Labour is one of the most painful experiences that women encounter in their lives,1 and can be associated with physiologic consequences and long-term emotional and psychological effects.2,3 Neuraxial analgesia is considered to be the criterion standard in the provision of pain relief in labour.4,5 But in the face of ongoing research,6 the optimal mode of delivering epidural solution to produce improved maternal, fetal, and neonatal outcomes has still not been determined. Continuous epidural infusion (CEI) can be defined as the constant infusion of epidural solution over the course of labour, while intermittent epidural boluses may be administered by many different techniques. They can be given at scheduled time intervals, that is by programmed intermittent epidural bolus (PIEB), or on demand by the clinician as demand intermittent epidural bolus (DIEB) or by the patient as patient-controlled epidural analgesia (PCEA). Computer-integrated CEI and PCEA (CI CEI + PCEA) and computer-integrated PIEB and PCEA (CI PIEB + PCEA) represent more novel delivery systems of epidural solution, in which the rate of infusion of CEI or the frequency of PIEB varies with the pattern of PCEA use through an autoregulatory feedback loop.7,8

In some parturients, however, an alternative to epidural analgesia may be needed. Contraindications to neuraxial modalities include allergy to local anesthetics, bleeding diathesis, hemodynamic instability, patient refusal, spinal abnormalities, and suspected infection. In such circumstances, the use of intravenous patient-controlled analgesia (PCA) with opioids is common.9 It has been suggested that remifentanil, in particular, might be the ideal systemic opioid for pain relief in labour as it has a fast onset of action, short context-sensitive half life, and a rapid offset owing to its metabolism by nonspecific plasma and tissue esterases.10

To differentiate between the various strategies of delivering epidural solution, meta-analyses have previously conducted pairwise comparisons to examine CEI vs PCEA;11 CEI + PCEA vs PCEA;12 CEI vs PIEB, both with or without PCEA;13,14,15,16,17 and CEI + PCEA vs PIEB + PCEA.18 The methodology of these systematic reviews was limited by the pooling of disparate interventions such as CEI, CEI + PCEA and CI CEI + PCEA and PIEB, PIEB + PCEA, and CI PIEB + PCEA. Further, meta-analyses have evaluated epidural analgesia vs remifentanil PCA,19,20,21 but none have looked at the possibility of other opioids such as fentanyl compared with neuraxial analgesia. Given the recent publication of several randomized controlled trials,6,22,23,24 a network meta-analysis would facilitate the simultaneous comparison of multiple competing interventions, substituting or supplementing direct comparisons with indirect ones, potentially increasing the precision of effect estimates.

In view of this, the aim of this systematic review and network-meta-analysis was to compare the influence of CEI, CI CEI, CI PIEB, DIEB, fentanyl PCA, PCEA, PIEB, and remifentanil PCA, either alone or in combination with each other, on outcomes related to the mother, fetus, and neonate.

Methods

We prospectively registered the protocol for this systematic review and network meta-analysis with PROSPERO (CRD42021254978), and followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines when presenting our findings.25 CENTRAL, CINAHL, Ovid Embase, Ovid Medline, and Web of Science were searched by two authors (D. O. and N. D.) from inception to 6 August 2022 for free-text keywords and subject headings associated with obstetric care, labour, analgesic modalities, and local anesthetics (Electronic Supplementary Material [ESM] eAppendix 1).

Inclusion criteria were met if a randomized controlled trial evaluated nulliparous and/or multiparous patients in spontaneous or induced labour and if they had any of the following analgesic interventions, either alone or in combination with each other: CEI; CI CEI; CI PIEB; DIEB; fentanyl PCA; PCEA; PIEB; and remifentanil PCA. Importantly, PIEB was defined as either the manual or automated scheduled administration of epidural solution, and we did not differentiate between these methods of delivery for PIEB. In the case of neuraxial analgesic techniques, the method of insertion could include epidural, combined spinal-epidural (CSE), or dural puncture epidural. Following the initiation of neuraxial analgesia, the maintenance epidural solution had to include a local anesthetic, at a concentration of bupivacaine or levobupivacaine less than or equal to 0.125% or ropivacaine less than or equal to 0.2%, and an opioid, with no difference in the constituents of this solution between the groups in a trial. Exclusion criteria were satisfied if the randomized controlled trial examined only a specific subgroup in the obstetric population such as preeclampsia, or it was published only as an abstract. We did not place any restrictions on language of publication.

Once the article citations had been deduplicated, two authors (S. W. and N. D.) used the aforementioned inclusion and exclusion criteria to screen their titles and abstracts independently in Rayyan (Qatar Computing Research Institute, 2016, Doha, Qatar).26 To find trials that had been missed from the database searches, the reference lists of included ones and previous systematic reviews were hand searched.

Characteristics of trials were extracted by two authors (N. D. and D. Z.) for the following: the parity of patients; neuraxial technique if applicable; method of initiation of regional or systemic analgesia; technique of maintenance of regional or systemic analgesia; and the management of breakthrough pain. Quantitative data were extracted by three authors (S. W., D. Z., and B. N.), and disputed differences were settled by a fourth author (N. D.). The coprimary outcomes were the need for rescue analgesia and the maternal satisfaction. It was our opinion that the efficacy of the different delivery modalities of epidural analgesia and intravenous analgesia from the perspective of pain relief in labour was best reflected by the need for rescue analgesia. Maternal satisfaction was thought to be multidimensional in nature, and have the potential to summarize the subjective influence of these various techniques on indices such as pain, lower limb blockade, duration of labour, mode of delivery, and fetal outcome. Secondary outcomes were related to the mother, fetus, or neonate, and included: the change in baseline pain score; pain score at 30 min, one hour, two hours, three hours, and four hours; overall pain score in labour; time to rescue analgesia; dose of bupivacaine-equivalent local anesthetic per hour; cumulative dose of bupivacaine-equivalent local anesthetic in labour; duration of first stage of labour; duration of second stage of labour; overall duration of labour; incidence of maternal respiratory depression, apnea, oxygen desaturation, bradycardia, sedation, nausea and/or vomiting, pruritus, shivering and lower limb motor blockade; rate of spontaneous vaginal delivery, instrumental delivery, and Cesarean delivery; maternal satisfaction; rate of fetal bradycardia; umbilical pH; and the rate of neonatal Apgar score < 7 at one and five minutes. If we required further information from trial authors, we emailed them up to three times.

We judged the methodological quality of trials with the revised Cochrane risk of bias tool (RoB 2),27 and two authors (D. O. and A. O.) determined the following risk of biases: randomization process; deviations from intended interventions; missing outcome data; measurement of the outcome; and selection of the reported result. Trials were provided with a summary risk of bias which was either low, of some concern, or high, and disputed differences were settled by a third author (N. D.).

Continuous data were extracted as means and standard deviations. In converting the data to these means and standard deviations, we followed recommendations from the Cochrane Collabouration.28 The mean was presumed to be the same as the median, and the standard deviation was calculated by dividing the interquartile range by 1.35 and the range by 4. Dichotomous data extracted as numbers of incidence. Of note, bupivacaine equivalents were calculated from levobupivacaine and ropivacaine using a factor of 0.6. Outcome data that were presented in graphical but not numerical format were quantified with the use of Plot Digitizer version 2.1 (Free Software Foundation, Boston, MA, USA).

Data were imported from Microsoft Excel (Microsoft Corporation, Redmond, WA, USA) into Stata version 16.1 (StataCorp LLC, College Station, Texas, USA) by two authors (N. D and D. Z.). Network meta-analysis with a common heterogeneity parameter and frequentist and multivariate methods was conducted for an outcome if the direct comparisons between trials of three or more interventions could be linked into a network.29,30 Empirical and simulation studies suggest no important differences in the performance, bias, or errors between the Bayesian and this frequentist model of network meta-analysis.31 Network plots were created in which the nodes indicated the interventions and the lines represented the direct comparisons between these interventions. Indirect comparisons, that is those which were not directly compared within trials, were derived mathematically. Local inconsistency between direct and indirect estimates was evaluated by separating indirect evidence from direct evidence, and global inconsistency was examined with the design-by-treatment interaction test. Network league tables were constructed to display the results of multiple comparisons between interventions. Continuous outcome data are presented as mean differences with 95% confidence intervals, and dichotomous outcome data are presented as odds ratios with 95% confidence intervals. In the absence of serious imprecision, competing interventions were ranked in order. The grading of recommendations assessment, development, and evaluation (GRADE) system for network meta-analysis, as recommended by Salanti et al.,32 was used by two authors (D. Z. and N. D.) to evaluate the certainty and quality of evidence for all outcomes, and the following domains were studied: risk of bias; indirectness; imprecision; inconsistency; and publication bias. This was supported by the use of CINeMA software (Institute of Social and Preventative Medicine, University of Bern, Bern, Switzerland), and the use of a comparison-adjusted funnel plot and the Egger’s linear regression test.

Results

In all, 73 trials were included in the systematic review,6,7,8,22,23,24,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99 and details of the screening process are illustrated in Fig. 1. The following interventions were compared: CEI vs DIEB in two trials;33,34 CEI vs PIEB in ten trials;35,36,37,38,39,40,41,42,96,97 CEI vs PCEA in eight trials;43,44,45,46,47,48,49,50 CEI vs DIEB vs PCEA in two trials;51,52 CEI vs CEI + PCEA in four trials;53,54,55,99 CEI vs CI CEI + PCEA in one trial;7 CEI vs CEI + PCEA vs PCEA in three trials;56,57,58 CEI + DIEB vs DIEB in one trial;59 CEI vs remifentanil PCA in four trials;60,61,62,63 CEI + PCEA vs DIEB in one trial;64 CEI + PCEA vs PCEA in ten trials;65,66,67,68,69,70,71,72,73,74 CEI + PCEA vs PIEB + DIEB in one trial;75 CEI + PCEA vs PIEB + PCEA in ten trials;6,22,23,76,77,78,79,80,81,82 CEI + PCEA vs CI PIEB + PCEA in three trials;8,83,84 CEI + PCEA vs CEI + PIEB + PCEA in two trials;85,86 CEI + PCEA vs remifentanil PCA in one trial;87 CI CEI + PCEA vs PCEA in one trial;88 CI CEI + PCEA vs CEI + PCEA in one trial;89 DIEB vs PCEA in three trials;90,91,92 PIEB vs remifentanil PCA in one tria;l93 PIEB + PCEA vs PCEA in two trials;24,98 PCEA vs fentanyl PCA in one trial;94 and fentanyl PCA vs remifentanil PCA in one trial.95 The findings of the risk of bias assessment are presented in Fig. 2, and none of the trials were evaluated as having an overall low risk. Some concerns were present in regard to the risk of bias in 43 trials6,7,8,22,24,33,34,35,36,38,39,40,41,44,48,50,51,52,55,57,59,61,62,64,65,66,68,69,70,77,78,83,84,85,86,87,88,89,91,94,95,96,98 and the risk of bias was high in 30 trials.23,37,42,43,45,46,47,49,53,54,56,58,60,63,67,71,72,73,74,75,76,79,80,81,82,90,92,93,97,99 To clarify the methodology and results of the trials, 33 authors were emailed and four responded with the requested information.49,71,72,82

Fig. 1
figure 1

PRISMA flow diagram summarizing the retrieved, included, and excluded randomized controlled trials. PRISMA = Preferred Reporting Items for Systematic Reviews and Meta-Analyses

Full size image
Fig. 2
figure 2figure 2

Risk of bias assessment of included trials using the revised Cochrane tool. ?, some concerns; !, some concerns; -, high risk; +, low risk

Full size image

Characteristics of the trials are detailed in Table 1. Participants were nulliparous in 41 trials,6,8,24,33,34,35,36,37,38,39,44,46,49,50,52,54,55,58,59,61,69,70,72,77,78,79,80,81,82,83,84,85,86,88,89,92,93,94,96,98,99 multiparous in one trial,76 nulliparous and multiparous in 28 trials,22,23,40,41,42,43,47,48,51,53,56,57,60,62,63,64,65,66,67,68,71,73,74,75,87,90,91,95 and not specified in three trials.7,45,97 Epidural, performed in 53 trials,22,23,33,34,37,39,40,41,42,43,44,45,46,47,48,49,50,51,53,54,55,56,57,60,62,63,64,65,66,67,71,72,73,74,75,77,78,79,80,81,82,85,86,87,90,91,92,93,94,96,97,98,99 was the main neuraxial technique. CSE was used in 16 trials,7,8,24,35,36,38,52,58,68,69,70,76,83,84,88,89 epidural or dural puncture epidural in one trial,6 and epidural or CSE in two trials.59,61 In the epidural analgesic solution, the local anesthetic was bupivacaine in 24 trials,22,39,41,42,43,44,45,49,51,52,53,56,58,59,64,65,71,76,87,90,91,92,93,94 levobupivacaine in 13 trials,34,36,48,50,61,72,73,77,82,85,86,97,98 ropivacaine in 33 trials,6,7,8,23,24,35,37,38,40,46,47,54,57,60,62,63,66,67,68,69,70,74,75,78,79,80,81,83,84,88,89,96,99, and bupivacaine or ropivacaine in two trials.33,55 The opioid was fentanyl in 55 trials7,8,22,23,33,34,35,36,37,38,39,40,41,42,43,44,45,46,48,49,50,52,53,54,55,56,57,58,59,61,62,64,65,69,70,71,72,73,74,76,79,83,84,85,86,87,88,89,90,92,93,94,96,97,99 and sufentanil in 17 trials.6,24,47,51,60,63,66,67,68,75,77,78,80,81,82,91,98

Table 1 Characteristics of the included trials
Full size table

Our first coprimary outcome, the need for rescue analgesia, was examined in 3,963 participants and 40 trials.6,7,8,22,23,24,36,37,38,39,40,41,42,47,48,49,50,51,57,58,65,66,68,71,72,73,76,77,82,83,84,87,89,91,92,94,95,96,97,98 The trials differed in their definition of the outcome and these definitions included: additional analgesia requested by the patient; inadequate analgesia; pain score greater than two, three, four or five out of ten; and not specified. In the network plot, 13 direct comparisons and 32 indirect comparisons were established between ten interventions (Fig. 3). Continuous epidural infusion was inferior to PIEB and PIEB + PCEA, and PIEB + PCEA was superior to PCEA. Fentanyl PCA was inferior to CEI + PCEA, CI CEI + PCEA, CI PIEB + PCEA, DIEB, PCEA, PIEB, and PIEB + PCEA (Table 2). Inconsistency was present between three comparisons, namely PCEA vs fentanyl PCA, CEI + PCEA vs remifentanil PCA, and fentanyl PCA vs remifentanil PCA. Overall inconsistency in the network was not revealed. Further, the standard deviation of the between-trials heterogeneity was 0.789. Publication bias was not uncovered on examination of the comparison-adjusted network plot and with the Egger’s test (P = 0.81) (Fig. 4). The quality of evidence was graded as low (ESM eAppendix 2), and the network ranking of interventions was not performed given the serious imprecision.100

Fig. 3
figure 3

Network plot of the need for rescue analgesia. Each intervention is depicted by a circle that is proportional in size to the number of patients who were randomized to that intervention. Lines connecting the circles indicate the direct comparisons of interventions, their width proportional to the number of trials evaluating the comparison, and their color representing the average risk of bias. Green, low risk; yellow, some concerns; red, high risk. CEI = continuous epidural infusion; CI = computer integrated; DIEB = demand intermittent epidural bolus; PCA = patient-controlled analgesia; PCEA = patient-controlled epidural analgesia; PIEB = intermittent epidural bolus

Full size image
Table 2 Network league table for all interventions in regard to need for rescue analgesia and maternal satisfaction
Full size table
Fig. 4
figure 4

Comparison-adjusted funnel plot in relation to the network for need for rescue analgesia. Different colors correspond to particular comparisons of interventions. The red line indicates the null hypothesis, which states that the comparison-specific pooled effect estimates do not differ from the respective trial-specific effect sizes

Full size image

Our second coprimary outcome, the maternal satisfaction, was examined in 5,779 participants and 29 trials.6,7,8,24,36,38,51,52,57,60,62,63,66,69,72,73,75,76,81,82,83,84,87,88,89,92,93,94,95 The trials varied in their definition of the outcome and these definitions included: satisfaction with labour analgesia; neuraxial analgesia or technique; and not specified. Further, maternal satisfaction was assessed on a scale out of five, ten, or 100 and recorded during labour, at an unspecified time point after delivery, within two hours or 24 hr of delivery, at 24 hr or the day following delivery, or prior to discharge from the delivery suite. In the network plot, 18 direct comparisons and 37 indirect comparisons were established between 11 interventions (Fig. 5). Programmed intermittent epidural bolus + PCEA was superior to CEI + PCEA and PCEA, and fentanyl PCA was inferior to CI CEI + PCEA, CI PIEB + PCEA, PIEB, PIEB + PCEA, and remifentanil PCA (Table 2). Inconsistency was present between one comparison, specifically CEI vs DIEB, and overall inconsistency in the network was not revealed. Moreover, the standard deviation of the between-trials heterogeneity was 0.589. Publication bias was not uncovered on evaluation of the comparison-adjusted network plot and with the Egger’s test (P = 0.36) (Fig. 6). The quality of evidence was graded as low, and the network ranking of interventions was not performed in view of the serious imprecision.

Fig. 5
figure 5

Network plot with respect to maternal satisfaction. Each intervention is depicted by a circle that is proportional in size to the number of patients who were randomized to that intervention. Lines connecting the circles indicate the direct comparisons of interventions, their width proportional to the number of trials evaluating the comparison, and their color representing the average risk of bias. Green, low risk; yellow, some concerns; red, high risk. CEI = continuous epidural infusion; CI = computer integrated; DIEB = demand intermittent epidural bolus; PCA = patient-controlled analgesia; PCEA = patient-controlled epidural analgesia; PIEB = intermittent epidural bolus

Full size image
Fig. 6
figure 6

Comparison-adjusted funnel plot in relation to the network for maternal satisfaction. Different colors correspond to particular comparisons of interventions. The red line indicates the null hypothesis that the comparison-specific pooled effect estimates do not differ from the respective trial-specific effect sizes

Full size image

Details of the results of the secondary outcomes are presented in Table 3 (ESM eAppendix 3). Compared with CEI, PIEB decreased the pain score at four hours and the overall duration of labour. Patient-controlled epidural analgesia was inferior to different modes of epidural analgesia with regard to analgesic indices, including PIEB + PCEA for the pain score at two hours, CEI, PIEB, and PIEB + PCEA for the pain score at four hours, and CEI, CEI + PCEA, PIEB, and PIEB + PCEA for the time to rescue analgesia. Further, PCEA lowered the dose of bupivacaine-equivalent local anesthetic per hour relative to CEI and PIEB, and reduced the cumulative dose of bupivacaine-equivalent local anesthetic in labour compared with CEI and CEI + PCEA. The overall duration of labour was lengthened with PCEA relative to PIEB, and the incidence of lower limb motor blockade was lower with PCEA compared with CEI. Patient-controlled epidural analgesia was superior to CEI with respect to the rate of spontaneous vaginal delivery and Cesarean delivery. Relative to CEI + PCEA, PIEB + PCEA reduced pain scores at two hours and four hours, the cumulative dose of bupivacaine-equivalent local anesthetic in labour, and the incidence of lower limb blockade. Computer-integrated CEI + PCEA increased the time to rescue analgesia compared with CEI, CEI + PCEA, CI PIEB + PCEA, PCEA, PIEB and PIEB + PCEA. No differences were present between CI and non CI equivalent modes of equivalent modes of epidural analgesia, other than the increased rate of instrumental delivery with CI CEI + PCEA relative to CEI + PCEA. Fentanyl and/or remifentanil PCA were inferior to various modes of epidural analgesia with respect to pain scores at 30 min, one hour, two hours, and three hours and the overall pain score in labour. Remifentanil PCA decreased the duration of first stage of labour and increased the incidence of oxygen desaturation. Moreover, epidural analgesia was superior to fentanyl and/or remifentanil PCA for the incidence of sedation and nausea and/or vomiting. Compared with fentanyl PCA, remifentanil PCA lowered the pain score at one hour, increased the incidence of pruritus, and reduced the rate of Apgar scores < 7 at one minute.

Table 3 Conclusion from the results of the analysis and GRADE quality of evidence assessment for the primary and secondary outcomes
Full size table

Discussion

Our systematic review and network meta-analysis demonstrated some differences between modalities for labour analgesia. For the first coprimary outcome, need for rescue analgesia, CEI was inferior to PIEB and PIEB + PCEA was superior to CEI and PCEA but not CEI + PCEA with a low certainty of evidence. The second coprimary outcome, maternal satisfaction, was improved by PIEB + PCEA relative to CEI + PCEA and PCEA alone with a low quality of evidence. Fentanyl PCA increased the requirement for rescue analgesia and decreased maternal satisfaction compared with many methods of delivering epidural solution.

Compared with CEI, PIEB decreased pain scores at four hours, reduced the need for rescue analgesia, and shortened the overall duration of labour, with no other identifiable differences. It is our opinion that the pain score at four hours is an important time point at which the initial method to achieve neuraxial analgesia has faded away and the strategy to maintain sensory blockade is well established. The significance of such time points has been previously recognized by other authors for similar reasons.101 For a technique of epidural solution delivery to produce optimal analgesia, the fluid must spread well given that the visceral pain in the first stage of labour is mediated by the T10 to L1 spinal segments, and the somatic pain in the second stage of labour is subserved by the S2 to S4 spinal segments.102 Fluid moves in the epidural space in a nonuniform manner through multiple channels between the dura, fat lobules, nerves, fascia of posterior longitudinal ligament, and the vertebral canal wall.103 In the case of a multiorifice epidural catheter, continuous infusion causes flow through only the proximal hole and bolus injection results in flow via the proximal, middle, and distal holes.104 It is hence likely that the increased efficacy of analgesia with PIEB was due to the increased engagement of channels and the extent and uniformity of spread with bolus rather than continuous infusion administration, as evidenced by previous experimental, animal, and cadaveric models.103,104,105 Interestingly, despite the improved analgesia provided by PIEB relative to CEI, it did not increase maternal satisfaction. Maternal satisfaction is complex and not only dependent on pain relief,106 but also on other factors such as personal expectations, involvement in decision-making, and support from clinical staff.107,108

PCEA is an attractive method of delivering epidural solution as it facilitates the autonomy of the parturient to individualize the analgesia as the pattern of labour changes, balancing pain relief with side effects such as motor blockade. Compared with CEI and PIEB, PCEA was inferior for the pain score at four hours and time to rescue analgesia. It did, however, decrease the consumption of local anesthetic, and it is possible that this explains the reduced incidence of lower limb blockade relative to CEI. The motor blockade of the lower limb may correlate with the muscles involved in labour, impairing the capacity and urge to bear down.101 Compared with CEI, PCEA therefore resulted in an increased and decreased rate of spontaneous vaginal delivery and Cesarean delivery, respectively. In addition to anesthetic practice, variability in institutional and obstetric factors might influence the rate of spontaneous vaginal delivery, instrumental delivery, and Cesarean delivery.109,110 No differences were revealed between DIEB and PCEA. In view of the significant cost of programmable and sophisticated pumps, DIEB may represent an alternative and appropriate strategy of epidural solution delivery, particularly in developing countries and low resource settings. Considerations relevant to the use of DIEB include the expense and time of clinical staff required for these manual boluses, possibility of infection associated with frequent opening of the closed epidural system,111 and the potential risk of inadvertent intravenous injection.112

Compared with CEI + PCEA, PIEB + PCEA decreased the pain score at two hours and four hours, without lowering the need for rescue analgesia, reduced the consumption of local anesthetic and incidence of lower limb motor blockade, and increased maternal satisfaction. This superiority of PIEB + PCEA may reflect the synergistic effect of these two methods of bolus administration in improving the extent and uniformity of spread. The decreased motor blockade of the lower limb translated into the increased rate of spontaneous vaginal delivery with PIEB + PCEA. Relative to CEI + PCEA, the use of CI CEI + PCEA resulted in a greater time to rescue analgesia, probably by increasing the rate of infusion of CEI with the frequency of PCEA boluses, and a higher rate of instrumental delivery, but no differences were shown between CI PIEB + PCEA and PIEB + PCEA.

Fentanyl and remifentanil PCA were inferior to all methods of delivering epidural solution in relation to the pain scores at 30 min and at one hour, two hours, and three hours, with a difference of as much as almost six on a pain scale out of ten. If it were possible to match the peak effect of remifentanil to the uterine contractions, then remifentanil may in theory be more effective for labour analgesia, although this is difficult in clinical practice as its effect-site concentration is highest at one to two minutes, while the uterine contraction might only last 60–80 sec.113 Compared with strategies of delivering epidural solution, the use of fentanyl and remifentanil PCA increased the pain scores as labour progressed, possibly indicative of more severe mechanisms of pain in later labour or of opioid-induced hyperalgesia. Remifentanil PCA was, however, superior to fentanyl PCA with respect to the pain score at one hour. Relative to various methods of delivering epidural solution, fentanyl and remifentanil PCA were associated with increased sedation and nausea and vomiting, fentanyl PCA decreased pruritus, and remifentanil PCA increased the incidence of oxygen desaturation. Remifentanil PCA was not found to increase the risk of apnea, but none of the included trials were powered for adverse events. Such undesirable occurrences were uncommon and lack of evidence of harm is not the same as evidence of absence of harm. The potential of remifentanil to result in apnea, bradycardia, and respiratory and cardiac arrest has been underlined in numerous case reports and surveys.114,115,116,117,118 In terms of neonatal outcomes, fentanyl PCA when compared with PCEA and remifentanil PCA increased the rate of Apgar scores < 7 at one minute, but it is important to note that this does not correlate with individual infant outcomes.119 Overall, despite its limitations relative to strategies of delivering epidural solution, remifentanil PCA was found to be superior to fentanyl PCA for analgesic indices. We opine that it should be considered in circumstances where neuraxial techniques are contraindicated, and only when safety can be maintained, in other words with one-on-one midwifery care, availability of supplemental oxygen, and the use of monitors for oxygenation and ventilation.

Our network meta-analysis is not comparable with other systematic reviews that pooled disparate interventions such as CEI, CEI + PCEA, and CI CEI + PCEA and PIEB, PIEB + PCEA, and CI PIEB + PCEA.13,14,15,16 Compared with a meta-analysis which evaluated CEI to PCEA,11 our findings are similar, although these results revealed differences in the mode of delivery between these two methods of delivering epidural solution. The findings for CEI + PCEA vs PIEB + PCEA from this meta-analysis are comparable to those from a previous one,18 but we did not uncover any difference in the rate of instrumental delivery. Unlike the systematic reviews that examined epidural analgesia relative to remifentanil PCA,19,20,21 the magnitude of the difference in the pain score at two hours was greater in this meta-analysis and the incidence of nausea and/or vomiting was higher with remifentanil. Moreover, we were able to add to the current evidence with the findings related to the pain score at three hours and the incidence of sedation. It is likely that some of the differences in the results were a reflection of the increased number of randomized controlled trials included in our systematic review, and the use of network meta-analysis to substitute or supplement direct comparisons with indirect ones.

Several limitations limit the findings of this systematic review. First, in our definition of PIEB, we did not distinguish between manual and automated programmed administration. It is possible that the automated delivery of scheduled epidural solution may be superior to manual administration or vice versa. Second, concerns were present with regard to the presence of bias in all trials and were particularly associated with the randomization process, deviations from intended interventions, and the selection of the reported result. Third, specific subgroups in the obstetric population, such as those with preeclampsia, were excluded and the results should not thus be generalized to their management in labour. Fourth, heterogeneity was present in the method of initiation of neuraxial or systemic analgesia and the nature, concentration, and dose of local anesthetic and opioid. Fifth, the initial strategy to achieve rather than maintain neuraxial analgesia was likely to influence the pain score at 30 min. Sixth, standardization was not present in the definition of outcomes such as motor blockade. Last, concerns related to imprecision for many outcomes precluded the ranking of studied interventions.

In conclusion, the results from this systematic review and network meta-analysis indicate that PIEB improves analgesic efficacy compared with CEI, and that PCEA decreases local anesthetic consumption at the expense of inferior analgesia relative to CEI and PIEB. Compared with CEI, PCEA reduces the incidence of lower limb motor blockade, increasing and decreasing the rate of spontaneous vaginal delivery and Cesarean delivery, respectively. No differences were found between PCEA and DIEB. Of significance, PIEB + PCEA has emerged as superior to CEI + PCEA with regard to the analgesic indices, consumption of local anesthetic, incidence of lower limb motor blockade, rate of spontaneous vaginal delivery, and the maternal satisfaction. Computer-integrated CEI + PCEA increases the time to rescue analgesia compared with CEI + PCEA while the results show no differences between CI PIEB + PCEA and PIEB + PCEA. Fentanyl and remifentanil PCA do not provide the same level of analgesia as all methods of delivering epidural solution, and analgesic ineffectiveness is increased with time spent in labour. They predispose to a higher incidence of side effects such as nausea and/or vomiting and sedation and, despite the results show remifentanil to be superior to fentanyl PCA for analgesia at an early time point, the former increases the incidence of oxygen desaturation relative to other strategies of delivering epidural solution. Overall, we interpret the findings of our systematic review as suggesting PIEB + PCEA to be the optimal delivery mode, although the potential differing importance of the various maternal, fetal, and neonatal outcomes in determining which is optimal has not, to our knowledge, been elucidated yet. In view of the extensive range of modalities for labour analgesia, further high-quality and large randomized controlled trials are required.