British Journal of Obstetrics and Gynaecology
July 2000, Vol 107, pp. 926-931
Anal sphincter tears: prospective study of obstetric
*E. Samuelsson Midwife, *L. Ladfors Consultant (Obstetrics), *U. B. Wennerholm Consultant (Obstetrics),
*B. GIreberg Midwife, tK. Nyberg Senior lecturer, *H. Hagberg Professor (Obstetrics and Pennatology)
*Pennatal Centel; Department of Obstetrics and Gynaecology, Sahlgrenska University HospitaVOstra, Goteborg;
?Department of Nursing, College of Health and Caring Sciences, University of Goteborg, Sweden
Objective To evaluate intrapartum risk factors for anal sphincter tear.
Design A prospective observational study.
Setting Delivery unit at the University Hospital in Goteborg, Sweden.
Participants 2883 consecutive women delivered vaginally during the period between 1995 and 1997.
Information was obtained from patient records and from especially designed protocols which were
completed during and after childbirth.
Main outcome measures Anal sphincter (third and fourth degree) tear.
Results Anal sphincter tear occurred in 95 of 2883 women (3.3%). Univariate analysis demonstrated
that the risk of anal sphincter tear was increased by nulliparity, high infant weight, lack of manual per-
ineal protection, deficient visualisation of perineum, severe perineal oedema, long duration of deliv-
ery and especially protracted second phase and bear down, use of oxytocin, episiotomy, vacuum
extraction and epidural anaesthesia. After analysis with stepwise logistic regression, reported as odds
ratio, 95% confidence interval, the following factors remained independently associated with anal
sphincter tear: slight perineal oedema (0-40, 0-26-0.64); manual perineal protection (0.49,
0.28-0.86); short duration of bear down (0.47, 0-24-0.91); no visualisation of perineum (2.77,
1.36-5.63); parity (0.59,0.40-0.89); and high infant weight (2.02, 1.30-3.16). Analysis of variance
showed that manual perineal protection had a stronger influence on lowering the frequency, and lack
of visualisation of perineum and infant weight had a stronger influence on raising the frequency, of
anal sphincter tears in nulliparous compared with parous women.
Conclusions Perineal oedema, poor ocular surveillance of perineum, deficient perineal protection dur-
ing delivery, protracted final phase of the second stage, parity and high infant weight all constitute
independent risk factors for anal sphincter tear. Such information is essential in order to reduce
perineal trauma during childbirth.
Vaginal delivery is the major cause of anal incontinence
in women’*2. Recent studies using anal endosonography
have revealed occult sphincter injury after vaginal deliv-
ery even in the absence of sphincter tears diagnosed at
birth”. However, women suffering from overt third or
fourth degree injury are at particular risk and even if
these tears are recognised and repaired at birth the out-
come is often unfa~ourable*,~-~.
30%-50% of these women suffer from chronic anal
incontinence, dyspareunia, faecal urgency or perineal
paink7. Therefore, attention should be focused primarily
Correspondence: Dr H. Hagberg, Perinatal Center, Department
of Obstetrics and Gynaecology, Sahlgrenska University Hospital/
Ostra, 41685 Goteborg, Sweden.
on improvements of obstetric practice to minimise per-
ineal trauma and, subsequently, reduce the number of
severe sphincter lacerations’s*.
The incidence of sphincter injury is estimated to
0.5%-2.5% in centres where mediolateral episiotomy is
practised2. In Sweden we have experienced a gradual
increase of the incidence of these injuries from 0.7% in
19824, to 2.9% in 1996 (Official Statistics of Sweden,
Medical Birth Registry, Stockholm, Sweden). Such a
marked increase of sphincter tears cannot be attributed
to altered frequency of previously recognised risk fac-
tors5.X-14 as no major change has occurred with regard to
instrumental delivery, high birthweight, nulliparity, epi-
siotomy, epidural anaesthesia, shoulder dystocia or fetal
presentation. There has been a major shift, however, in
obstetric practice where alternative birth positions pre-
dominate, the woman is encouraged to choose her own
0 RCOG 2000 British Journal o f Obstetrics and Gynaecology
RISK FACTORS FOR ANAL SPHINCTER TEARS
way of delivery, and manual perineal protection is not
practised to the same extent as before. Our hypothesis
was that these modifications in obstetric practice, includ-
ing reduced ocular surveillance of perineum, lack of
manual perineal protection, complicated birth positions
and deficient support and assistance at delivery, are asso-
ciated with an increased risk of sphincter lacerations.
Most previous studies are retrospective and detailed
information about obstetrical practice, progress of deliv-
ery or birth position is often lacking. In this study we
report 2883 consecutive deliveries which were followed
prospectively with meticulous registration of the pro-
cess of labour. The aim was to find novel risk factors
and based on that information, ultimately, strategies for
prevention of severe sphincter tears.
Delivery was assisted by midwives under ordinary cir-
cumstances, instrumental deliveries were all performed
by obstetricians, and in case of a perineal laceration in
proximity to the anal sphincter the midwives consulted
the obstetrician for assessment of the extent of injury.
Rupture of the anal sphincter (partial or complete) was
diagnosed as third degree and a complete tear combined
with laceration of the anal canal or rectum as a fourth
degree tear, but in our statistical analysis, no distinction
was made between third and fourth degree lacerations.
The study was not started until all midwives/obstetri-
cians at the unit were well informed.
The following information was obtained from
records and the especially designed protocols: age; par-
ity; time passed after previous birth; previous epi-
siotomy; previous sphincter tear; time of first stage of
labour (start of delivery defined as cervix dilated
2 3 cm); time of second stage of labour; time of bear
down; time from point when the presenting part was
visible during contraction to delivery; time from pre-
senting part visible during and between contractions
until delivery; number of uterine contractions from pas-
sage of the fetal head through perineum to delivery of
the infant; delivery position (there were three pre-deter-
mined positions: lateral, semi-recumbent or kneeling);
use of manual perineal protection, ocular surveillance
of perineum (excellentlpartiaho ocular surveillance);
degree of perineal oedema (slight/moderate/severe);
manual assistance during delivery of the fetal head
and/or shoulders; year when the midwife graduated;
professional experience (counted in years) of the mid-
wife; use of vacuum extractiodforceps; expeditious
delivery due to fear of fetal asphyxia; the woman’s abil-
ity to relax between contractions (slightlacceptableho
relaxation); use of oxytocin during first stage and/or
second stage; mode of anaesthesia or other methods to
relieve pain (epidural, nitrous oxide, acupuncture,
pudendal blockade, local anaesthesia and/or warm tow-
els placed towards perineum), vaginallperineal tear
including thirafourth degree laceration; episiotomy;
indication for episiotomy ; suture technique; fetal pre-
sentation; the infants weight and head circumference.
The association between the above factors and the occur-
rence of anal sphincter tears was tested with univariate
logistic regression. Stepwise logistic regression analysis
was used to suggest the predictor variables, which con-
sisted of apparently independent and significant predic-
tors of sphincter tears. Odds ratios with 95% confidence
intervals were calculated. All factors were processed by
stepwise logistic regression analysis, including those
factors that were not significantly associated with out-
come in the univariate model. Analysis of variance was
used to compare outcomes between nulliparous and mul-
tiparous women. Statistical software (SAS, Version 6.12)
was used for the analysis. Differences between women
who were included and those who were not included
were analysed using Wilcoxon rank-sum test, and pro-
portions were compared using Fisher’s exact test.
A l l women (n = 3723) who planned for a vaginal deliv-
ery at Sahlgrenska University Hospital in Goteborg were
recruited prospectively between 1995 and 1997, includ-
ing multiple births and breech deliveries. A total of 2883
women were admitted corresponding to 77.4% of all
those admitted during the time period. Those who were
not included tended to be women for whom detailed
information was not completed by the hospital unit.
No significant differences were found between
women recruited and not recruited concerning: maternal
age, gestational age, duration of delivery, use of oxy-
tocin, occurrence of sphincter tears and birthweight.
P a r i t y was lower and the rate of episiotomy higher in
women not included, but the differences were small.
The use of vacuum extraction was twice as high in the
group not included, compared with the study group
(Table 1). However, the rate of anal sphincter tear in
those delivered by vacuum was similar in those who
were included (10.6%) and not included (9.1%), sug-
gesting that not only vacuum deliveries with a low risk
of sphincter laceration were selected to the study group.
The obstetric characteristics of the cases included are
given in Table 2. Anal sphincter tear occurred in
95/2883 women (3-3%), including 88 cases of third
degree and seven of fourth degree tear. There was no
significant association between sphincter tears and the
following variables: age of woman; number of contrac-
tions between passage of the head and delivery of the
0 RCOG 2000 Br J Obstet Gynaecol 107,926-931
928 E. SAMUELSSON ET AL.
Table 1. Comparison between women included i n the study and those who were excluded. Values are given as % or mean (SD), unless
Gestational age at birth (weeks)
Maternal age (years)
Duration of delivery (hours)
Use of oxytocin during first stage of delivery
Use of oxytocin during second stage of delivery
Rate of thirdfourth degree tears
Rate of episiotomy
Rate of vacuum extraction
(n = 2883)
(n = 840)
infant; manual assistance for delivery of shoulders; pro-
fessional experience of the midwife; expeditious delivery
due to suspected asphyxia; the woman’s ability to relax
between contractions; fetal presentation; and head cir-
cumference. Most women (90.2%) were delivered in the
three pre-determined positions, but birth position did
not appear to be a risk factor (Table 3).
The univariate analysis (Table 3) demonstrated that
the estimated risk of anal sphincter tear was increased
by nulliparity, long duration of the first stage of labour
and especially of the second stage, bearing down or the
very last phase of the pushing phase. Risk of tearing was
increased by lack of manual perineal protection, no
visualisation of perineum, severe perineal oedema, vac-
uum extraction, use of oxytocin, epidural anaesthesia,
episiotomy and high infant weight (Table 3). All factors
Table 2. Obstetric characteristics of women included in the study.
Values are given as n (%).
(n = 1296)
(n = 1587)
were processed by stepwise logistic regression analysis,
including those factors that were not significantly asso-
ciated with outcome in the univariate model (Table 4 ) .
Slight perineal oedema, short duration of bearing down,
manual perineal protection, and parity all reduced the
risk of sphincter tear whereas no visualisation of per-
ineum and high infant birthweight independently
increased the risk.
Table 4 shows the comparison in outcomes between
nulliparous and multiparous women. Significant inter-
actions were found between anal sphincter tears and no
visualisation of perineum, manual perineal protection
and birthweight. Manual perineal protection had a
stronger influence on lowering the frequency of anal
sphincter tears in nulliparous women. No visualisation
of perineum and high birthweight had a stronger influ-
ence of raising the frequency of anal sphincter tears in
nulliparous than in parous women.
The prospective design of the study was a prerequisite
in order to obtain critical information concerning mode
of delivery and the way labour was assisted. Such data
are usually not possible to retrieve from standard birth
records and much effort was put into the preparation of
the protocols with active participation of the staff to
ensure high quality of the acquired information. The aim
was to recruit all women admitted consecutively during
the study period to avoid selection bias, but only 77.4%
of those admitted were entered into the study due to
shortage of staff during busy hours and the cumbersome
nature of the study protocols. Analysis of those
excluded from the study showed that this group was
comparable to the study group with regard to most
important characteristics. There was, however, a some-
what higher frequency of vacuum deliveries among the
0 RCOG 2000 Br J Obstet Gynaecol 107,926-93 1
RISK FACTORS FOR ANAL SPHINCTER TEARS
Table 3. Univariate analyses of the association between intrapartum variables and sphincter tear. Values are given as n/nTota,
Sphincter tear incidence
OR (95% CI)
t 2 previous deliveries
1 previous delivery
Infant weight (kg)
Duration of first stage of labour
Duration of second stage of labour
< 30 min
2 90 min
Duration of bear down
2 60 min
Time from presenting part visible in vulva during contraction to delivery
< 5 min
2 25 min
Manual perineal protection
Visualisation of perineum during last phase of bear down
Use of oxytocin during first stage
Use of oxytocin during second stage
4/489 (0.8 )
2.40 ( 1.23-4.69)
1 s o
1 s o
2.9 1 (1.8M.70)
0 RCOG 2000 Br J Obstet Gynaecol 107,92693 1
930 E. SAMUELSSON ET AL.
Table 4. Risk factors significantly associated with the occurrence of anal sphincter tear as analysed with logistic multiple regression (columns
2 and 3). The difference between nulliparous and multiparous women was analysed with analysis of variance and expressed as main effect
(difference in the level of the association (Main)) and interaction effect (difference in the slope of the interaction (Inter)) in columns 4-7.
OR (95% CI)
Main P Inter P
No visualisation of perineum
Slight perineal oedema
Pushing time < 30 min
Manual perineal protection
Infant weight (kg)
women who were excluded which may have constituted
significant bias, as vacuum deliveries during busy hours
(not included) may be associated with a higher risk of
sphincter tear than vacuum deliveries that took place
during less busy circumstances (included). A similar
occurrence of sphincter tear in both groups suggests,
however, that the study group was likely to be represen-
Some of the risk factors presently found to be associ-
ated with sphincter tear confirm previous reports: high
birthweight5,l4; episiotomy".'5.'h; vacuum
t i ~ n ~ ~ " ~ ~ ' ; epidural anaesthe~ial~;
However, stepwise logistic regression analysis revealed
that only high birthweight remained an independent risk
factor. Surprisingly, vacuum extraction was not an inde-
pendent risk factor". A comparison showed that vacuum
deliveries differed from nonvacuum deliveries in the
sense that duration of the first stage, second stage and
bearing down were longer, and manual perineal protec-
tion was not provided as often, suggesting that vacuum
delivery per se does not contribute as a cause of sphinc-
ter tear. This agrees with recent report^^^'".'^^'^, implying
that modification of obstetric practice may indeed
reduce the risks of the vacuum procedure. The use of
forceps was too low in our population to allow analysis
of its importance as a risk factor of anal sphincter tear.
Lack of manual protection or suboptimal visualisa-
tion of the perineum and perineal oedema were also sig-
nificantly associated with the occurrence of anal
sphincter tears in both the univariate and multiple logis-
tic regression analysis. The duration of the second stage
of labour has previously been reported to be unrelated to
sphincter tear~"J~.~~. However, in the paper by Bek and
Laurberg" there was an association with an unadjusted
odds ratio of 4.06 (2-5-6.6) which was adjusted to 1.6
(0.9-2.3) in the multiple logistic regression analysis.
This is in agreement with our data where only the asso-
ciation between duration of bearing down and sphincter
tear remained significant in the logistic regression
model (Table 4), whereas the duration the entire second
phase was not an independent factor of importance. It is
also reasonable to assume that the very last phase of sec-
and use of oxytocin",
ond stage is most critical with respect to perineal
trauma, and our study is the first to distinguish the dif-
ferent sub-phases of the second stage.
It is interesting to note that the risk of sphincter tear
increased after a phase of pushing exceeding 30 min
(Tables 3 and 4). The study does not provide informa-
tion on whether the risks are reduced by earlier delivery
by use of oxytocin at that point or instrumental delivery.
In order to allow analysis of the importance of deliv-
ery position, more than 90% of women gave birth in one
of three positions. We did not, however, find any corre-
lation between delivery position and outcome. Accord-
ing to a recent meta-analysis of four studiesz0, the risk of
sphincter tear was lower in upright, compared with
recumbent position, during second stage of labour. In a
previous study2I we found a markedly higher Occurrence
of sphincter lacerations in upright standing, compared
with upright sitting delivery positions. Further studies
are needed to clarify the impact of delivery positions on
There are data to suggest that nulliparous women are
at particular risk of sphincter injury during delivery
(high risk of third/fourth degree tears and occult
sphincter injury) whereas the risk of additional injury
during subsequent deliveries may be rather limited3.22.
We used analysis of variance to assess the comparative
effects of risk factors in parous and nulliparous women
in the present study. A stronger influence of perineal
protection, visualisation of perineum and high birth-
weight on the frequency of sphincter tear in nulli-
parous compared with parous women was found.
Apparently, high birthweight is not as important a risk
factor in parous women, which agrees with a recent
report14 which also used a model of stepwise logistic
Our data also suggest that visualisation and manual
protection of perineum are important in preventing
sphincter injury, especially in nulliparous women. Two
decades ago this information would have been considered
as common sense to most practitioners and midwives in
ScanQnavia. There is, however, so far no evidence for
this assumption, and according to policies dictated by
0 RCOG 2000 Br I Obstet Gynaecol 107,926-93 1
the World Health Organi~ation~~ Download full-text
nancy and Chlldbirth database24, manual protection of the
perineum is not specifically recommended.
Unfortunately, the present data do not provide infor-
mation on how perineal guarding should be performed
in practical terms. In a recent uncontrolled
deliveries at the University Hospital in Turku, Finland
were compared with those at the University Hospital in
Malmo, Sweden. In Turku the perineum was actively
protected by use of Ritgens manoeuvre, whereas in
Malmo a more passive approach was practised. The fre-
quency of anal sphincter tear was 13 times higher in
Malmo than in Turku. A recently published randomised
study26 by McCandlish compared the approach of
‘hands-on’ with ‘hands-poised’ during delivery of the
head. No difference with regard to sphincter tear was
noted. It is important to point out, however, that hands-
on was practised also in the hands-poised group when
considered necessary. Therefore, that study is not in
contradiction with our study demonstrating that no visu-
alisation of perineum or complete lack of perineal pro-
tection was associated with an increased risk of
sphincter tear. Further randomised controlled trials are
needed to find out the optimal way of protecting the
perineum during childbirth.
Perineal oedema, poor ocular surveillance of per-
ineudlack of manual perineal protection, protracted
final phase of second stage and high infant weight all
constituted independent risk factors of sphincter tears.
Such information may be useful for finding novel
strategies for the prevention of perineal trauma during
and the Cocluane Preg-
This work was supported by the Virdal Stiftelsen, the
Swedish Medical Research Council (grant no. 09455),
the Goteborg Medical Society and The Medical Faculty
of Goteborg University.
1 Swash M. Faecal incontinence. EMJ 1993; 307 636-637.
2 Sultan AH. Anal incontinence after childbirth. Curr Opinion Obstet
Gynecoll997; 9: 320-324.
3 Sultan AH, Kamm MA, Hudson CN, Thomas JM, Bartram CI. Anal-
sphincter disruption during vaginal delivery. N Engl J Med 1993; 329:
4 Haadem K, Ohrlander S, Lingman G. Long-term ailments due to anal
sphincter rupture caused by delivery-a
Gynecol Reprod Biol1988; 27: 27-32.
hidden problem. Eur J Obstet
RISK FACTORS FOR ANAL SPHINCTER TEARS
Sultan AH, Kamm MA, Hudson CN, Bartram CI. Third degree obstet-
ric anal sphincter tears: risk factors and outcome of primary repair.
BMJ 1994; 308: 887-891.
Fornell EKU, Berg G, Hallbook 0, Matthiesen LS, Sjodahl R. Clini-
cal consequences of anal sphincter rupture during vaginal delivery. J
Am CollSurg 1996; 183: 553-558.
Crawford LA, Quint EH, Pearl ML, DeLancey JOL. Incontinence fol-
lowing rupture of the anal sphincter during delivery. Obsrer Gynecol
1993; 82: 527-531.
Borgatta L, Piening SL, Cohen WR. Association of episiotomy and
delivery position with deep perineal laceration during spontaneous deliv-
ery in nulliparous women. Am J Obstet Gynecol1989; 160: 294-297.
Combs CA, Robertson PA, Laros RK Jr. Risk factors for third-degree
and fourth-degree perineal lacerations in forceps and vacuum deliver-
ies. Am J Obster Gynecol1990.163. 100-104.
Walker MPR, Farine D, Rolbin SH. Ritchie JWK. Epidural anaes-
thesia, episiotomy, and obstetric laceration. Obstet Gynecol 1991 :
Bek KM, Laurberg S. Intervention during labor: risk factors associ-
ated with complete tear of the anal sphincter. Acfa Obstet Gynecol
Scand 1992; 71: 520-524.
Bek KM, Laurberg S. Risks of anal incontinence from subsequent
vaginal delivery after a complete obstetric anal sphincter tear. Br J
Obster Gynaecoll992; 99 724-726.
Sultan AH, Kamm MA, Bartram CI, Hudson CN. Anal sphincter
trauma during instrumental delivery. Int J Cynecol Obster 1993; 43
Poen AC, Felt-Bersma RJF, Dekker GA, Deville W, Cuesta MA,
Meuwissen SGM. Third degree obstetric perineal tears: risk factors
and the preventive role of mediolateral episiotomy. Br J Obsret
Gynaecoll997; 104 563-566.
Buekens P, Lagasse R, Drarnaix M, Wollast E. Episiotomy and third
degree tears. BrJ Obsfef Cynaecoll985; 92: 820-823.
Rockner G, Wahlberg V, Olund A. Episiotomy and perineal trauma
during childbirth. JAdv Nursing 1989; 14: 264-268.
Johanson RB, Rice C, Doyle M et al. Arandomised prospective study
comparing the new vacuum extractor policy with forceps delivery. Er
J Obstet Gynaecoi 1993; LOO: 524-530.
Green JR, Soohoo SL. Factors associated with rectal injury in sponta-
neous deliveries. Obstet Gynecoll989; 7 3 732-738.
Wilcox LS, Strobino DM, Baruffi G, Dellinger WS Jr. Episiotomy and
its role in the incidence of perineal lacerations in a maternity center
and a tertiary hospital obstetric service. Am J Obsfer Gynecol 1989;
Nikodem VC. Upright vs recumbent position during the second stage of
labour. Review No. 03335, Issue 2. Oxford: Update Software; 1994.
Gheberg B, Magnusson B, Sultan B, Wennerholm U-B, Wennergren
M, Hagberg H. Birth in standing position: a high frequency of third
degree tears. Acta Obstet Gynecol Scand 1994; 73: 63G633.
Sultan AH, Stanton SL. Preserving the pelvic floor and perineum dur-
ing childbirth--elective caesarean section? Br J Obstet Gynaecol
1996; 103: 731-734.
Care in Normal Birth. Geneva: WHO, 1996.
Chalmers T , Enkin M, Keirse M. Ejjective Care in Pregnancy and
Childbirth Vol2. Oxford: Oxford University Press, 1993.
Pirhonen JP, Grenman SE, Haadem K et al. Frequency of anal sphinc-
ter rupture at delivery in Sweden and Finland-result of difference in
manual help to the baby’s head. ACZU Obsfef Gynecol Scund 1998; 77
McCandlish R, Bowler U, van Asten H et al. A randomised controlled
bid of care of the perineum during second stage of normal labour. Br
J Obstet Gynaecol1998; 105: 1262-1272.
Accepted 19 January 2000
0 RCOG 2000 Br J Obstet Gynuecol 107,926-931