analgesia in labour and fetal acid–base balance: a meta-analysis comparing epidural with systemic...

Analgesia in labour and fetal acid–base balance: a meta-analysiscomparing epidural with systemic opioid analgesia

Felicity Reynoldsa,*, Shiv K. Sharmab, Paul T. Seedc

Objective To assess the effect of epidural versus systemic labour analgesia on funic acid–base status at birth.

Design A systematic review of trials, both randomised and non-randomised, comparing epidural with systemicopioid analgesia.

Population Babies of 2102 mothers taking part in trials comparing epidural with systemic analgesia in fivecountries.

Methods From the published and unpublished figures obtained from authors, fetal pH data from 12 studies(eight randomised) (1098 babies in the epidural group þ 1004 controls) and base excess from 8 studies (fourrandomised) (856 epidural þ 842 controls) were subjected to random effect meta-analysis.

Main outcome measures Umbilical artery pH and base excess values.

Results Fetal pH was higher in the epidural than in the control group in the randomised trials (differenceþ0.009, 95% CI þ0.002 to þ0.015), but when all studies were included, the difference was not significant(þ0.004, 95% CI �0.005 to þ0.014). Fetal base excess was higher in the epidural group in the fourrandomised studies (difference þ0.779 mEq/L, 95% CI þ0.056 to þ1.502) and in all eight studies(difference þ0.837 mEq/L, 95% CI þ0.330 to þ1.343).

Conclusion Umbilical artery pH is influenced by maternal hyperventilation. Base excess is therefore a betterindex of metabolic acidosis after labour. Epidural analgesia is associated with improved neonatal acid–basestatus, suggesting that placental exchange is well preserved in association with maternal sympatheticblockade and good analgesia. Although epidural analgesia may cause maternal hypotension and fever,longer second stage of labour and more instrumental vaginal deliveries, these potentially adverse factorsappear to be outweighed by benefits to neonatal acid–base status.

INTRODUCTION

Epidural analgesia produces better pain relief and more

maternal satisfaction than other methods of analgesia in

labour1, but its possible adverse effects have been the focus

of much attention in recent years. Many perceived compli-

cations stemmed from the fact that, appropriately, it was

used more in induced and high risk than in normal spon-

taneous labours. A number of prospective studies, random-

ised trials and meta-analyses comparing it with systemic

opioid analgesia have refuted concern that it may provoke

an increase in the caesarean section rate1,2 and in the

prevalence of postpartum backache3, and that it may

adversely affect breastfeeding4. The most recent and com-

prehensive systematic review1, however, has confirmed an

association between epidural analgesia and maternal hypo-

tension, an increase in the duration of the second stage of

labour, in the need for oxytocin and in instrumental vaginal

delivery rate and a rise in maternal temperature during

labour, although the latter only amounts to approximately

0.07jC/hour5.

Many British obstetricians and midwives assume that

because epidural analgesia is associated with these mater-

nal and obstetric changes, it must have adverse effects on

the fetus and newborn, without pausing to consider whether

there is direct evidence of any such detriment. Indeed, scant

attention has been paid to indices of the baby’s wellbeing,

one meta-analysis even making no mention of neonatal

effects2. If these potentially adverse intrapartum effects of

epidural analgesia were harmful to the baby, this would be

reflected in the neonatal acid–base status. Early studies

published in 1974, however, reporting funic pH and base

excess, suggested that epidural analgesia was associated

with reduced fetal/neonatal acidosis and even appeared to

protect the fetus from the detrimental effects of a prolonged

second stage of labour6 – 9.

Acid–base balance measured in cord blood at delivery is

a useful index of the recent intrauterine environment.

The commonly recorded umbilical artery pH, however,

BJOG: an International Journal of Obstetrics and GynaecologyDecember 2002, Vol. 109, pp. 1344–1353

D RCOG 2002 BJOG: an International Journal of Obstetrics and Gynaecology

PII: S1 4 7 0 - 0 3 2 8 ( 02 ) 0 1 9 6 1 - 4 www.bjog-elsevier.com

aAnaesthetic Department, St Thomas’ Hospital, London,

UKbDepartment of Anesthesiology and Pain Management,

University of Texas Southwestern Medical Center, Dallas,

Texas, USAcMaternal and Fetal Research Unit, Department of

Obstetrics and Gynaecology, Guy’s Kings and St

Thomas’ School of Medicine, King’s College, London, UK

* Correspondence: Professor F. Reynolds, Department of Anaesthetics,

St Thomas’ Hospital, London SE1 7EH, UK.

is influenced by respiratory as well as metabolic factors.

Maternal hyperventilation in painful labour tends to raise

arterial pH in both maternal and fetal blood and may there-

fore mask fetal acidosis, but with epidural blockade, exces-

sive hyperventilation is prevented by adequate analgesia,

counteracting any possible respiratory alkalosis. Base ex-

cess is a more specific index of the metabolic component of

acid–base balance, but unfortunately, although derived from

pH, PCO2 and haemoglobin, and usually calculated automat-

ically, it has rarely been reported in more recent published

work. We have therefore explored whether there is any up-

to-date evidence from randomised studies, both published

and unpublished, that supports or refutes the earlier findings

of a favourable effect of epidural analgesia on umbilical

artery pH and base excess values. It is hoped that this work

may improve understanding among those who inform preg-

nant women of this crucial aspect of neonatal welfare.

METHOD

A search was conducted, using a personal card index,

published meta-analyses1,2, earlier reviews10 and also Med-

line and Embase using key words ‘epidural AND (meper-

idine OR opioid analgesia) AND (umbilical OR neonatal

OR fetal) AND pH OR acid–base balance’, to identify

studies, both randomised and observational, comparing

epidural with other forms of labour analgesia, in which it

appeared that umbilical cord blood was sampled at birth.

The original papers were scrutinised and when incomplete

data were published, we attempted to contact the authors to

obtain additional unpublished umbilical artery pH and base

excess values.

Where raw data were available, the effect of epidural use

was re-estimated using linear regression with robust standard

errors11. Where the authors gave separate data for different

subgroups (different delivery types or second stage duration),

the mean and standard deviation were calculated for the

combined data and unpaired t tests with Satterthwaite’s

correction for unequal variance conducted. The treatments

effects were combined using random effect meta-analysis

both for randomised studies and for all studies12. Funnel plots

and Eggar’s test were used to check for heterogeneity and

publication bias13. All calculations were carried out in Stata

version 6.0 (Stata, College Station, Texas).

Table 1 describes the studies that have been included in

the meta-analysis6 – 9,14 – 22. These include some studies in

which old-fashioned epidural analgesia using local anaes-

thetic without opioid was used, and others using the more

up-to-date low dose combinations. Four of the early studies

were not randomised7 – 9,21,22 and one was quasi-random

(alternate allocation)14, but they have been included in the

full analyses for reasons that are given in the Discussion. In

one early study6, acid–base balance was measured in blood

drawn from ‘umbilical vessels’ at birth. In another, absolute

values were reported for fetal scalp blood during labour7,

while absolute values for cord blood data were not reported,

but only trends during the second stage were given8. The

author was able to provide the actual umbilical artery pH

and base excess values for many more babies than were in

the published series. All those for whom information on

parity, delivery type, treatment group and umbilical artery

acid–base data were available have been included in the

present analysis.

In one randomised trial, actual pH values were not

reported, but only numbers of babies with pH values less

than a particular cutoff point, but it was possible to obtain

the absolute values from the authors16. It was also possible

to obtain base excess values for individual patients in

this and another large randomised trial17. In the latter trial,

unpublished data for babies of non-compliant mothers were

also obtained. Most of the data presented for these two large

randomised trials are therefore previously unpublished.

In the study by Zador and Nilsson9, mean pH and base

excess values were given separately for those in whom the

second stage of labour was longer or shorter than 60

minutes. In the study reported by Shyken et al.22, results

were separated into those delivering vaginally and those by

caesarean section. The authors did not state whether the

women who delivered by caesarean section under epidural

anaesthesia had received epidural analgesia during labour.

For the purpose of this meta-analysis in both these studies

the data for all subjects were combined within each

treatment group to avoid bias by omitting babies delivered

by caesarean section.

RESULTS

Table 2 gives the values for pH in individual studies and

unpublished series. Meta-analysis of pH data is shown in

Fig. 1 for randomised trials and Fig. 2 for all studies. In the

randomised studies, there is a difference of þ0.009 (95%

CI þ0.002 to þ0.015, P ¼ 0.007) in umbilical artery pH

values between the epidural and systemic analgesia groups,

favouring epidural analgesia. There is no significant het-

erogeneity between treatments. When all studies are

included, the difference between treatment groups is not

significant (þ0.004; 95% CI �0.005 to þ0.014), but there

is significant heterogeneity ( P ¼ 0.015), revealing incon-

sistencies with the non-randomised studies, the main dif-

ference being between Shyken et al.22 and the rest, and a

lesser difference (in the opposite direction) between Deck-

ardt et al.21 and the rest.

Table 3 gives the values for base excess in individual

studies and unpublished series. Analyses of data for base

excess are shown in Figs 3 and 4. Only the small quasi-

randomised study of Jouppila and Hollmen14 does not show

improvement in base excess with epidural analgesia. The

difference between base excess values in the two treatment

groups is þ0.779 mEq/L (CI þ0.056 to þ1.502) for the

four randomised studies only and þ0.837 mEq/L (CI þ0.33

ANALGESIA IN LABOUR AND NEWBORN ACID –BASE BALANCE 1345

D RCOG 2002 Br J Obstet Gynaecol 109, pp. 1344–1353

Table 1. Studies included in meta-analysis.

Source Type of trial and subjects Nature of epidural n Nature of control analgesia n

Randomised trials

Thalme et al.6, Sweden Randomised, uncomplicated,

nullipara, established labour

Bupivacaine 0.25% þ epinephrine boluses 14/14 Intramuscular pethidine 100 mg þchlorpromazine

14/14

Jouppila and Hollmen14, Finland Alternate, induced, mixed

parturients

Bupivacaine 0.5% þ epinephrine boluses 14/14 Intramuscular pethidine 50 mg or

nothing

14/14

Thorp et al.15, USA Randomised, uncomplicated,

nullipara, spontaneous labour

Bupivacaine 0.25% bolus and 0.125%

infusion

48/48 Intravenous pethidine 75 mg þ promethazine

every 90 minutes, prn

45/45

Ramin et al.16, USA Randomised, uncomplicated,

mixed, spontaneous labour

Bupivacaine 0.25% bolus þ infusion of

0.125% with fentanyl

432/664 Intravenous pethidine 50 mg þ promethazine, prn 437/666

Sharma et al.17, USA Randomised, uncomplicated,

mixed, spontaneous labour

Bupivacaine 0.25% bolus þ infusion of

0.125% with fentanyl

243/358 PCA pethidine þ initial promethazine 259/357

Bofill et al.18, USA Randomised, uncomplicated,


Bupivacaine 0.25% boluses F fentanyl þinfusion of 0.125% with fentanyl

47/49 Intravenous butorphanol F promethazine

hourly, prn

39/51

Nikkola et al.19, Finland Randomised, uncomplicated,


Bupivacaine 0.5% boluses 10/10 PCA fentanyl 7/10

Clark et al.20, USA Randomised, uncomplicated,


Bupivacaine 0.25% with fentanyl bolus þinfusion of 0.125% with fentanyl

147/156 Intravenous pethidine every 90 minutes, prn 78/162

Non-randomised trials

Pearson and Davies7,8, UK Uncomplicated, mixed, induced

and spontaneous

Plain bupivacaine 0.25% boluses 17 Intramuscular pethidine 100 mg or more

F promazine

22

Pearson (unpublished) Mixed parturients, induced and

spontaneous

Plain bupivacaine 0.25% boluses 19m, 36p Intramuscular pethidine 100 mg or more

F promazine or nil

31m, 24p

Zador and Nilsson9, Sweden Uncomplicated, mixed parturients,

spontaneous labour

1% or 0.5% lidocaine þ epinephrine

boluses

49 Nitrous oxide F pethidine 50 mg in early

labour F diazepam

71

Deckardt et al.21, Germany Uncomplicated, mixed parturients,

induction not stated

Plain bupivacaine 0.25% boluses 15p Intramuscular pethidine/nitrous oxide or

no analgesia

16m, 15p

Shyken et al.22, USA Uncomplicated, mixed parturients,

spontaneous labour

Bupivacaine 0.125 or 0.25% F fentanyl

boluses

49 Narcotic/unspecified/general for

caesarean (n ¼ 16)

45

In the randomised trials, n is given as numbers that were protocol compliant over total numbers randomised to the treatment group.

‘mixed’ refers to parity; m ¼ multipara; p ¼ primipara.

In the study by Deckardt et al.21, the epidural group contained only primipara, whereas the control group was of mixed parity.

13

46

F.

RE

YN

OL

DS

ET

AL

.

DR

CO

G2

00

2B

rJ

Ob

stetG

yna

ecol

10

9,

pp

.1

34

4–

13

53

to þ1.343) for all eight studies. Again there is significant

heterogeneity with the non-randomised studies ( P ¼ 0.028).

In Tables 2 and 3, the data as originally published are

presented as well as the additional material. This reveals a

difference between the groups with long and short second

stages in the study by Zador and Nilsson9. The inclusion

of data for non-compliant as well as compliant women

in Sharma’s study has little impact on the results. In two

small randomised studies, neonatal data were also analysed

by intention-to-treat18,19. All participants are included by

intention-to-treat in the meta-analysis. In the study by

Pearson and Davies7,8, only results for those having appa-

rently normal labours and without major neonatal acidosis

were originally selected for publication. The results are thus

akin to those of a randomised trial, where it is normal

practice for uncomplicated labours only to be selected. The

110 participants included in the present analysis are not

exclusive and, as in real life, the epidural group included

more primiparae (Table 1), more inductions, more compli-

cated labours and more abnormal deliveries. Despite this

bias against epidurals, the results remain favourable,

although less so, to the epidural group.

Certain other studies that gave some data about funic

acid–base balance in epidural and non-epidural labours23 – 27

have been excluded for the reasons given in Table 4. The

findings in these excluded studies were, however, consistent

with those of this meta-analysis. A second randomised trial

by Sharma et al.28 that included umbilical artery pH was

published after this meta-analysis was completed.

DISCUSSION

Arterial pH reflects both a respiratory and a metabolic

component; thus, during labour, both maternal and fetal

values are influenced by the extent to which a mother in

pain is stimulated to hyperventilate. Systemic analgesia is

often not sufficiently effective to prevent this hyperventi-

lation. Base excess is therefore a more specific index of

metabolic acidosis, and hence, fetal hypoxia. Once born, a

baby can no longer rely on maternal ventilation to maintain

acid – base balance, and neonatal respiration may be

depressed following maternal systemic opioid analgesia6.

The presence of adequate buffer base is therefore important

to the newborn. Base excess is designed to be zero in

normal circumstances, but it is difficult to say what con-

stitutes normality in the perinatal period. The normal

umbilical artery pH is said to be >7.2 and base excess

Table 2. Umbilical artery pH values (mean [standard deviation]) recorded after labour with or without epidural analgesia.

n þ n Epidural Controls Differencey (95% CI)

Randomised trials

Thalme et al.*6 12 þ 12 7.28 [0.062] 7.27 [0.057] þ0.01 (�0.044 to þ0.064)

Jouppila and Hollmen14 14 þ 14 7.26 [0.08] 7.28 [0.06] �0.02 (�0.078 to þ0.038)

Thorp et al.15 47 þ 41 7.26 [0.06] 7.25 [0.06] þ0.01 (�0.016 to þ0.036)

Ramin (unpublished) 423 þ 411 7.27 [0.074] 7.26 [0.078] þ0.003 (�0.007 to þ0.014)

Sharma (compliant) 235 þ 214 7.26 [0.076] 7.25 [0.081] þ0.011 (�0.003 to þ0.026)

Sharma (all) 302 þ 271 7.26 [0.073] 7.24 [0.081] þ0.020 (þ0.007 to þ0.032)

Bofill et al.§18 49 þ 51 7.27 [0.06] 7.27 [0.08] þ0 (�0.028 to þ0.028)

Nikkola et al.§19 10 þ 8 7.24 [0.06] 7.23 [0.11] þ0.01 (�0.081 to þ0.101)

Clark et al.20 142 þ 66 7.24 [0.05] 7.23 [0.05] þ0.01 (�0.002 to þ0.002)

Observational studies

Pearson and Davies**7,8 16 þ 13 7.36 [0.044] 7.31 [0.036] þ0.05 (�0.006 to þ0.106)

Pearson (unpublished) 55 þ 55 7.26 [0.073] 7.25 [0.057] þ0.01 (�0.016 to þ0.033)

Zador and Nilsson9

Short 21 þ 56 7.26 [0.04] 7.27 [0.05] þ0.01 (�0.033 to þ0.013)

Long 26 þ 15 7.22 [0.05] 7.19 [0.08] þ0.03 (�0.019 to þ0.079)

Deckardt et al.21 15 þ 16? 7.29 [0.06] 7.21 [0.10] þ0.08 (þ0.013 to þ0.147)

Shyken et al.22

Vaginal 25 þ 29 7.26 [0.04] 7.29 [0.04] �0.03 (�0.05 to �0.008)

CS 24 þ 16 7.27 [0.04] 7.27 [0.04] 0 (�0.032 to þ0.032)

Numbers differ from those in Table 1, as pH was not measured in all cases.

Deckardt et al.21 data for primipara only, as multipara consisted of ‘controls’ only. Numbers analysed in the control group are uncertain.

Shyken et al.22 separated results by mode of delivery: vaginal or caesarean section (CS). All were labouring women, but it is not stated whether women given

epidural anaesthesia for caesarean section received it also during labour.

The results of Zador and Nilsson9 were separated into those with a second stage <60 minutes or > 60 minutes.

The results for Ramin (unpublished) and Sharma (unpublished) are drawn from the populations in the published trials15,19.

The results for Pearson in this table and Table 3 are unpublished and drawn from a larger population than in the published series.

* Sample from ‘umbilical vessels’.

** Fetal scalp blood sampled at full dilatation.y Epidural minus control.§ Analysed by intention-to-treat.



�10 to 0 mEq/L29. Table 3 shows that mean figures for

base excess vary widely between studies, while many

individual values fall outside the ‘normal’ range.

Epidural analgesia is used in the UK more frequently in

longer and more complex labours; thus inclusion of unran-

domised studies would be expected to introduce bias against

epidural analgesia, and reduce its apparent benefit to the

baby. When considering only randomised studies, epidural

analgesia has a favourable effect on funic pH, while if

unrandomised studies are included, the difference in pH

between treatment groups is not significant. Yet there is a

statistically significant improvement in base excess both in

randomised studies and when all studies are included. In

the control groups, therefore, the fetus suffers a more

severe metabolic acidosis, which is not fully compensated

by maternal hyperventilation. The small but significant

improvement with epidurals could represent a crucial

advantage for the at risk fetus, in whom surveys have

shown improvement in mortality with epidurals30, and it

certainly demonstrates an absence of any fetal detriment.

Methods of assessing the newborn

Many methods have been used to assess the wellbeing of

fetus and newborn. Perinatal mortality is now too low to be

a useful yardstick, particularly since it is common practice

to select only normal labours for randomised controlled

trials. An early observational study, however, found

reduced early neonatal mortality among low birthweight

infants in the epidural group compared with the non-

epidural group30. The Apgar score, a method that has been

used universally for many years, can be applied only within

a few minutes of birth, a stimulating time for the newborn.

A baby who has been exposed in utero to systemic drugs

may therefore perform well at birth but may later become

drowsy31. Nevertheless, a meta-analysis has demonstrated

that Apgar score may be adversely affected by systemic

compared with epidural analgesia1.

More recently, investigators have turned to neonatal

neurobehavioural scoring, which is applicable in the first

few hours or days of life rather than minutes, in the hope that

it may be more relevant and more sensitive. Doubt, however,

has been cast on the validity of the most popular approach32,

the Neurological and Adaptive Score (NACS), which was

designed to discriminate between drug effects and neonatal

asphyxia33. The differences between neurobehavioural

scores for epidural and systemic analgesia are variable and

inconsistent34 and a meta-analysis yielded no significant

difference35. The number of babies needing resuscitation

with naloxone is also sometimes reported17,27,28, which

reflects the doses of opioid used in the control groups.

The numbers needing intensive care are rarely reported in

randomised trials, and are also too few (0.065%)17–19 to

Fig. 1. Meta-analysis of difference (epidural minus control) between umbilical artery (**umbilical cord) pH values in eight randomised controlled trials of

epidural versus systemic opioid analgesia. The vertical line represents unity, the size of the box gives an estimate of the weighting of the study and the horizontal

lines are the 95% confidence intervals. [Weighting is based on the amount of information provided by the study, defined as 1/the square of the standard error of

the estimate.] There is a significant difference ( P ¼ 0.007) between the two treatment groups, favouring epidural analgesia. þIncludes unpublished data.

1348 F. REYNOLDS ET AL.


Fig. 2. Meta-analysis of difference (epidural minus control) between umbilical artery (**umbilical cord) pH values in all 12 studies of epidural versus

systemic opioid analgesia. See caption to Fig. 1 for further explanation. The difference is no longer significant. þIncludes unpublished data.

Table 3. Umbilical artery base excess (mEq/L) recorded after labour with or without epidural analgesia. Values are given as mean [SD].

n þ n Epidurals Controls Differencey (95% CI)

Randomised trials

Thalme et al.*6 12 þ 11 �7.9 [2.7] �10.1 [2.2] þ2.2 (�0.011 to þ4.411)

Jouppila and Hollmen14 14 þ 14 �7.6 [3.9] �6.8 [2.7] �0.80 (�3.55 to þ1.95)

Ramin (unpublished) 423 þ 413 �3.71 [2.55] �4.08 [2.96] þ0.38 (þ0.003 to þ0.75)

Sharma (compliant) 241 þ 217 �4.64 [3.19] �5.87 [3.21] þ1.23 (þ0.64 to þ1.82)

Sharma (all) 308 þ 274 �4.61 [3.02] �5.76 [3.08] þ1.14 (þ0.65 to 1.64)

Observational studies

Pearson and Davies**7,8 17 þ 13 �2.2 [2.06] �4.5 [1.80] þ2.3 (�0.58 to þ5.18)

Pearson (unpublished) 55 þ 55 �6.81 [3.92] �8.57 [3.16] þ1.76 (þ0.42 to þ3.11)

Zador and Nilsson9

Short 21 þ 56 �3.94 [1.7] �4.23 [1.94] þ0.29 (�0.64 to þ1.22)

Long 26 þ 15 �5.0 [2.26] �8.27 [2.35] þ3.27 (þ1.68 to þ4.86)

Deckardt et al.21 15 þ 16? �6.4 [2.2] �9.5 [4.5] þ3.0 (þ0.08 to þ5.92)

Shyken et al.22

Vaginal 25 þ 29 �3.9 [1.8] �4.8 [2.6] þ0.90 (�0.34 to þ2.14)

CS 24 þ 16 �4.4 [1.9] �3.7 [2.2] �0.7 (�2.02 to þ0.62)

Numbers differ from those in Table 1, as base excess was not measured in all cases.

Deckardt et al.21: data for primipara only, as multipara consisted of ‘controls’ only.

Shyken et al.22 separated results by mode of delivery: vaginal or caesarean section (CS). All were labouring women, but it is not stated whether women given

epidural anaesthesia for caesarean section received it also during labour.

The results of Zador and Nilsson9 were separated into those with a second stage <60 minutes or >60 minutes.

The results for Ramin (unpublished) and Sharma (unpublished) are drawn from the populations in the published trials15,19.

* Sample from ‘umbilical vessels’.

** Fetal scalp blood sampled at full dilatation.y Epidural minus control.



provide a sensitive index of neonatal welfare. Among the

more widely used methods of neonatal assessment, this only

leaves funic acid–base status, which has the additional

advantage that it is applicable to all labours.

The recent systematic review by Leighton and Halpern1

demonstrates that, despite a greater risk of maternal hypo-

tension, a mean increase of 15 minutes in the duration of

the second stage of labour, a 2-fold increase in the rate of

instrumental delivery, a 2.8-fold increase in the need for

oxytocin and a 5.6-fold increase in the risk of maternal

fever with the epidural analgesia, the odds ratio for a

1-minute Apgar score <7 was 0.54 ( P < 0.05), for a

5-minute Apgar score <7 was 0.54 (NS) and for needing

naloxone was 0.20 ( P < 0.01)—consistently in favour of

epidural analgesia. The incidence of low umbilical artery

pH or severe asphyxia did not differ significantly between

groups. That we found a significant difference in pH

values between groups may reflect the greater sensitivity

of using mean and standard deviations for meta-analysis

involving continuous variables.

Mechanisms by which labour analgesia may affect thebaby

Analgesia may affect the baby either because the drugs

reach active concentrations in the plasma (as is the case

with systemic analgesia), or because they affect maternal

physiology or biochemistry. The latter is the more likely

mechanism with epidural analgesia, which does not

depend for its effect on the presence of a drug in maternal

blood. The acid–base status of umbilical arterial blood

reflects the intrauterine environment and the efficiency of

transplacental exchange. It is therefore an appropriate

yardstick for the potential effects of epidural analgesia.

Several studies that followed the introduction of ion-

sensitive electrodes in the 1970s showed that painful labour

led not only to maternal respiratory alkalosis, but also to

progressive metabolic acidosis, which was transmitted to

the fetus during the first stage of labour6,7,36, while the

second stage was associated with further deterioration in

fetal acid–base balance6,8. These early studies, however,

suggested that in the presence of epidural analgesia, both

maternal and fetal metabolic acidosis were less severe and

the second stage was not associated with the same deteri-

oration in maternal or fetal acid–base balance8. Indeed, the

study by Zador and Nilsson9 showed that this apparent

protective effect of epidural analgesia was seen in the

subgroup in which the second stage lasted longer than 60

minutes. These findings suggested that, although epidural

analgesia may prolong the second stage1, this does not

necessarily result in worsening fetal acidosis.

Although hypotension may occur following regional

anaesthesia for caesarean section, and if prolonged may

Fig. 3. Meta-analysis of difference (epidural minus control) between umbilical artery (**umbilical cord) base excess data from four randomised controlled

trials of epidural versus systemic opioid analgesia. See caption to Fig. 1 for further explanation. There is a significant difference ( P ¼ 0.035) between the two

treatment groups, favouring epidural analgesia. þ Includes unpublished data.



have a detrimental effect on newborn acid–base bal-

ance37,38, the frequency and degree of hypotension are

less following the smaller doses required for analgesia in

labour. Numerous studies, cited by Hollmen39, have shown

using various techniques that sympathetic blockade from

lumbar epidural analgesia tends to improve maternal pla-

cental blood flow, and to have no detrimental effect

on umbilical blood flow40,41. Respiratory gas exchange,

which is flow dependent, should therefore be favoured.

Moreover, epidural analgesia without opioid prevents

episodes of maternal desaturation that otherwise occur

during painful labour, particularly when systemic analgesia

is used21,42. The reduction in maternal hyperventilation36,43

and in the levels of maternal stress hormones44,45, which

are both associated with regional analgesia, may also

improve transplacental gas exchange. Similar improve-

ments in stress responses are seen when low dose local

anaesthetic and opioid combinations are used46,47.

Thus, there are theoretical reasons why fetal acid–base

status during labour may both improve and deteriorate with

Fig. 4. Meta-analysis of difference (epidural minus control) between umbilical artery (**umbilical cord) base excess data from all eight studies of epidural

versus systemic opioid analgesia. See caption to Fig. 1 for further explanation. There is a significant difference ( P ¼ 0.001) between the two treatment

groups, favouring epidural analgesia. þIncludes unpublished data.

Table 4. Excluded studies.

Source Type of trial and reason for exclusion Nature of epidural n Nature of control analgesia n

Wiener et al.23, UK Unrandomised, retrospective selection

of easy vaginal deliveries of fit babies

only. UV pH

Plain bupivacaine 0.5%

boluses

11 Intramuscular pethidine 100– 300 mg

(total dose) only (n ¼ 18) or þ naloxone

at delivery (n ¼ 15)

33

Swanstrom and Bratteby24,

Sweden

Randomised but controls included

regional blocks. ‘Asphyxiated’ babies

in separate group. UA pH and BE

Lidocaine bolus or

infusion

35 ‘Control’ (nitrous oxide or nil)

Paracervical F pudendal block

Asphyxia

25

15

10

Philipsen and Jensen25,

Denmark

Randomised, UV pH: means and

ranges but no standard deviations*

Plain bupivacaine 0.375%

boluses

57 Pethidine 75 mg prn F nitrous

oxide

55

Muir et al.26, USA Randomised, reported only numbers

with ‘cord’ pH <7.15*

Patient controlled epidural

bupivacaine plus pethidine

28 Patient-controlled intravenous

pethidine

22

Gambling et al.27, USA Randomised, reported only numbers

with ‘cord’ pH <7.20*

Intrathecal sufentanil then

epidural infusion of

bupivacaine þ fentanyl

616 Intravenous pethidine 50 mg þpromethazine then meperidine

50 mg prn hourly

607

* Further details requested but not obtained.



epidural analgesia, and the present analysis suggests that, in

practice, it is the beneficial effects that prevail.

Improvements in analgesic techniques

In the past 20 years, the practice of epidural analgesia has

changed in that other drugs such as opioids are added to the

local anaesthetic. This enables a smaller dose of local

anaesthetic to be used, thereby reducing the local anaes-

thetic side effects of hypotension and muscle weakness and

also reducing the danger from misplacement of a dose.

Although large doses of opioids given epidurally or intra-

thecally have been known to cause neonatal depression34,

the modest doses in current use have no measurable adverse

neonatal effects48. The comparative lack of neonatal effect

of epidural as opposed to systemic opioids is reflected in the

greater need for resuscitation with naloxone in the non-

epidural groups1. Moreover, although the improvement in

funic acid–base balance was first demonstrated at a time

when local anaesthetic was used without opioid6 – 9, similar

beneficial effects have been found in the more recent

randomised trials in which low dose combinations were

used16 – 18,20. These four studies contributed 916 of the total

of 952 women in this meta-analysis who were randomised. It

appears that replacing systemic with modest doses of

neuraxial opioids not only produces superior analgesia but

also reduces neonatal complications. The COMET study49

did not compare epidural with systemic analgesia, but

rather, compared using plain bupivacaine epidural analgesia

with combined spinal epidural analgesia using epidural top-

up injections of local anaesthetic and opioid solutions, and

also with a continuous infusion of the mixture of drugs. The

continuous infusion of bupivacaine and opioid was associ-

ated with an increased need for neonatal resuscitation.

Women in this group, however, received the same total

dose of bupivacaine as in the first group, but received opioid

as well, which may explain these findings. These findings

disagree with those of Porter et al.48.

Changes in practice might also affect the control groups.

Table 1 shows that intramuscular pethidine was used in

only 28 women in the randomised trials, and various more

sophisticated intravenous and patient controlled analgesia

regimens were used in the remainder. These were success-

ful in improving compliance and reducing crossover, but

possibly at a cost of greater need for naloxone.

CONCLUSION

Expectant mothers can be reassured that, although epi-

dural analgesia may be associated with some short term

maternal side effects, it does not exacerbate fetal acidosis,

and if anything, may partially protect the fetus from fetal

hypoxia. It is important to dispel the notion that epidural

analgesia is in some way harmful to babies.

Acknowledgements

The authors would like to thank Jim Pearson for

entrusting to us his original lab book and for use of his

raw data.

References

1. Leighton BL, Halpern SH. The effects of epidural analgesia on the

progress of labor, maternal and neonatal outcomes: a systematic

review. Am J Obstet Gynecol 2002;186:S69– S77.

2. Zhang J, Klebanoff MA, DerSimonian R. Epidural analgesia in asso-

ciation with duration of labor and mode of delivery: a quantitative

review. Am J Obstet Gynecol 1999;180:970 –977.

3. Russell R, Reynolds F. Back pain, pregnancy and childbirth. BMJ

1997;314:1062–1063.

4. Halpern SH, Levine T, Wilson DB, MacDonell J, Katsiris SE,

Leighton BL. Effect of labor analgesia on breastfeeding success. Birth

1999;26:83– 88.

5. Polley LS. What’s new in obstetric anesthesia, 1999? Int J Obstet

Anesth 2001;10:46–54.

6. Thalme B, Belfrage P, Raabe N. Lumbar epidural analgesia in labour.

I: Acid– base balance and clinical condition of the mother, fetus and

newborn child. Acta Obstet Gynecol Scand 1974;53:27– 35.

7. Pearson JF, Davies P. The effect of continuous lumbar epidural

analgesia upon fetal acid–base status during the first stage of labour.

J Obstet Gynaecol Br Commonw 1974;81:971 –974.


analgesia upon fetal acid – base status during the second stage of

labour. J Obstet Gynaecol Br Commonw 1974;81:975 –979.

9. Zador G, Nilsson BA. Low dose intermittent epidural anaesthesia

with lidocaine for vaginal delivery. Acta Obstet Gynecol Scand

1974;(Suppl 34):17– 30.

10. Downing JW, Ramasubramanian R. Effects of analgesia and anaes-

thesia on fetal acid– base balance and respiratory gas exchange. In:

Reynolds F, editor. Effects on the Baby of Maternal Analgesia and

Anaesthesia. London: Saunders, 1993:125–147.

11. Gail MH, Tan WY, Piantadosi S. Tests for no treatment effects in

randomised clinical trials. Biometrika 1988;75:57– 64.

12. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin

Trials 1986;7:177 –188.

13. Egger M, Smith GD, Schneider M, Minder C. Bias in meta-analysis

detected by a simple, graphical test. BMJ 1997;315:629–634.

14. Jouppila R, Hollmen A. The effect of segmental epidural analgesia on

maternal and foetal acid– base balance, lactate, serum potassium and

creatinine phosphokinase during labour. Acta Anaesthesiol Scand

1976;20:259 –268.

15. Thorp JA, Hu DH, Albin RM, et al. The effect of intrapartum epidural

analgesia on nulliparous labor: a randomized controlled prospective

trial. Am J Obstet Gynecol 1993;169:851 –858.

16. Ramin SM, Gambling DR, Lucas MJ, Sharma SK, Sidawi JE, Leveno KJ.

Randomized trial of epidural versus intravenous analgesia during labor.

Obstet Gynecol 1995;86:783–789.

17. Sharma SK, Sidawi JE, Ramin SM, Lucas MJ, Leveno KJ, Cunning-

ham FG. Cesarean delivery: a randomized trial of epidural versus pa-

tient-controlled meperidine analgesia during labor. Anesthesiology

1997;87:487 –494.

18. Bofill JA, Vincent RD, Ross EL, et al. Nulliparous active labor,

epidural analgesia and cesarean delivery for dystocia. Am J Obstet

Gynecol 1997;177:1465– 1470.

19. Nikkola EM, Ekblad UU, Kero PO, Alihanka JJM, Salonen MAO.

Intravenous fentanyl PCA during labour. Can J Anaesth 1997;

44:1248– 1255.

20. Clark A, Carr D, Lloyd G, Cook V, Spinnato J. The influence of



epidural analgesia on cesarean delivery rates: a randomized prospec-

tive clinical trial. Am J Obstet Gynecol 1998;179:1527– 1533.

21. Deckardt R, Fernbacher PM, Schneider KTM, Graeff H. Maternal

arterial oxygen saturation during labor and delivery: pain dependent

alterations and effects on the newborn. Obstet Gynecol 1987;70:

21–25.

22. Shyken JM, Smeltzer JS, Baxi LV, et al. A comparison of the effect

of epidural, general and no anesthesia on funic acid–base values by

stage of labor and type of delivery. Am J Obstet Gynecol 1990;

163:802– 807.

23. Wiener PC, Hogg MI, Rosen M. Neonatal respiration, feeding and

neurobehavioural state. Anaesthesia 1979;34:996– 1004.

24. Swanstrom S, Bratteby LE. Metabolic effects of regional analgesia

and asphyxia in the newborn infant during the first 2 hours after birth.

III: Adjustment of arterial blood gases and acid– base balance. Acta

Paediatr Scand 1981;70:811– 818.

25. Philipsen T, Jensen NH. Epidural block or parenteral pethidine as

analgesic in labour; a randomized study concerning progress in labour

and instrumental deliveries. Eur J Obstet Gynecol Reprod Biol 1989;

30:27– 33.

26. Muir HA, Shukla R, Liston R, Writer D. Randomized trial of labor

analgesia: a pilot study to compare patient-controlled intravenous

analgesia with patient-controlled epidural analgesia to determine if

analgesic method affects delivery outcome [abstract]. Can J Anaesth

1996;43:A60.

27. Gambling DR, Sharma SK, Ramin SM, et al. A randomized study of

combined spinal – epidural analgesia versus intravenous meperidine

during labor. Anesthesiology 1998;89:1336–1344.

28. Sharma SK, Alexander JM, Messick G, et al. A randomized trial of

epidural analgesia versus intravenous meperidine analgesia during

labor in nulliparous women. Anesthesiology 2002;96:546 –551.

29. Martin RW, McColgin SG. Evaluation of fetal and neonatal acid –

base status. Obstet Gynecol Clin North Am 1990;17:223– 233.

30. David H, Rosen M. Perinatal mortality after epidural analgesia.

Anaesthesia 1976;31:1054– 1059.

31. Brazelton TB. Effects of prenatal drugs in the behavior of the neonate.

Am J Psychiatry 1970;126:1261– 1266.

32. Camann W, Brazelton TB. Use and abuse of neonatal neurobehavio-

ral testing [editorial]. Anesthesiology 2000;92:3– 5.

33. Amiel Tison C, Barrier G, Shnider SM, et al. A new neurologic and

adaptive capacity scoring system for evaluating obstetric medications

in full term newborns. Anesthesiology 1982;56:340– 350.

34. Reynolds F, Porter J. Neonatal effects of regional analgesia. In:

Reynolds F, editor. Regional Analgesia in Obstetrics: A Millennium

Update. London: Springer, 2000:237–252.

35. Halpern SH, Leighton BL, Ohlsson A, Barrett JFR, Rice A. Effect of

epidural vs parenteral opioid analgesia on the progress of labor. JAMA

1998;280:2105– 2110.


analgesia on the acid–base status of maternal arterial blood during

the first stage of labour. J Obstet Gynaecol Br Commonw 1973;

80:218– 224.

37. Antoine C, Young BK. Fetal lactic acidosis with epidural anesthesia.

Am J Obstet Gynecol 1982;142:55 –59.

38. Corke BC, Datta S, Ostheimer GW, Weiss JB, Alper MH. Spinal

anaesthesia for caesarean section. The influence of hypotension on

neonatal outcome. Anaesthesia 1982;37:658– 662.

39. Hollmen A. Effect of regional anaesthesia on placental blood flow. In:

Reynolds F, editor. Effects on the Baby of Maternal Anaesthesia and

Analgesia. London: Saunders, 1993:67– 87.

40. Hughes AB, Devol LD, Wakefield ML, et al. The effect of epidural

anesthesia on the Doppler velocimetry of umbilical and uterine

arteries in normal term labor. Obstet Gynecol 1990;75:809– 812.

41. Marx GF, Patel S, Berman JA, Farmakides G, Schulman H. Umbili-

cal blood flow velocity waveforms in different positions and with

epidural analgesia. Obstet Gynecol 1986;68:61 –64.

42. Griffin R, Reynolds F. Maternal hypoxaemia during labour and de-

livery; the influence of analgesia and effect on neonatal outcome.

Anaesthesia 1995;50:151– 156.

43. Motoyama EK, Rivard G, Acheson F, Cook CD. Adverse effect of

maternal hyperventilation on the foetus. Lancet 1966;i:286– 288.

44. Shnider SM, Abboud TK, Artal R. Maternal catecholamines decrease

during labor after epidural anesthesia. Am J Obstet Gynecol 1983;147:

13–15.

45. Westgren M, Lindahl SGE, Norden NE. Maternal and fetal endocrine

stress response at vaginal delivery with and without epidural block.

J Perinat Med 1986;14:235– 241.

46. Cascio M, Pygon B, Bernett C, Ramanathan S. Effects of intrathecal

fentanyl on plasma catecholamine levels in term laboring paerturients.

Anesthesiology 1995;83:A493.

47. Eberle R, Kinsella SM, Arrison E, et al. Maternal plasma catechol-

amine concentrations after labor analgesia with intrathecal sufentanil.

Anesthesiology 1995;83:A971.

48. Porter JS, Bonello E, Reynolds F. The effect of epidural fentanyl on

neonatal respiration. Anesthesiology 1998;89:79– 85.

49. COMET study group. Effect of low-dose mobile versus traditional

epidural techniques on mode of delivery: a randomised controlled

trial. Lancet 2001;358:19– 23.

Accepted 2 October 2002



analgesia in labour and fetal acid–base balance: a meta-analysis comparing epidural with systemic...

Documents