Accelerat ing t he world's research.

A randomised controlled trial of care

of the perineum during second stage
of normal labour
Jo Garcia
Bjog-an International Journal of Obstetrics and Gynaecology

Cite this paper Downloaded from Academia.edu 

Get the citation in MLA, APA, or Chicago styles

Related papers Download a PDF Pack of t he best relat ed papers 

Perineal t echniques during t he second st age of labour for reducing perineal t rauma
Liv Reinar, Mirjam Lukasse

Perineal t echniques during t he second st age of labour for reducing perineal t rauma (Review) Perineal …
cahla moloughney

Anal sphinct er t ears: prospect ive st udy of obst et ric risk fact ors
lars ladfors
 zyxwvut
British Journal of Obstetrics and Gynaecology
December 1998, Vol. 105, pp. 1262-1272

zyxwvutsrq
A randomised controlled trial of care of the perineum

zyxwvutsr
during second stage of normal labour

zyxw
*Rona McCandlish Research Fellow, *Ursula Bowler Administrative Co-ordinator, *Hedwig van Asten Data Manager,
*Georgina Berridge Computer Programmer, **Cathy Winter Local Midwife Co-ordinator, ?Lesley Sames Local
Midwife Co-ordinator, *Jo Garcia Social Scientist, $Mary Renfrew Professor, §Diana Elbourne Senior Lecturer
*National Perinatal Epidemiology Unit, Oxford; **Southmead Health Services N H S Trust, Bristol; t Royal Berkshire and Battle
Hospital NHS Trust, Reading; $Mother and Infant Research Unit, University of Leeds; $Medical Statistics Unit, London School
of Hygiene and Tropical Medicine, London

Objective To compare the effect of two methods of perineal management used during spontaneous
vaginal delivery on the prevalence of perineal pain reported at 10 days after birth.
Design Randomised controlled trial.
Setting Two English maternity care units.
Sample 5471 women who gave birth between December 1994and December 1996.
Methods At the end of the second stage of labour women were allocated to either the ‘hands on’
method, in which the midwife’s hands put pressure on the baby’s head and support (‘guard’) the

zyx
perineum; lateral flexion is then used to facilitate delivery of the shoulders, or the ‘hands
poised’ method, in which the midwife keeps her hands poised, not touching the head or
perineum, allowing spontaneous delivery of the shoulders.
Main outcome measure Perineal pain in the previous 24 hours reported by women in self-
administered questionnaire 10 days after birth.

zyxwv
Results Questionnaires were completed by 97% of women at 10 days after birth. 910 (34-1%u)
women in the ‘hands poised’ group reported pain in the previous 24 hours compared with 823
(31.1%) in the ‘hands on’group (RR 1-10,95%CI 1.01 to 1.18: absolute difference 3%’ 0.5% to
5%, P = 0.02). The rate of episiotomy was significantly lower in the ‘hands poised’ group (RR
0.79, 99% C1 0.65 to 0.96, P = 0.008) but the rate of manual removal of placenta was
significantly higher (RR 1-69,99%CI 1.02 to 2.78; P = 0.008). There were no other statistically
significant differences detected between the two methods.
Conclusion The reduction in pain observed in the ‘hands on’ group was statistically significant and
the difference detected potentially affects a substantial number of women. These results provide
evidence to enable individual women and health professionals to decide which perineal
management is preferable.

INTRODUCTION postnatal pain and morbidity. Women and health

professionals place a high value on minimising
Few women escape perineal pain or discomfort in the
trauma and reducing adverse outcome^',^. A range of
early days after a vaginal birth. This pain can make
approaches to perineal care during labour are in
ordinary activities such as walking and sitting diffi-
use5,6but most have never been adequately evalu-
cult, breastfeeding problematic, and the physical
ated5y7. Midwifery and obstetric textbooks reflect
demands of caring for a new baby harder’. For some
conflicting opinions about which are In the
women the misery of pain can continue for weeks or
UK, midwives are the principal caregivers at the
even months following birth and can delay the
majority of vaginal births and the study reported
resumption of intercourse*,’.
here was developed in response to midwives’ concern
Trauma incurred during delivery contributes to
about the lack of evidence regarding the impact on

zyxwvutsrq
Correspondence: Ms R. McCandlish, National Perinatal
EpidemiologyUnit, Radcliffe Infirmary, Oxford OX2 6HE, UK.

1262
postnatal pain of two widely used methods of per-
ineal management. The techniques were ‘hands on’
and ‘hands poised’, defined below.

0 RCOG 1998 British Journal of Obstetrics and Gynaecology

METHODS
Protocol zyxwvutsrq C A R E O F T H E P E R I N E U M D U R I N G SECOND STAGE L A B O U R
The study was designed and co-ordinated by the
National Perinatal Epidemiology Unit (NPEU),
Oxford. The two methods of perineal care compared
were identified in a systematic review of the
literaturelo and clarified at a meeting with a multi-
disciplinary group of maternity care professionals
and consumer representatives. They are:
1. The ‘hands on’ method, in which the midwife’s
hands are used to put pressure on the baby’s head
in the belief that flexion will be increased, and to
support (‘guard’) the perineum, and to use lateral
flexion to facilitate the delivery of the shoulders.
2. The ‘hands poised’ method, in which the midwife
keeps her hands poised, prepared to put light pres-
sure on the baby’s head in case of rapid expulsion,
but not to touch the head or perineum otherwise
and to allow spontaneous delivery of the shoulders.
The primary hypothesis to be tested was that a
policy in which the midwife keeps her ‘hands poised’
at the time of delivery reduces the occurrence of per-
ineal pain in the previous 24 hours reported by the
mother ten days after birth when compared with a
zyxwvutsr
policy of ‘hands on’.
Based on an estimate of 23% of women reporting
perineal pain at 10 days after birth” 8500 women
would need to be randomised to detect a reduction
from 23% to 20.5% (relative risk 0.89) with power of
80%(alpha = 0-05,two sided test).
Outcomes
The pre-specified primary outcome was any perineal
pain during the previous 24 hours as reported by
zyxw
women on a questionnaire to be completed at 10
days after birth. Secondary measures of outcomes
were maternal pain 2 days, 10 days and 3 months
after birth, urinary and bowel problems at 10 days
and 3 months, dyspareunia at 3 months, and other
maternal and neonatal morbidity.
Recruitment and postnatal follow up was carried
out by midwives working in three NHS Trusts and
randomisation took place in the two hospitals which
served these Trusts. In both hospitals there were
approximately 5500 births per year. A local midwife
was employed half-time to co-ordinate the trial in
z
each centre. Approval for the study was given by the
relevant research ethics committees.
zyxwvutsrqpon
Clinical preparation
For three months before randomisation began a
preparation programme for midwifery staff was
0 RCOG 1998 Br J Obstet GynaecollO5, 1262-1272
organised at each centre. This included showing staff
a video about the trial which described the back-
ground to the study and showed two births featuring
the ‘hands on’ method and two featuring the ‘hands
poised’ method. Clinical preparation was led by the
local midwife co-ordinator. In addition midwives
who felt they needed support to improve their skills
1263
in either technique asked colleagues confident about
using either ‘hands on’ or ‘hands poised’ to attend
births with them and mentor them in developing
competence.
Around the time randomisation began, all mid-
wives in both centres received a postal questionnaire
from the NPEU asking whether in their current prac-
tice (i.e. outwith the trial) they would be most likely
to use the ‘hands on’ or ‘hands poised technique.
When randomisation ended the survey was repeated.
Eligibility and recruitment
Written information about the trial was given to
women by midwives at routine antenatal visits and
classes. Women were eligible to join if they had a sin-
gleton pregnancy with cephalic presentation, antici-
pated a normal birth, did not plan a waterbirth, had
no elective episiotomy prescribed, did not plan adop-
tion and gave oral consent to participate. Once the
midwife had discussed participation, she attached a
label to the woman’s hand-held maternity notes indi-
cating whether or not the woman had been recruited
to the trial. Women were assured of the right to with-
draw from the trial at any point without prejudice to
their care or relationship with their care-givers.
Eligibility and consent to participate were re-checked
once a woman was in established labour and those
who were less than 37 weeks of gestation were
excluded at that point.
Randomisation
Randomisation was in a ratio of 1/1 within balanced
blocks of 4 to 8, stratified by centre. Details of the
allocated group were given on coloured cards con-
tained in sequentially numbered, opaque, sealed
envelopes. These were prepared at the NPEU and
kept in an agreed location on each labour ward.
Randomisation took place at the end of second stage
of labour when the midwife considered a vaginal
birth was imminent. To enter a woman into the
study, the midwife opened the next consecutively
numbered envelope. If an envelope was not opened,
the reason for non-use was recorded by the midwife
who had drawn it. All envelopes, whether used or
unused, were returned to the NPEU. Unopened
envelopes were not returned for re-use in the centres.
 zyxwvutsrq
zyxwvutsrqpo
zyxwvuts
1264 R. M C C A N D L I S H E T A L .
Concealmentand compliance
Allocation could not be concealed from midwives who
randomised women; however staff were asked not to
volunteer information about allocation to women. If a
woman asked and was told this information, the mid-
wife recorded this in the trial data form. The allocated
group was not recorded in postnatal notes and mid-
wives involved in postnatal follow up were unlikely to
have been present when a woman was randomised.
Midwives were instructed to follow the allocated
policy wherever possible, while maintaining responsi-
bility to manage care safely and appropriately. Where
the management was other than that allocated the
midwife documented the reason.
Data collection and management
Prior to the start of the trial, data collection tools,
some of which were modified from previous stud-
ies"*'*, and written information were tested with 25
women who had recently given birth and 30 mid-
wives working in a clinical centre which otherwise
did not take part in the study.
A register was kept on both labour wards in which
midwives recorded whether or not each woman
admitted was randomised and reasons for non-inclu-
sion. Midwives collected data for each woman ran-
domised immediately after delivery, at 2 days and at
between 9-1 1 days after birth. Data were collected by
zyxwvuts
midwives after delivery and at 9-11 days after birth
for any babies who required admission to any cate-
gory of care other than routine postnatal care; this
included high dependency/intensive/special/transi-
tional care. The reason a baby required such care was
not recorded on the data forms so this information
was later extracted from routine hospital records by
one of four midwives. Each woman was also asked to
complete a questionnaire at 2 days and 10 days after
birth and a system was established to try to ensure
that women who had a neonatal death were not
asked to complete any trial questionnaires. All ques-
tionnaires administered within the first 11 days were
returned to the local midwife co-ordinator and once
follow up for a participant was as complete as possi-
ble, questionnaires were transferred to the NPEU.
Three months after birth, a postal questionnaire
was sent from the NPEU to each participant. All
data were double entered at the NPEU.
Data analysis
An independent Data Monitoring Committee
reviewed data collected up to 10 days after birth from
the first 1000 women randomised. Subsequently the
target sample size was adjusted to 5500 women. This
reduction was based on the overall prevalence of
pain at 10 days being higher (34%) than had origi-
nally been expected and the fact that accrual was less
than anticipated. To achieve the revised sample size
funding was awarded to extend the randomisation
zyxw
phase in both centres.
Analysis of the primary and secondary measures
of outcome was based on the groups as randomly
allocated in an 'intention-to-treat' comparison. For
the primary outcome, statistical significance was
taken at the 5% level, and for secondary measures
significance was taken at the 1% level. Relative risks
(RR), percentage differences, and confidence inter-
vals (CI) are shown for the primary outcome, and
thereafter only for differences reaching significance
at 99% level where appropriate. Percentages in the
tables are based on the numbers at the top of each
table, unless otherwise stated. Numbers do not
always add exactly to those totals because small
numbers of missing answers have been excluded.
Pre-specified stratified analyses were carried out
according to the centre in which randomisation took
place, by previous vaginal delivery or not, by use of
epidural anaesthesia during second stage, by the
attending midwife's pre-stated preference for the
techniques, and by the first 6 months of randomisa-
tion and the remainder of randomisation.
RESULTS
In one centre randomisation took place over a 25
month period between December 1994 and December
1996 and in the other during 20 months from March
1995 until November 1996. During the overall ran-
domisation period, 18,458 women gave birth in the
collaborating centres of whom 547 1 (2906%) were ran-
domised into the trial. Most women not entered into
the trial were either ineligible and/or had not been
recruited antenatally (Fig. 1). Data were collected by
the attending midwife for all 547 1 women immediately
after delivery. Nearly all subsequent data from both
midwives and mothers were collected at, or close to,
the planned times of two days, ten days and three
months after birth (data available, not shown here).
Comparabilityof trial groups
The two groups were well balanced in respect of
socio-demographic characteristics and other prog-
zyxw
nostic factors at trial entry (Table 1). One fifth of the
women had epidural analgesia. Nearly all the deliver-
zy
ies were vaginal and spontaneous, and conducted
by midwives, of whom 77.8% expressed a practice
preference at the time of beginning participation in
the trial.
0 RCOG 1998 Br J Obstet Gynaecol105, 1262-1272
 z
zyxw
C A R E OF THE P E R I N E U M D U R I N G SECOND STAGE L A B O U R

Total deliveries
1265

zyxw
n=18458(100)

Not randomised’ Not randomised

n = 12 987 (70.4) n=5471 (29.6)

Rendomised

Non recruited antenatally 4529

Planned instrumental 2213

delivery Hands poised Hands on

I
Planned caesarean 2152 Data avallable from midwives’
n = 2740 (100) /I = 2731 (1 00)
section questionnaireat entry

zyx
Maternal refusal 1312

Non-cephalic
presentation
462

U Izy
zyxwvuts
n= 2685 (98.2)
Data available from mothers‘
questlonnaire at 2 days
n= 2686 (98.4)

wzyxwvu
Multiple pregnancy 227

Planned birth in water 148 Data available from midwives’

n=2685 (98.1)
questionnaireat 2 days
n= 2680 (98.1)
Intrauterine death 65 .

Episiotomyprescribed 51

Other 2672

more than ane reason was given for

non-randomisationfor many women
1 n= 2669 (97.4)

n=2689 (98.1)
Data available from mothers’
questionnaireat 10 days

Data available from midwives’

questionnaire at 10 days

Data available from mothers’

n= 2519 (91.9) questionnaire at 3 months

zyxw
Fig. 1. Participation and response rates - n (YO)).

Compliance

zyxwv
group (953%). The main reasons reported for actual
Compliance with the allocated management, both as care differing from the allocated policy in the ‘hands
reported directly by the attending midwife and as cal- poised’ group related to concerns about fetal distress

zyxwvutsrqpon
culated by her report of specific components of care,
is shown in Table 2. Full compliance was lower in the
‘hands poised’ group (70.1%) than in the ‘hands on’
0 RCOG 1998 Br J Obstet Gynaecol 105, 1262-1272
and the midwives’ perception of the need to facilitate
delivery. The main reason given in the ‘hands on’
group was too rapid a delivery.
 zyxwvutsrq
zyxwvutsrqpon
1266 R. zyxwvutsrqpo
zyxwvutsrqp
zyxwvutsrq
zyxwvu
M C C A N D L I S H ET AL.

Table 1. Characteristics of the women at trial entry. Values are given as n (YO),
mean [SD] and median { IQR}.

Random allocation
~

Poised n = 2740 On n = 2731

Age (years) 29.1 [5.1] 29.2 [4.9]

Living with husbandpartner 2588 (94.5) 2579 (94.4)
Gestation (completed weeks) 40.0 [1.0] 40.0 I0.91
Primiparous 1051 (38.4) 997 (36.5)
No previous vaginal delivery 1086 (39.6) 1037 (38.0)
Previous vaginal delivery 1488 (54.3) 1519 (55.6)
When randomised
1st stage 170 (6.2) 167 (6.1)
2nd stage 2566 (93.6) 2552 (93.4)
Envelope opened in error* 1 (0) 3 (0.1)
Not known 3 (0.1) 9 (0.3)
Length of 1st stage (minutes) 330 (205-510) 330 (207-505)
Delivery** conducted by midwife whose
initial practice preference was:
On 1133 (51.4) 1151 (50.0)
Poised 487 (21.5) 463 (20.6)
Both equally 157 (6.9) 121 (5.4)
Not known 491 (21.6) 514 (22.8)

*This occurred due to over enthusiasm by midwives at the early stage in the study. As opening the envelope had been deemed to be the
point of randomisation these women have remained in the analysis. Re-analysis excluding these women did not, however, alter any
outcomes.
**n = 2249,2268, not all deliveries conducted by midwives.

Table 2. Self-reported management for delivery. Values are given as n (%).

Random allocation

Poised n = 2740 On n = 2731

~~~~ ~

Midwife reported compliance with allocated policy 2007 (73.2) 2608 (95.5)
Components of management reported by midwife
Hands on the perineum before head delivered 328 (12.0) 2645 (96.9)
Hands on the baby’s head before head delivered 411 (15.0) 2684 (98.3)
Hands on the baby’s head to facilitate delivery of shoulders 636 (23.2) 2662 (97.5)
Compliance (as calculated from components above)
Hands on 204 (7.4) 2603 (95.3)
Hands poised 1922 (70.1) 24 (0.9)
Mixed 612 (22.3) 101 (3.7)

lower in the ‘hands poised’ group (RR 0-79, 99% CI

Labour and delivery
Outcomes around the time of labour and delivery are
shown in Table 3. The length of time between ran-
domisation and delivery was slightly shorter in the
‘hands poised’ group, and the length of second stage
slightly longer, but the differences did not reach sta-
tistical significance at the 1% level. The rate of man-
0.65 to 0.96 P = 0.003). The main reasons given by
midwives for performing an episiotomy were fetal
distress, a wish to avoid uncontrolled trauma, and a
tight or rigid perineum.
Effect on perineal pain

ual removal of placenta was significantly higher in
the ‘hands poised’ group (RR 1.69, 99% CI 1.02 to
2.78; P = 0.008) in the ‘hands poised’ group. The inci-
dence of any perineal and genital tract trauma was
similar in the two trial arms in terms of site, severity,
and repair. Use of episiotomy was low overall, and
z
Table 4 gives information about perineal pain. For
the primary outcome based on questionnaire com-
pletion around the 10th day, 910 (34.1%) women in
the ‘hands poised’ group reported having experi-
enced some pain in the previous 24 hours at the time
of questionnaire completion compared with 823
0 RCOG 1998 Br J Obstet Gynaecol 105, 1262-1272
 zyxwvut
zyxwvutsrqp
zyxwvutsrqpo
zyxwvut C A R E OF T H E PERINEUM D U R I N G S E C O N D STAGE LABOUR

Table 3. Labour and delivery. Values are given as n (%), mean [SE] and median { IQR} .
1267

Random allocation

Poised n = 2740 On n=2731

Interval between randomisation and delivery (minutes) 8 {5-15) 9 {5-15)

Main method of pushing at end of 2nd stage
Directed 832 (30.4) 894 (32.7)
Spontaneous 652 (23.8) 597 (21.9)
Both 1249 (45.6) 1229 (45.0)
Epidural analgesia during 2nd stage of labour 601 (21.9) 562 (20.6)
Method of delivery
Spontaneous vaginal 2717 (99.2) 2706 (99.1)
Forceps, Ventouse and breech 19 (0.7) 25 (0.9)
Caesarean section 4 (0.2) 0
Maternal position at vaginal delivery*
Upright (standing, kneeling, squatting, sitting upright) 506 (18.5) 493 (18.1)
Semi recumbent 1875 (68.4) 1869 (68.4)
Lying down (lateral, supine and lithotomy) 273 (10.0) 305 (11.2)
All fours 68 (2.5) 58 (2.1)
14 (0.5) 6 (0.2)

zy
Other
Baby delivered by:
Midwife 2268 (82.8) 2249 (82.4)
Student midwife 421 (15.4) 430 (15.7)
Duration of 2nd stage (minutes) 23 (10-56) 22 {10-52)
Duration of 3rd stage (minutes) 6 (5-9) 6 {5-8)
Blood loss at delivery 2 500 mL 143 (5.2) 123 (45)
Manual removal of placenta 71 (2.6) 42 (1.5)
Any trauma 2356 (86.0) 2321 (85.0)
Perineal trauma* *
1st degree 802 (29.3) 813 (29.8)
2nd degree 1011 (36.9) 1002 (36.6)
3rd4th degree 40 (1.5) 31 (1.2)
Location of other trauma
Vaginal trauma 1686 (61.5) 1671 (61.2)
Labial trauma 987 (36.0) 935 (34.2)
Clitoral trauma 39 (1.4) 34 (1.3)
Other trauma? 38 (1.4) 36 (1.3)
Trauma sutured 1636 (59.7) 1605 (58.8)
Suture materials used for repair
Catgut 1056 1037
Other (including Dexon, Vicryl) 63 1 617
Interrupted suture to repair skin 1183 1189
Time spent suturing trauma (minutes) 20.5 [0.3] 20.0 [0.3]

zyxwv
*n = 2736,2731.
**Including episiotomy n = 280 (10.2), 351 (12.9).
+Includes:cervical, urethral trauma.

(31.1%) in the ‘hands on’ group (RR 1.10 95% CI

1-01 to 1.18: absolute difference 3%’ 0.5% to 5%’
Neonatal outcomes
P = 0.02). The difference was mainly in the ‘mild’cat- There was no significant difference in baby outcomes
egory of pain. Ninety-nine women (3.70/0)in the (Table 5 ) .
‘hands poised’ group and 89 (3.4%) in the ‘hands on’
group said they experienced pain ‘all the time’. At
Maternal postnatal outcomes

zyxwv
around two days postpartum pain during the previ-
ous 24 hours was reported by 1871 (70%) women in Table 6 presents maternal outcomes (other than

zyxwvutsrqp
the ‘hands poised’ group compared with 71.3% in the
‘hands on’ group. Pain in the past week was reported
at three months by 171 (6.8%) women in the ‘hands
poised’ group and 176 (7%) in the ‘hands on’ group.
0 RCOG 1998 Br J Obstet Gynaecol 105, 1262-1272
pain) at around two days and ten days reported by
women and midwives and outcomes reported by
women at three months after the birth. Trauma and
healing assessed by midwives were similar in both
1268

zyxwv
zyxwvutsrqpo
zyxw
R. M C C A N D L I S H ET AL.

Table 4. Postnatal pain in and around the perineum as reported by the women. Values are given as n (“/)*.

zyxwvutsrqp n = 2669
10 days
Random allocation

Poised On
n = 2647
Poised
2 days
Random allocation

n = 2685
On
n = 2686
Poised
n = 2519
3 months
Random allocation

On
n = 2486

zyxwv
In previous 24 hours
None 1748 (65.5) 1816 (68.6) 807 (30.0) 761 (28.3)
Somet 910 (34.1) 823 (31.1) 1871 (70.0) 1915 (71.3)
Mild 627 (23.5) 554 (20.9) 738 (27.5) 773 (28.8)
Moderate 246 (9.2) 233 (8.8) 994 (37.0) 1004 (37.4)
Severe 37 (1.4) 36 (1.4) 139 (5.2) 138 (5.1)
Current
None 2098 (78.6) 2119 (80.1) 1200 (44.7) 1125 (41.9)
Some 568 (21.3) 524 (19.8) 1481 (55.2) 1555 (57.9)
Mild 455 (17.0) 405 (15.3) 993 (37.0) 1078 (40.1)
Moderate 101 (3.8) 109 (4.1) 465 (17.3) 456 (17.0)
Severe 12 (0.4) 10 (0.4) 23 (0.9) 21 (0.8)
In last week
None 2314 (91.9) 2296 (92.4)
Some 171 (6.8) 176 (7.1)
Mild 113 (4.5) 124 (5.0)
Moderate 53 (2.1) 46 (1.9)
Severe 5 (0.2) 6 (0.2)
Pain ‘all the time’ 99 (3.7) 89 (3.4) 311 (11.6) 313 (11.7) 9 (0.4) 11 (0.4)

*The questionnaire for mothers administered at 2 days, 10 days and 3 months after birth included the following information about the
questionnaire ‘...A lot of the questions are about any pain you may have had in or around your perineum (the area around your birth
canal) ...’ a) In the last 24 hours, (in the last week at 3 months) have you had any pain in or around your perineum? b) If yes, is it painful
all the time? c) Would you describe the pain you are feeling right now as: nonelmildlmoderatelsevere. d) Would you describe the worst
pain you have had in the past 24 hours (in the last week at 3 months) as: mild/moderate/severe.
tRR 1.10,95%CI 1.01 to 1.18,P=0.02.

groups (data available, not shown) and no statisti-
cally significant differences were observed in these or
any of the outcomes recorded by midwives. Similarly,
no statistically significant differences were observed
in outcomes reported by women at two days, ten days
or three months after birth.
Stratified analyses
The stratified analyses for the primary outcome are
shown in Fig. 2. In general these reflect the results for
the primary outcome with women in the ‘hands on’
group reporting less pain than those in the ‘hands
poised’. The differences between the randomly allo-
cated policies are statistically significant only in the
later of the two randomisation time periods, among
allocated groups) the proportion of women who had
pain at the time of completing the ten day question-
naire was higher in one centre than the other, in the
second time period than the first six months, in
women who had not had a previous vaginal birth
compared with those who had, in those with an
epidural during the second stage of labour compared
with those without, and in those whose midwife had
an initial practice preference for ‘hands on’ com-
pared with those who had a preference for ‘hands
poised’.
DISCUSSION
The results of this trial indicate that the ‘hands on’

zyxw
women with no previous vaginal birth, and for policy of perineal management is associated with less
women with no epidural analgesia in the second perineal pain in the previous 24 hours at 10 days
stage of labour. There is a trend in the opposite
direction for the stratum in which care was delivered
by a midwife whose practice preference at the start of
randomisation was for the ‘hands poised’ method,
with evidence that such preferences may modify the
effects of the policies (heterogeneity test P = 0.03).
Comparing between the strata (ie not between the
zyxw
after birth compared with the ‘hands poised’ method

z
(P= 0.02). This benefit was also reflected in the strati-
fied analyses; even though there was a trend in favour
of ‘hands poised’ in the stratum in which the midwife
had stated an a priori preference for the ‘hands
poised’ method of care, the confidence interval also
included the possibility of benefit from ‘hands on’,
0 RCOG 1998 Br J Obstet Gynaecd 105, 1262-1272
 zyxwvut z
zyxwvutsrq
zyxwvutsrq C A R E O F T H E PERINEUM D U R I N G SECOND STAGE LABOUR

Table 5. Neonatal outcomes. Values are given as n (%).

1269

zyxwvutsrqp
Random allocation

Poised n = 2740 On n = 2731

Apgar score
At 1 minute I 6 90 (3.3) 100 (3.7)
At 5 minutes I 6 9 (0.3) 9 (0.3)
Care at birth*
Oxygen 479 457
Ambubag 109 1 I5
Suction 571 57 1
Intubation 26 25
Narcan 106 84
Admitted to additional care category** within
first 11 dayst 132 118
Reason1
Asphyxia 8 5
Respiratory complications 25 23
Meconium aspiration 8 4
Other (includes congenital anomaly, jaundice, IUGR) 84 84
Not known 7 2
Breastfeeding as reported by mother at 2 days after birth8
Full breastfeeding 1515 (56.4) 1516 (56.4)
Partial breastfeeding 371 (13.8) 321 (11.9)
Breastfeeding as reported by mother at 10 days after birth@
Full breastfeeding 1483 (52.9) 1428 (53.9)
Partial breast feeding 233 (8-8) 216 (8.2)
Method of feeding baby milk feeds at time of
questionnaire completion as reported by mother
at 3 months after birth1
Fully breastfeeding 571 (22.7) 594 (23.9)
Partial breastfeeding 396 (15.7) 364 (14.6)

zyxwvuts
*Some babies had more than one type of care.
**Additional care category: that is any category above ‘routine postnatal’ care. This includes highlintensive care, SCBU,
transitional care.
+14babies had 2nd admissions in 1st 1 1 days (7, 11) 4 babies had 3 admissions in 1st 11 days (1,3).
tCollected by one of 4 midwives from routine hospital records.
@n = 2685,2686.
% = 2669,2647.
In = 2519,2486.

In interpreting these results, a number of issues
require consideration. Staff participating in this trial
integrated recruitment and randomisation as part of
routine midwifery care. A third of women who gave
birth in the collaborating centres took part. Reasons
given for non-randomisation were dominated by the
fact that many had not been recruited during preg-
nancy or were not eligible for the trial. Unfortunately
these data were collected in such a way that it was not
possible to separately identify these reasons and/or
maternal refusals. Despite this it is likely that the
results of the HOOP study are generally applicable
use were monitored. Very few envelopes were opened
out of numerical order. Randomisation produced
groups which were very similar at trial entry in terms
of descriptive and known prognostic factors. Levels
of data collection and completeness were very high
and similar in both groups.
It was impossible to conceal the allocated method
from the midwife who randomised. Thus the poten-
tial existed for reporting bias in data collected imme-
diately after the birth. However, given the large num-
ber of midwives who contributed data, systematic
distortion is unlikely. The principal outcome was

zyxwv
to routine practice among similar women in other pain as reported by the women at 10 days and
care settings. midwives reported that a third in each group had

zyxwvutsrqp
The results are unlikely to be explained by selec-
tion biases. Randomisation details were obtained by
opening an opaque, sealed envelope and the date and
time of opening all envelopes or the reason for non-
0 RCOG 1998 Br J Obstet Gynaecol 105, 1262-1272
been told about their allocation. If knowledge about
allocation influenced maternal assessment of pain it
is unlikely to have been significant enough to obscure
a large treatment effect in either direction.
 zyxwvutsrq
zyxwvutsrqpo
zyxwvutsrq
zyxwvutsrqp
zyxwvut
1270 R. M C C A N D L I S H ET AL.

Table 6. Maternal morbidity. Values are given as n (%).

Random allocation

Poised n = 2519 On n = 2486

Urinary problem(s) as reported by women at 10 days* 238 (8.9) 197 (7.4)

Incontinence 61 48
Urgency 46 38
Frequency 48 42
Other 143 111
Bowel problem(s) as reported by women at 10 days* 676 (25.3) 604 (22.8)
Pain on opening bowels 337 288
Constipation 366 322
Flatus 214 184
Haemorrhoids 294 285
Incontinence 14 20
Other 38 39
Outpatient visit for any perineaYlabiaYvagina1
problem(s) as reported by women at 3 months 23 (0.9) 18 (0.7)
Treatment for perineaYlabiaYvagina1trauma between 10 days
and 3 months as reported by women at 3 months 108 (4.3) 116 (4.7)
Antibiotics for infection 99 100
Dyspareunia as reported by women at 3 months
Severe 17 (0.7) 14 (0.6)
Moderate 148 (5.9) 108 (4-3)
Mild 211 (8.4) 220 (8.8)
Dyspareunia at first, now resolved 559 (22.2) 557 (22-4)
No sex now 85 (3.3) 113 (45)
No sex yet 331 (13.1) 346 (13.9)

zyxwvutsrqp
No pain at all 970 (38.5) 941 (37.9)
Not known 192 (7.6) 184 (7.4)
Urinary problem(s) in the past week as reported by women
at 3 months 607 (24.0) 602 (24.2)
Bowel problem(s) in past week as reported by women at

zyxwvutsr
3 months 414 (16.4) 392 (15.8)
EPDS** 2 13 as reported by women at 3 months 179 (7.1) 180 (7.2)

*n = 2669,2647.
**Edinburgh Postnatal Depression Scale. Cox J, Holden J, Sagovsky R. Detection of postnatal depression: development of the 10 item
Edinburgh Depression Scale (EDPS). Br JPsych 1987; 150: 782-786.

Hands poised Hands on

RR,95% CI
All women 92112669 83112647 1.10 (1.01, 1.18)
RR, 99% CI
Centre A 58811681 52111663 1.12 (0.98, 1.27)
Centre B 3331988 3101984 1.07 (0.91, 1.26)
Randomisation
1st 6 months 2081637 2141640 0.98 (0.80, I .20)
Remainder 713/2032 61712007 1.14 (1.02, 1.28)
Vaginal birth
No previous 55111088 45911043 1.15 (1.02, 1.30)
Previous 37011581 37211604 1.01 (0.86, 1.19)
Epidural in 2nd stage
No 67312084 60312100 1.12 (1.00, 1.27)
Yes 2481585 2281547 1.02 (0.85, 1.22)
Midwives preference at start of trial
On 409/1 I 1 1 34611125 1.20 (1.03, 1.40)
i 1

zyxwv
Poised 1491495 1481461 0.94 (0.73. 1.20)

o,5 Hands poised better Hands on better

1.5

Fig. 2. Pain in previous 24 hours at 10 days stratified by centre, period of randomisation, previous vaginal birth, epidural in 2nd stage,
practice preference at start of randomisation among midwives who randomised.

0 RCOG 1998 Br J Obstet Gynaecol 105, 1262-1272

 C A R E OF THE P E R I N E U M D U R I N G SECOND STAGE L A B O U R
In terms of compliance with the allocated policies,
the study relied on self-reporting of practice. Direct
observation of care during 28 deliveries was carried
out by two independent observers in an attempt to
develop a tool to validate self-reported care. The fea-
sibility of observing a large number of deliveries as
well as the likelihood of an observer effect curtailed
this work. Self-reported compliance was higher in the
‘hands on’ compared with the ‘hands poised’ group,
reflecting the pragmatic approach of this trial.
Midwives were expected to attempt to use the poli-
cies in the context of normal care and therefore in
some cases it was necessary to modify care from the
allocated policy. The reasons given for deviation were
clinically plausible, for example most ‘deviations’
from the ‘hands poised’ allocation involved the mid-
wife placing her hands on the baby’s head to deliver
the shoulders. It is possible that the effect on pain
might have been different if the level of compliance
in the ‘hands poised’ group had been higher than the
70% reported. Any estimate of a change either in
the size or direction of effect can, however, only be
speculative.
Before the trial began most midwives were more
familiar with the ‘hands on’ method and less confi-
dent about the ‘hands poised’ method, although it is
unlikely that the beneficial effect of the ‘hands on’
policy was entirely due to midwives unfamiliarity
with the ‘hands poised’ approach. We considered this
issue in two main ways. Firstly, in the stratified analy-
sis presenting midwives’ practice preference at the
start of the trial, the benefit of ‘hands on’ is less
evident in the stratum in which practitioners pre-
ferred ‘hands poised’ (Fig. 2). However the wide con-
fidence intervals surrounding this estimate do not
exclude the possibility of this benefit. Secondly, the
survey of preference conducted at the end of the trial
indicated that, overall, the midwives in the two units
at that point in time (there had been some staff
changes over the period) felt more inclined to use
‘hands poised’ than those surveyed at the start (data
available, not shown). One might expect then, that
‘hands on’ would be of greater benefit in the early
part of the trial when midwives were more familiar
with it. This is not supported by the stratified analy-
sis comparing the first six months of randomisation
with the rest of the trial (Fig. 2).
One explanation for the similarity in outcomes, for
z
example in the duration of second stage and the
degree and severity of trauma, is that there was a
dilution of the effects of the policies on pain because
zyxwvutsrq
the components of care converged. The ‘hands on’
zyxwvutsrqpo
policy may have become less intrusive involving per-
haps, only a gentle touch to aid assessment of
progress. The ‘light touch on the baby’s head in case
0 RCOG 1998 Br J Obstet GynuecollO5, 1262-1272
of rapid expulsion’ sanctioned in the ‘hands poised’
policy could have become firmer. Alternatively the
method of delivering the baby’s head may have less
1271 zy
impact on postnatal morbidity than management
used to deliver the shoulders. As described above,
midwives cited the need to deliver the shoulders most
often as the reason for deviating from the ‘hands
poised’ policy. It could be that a mixture of the two
approaches is associated with more postnatal pain.
It is possible that the levels of morbidity observed
were influenced by the effect of participation in a
study. Midwives were required to make a particularly
thorough record of genital tract and perineal trauma
and, in common with previous researchI3,the preva-
lence of any trauma may appear higher than is
routinely recorded. Certainly in HOOP the preva-
lence of perineal pain in the previous 24 hours
reported at 10 days after birth is higher than previous
studies about perineal management” but the propor-
tion of moderate and severe pain is in keeping with
those trials. It may be that the detailed self-adminis-
tered questionnaires disposed women to report ‘dis-
comfort’as mild pain.
The hypothesis that a policy of ‘hands poised’
would result in longer second stage of labour and
therefore more asphyxial problems for neonates was
not confirmed in this trial. The methods used for col-
lecting these neonatal data might however have
resulted in some mis-reporting of morbidity.
Unexpected differences between the trial groups in
terms of use of episiotomy and manual removal of
the placenta were observed. In fact the use of epi-
siotomy recorded in HOOP was strikingly low over-
all, and the higher episiotomy rate in the ‘hands on’
group is consistent with the hypothesis that a policy
of using a particular intervention brings with it an
increased risk that other interventions will ensueI4.
Reviews of randomised controlled trials comparing
liberal and restricted use of episiotomyI5 show that
more liberal use of episiotomy is associated with a
different pattern of trauma but no difference in pain.
It is possible that the slightly lower episiotomy rate in
the ‘hands poised’ group could have led to more pain
but this is not easy to support from the systematic
reviews.
Manual removal of placenta was significantly
higher in the ‘hands poised’ than ‘hands on’ group
but surprisingly was not associated with a signifi-
cantly higher rate of postpartum haemorrhage. This
may be a chance finding, although the fact that it was
consistent in both centres makes this less likely.
Practice guidelines in both centres recommended
routine active management of the third stage of
labour but information about the timing of oxytocic
administration and about other aspects of third stage
 zyxwvutsr
zyxwvutsrqpon
zyxwvutsrq
1272 R. MCCANDLISH ET AL.
management was not collected in the trial. Therefore
whether or not a ‘hands poised’ delivery resulted in a
more or less interventionist approach to third stage
care, for example by delaying administration of an
oxytocic, is not known, and such data cannot be
obtained from hospital records.
In conclusion, the reduction in pain observed in the
‘hands on’ group was statistically significant and the
difference detected potentially affects a substantial
number of women. The difference related mainly to
mild pain at 10 days. At three months, pain and other
outcomes were not found to differ between the two
groups. Nevertheless, it is hard to recommend that
units move to a policy of ‘hands poised’. If ‘hands
poised’ care is used, audit, particularly of third stage
care, should be maintained and midwifery education
should prepare practitioners to use either technique.
Midwives who are currently in practice and who are
requested to offer ‘hands poised’ care should have
access to clinical support to enable them to develop
appropriate skills. Overall the results provide evidence
to enable individual women and health professionals
to make informed decisions about which of these
perineal managements is preferable for them.
Acknowledgements
We thank the women and the midwives who partici-
pated in the trial. The trial was conducted by staff in
Southmead Health Services NHS Trust, Frenchay
Healthcare NHS Trust, Royal Berkshire and Battle
Hospital NHS Trust, West Berkshire Priority Care
Service NHS Trust, Severn NHS Trust, United
Bristol Healthcare NHS Trust, Weston Area Health
NHS Trust and Glan Hafren NHS Trust. The trial
was funded by the Medical Research Council; during
the first 20 months partial funding of the salary for
the Local Midwife Co-ordinator in Bristol was pro-
vided by Southmead Health Services NHS Trust. Jo
Garcia is funded by the Department of Health,
Diana Elbourne is employed by the London School
of Hygiene and Tropical Medicine and Mary
zyx
Renfrew by the University of Leeds.
zyxw
References
Greenshields W, Hulme H. A survey of women’s experiences and
zyxw
midwives’ practices. In: Oliver S, editor. The Perineum in
Childbirth. London: National Childbirth Trust, 1993.
Sleep J, Grant A. West Berkshire perineal management trial: three
year follow up. BMJ 1987; 295: 749-751.
Glazener CM. Sexual function after childbirth: women’s experi-
ences, persistent morbidity and lack of professional recognition.
Br J Obstet Gynaecoll997; 104: 330-335.
Stamp G. Care of the perineum in the second stage of labour: a
study of views and practices of Australian midwives. Midwifery
1997; 13: 10CL104.
Floud E. Protecting the perineum in childbirth 3: perineal care
today. Br JMidwifery 1994; 2: 356361.
Albers LL, Anderson D, Cragin L et al. Factors leading to per-
ineal trauma in childbirth. J Nurse Midwifery 1996; 41: 269-276.
zyx
McCandlish R. Care during the second stage of labour. In:
Alexander J, Levy V, Roth C, editors. Midwifery Practice Core
Topics. London: Macmillan, 1997: 103-108.
Bennett VR, Brown LK, editors. Myles Textbook for Midwives.
12th edition. Edinburgh: Churchill Livingstone, 1993: 208-210.
Cunningham FG, MacDonald PC, Gant NF et al., editors.
William’s Obstetrics. Stamford, C T Appleton & Lange, 1997:
333-335.
10 Renfrew MJ, Hannah W,Albers L, Floyd E. Practices that min-
imise trauma to the genital tract in childbirth: a systematic review
of the literature. Birth 1998; 25: 143-160.
11 Sleep J, Grant A, Garcia J, Elbourne D, Spencer J, Chalmers I.
West Berkshire perineal management trial. BMJ 1984; 289:
587-590.
12 Gordon B, Mackrodt C, Fern E, Truesdale A, Ayers S, Grant A.
The lpswich Childbirth Study: 1. A randomised controlled evalua-
tion of two stage postpartum perineal repair leaving the skin
unsutured. Br J Obstet Gynaecol1998; 105: 4 3 5 4 0 .
13 Stewart P, H a a n E, Calder AA. A randomised trial to evaluate
the use of a birth chair for delivery Lancet 1983; 1: 1296-1299.
14 Chalmers I. Obstetric practice and outcome of pregnancy in
Cardiff residents, 1965-1973. BMJ 1976; 1: 735-738.
15 Carroli G, Belizan J, Stamp G. Episiotomy policies in vaginal
zyxwvu
births. (Cochrane Review) in The Cochrane Library, Issue 2.
zyxw
Oxford: Update Software; 1998. Updated quarterly.
ReceivedZS February I998
Returned f o r revision 12 May 1998
Accepted 28 Jury I998
0 RCOG 1998 Br J Obstet GynaecollO5, 1262-1272