Postoperative pain after different doses of remifentanil infusion during anaesthesia: a meta-analysis

Background This meta-analysis aimed to explore the correlation between the different doses of remifentanil-based anaesthesia and postoperative pain in randomised trials. Methods The electronic databases including PubMed, Cochrane, clinical trial registries, and Google Scholar were searched up to November 2022 for randomised controlled trials (RCTs) that assessed the dose dependent efficacy of remifentanil for postoperative pain intensity and hyperalgesia. Results 31 studies involving 2019 patients were included for analysis. Compared with the high remifentanil dose administration, patients in low doses showed less postoperative pain intensity at 1-2 h (weighted mean differences (WMD): 0.60, 95% CI, 0.05 to 1.15), 3-8 h (WMD: 0.38, 95% CI, 0.00 to 0.75), 24 h (WMD: 0.26, 95% CI, 0.04 to 0.48) and 48 h (WMD: 0.32, 95% CI, 0.09 to 0.55). Remifentanil-free regimen failed to decrease the pain score at 24 h (WMD: 0.10, 95% CI, -0.10 to 0.30) and 48 h (WMD: 0.15, 95% CI, -0.22 to 0.52) in comparison with remifentanil-based anaesthesia. After excluding trials with high heterogeneity, the dose of the remifentanil regimen was closely correlated with the postoperative pain score (P=0.03). In addition, the dose of the remifentanil regimen was not associated with the incidence of postoperative nausea and vomiting (PONV) (P=0.37). Conclusions Our meta-analysis reveals that the low dose of remifentanil infusion is recommendable for general anaesthesia maintenance. No evidence suggests that remifentanil-free regimen has superiority in reducing postoperative pain. Moreover, remifentanil doesn’t have a dose dependent effect in initiating PONV. Trial registration The protocol of present study was registered with PROSPERO (CRD42022378360). Supplementary Information The online version contains supplementary material available at 10.1186/s12871-023-02388-3.


Introduction
Opioids are commonly used to alleviate perioperative pain during surgery.However, opioid, especially remifentanil use, can cause opioid tolerance and induce paradoxical pain [1].Remifentanil was associated with primary and secondary hyperalgesia and can lead to opioid addiction.
In recent years, opioid-free general anaesthesia has been introduced to avoid unexpected pain.Opioidfree anaesthesia using dexmedetomidine or propofol [2][3][4] has been associated with less postoperative pain, † Xinyi Huang, Jinxia Cai and Zhu Lv These authors contributed equally to this manuscript.
resulting in less postoperative opioid consumption.However, the absence of remifentanil or other opioids during the surgery increases the amount of sedative infusion and results in delayed recovery [5].
It was found that remifentanil had a dose-dependent correlation with postoperative pain threshold [6,7].However, these findings are contrary to those of other studies that did not show an effect of the intraoperative opioid dose on postoperative pain intensity and rescue morphine consumption [8].There is still a pending question about whether remifentanil infusion should be abandoned.There are a limited number of studies that have evaluated different doses of remifentanil and their relationship to postoperative pain intensity.
This meta-analysis aimed to evaluate dose dependent effect of remifentanil on the postoperative analgesic effect, secondary hyperalgesia, and side effects after general anaesthesia.

Materials and methods
This meta-analysis of randomised, controlled trials (RCTs) was performed in accordance with the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (Additional file 1).The protocol was registered with PROSPERO (CRD42022378360).

Eligibility criteria
Inclusion Criteria: randomised controlled trials were based on remifentanil anaesthesia or remifentanil free anaesthesia and focused on postoperative pain intensity and hyperalgesia in adults.
In order to exclude the impact of dexmedetomidine, the trials with dexmedetomidine only used in the remifentanil-free group were not included in the metaanalysis.So the exclusion criteria were as follows: dexmedetomidine was only applied in remifentanil-free group, general anaesthesia with epidural analgesia or nerve block, observational studies, non-randomised controlled trials, studies published as abstracts, duplicate articles, populations with chronic opioid use, and articles reporting no indispensable data.

Search strategy
PubMed, Cochrane, clinical trial registries, and Google Scholar were searched to retrieve studies published up to November 2022 without language restrictions (by XH and JS).The following search string was used ("remifentanil" OR "remifentanyl" OR "opioid" OR "opiate") AND ("hyperalgesia" OR "hyperalgesia" OR "hyperalgesias" OR "hyperanalgesia" OR "nociception" OR "nociceptive" OR "pronociception" OR "pronociceptive" OR "allodynia" OR "tolerance") (Full links are given in Additional file 2).The searches were limited to human trials.A manual search of the references listed in the reports and reviews was performed.We reviewed the trial registries when available.In the case of secondary publications, the original papers were reviewed.

Selection of included studies
Three reviewers (XH, JC, and ZL) independently screened the titles and abstracts obtained by the literature search.The remaining full texts were independently retrieved and evaluated by the authors to determine whether the retrieved trials met the inclusion criteria.Disagreements were discussed among the investigators to reach a consensus.

Data extraction
The following data were extracted from the included studies: participant demographics, type of surgery, anaesthetic selection, intraoperative remifentanil regimens, pain scores at all reported times, postoperative allodynia, time to the first analgesic request, and opioid-related side effects.Pain scores on different scales were converted to a standardized 0-10 analogue scale.Any differences resulting from discrepant assessments during data extraction and analysis were resolved through discussion among the study authors.Data reported in the form of a graph were extracted with the assistance of graphics processing software (Web plot Digitalise, HTML5 Software, University of Notre Dame, USA).
Secondary outcomes: Periincisional wound allodynia and forearm allodynia, time to first postoperative analgesic requirement, postoperative consumption of rescue analgesics in milligrams of morphine equivalence, postoperative nausea and vomiting (PONV), and postoperative shivering.
Assessment of Methodological Quality and Risk of Bias: The risk of bias was independently assessed using the Cochrane Collaboration tool [9].Studies with a dropout rate of less than 20% were considered "low-risk" of attrition bias; otherwise, they were assessed as "high risk of bias"."Other potential sources of bias" were assessed as high-risk in studies that had fewer than 15 participants per arm.However, there is currently no consensus on the trial size in this setting.
For trials that did not report the results in the form of mean ± standard deviation (SD), the corresponding authors were contacted thrice by mail to supply the missing data.If no response was obtained, the sample size (n), median (m), minimum value (a), first quartile (q1), third quartile (q3), maximum value (b), were converted to mean ± SD by the specific formula [10,11].Note that the data may not always be given in full.The three frequently encountered scenarios are: C1 = {a, m, b; n}, C2 = {a, q1, m, q3, b; n}, C3 = {q1, m, q3; n}.The skew data can be diagnosed and transformed automatically based on the formular link: https:// www.math.hkbu.edu.hk/ ~tongt/ papers/ media n2mean.html.
We estimated the weighted mean differences (WMD) or standardised mean differences (SMD) with 95%CI for continuous data and the odds ratio (OR) for categorical data among the groups, with an overall estimate of the pooled effect.Forest plots were used to present the results graphically.Statistical heterogeneity across trials was assessed using the I 2 value.A value of I 2 >50% or P<0.1 was considered as high heterogeneity.A randomeffects model was applied in the case of high heterogeneity; otherwise, a fixed-effects model was adopted.For the primary outcome (pain score at postoperative 1-2, 3-8, 24 and 48 h), a priori sensitivity analysis was performed by removing the studies with a high risk of bias.
Mixed meta-regression was used to explore any potential dose-related interaction between the intraoperative remifentanil dose and postoperative pain intensity / PONV.In volunteers, remifentanil infusion at a rate of 0.10 μg/kg/min was reported [12] to provoke hyperalgesia, while opioid infusion at a rate of 0.05 μg/kg/min failed to induce RIH after discontinuation.The infusion rate at 0.1 µg/kg/min was proved to achieve a stable plasma concentration ranging between 2.7 and 2.9 ng/ml [13].In addition, remifentanil plasma concentrations of 1.6 and 3.2 ng/ml correspond to steady-state concentrations achieved when infusing remifentanil at a constant rate of about 0.065 and 0.13 µg/kg/min.The result partially proved the linear correlation between the plasma concentration and constant infusion rate [14].According to the linear formula, the trial which used a dose with 0.05 µg/kg/min was equally with the remifentanil concentrations of 1.2 ng/ml.Therefore, studies with remifentanil infusion less than 0.05 μg/kg/min or 1.2 ng/ml were allocated to the control group when performing metaregression analysis.Based on the outcome of the mixed meta-regression analysis, piecewise linear regression was performed to define a cutoff value of the remifentanil dose to induce postoperative pain intensity.
Trial sequential analysis.In order to estimate the number of patients needed to allow for reliable statistical inference, we performed a sample size calculation to ensure that a sufficient number of patients were included in the meta-analysis.The random effect model using DerSimonian-Laird method was selected for the Trial sequential analysis program to integrate effective sizes.The required information size and the adjusted significance threshold for the postoperative pain score were calculated, with an anticipated 20% reduction of mean difference in pain score (mean difference=0.4)and variance of 0.4 with a 5% risk of type 1 error (β=0.8), and model variance-based heterogeneity correction.

Assessment of publication bias
The risk of potential publication bias was evaluated using the Egger's regression test.

Results
The literature search yielded 7633 results.31 studies with a total of 2019 patients between 2000 and 2020 met our inclusion criteria, and were included in the meta-analysis [5][6][7][8] (Fig. 1).No unpublished data were identified from clinical register or major annual meetings of anaesthesiology.

Study characteristics
4 studies were at a high risk of attrition bias [23,32,39,40].The other studies were all at low risk or unclear (Fig. 2).The characteristics of the included studies [5][6][7][8] are shown in Table 1.The quality of the included studies is shown in Additional file 3, Additional file 4 and Table 2.

Table 2 Summary of findings and quality of evidence (GRADE)
CI Confidence interval, GRADE Grading of Recommendations Assessment, Development and Evaluation, OR Odds ratio, PONV Postoperative nausea and vomiting, SM, Standardized mean differences, WMD Weighted mean differences The level of evidence was assessed by the GRADE method.⊕⊕⊕⊕(High quality): Further research is very unlikely to change our confidence in the estimate of effect.⊕⊕⊕(Moderate quality): Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.⊕⊕(Low quality): Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.⊕(Very low quality): We are very uncertain about the estimate. 1Downgraded for imprecision: optimal information size not reached. 2Downgraded for insufficient data quality. 3 Downgraded for inconsistency (I 2 > 50%).
After exclusion of studies with a high risk of bias, meta-regression analysis found that there was no association between remifentanil dose and the pain intensity at 24 h after the surgery (1264 participants in 17 studies, Tau 2 =0.12, slope of regression line: 0.39; P=0.57; 95% CI, -0.96 to 1.74) (Additional file 7).However, after excluding two trials [17,18] with extreme data that generated high heterogeneity, the intensity of the pain score at 24 h was closely correlated with the dose of the intraoperative remifentanil infusion.(1085 participants in 15 studies, Tau 2 =0.09, slope of the regression line: 1.96; 95% CI, 0.19 to 3.72; P=0.03) (Fig. 4A).After piecewise linear regression analysis, the cutoff dose of the remifentanil to initiate postoperative pain was 0.1 μg/kg/min.

Discussion
In this study, low dose of remifentanil was correlated with lower pain score and less allodynia.Compared with the remifentanil regimen, the remifentanil-free group showed no benefit in inhibiting pain at 24 and 48 h.The meta-regression analysis found that the intensity of postoperative pain at 24 h was correlated with the dose of remifentanil infusion.
In a previous animal study [42], remifentanil ranged between 0.66 and 3.33 μg/kg/min has been reported to induce a dose-dependent pronociceptive effect.However, the drug concentrations were far exceeded the clinical demand.In the present study, the maximum remifentanil dose was 1.2 μg/kg/min.A previous meta-analysis [43,44] with low certainty of evidence has shown that high doses of remifentanil are associated with acute pain after surgery.The use of opioid-free anaesthesia was reported to be associated with a reduction in PONV [45].Moreover, neither of the papers [43][44][45] explored the dosedependent association between remifentanil exposure and the incidence of postoperative pain or PONV.In the current meta-analysis that included more studies, quantitative and meta-regression analyses were introduced to conclude that the incidence of PONV was not correlated with the dose of remifentanil regimen, which was in contrast to a previous report [45].A possible explanation for the higher pain score after remifentanil infusion is remifentanil-induced acute tolerance and hyperalgesia.The exact mechanism underlying opioid-induced hyperalgesia remains unclear.N-methyl-d-aspartate (NMDA) receptors have been shown to play a key role in opioid-induced hyperalgesia [46].The reason for the inadequate postoperative pain control in the remifentanil protocol was attributed to NMDA activation.Sevoflurane, which is widely used in remifentanil-free groups, was reported to prevent central sensitisation through NMDA receptor antagonistic properties [46].
Opioid-induced thermal hyperalgesia can last for 2-7 days in rats [47].Several studies have demonstrated that RIH occurred at 2 h and reached maximal at 24-48 h [47,48].In the present study, the higher pain scores in the high-dose remifentanil regimen lasted for 2 days after surgery.
Owing to the absence of opioid exposure, remifentanil-free anaesthesia can theoretically provide better resistance against the pathogenesis of hyperalgesia.Nevertheless, we found that there was no superiority over the remifentanil group in terms of postoperative pain scores at 24 and 48 h.Correspondingly, no improvement was detected in the rescue analgesic consumption at 24-48 h in the remifentanil-free group.Caution should be observed when using a remifentanilfree protocol in clinics.
The low-dose group inhibited postoperative pain during the first 48 h after surgery.Postoperative analgesic consumption, pain intensity and secondary hyperalgesia in the low-dose group were less evident than those in high-dose group.Notably, the degree of hyperalgesia was closely correlated with the amount of remifentanil infused.Compared with the remifentanil-free group, a small dose of remifentanil regimen seems recommendable.
The perioperative application of opioids is considered as a major factor in inducing PONV.The risk of PONV increases in direct proportion with the perioperative amount of opioid consumption [49].Moreover, opioid-induced hyperalgesia requires more rescue opioids, which in turn aggravates PONV.A retrospective observational study [50] reported a dose-dependent association between the dose of intraoperative remifentanil and an increase in the risk of PONV.However, this meta-analysis upends our basic assumption.Regardless of remifentanil-free group or low-dose remifentanil regimen, the incidence of PONV did not decrease in comparison with the high-dose group.In other words, the incidence of PONV was not correlated with the amount of remifentanil infusion.
Postoperative shivering increases oxygen consumption, leading to an increased incidence of cardiovascular and neurological complications.It has been proposed that shivering results from rapid opioid withdrawal [51].The present meta-analysis revealed an increased incidence of shivering after high dose of remifentanil.

Limitation
Firstly, the decrease in pain scores (0-10) was on average less than one point if there was a comparison between remifentanil and remifentanil-free regimens or high and low doses of remifentanil regimens.Even if this was statistically significant, it is doubtful that the difference was clinically significant as expected.This will not restrain the use of higher intraoperative remifentanil regimen, especially with the proper use of intraoperative pain monitoring.Secondly, most of the included studies were conducted without using any nociception monitoring.In the absence of a monitoring device, such as the Nociception Level index, the remifentanil dose was not known to be adequate or insufficient during surgery.Furthermore, the low and high doses of opioids overlapped among the trials.Therefore, there is a high heterogeneity for most pain scores comparisons between high and low doses of remifentanil regimens.The meta-regression and sensitivity analysis were performed to exclude the impact of the high heterogeneity and correct selective bias.We believe that this limitation did not affect the validity of our results.

Conclusion
Remifentanil-free anaesthesia has shown insufficient benefits in inhibiting postoperative pain.Patients receiving low dose remifentanil were correlated with lower pain scores, less allodynia and less shivering than those who received high dose remifentanil.In view of the current opioid epidemic, low-dose remifentanil anaesthesia should be recommended.These findings can be broadly generalised to patients across surgical disciplines.

Fig. 3
Fig.3Forest plots for the periincisional wound and forearm allodynia.Data were pooled using a random-effects model to calculate the SMD and 95% CI for each outcome.Intervention refers to the high dose remifentanil, and control refers to the low dose remifentanil group.CI indicates confidence interval; df, degrees of freedom; M-H, Mantel-Haenszel; SMD, standardized mean differences

Table 1
Characteristics of included studies ①Pain score at rest ②Pain score at movement ③Periincisional mechanical pain threshold ④Forearm mechanical pain threshold ⑤Time to first postoperative analgesic requirement ⑥Need for rescue analgesic ⑦Analgesic consumption ⑧Emergence time ⑨PONV ⑩Shivering

Table 3
Summary of findings by secondary outcomes CI Confidence interval, OR Odds ratio, PONV Postoperative nausea and vomiting, SMD Standardized mean differences, WMD Weighted mean differences