Routine labour epidural analgesia versus labour analgesia on request: a randomised non-inferiority trial

Authors

  • MMLH Wassen,

    Corresponding author
    1. Department of Obstetrics and Gynaecology, Maastricht University Medical Centre+, GROW – School for Oncology and Developmental Biology, Maastricht, The Netherlands
    • Correspondence: MMLH Wassen, Department of Obstetrics and Gynaecology, Maastricht University Medical Centre +, GROW – School for Oncology and Developmental Biology, PO Box 5800, 6202 AZ, Maastricht, The Netherlands. Email martine_wassen@hotmail.com

    Search for more papers by this author
  • LJM Smits,

    1. Department of Epidemiology, Maastricht University Medical Centre+, CAPHRI – School for Public Health and Primary Care, Maastricht, The Netherlands
    Search for more papers by this author
  • HCJ Scheepers,

    1. Department of Obstetrics and Gynaecology, Maastricht University Medical Centre+, GROW – School for Oncology and Developmental Biology, Maastricht, The Netherlands
    Search for more papers by this author
  • MAE Marcus,

    1. Department of Anaesthesiology and Pain Treatment, Maastricht University Medical Centre+, Maastricht, The Netherlands
    Search for more papers by this author
  • J Van Neer,

    1. Department of Obstetrics and Gynaecology, Atrium Medical Centre, Parkstad, Heerlen, The Netherlands
    Search for more papers by this author
  • JG Nijhuis,

    1. Department of Obstetrics and Gynaecology, Maastricht University Medical Centre+, GROW – School for Oncology and Developmental Biology, Maastricht, The Netherlands
    Search for more papers by this author
  • FJME Roumen

    1. Department of Obstetrics and Gynaecology, Atrium Medical Centre, Parkstad, Heerlen, The Netherlands
    Search for more papers by this author

Abstract

Objective

To assess the effect on mode of delivery of the routine use of labour epidural analgesia (EA) compared with analgesia on request.

Design

Randomised non-inferiority trial.

Setting

One university and one non-university teaching hospital in The Netherlands.

Population

Women with a singleton pregnancy in cephalic presentation beyond 36 + 0 weeks' gestation.

Methods

Participants were randomly allocated to receive either routine EA or analgesia on request. Intention-to-treat (ITT) and per-protocol (PP) analyses were performed, with confidence intervals (CI) calculated for the differences in percentages or means.

Main outcome measures

Rate of operative delivery (instrumental vaginal or caesarean), labour characteristics, and adverse labour and neonatal outcomes.

Results

A total of 488 women were randomly allocated to the routine EA (= 233) or analgesia on request group (= 255). In the routine EA group, 89.3% (208/233) received EA. According to ITT analysis, 34.8% (81/233) women in the routine EA group had an operative delivery, compared with 26.7% (68/255) in the analgesia on request group (difference 8.1%, 95% CI −0.1 to 16.3). The difference in rate of operative deliveries according to the PP analysis was statistically significant (difference 8.9%, 95% CI 0.4 to 17.4). Inferiority of EA could not be rejected, as in both analyses the upper bound of the confidence interval exceeded the pre-specified inferiority criterion of +10%. Women in the routine EA group had more adverse effects, including hypotension (difference 9.5%, 95% CI 4.2 to 14.9), and motor blockade (difference 6.8%, 95% CI 1.1 to 12.5).

Conclusion

Non-inferiority of routine EA could not be demonstrated in this trial. Routine EA use is likely to lead to more operative deliveries and more maternal adverse effects. The results of our study do not justify routine use of EA.

Introduction

Epidural analgesia (EA) is the most effective method of pain relief but also has unintended adverse effects on mode of delivery and labour outcome.[1, 2] Labour EA is associated with an increased risk of an instrumental vaginal delivery (IVD) but not with an increased risk of a caesarean section (CS).[1] The question remains whether these adverse effects are caused by EA itself or whether they are confounded by indication. In the absence of a medical contraindication, maternal request is a sufficient medical indication for pain relief during labour.[3] All studies published to date have been conducted in women who had a strong need for pain relief.[1, 2] In general, there is a correlation between women with a strong need for pain relief and failure of progress in labour, for example, in the case of labour dystocia.[4] Thus, a request for EA for pain relief may be an indication of obstructed labour. At present, there is a lack of randomised controlled trials investigating the effect of EA on labour outcome in women without a strong need for pain relief.

In The Netherlands, and throughout the world, maternal request for EA during labour is increasing.[5-8] Recently, a systematic review showed no increased risk of an IVD or CS in women receiving early EA at a cervical dilation of 3 cm or less compared with late EA at a cervical dilation of at least 4 cm.[9]

The objective of the present study was to assess the effect of routine use of EA compared with analgesia on maternal request on operative delivery, and labour and neonatal outcomes. This non-inferiority trial was designed to test the hypothesis that routine EA is not unacceptably inferior to analgesia on request regarding the risk of an IVD or unplanned CS.

Methods

We performed a bicentre, open-label, randomised, non-inferiority trial. The institutional review boards of the participating centres approved the study. Participants were recruited from the Atrium Medical Centre Parkstad (Atrium MCP), a non-university teaching hospital, and the Maastricht University Medical Centre (MUMC), The Netherlands. The recruitment period was from September 2008 until May 2012. The trial was reported in concordance with the CONSORT statement.[10]

Participants

Women were eligible to participate in the study if they were 18 years or older, pregnant with a singleton in vertex presentation and a gestational age of 36 weeks or more, had the intention to deliver vaginally, and did not have a contraindication for EA. For this study, a contraindication for EA was defined as: use of coumarin derivatives, use of low molecular weight heparin (LMWH) in therapeutic doses, use of LMWH in prophylactic doses less than 10 hours before, low platelet count <80 × 109/l, use of blood platelet aggregation inhibitors, a history of increased bleeding tendency, history of blood clotting disorders, an allergy for anaesthetics used, or history of spine disorders or spinal infection.

From 32 weeks' gestation on, eligible pregnant women were given oral and written information about the aims, methods, reasonably anticipated benefits, and potential hazards of the study. Advantages and disadvantages of epidural analgesia as described in the Cochrane review,[1] were explained to them by the research coordinator and/or the staff of the participating hospital. The rationale of the study was discussed, as was the hypothesis that routine epidural analgesia was not unacceptably inferior to analgesia on request regarding the risk of an operative delivery. Subjects were informed that their participation was voluntary and that they might withdraw consent to participate at any time during the study without any consequences for their obstetrical care. This was done in the outpatient clinic, but a number of women were also informed in the labour room before induction of labour was started. After a period of consideration, both oral and written informed consent was obtained.

Randomisation

Participating women were randomised before labour started. Women were randomly allocated to either routine EA or analgesia on request in a 1:1 ratio with stratification for centre and parity (nulliparous or parous women). Randomisation was performed using sequentially numbered opaque envelopes that were opened by study personnel after obtaining written informed consent. Although allocation of group was concealed, the participants, obstetricians, and outcome assessors were not blinded to the eventual treatment (open-label). Data-analysing members were blinded to group assignment.

Interventions

Participants allocated to the routine EA group immediately received EA when they were in labour, as determined by the attending obstetrician. Labour was defined as having an effaced cervix with at least 2 cm dilation in combination with regular contractions.

At Atrium MCP, EA was maintained by a continuous background infusion of ropivacaine (0.125%) with sufentanil (1 μg/ml) (from 1 September 2008 to 12 May 2011) and bupivacaine (0.125%) with sufentanil (1 μg/ml) (from 12 May 2011 to 31 May 2012) at a rate of 7–10 ml/hour. At MUMC+, EA was initiated with an 8-ml bolus of ropivacaine (0.180%) and sufentanil (0.5 μg/ml) and subsequently was patient-controlled with a 4-ml bolus of ropivacaine (0.180%) and sufentanil (0.5 μg/ml) without background infusion. In the analgesia on request group, labouring women received pain relief only on request. According to local hospital protocol and patient preference, the women were given opiates intramuscularly or EA (at first or second pain relief request after opiates) for labour pain control.

Outcomes

The primary non-inferiority outcome was the rate of operative deliveries (defined as either IVD or unplanned CS). Secondary outcomes were maternal and labour characteristics, including interval between rupture of membranes and birth, use of oxytocin augmentation, maternal fever during labour (defined as temperature ≥38.0°C), maternal labour antibiotic use (excluding prophylactic use), length of second stage of labour, and adverse labour outcome (e.g. shoulder dystocia, postpartum haemorrhage, manual removal of the placenta, and third or fourth degree perineal rupture). Neonatal data were also analysed: birthweight, 1- and 5-minute Apgar scores, umbilical artery pH, umbilical artery base excess, neonatal fever (defined as temperature ≥38.0°C) and neonatal admission. EA-related complications were also recorded, including maternal hypotension, postpartum urinary retention (need for bladder catheterisation) and motor blockade. The degree of motor block was assessed using the Bromage scale, with 0 indicating free movement of legs and feet; I, just able to flex knees with free movement of feet; II, unable to flex knees, but with free movement of feet; and III, unable to move legs or feet.

Study design, sample size and statistical analysis

This non-inferiority trial was designed to test the hypothesis that routine EA is not inferior to analgesia on request to an acceptable extent regarding the risk of having an operative delivery. The largest clinically acceptable difference (degree of inferiority, Δ) was set at 10%. The statistical null hypothesis was that the risk of an operative delivery in the routine EA group would exceed that in the analgesia on request group by more than 10%. If the non-inferiority margin (+Δ = 10%) is exceeded, routine EA is defined to be unequivocally inferior to analgesia on request. To reject the null hypothesis, the upper limit of the CI around the observed difference should be less than the non-inferiority margin.

For sample size calculation, we used the following input values: chance that delivery is non-instrumental, 72.9%;[11] power 0.8; alpha 0.05; delta (maximally acceptable difference) 10%. We calculated that 244 women were required in each of the study arms. Based on a preliminary study in 2008 in the Atrium MCP, which showed that six of 50 eligible women (12%) would participate in this randomised trial, it was estimated that more than 4000 women had to be approached. The primary analysis was according to intention-to-treat (ITT). Although ITT-analysis is the preferred type of analysis in superiority trials, in a non-inferiority trial it may unjustifiably lead to the conclusion of non-inferiority. Therefore, a per-protocol (PP) analysis is needed to cross-validate the results of the ITT analysis.[10, 12] In our PP analysis we excluded all women who did not receive EA from the EA group. Continuous variables were summarised as medians with interquartile ranges (IQR). Treatment effects were presented as differences in means or percentages with 95% confidence intervals (CI). The chi-square test was used for categorical variables. Student's t-test or the non-parametric Mann–Whitney U-test was used to analyse continuous variables. Non-inferiority was evaluated by comparing the upper bound of the CI around the difference between the number of operative deliveries in the routine EA group and the analgesia on request group, with Δ (10%). Statistical significance of any differences between the two study arms was evaluated by checking whether the 95% CI included 0 (for differences). Three variables had more than 5% missing data: neonatal fever (17%), ethnicity (17.8%) and education (18.4%). Missing data were imputed using single imputation. Statistical analysis was performed using spss® software (version 18.0; SPSS Inc., Chicago, IL, USA).

Results

A total of 493 women were randomised, and five women were excluded after randomisation. Figure 1 shows the 488 participants assigned to either receive routine EA (= 233) or analgesia on request (= 255). EA was initiated at a median cervical dilatation of 3 cm (IQR 2–4) in the routine EA group and at 4 cm (IQR 3–5) in the analgesia on request group (difference in mean −1.0; 95% CI −1.3 to −0.7). Median duration of EA until birth was 448 (IQR 300–648) minutes and 441 (IQR 275–602) minutes in the routine EA group and analgesia on request group, respectively. In the routine EA group, 70/208 (33.7%) women received patient-controlled EA, and 44/120 (36.7%) women in the analgesia on request group received it (difference in percentage −3.0; 95% CI −13.8 to 7.7).

Figure 1.

Participant flow diagram.

As shown in Table 1, baseline characteristics of the participants in both groups were comparable. In the ITT analysis, 81/233 (34.8%) women in the routine EA group experienced an operative delivery, compared with 68/255 (26.7%) in the analgesia on request group (difference 8.1%, 95% CI −0.1 to 16.3, Table 2, Figure 2). According to the PP analysis, 74/208 (35.6%) women in the routine EA group and 68/255 (26.7%) in the analgesia on request group had an operative delivery (difference 8.9%, 95% CI 0.4 to 17.4). The difference in rate of operative deliveries according to the PP analysis was statistically significant. However, in both the ITT and PP the upper bound of the CI exceeded the pre-specified inferiority criterion of +10%, as also shown in Figure 2, and non-inferiority could therefore be rejected. However, the result is inconclusive regarding possible inferiority of magnitude of delta. Mode of operative delivery (IVD or CS) and the indication to perform an operative delivery were not significantly different between both groups. There was also no important difference between the groups in continuing EA during the second stage of labour (73.8 versus 71.3%; difference 2.6%, 95% CI −8.7 to 13.8). ITT analysis showed no significant differences in labour characteristics (Table 2). In the PP analysis, the routine EA group used significantly more oxytocin augmentation (59.1 versus 45.5%; difference 13.6%, 95% CI 4.6 to 22.7) and experienced significantly more maternal fever (23.1 versus 15.7%; difference 7.4%, 95% CI 0.1 to 14.7). In the routine EA group, significantly more women experienced EA-related adverse events, such as maternal hypotension (15.0 versus 5.5%; difference 9.5, 95% CI 4.2 to 14.9) and motor blockade (15.0 versus 8.2%; difference 6.8, 95% CI 1.1 to 12.5). When analysing these EA-related adverse effects only in women who actually received EA in both groups, there was no significant difference between the groups. Other epidural-related adverse events such as bleeding, infection or postpuncture headache did not occur.

Table 1. Baseline characteristics of the study groups
 Routine epidural analgesia Per-protocol (= 208)Routine epidural analgesia Intention-to-treat (= 233)Analgesia on request (= 255)
  1. Data presented as median and interquartile range (25–75) or n (%).

  2. Highest completed education level was divided into primary (elementary school), secondary (high school, vocational school and preparatory school), and tertiary education (higher professional education or Master's/Bachelor's level).

  3. a

    = 166 routine epidural analgesia group according to per-protocol analysis, = 186 routine epidural analgesia group according to intention-to-treat analysis, = 215 analgesia on request group.

  4. b

    = 167 routine epidural analgesia group according to per-protocol analysis, = 187 routine epidural analgesia group according to intention-to-treat analysis, = 211 analgesia on request group.

Maternal age (years) 30.0 (26.0–33.0)30.0 (27.0–34.0)30.0 (26.0–33.0)
BMI at start pregnancy (QI) 24.2 (21.6–29.4)24.0 (21.6–29.0)24.6 (21.7–28.6)
Parity
Nulliparous106 (51.0)115 (49.4)122 (47.8)
Multiparous, history102 (49.0)118 (50.6)135 (52.9)
  Vaginal delivery69 (67.6)82 (69.5)93 (68.9)
  Planned caesarean section9 (8.8)10 (8.5)16 (11.9)
  Unplanned caesarean section19 (18.6)21 (17.8)14 (10.4)
  Vaginal and caesarean section5 (4.9)5 (4.2)1 (0.7)
Caucasian a 143 (86.1)161 (86.6)191 (88.8)
Education b
Primary1 (0.5)1 (0.5)3 (1.4)
Secondary107 (64.1)118 (63.1)140 (66.4)
Tertiary59 (35.3)68 (36.4)68 (32.2)
Table 2. Maternal and labour characteristics
 Routine epidural analgesia Intention-to-treat (= 233)Analgesia on request (= 255)Difference in percentage or mean (95% CI)
  1. CI, confidence interval.

  2. Data presented as median and interquartile range (25–75) or n (%).

  3. a

    = 81 for routine epidural analgesia group, = 68 for analgesia on request group.

Spontaneous onset of labour 83 (35.6)104 (40.8)−5.2 (−13.8 to 3.5)
Gestational age at birth (days) 280 (273–287)280 (272–288)0.2 (−1.5 to 1.9)
Interval between rupture of membranes and birth (hours) 7.9 (4.3–14.4)7.7 (2.7–14.1)1.5 (−5.1 to 8.2)
Oxytocin augmentation 125 (53.6)116 (45.5)8.2 (−0.7 to 17.0)
Maternal fever (≥38.0°C) 49 (21.0)40 (15.7)5.3 (−1.5 to 12.2)
Maternal labour antibiotic use 15 (6.4)17 (6.7)−0.2 (−4.6 to 4.2)
Number digital vaginal examination 6 (5–9)6 (4–8)0.2 (−3.5 to 0.8)
Length of second stage (minutes) 20.0 (10.0–49.5)19.0 (8.0–45.0)2.0 (−3.2 to 7.1)
Mode of delivery
Operative81 (34.8)68 (26.7)8.1 (−0.1 to 16.3)
  Vaginal instrumental37 (15.9)29 (11.4)4.5 (−1.6 to 10.6)
  Unplanned caesarean section44 (18.9)39 (15.3)3.6 (−3.1 to 10.3)
Indications for instrumental delivery a
Suspected fetal distress22 (27.2)13 (19.1)8.0 (−5.4 to 21.5)
Arrest of labour45 (55.6)47 (69.1)−13.6 (−29.0 to 1.9)
Suspected fetal distress and arrest of labour11 (13.6)7 (10.3)3.3 (−7.1 to 13.7)
Other3 (3.7)1 (1.5)2.2 (−2.8 to 7.2)
Adverse labour outcome
No203 (87.1)215 (84.3)2.8 (−3.4 to 9.0)
Shoulder dystocia5 (2.1)8 (3.1)−1.0 (−3.8 to 1.8)
Postpartum haemorrhage (≥1000 ml blood loss)14 (6.0)11 (4.3)1.7 (−2.2 to 5.6)
Postpartum haemorrhage and manual removal of placenta6 (2.6)8 (3.1)−0.6 (−3.5 to 2.4)
Manual removal of placenta3 (1.3)1 (0.4)0.9 (−0.7 to 2.5)
Third/fourth degree perineal rupture2 (0.9)4 (1.6)−0.7 (−2.6 to 1.2)
Adverse events epidural analgesia
Maternal hypotension35 (15.0)14 (5.5)9.5 (4.2 to 14.9)
Motor blockade35 (15.0)21 (8.2)6.8 (1.1 to 12.5)
Postpartum urinary retention2 (0.9)2 (0.8)0.1 (−1.5 to 1.7)
Figure 2.

Difference in percentage of operative delivery according to the per-protocol and the intention-to-treat analysis. The black dashed line at ∆ = 10% indicates the non-inferiority margin. The black dots and the error bars indicate the mean difference and 2-sided 95% confidence intervals (CI) in percentage of operative delivery between routine EA and analgesia on request. If the CI includes the non-inferiority margin and zero, as is the case in the ITT analysis, the difference is non-significant, and the result regarding non-inferiority is inconclusive. If the CI includes the non-inferiority margin and is wholly to the right of zero, as is the case in the PP analysis, the difference is statistically significant, but the result regarding non-inferiority is still inconclusive in that it is still plausible that the true treatment difference is less than ∆ = 10%.

The incidence of adverse labour outcome (Table 2) and neonatal outcome (Table 3) was comparable between groups in the ITT analysis and PP analysis.

Table 3. Neonatal outcome
 Routine epidural analgesia Intention-to-treat (= 233)Analgesia on request (= 255)Difference in percentage or mean (95% CI)a
  1. Data presented as median and IQ (25–75) or n (%).

  2. a

    n = 232 for routine epidural analgesia group, n = 255 for analgesia on request group.

Birthweight (g) 3540 (3148–3873)3450 (3130–3800)39 (−53 to 132)
Apgar score a
1 minutes <714 (6.0)17 (6.7)−0.6 (−5.0 to 3.7)
5 minutes <74 (1.7)5 (2.0)−0.2 (−2.6 to 2.2)
Arterial pH a
<7.051 (0.4)5 (2.0)−1.5 (−3.4 to 0.4)
7.05–7.1530 (12.9)39 (15.3)−2.4 (−8.5 to 3.8)
>7.15201 (86.6)211 (82.7)3.9 (−2.5 to 10.3)
Arterial base excess <−10 a 30 (12.9)40 (15.7)−2.8 (−9.0 to 3.5)
Neonatal fever (≥38.0°C)a 30 (12.9)24 (9.4)3.5 (−2.1 to 9.1)
Neonatal admission a 49 (21.1)63 (24.7)−3.6 (−11.0 to 3.9)

Discussion

Main findings

Non-inferiority of routine EA could not be demonstrated in this trial, but routine labour EA may result in an increased rate of operative deliveries. Routine use of EA results in significantly more EA-related adverse events such as maternal hypotension and motor blockade. No differences in adverse labour and neonatal outcomes were found between routine use of EA and analgesia on request.

Strengths and limitations

This study is the first randomised controlled trial investigating the effect on labour outcome of routine use of EA in women without a request for pain relief versus pain relief with opiates or EA only at the woman's request. All studies published until now have been performed in women with a request for pain relief, which may have provided a bias in the study results.[1, 2]

The study also has some limitations. First, the techniques used for EA varied by institution, which may influence the outcome of labour. However, despite this difference, a meta-analysis of patient-controlled EA versus continuous infusion found no difference in the rate of IVD or CS.[13] A recent systematic review and meta-analysis also found no statistical difference in the rate of CS between intermittent EA bolus and continuous epidural infusion.[14] Secondly, different local anaesthetics (bupivacaine and ropivacaine) were used in this study. However, a double-blinded multicentre study by Gogarten showed that both local anaesthetics provided equally effective analgesia at equal doses without a difference in adverse effects.[15]

Thirdly, some selection bias occurred. The estimation is that around 10–15% of all eligible women were included. This was based on both the data of the preliminary results and the total number of women delivering in the study period. Furthermore, the use of EA in the analgesia on request group was higher than in the normal population in the Netherlands (47.1 versus 26%[16]), suggesting that women who already favoured EA for pain treatment were more likely to participate. If EA resulted in an increased risk of operative deliveries, the rate of operative deliveries should have been higher in this analgesia on request group because of the selection of women favouring EA. However, the rate of operative deliveries in the analgesia on request group was 26.7%, comparable with national data.[11]

Finally, study enrolment was slow and the inclusion period took almost 4 years. Based on the hospital birth registers, the exact number of eligible women in both hospitals was estimated to be around 8000. Around 4000 of them were asked to participate in the study.

Interpretation

To the best of our knowledge, this study is the first randomised trial on routine use of EA during labour in women without request for pain relief. When compared with the EA group of a Cochrane systematic review,[1] which compared EA with non-epidural or no pain relief during labour, the IVD rate was comparable to this study's routine EA group (15.9% [37/233] versus 17.0% [675/3981]). The CS rate, however, was substantially higher in the early EA group in our study (18.9% [44/233] versus 10.8% [455/4223, respectively]).[1] The Cochrane review found no significant increased risk for CS delivery in women receiving labour EA. Maternal hypotension and motor blockade were recorded significantly more often in the routine EA group. These results are in agreement with earlier results.[1] These adverse effects were comparable between groups when only analysing women receiving EA.

Conclusion

In this study, with the regimens for analgesia used, non-inferiority of routine EA could not be demonstrated. Routine labour EA is likely to increase the rate of operative deliveries and is also associated with more adverse maternal effects such as maternal hypotension and motor blockade. On the basis of these results, we conclude that routine use of EA should not be advised.

Disclosure of interests

No relevant financial, personal, political, intellectual or religious interests were disclosed.

Contribution to authorship

All authors approved the final version of the manuscript. MMLHW was the first and main author of the manuscript. LJMS contributed (methodology) to the study design, statistical data analysis, interpretation and writing of the manuscript. HCJS participated in the conception of the study, acquisition of data, interpretation of data and revised the article critically. MAEM as anaesthesiologist contributed to study design, data analysis and writing of the article. JvN participated in data acquisition, analysis and drafting the article. JGN participated in design of the study, enrollment of women, interpreting data and writing of the article. FJMER participated as research coordinator to all phases (design, acquisition, data entry, data analysis, interpretation and writing of the manuscript).

Details of ethics approval

The procedures of the study received ethics approval (date of approval 03-04-2008, reference number 08-T-26) from the institutional ethics committee responsible for human experimentation. The institutional review board of the Atrium MCP approved the study.

Funding

There was no funding by a commercial company, charity or government department for writing this systematic review.

Acknowledgements

No specific acknowledgements.

Ancillary