in the literature.

Conclusion
This meta-analysis provides support for the administration of peri-operative IVL in elective colorectal surgery, in terms of earlier return of gastrointestinal function and reduced hospital LOS, with no difference in complication rates or apparent issues surrounding local anaesthetic toxicity. The pain scores at early time points were significantly lower in those receiving the IVL infusion, although this did not meet the threshold for a clinically relevant difference. This meta-analysis supports the use of peri-operative IVL infusion as a good choice for analgesia, particularly in laparoscopic colorectal surgery, with benefits most pronounced in the first 24 h following surgery, which may help promote early mobilisation and nutrition, key components of enhanced recovery pathways. There remains the option to give regular multimodal analgesia alongside IVL in the postoperative setting, thus reserving opioid analgesia for breakthrough pain alone. Further research is needed to explore whether the anti-inflammatory and natural killer cell effects result in improved oncological outcomes. In addition, further evidence would be beneficial in light of the high levels of study heterogeneity as well as to understand the optimal dosage and timing of the IVL infusion in order to standardise further studies and maximise the potential benefits of its administration.