DOI: 10.1111/1471-0528.

14145
www.bjog.org

Do pelvic floor muscle exercises reduce

postpartum anal incontinence? A randomised
controlled trial
HH Johannessen,a A Wibe,b,c A Stordahl,d L Sandvik,e S Mørkvedf,g
a
Department of Physiotherapy, Østfold Hospital Trust, Grålum, Norway b Department of Cancer Research and Molecular Medicine,
Norwegian University of Science and Technology, Trondheim, Norway c Department of Surgery, St Olavs Hospital, Trondheim University
Hospital, Trondheim, Norway d Department of Surgery, Østfold Hospital Trust, Gr
alum, Norway e Centre of Biostatistics and Epidemiology,
f
Oslo University Hospital, Oslo, Norway Department of Public Health and General Practice, Norwegian University of Science and
Technology, Trondheim, Norway g Department of Clinical Services, St Olavs Hospital, Trondheim University Hospital, Trondheim, Norway
Correspondence: H Hølmo Johannessen, Department of Research and Innovation, Østfold Hospital Trust, P.O. Box 300, 1714 Grålum,
Norway. Email hege.holmo.johannessen@so-hf.no

Accepted 25 April 2016. Published Online 7 June 2016.

Objective To evaluate the effect of pelvic floor muscle exercises
(PFME) for postpartum anal incontinence (AI).
Design A parallel two-armed randomised controlled trial stratified
on obstetrical anal sphincter injury with primary sphincter repair
and hospital affinity.
Setting Ano-rectal specialist out-patient clinics at two hospitals in
Norway.
Population One hundred and nine postpartum women with AI at
baseline.
Methods The intervention group received 6 months of individual
physiotherapy-led PFME and the control group written
information on PFME. Changes in St. Mark’s scores and
predictors of post-intervention AI were assessed by independent
samples t-tests and multiple linear regression analyses,
respectively. The study was not blind.
Main outcome measures The primary outcome measure was
change in AI symptoms on the St. Mark’s score from baseline to
post-intervention. Secondary outcome measures were manometry
measures of anal sphincter length and strength, endoanal
ultrasound (EAUS) defect score and voluntary pelvic floor muscle
contraction.
Results There was a significant difference in the reduction of St.
Mark’s scores from baseline to post-intervention in favour of
the PFME group ( 2.1 versus 0.8 points, P = 0.040). No
differences in secondary outcome measures were found
between groups. Baseline St. Mark’s, PFME group affinity and
EAUS defect score predicted post-intervention St. Mark’s score
in the imputed intention-to-treat analyses. The analysis on un-
imputed data showed that women performing weekly PFME
improved their AI scores more than women in the control
group did.
Conclusions Our results indicate that individually adapted PFME
reduces postpartum AI symptoms.
Keywords Anal incontinence, pelvic floor muscle exercises,
randomised controlled trial.
Tweetable abstract Performing regular pelvic floor muscle
exercises may be an effective treatment for postpartum anal
incontinence.

Please cite this paper as: Hølmo Johannessen H, Wibe A, Stordahl A, Sandvik L, Mørkved S. Do pelvic floor muscle exercises reduce postpartum anal
incontinence? A randomised controlled trial. BJOG 2016; DOI: 10.1111/1471-0528.14145.

of postpartum women may lose some of the supporting

Introduction
Pregnancy and childbirth increase the risk of weakening of
and injury to the perineum and pelvic floor muscles
(PFM), and it has been estimated that approximately half
Source of study: Østfold Hospital Trust and St Olavs Hospital, Trondheim
University Hospital, Trondheim, Norway.
function of the PFM after delivery.1 Loss in PFM function
may cause defecation dysfunction, sexual dysfunction, pel-
vic organ prolapse or involuntary loss of urine (urinary
incontinence, UI) or stool and gas (anal incontinence,
AI).2,3 Pelvic floor muscle exercises (PFME) are effective as
first line prevention and treatment for UI in the adult
female population, and despite scarce documentation, it is

ª 2016 Royal College of Obstetricians and Gynaecologists 1

 Hølmo Johannessen et al.
hypothesised that PFME may also be effective in treating
AI in pregnant and postpartum women.4,5 Few intervention
studies have been designed with AI as the main outcome
measure, and studies reporting on the effect of PFME for
postpartum AI may thus be underpowered to evaluate
PFME as prevention or treatment for AI.4,6,7 However, two
studies designed to evaluate the efficacy of PFME in combi-
nation with other treatment aids such as biofeedback and
electrical stimulation, showed a reduction in AI symp-
toms.8,9 Moreover, the women in the intervention groups
also had increased ability to perform voluntary pelvic floor
muscle contractions (VPFMC), and both studies had low
drop-out rates.
The aim of the present study was to evaluate the treat-
ment effect of individualised PFME for postpartum AI. The
null-hypothesis assumed no difference in reduction of post-
partum AI symptoms between the intervention group
receiving physiotherapy-guided PFME for 6 months and
the control group receiving written information on PFME
but no other intervention.
Methods
A two-armed randomised controlled trial was conducted at
Østfold Hospital Trust and St Olavs Hospital in Norway
during 2010–2014. The intervention group (PFME group)
received 6 months of individual physiotherapy-guided
PFME and the control group received written information
on PFME only. The study population consisted of parous
Norwegian-speaking women over the age of 18 who
reported AI on inclusion (fecal incontinence monthly or
more often or St. Mark’s score ≥3 points).10 The study
included women with and without obstetrical anal sphinc-
ter injury (OASI) and primary sphincter repair. Exclusion
criteria were inadequate Norwegian language skills, neuro-
logical conditions such as multiple sclerosis and polio,
women already receiving PFME treatment before inclusion
due to severe postpartum AI or pelvic floor dysfunction,
PFM pain/dysfunction, secondary sphincter repair, and
being unable to attend treatment and follow up at the
nearest available anorectal out-patient clinic or community
pelvic floor physiotherapist.
The randomisation was performed using an internet-
based computerised procedure at the Unit for Applied
Clinical Research at the Norwegian University of Science
and Technology, Norway. Randomisation was stratified on
hospital affinity and presence of OASI with primary
sphincter repair, to ensure balance in the intervention
groups. The staff involved in the clinical assessments or
PFME had no influence on the randomisation procedure.
Assessors and participants were not blinded due to the nat-
ure of the study and financial restraints.
2 ª 2016 Royal College of Obstetricians and Gynaecologists
Sample size estimation
Sample size estimation was based on findings from previ-
ous studies and clinical experience. Any reduction of 2
points or more on the St. Mark’s score was defined as
improvement based on the previously documented correla-
tion between the St. Mark’s score and patients’ subjective
improvement following PFME for AI symptoms. Thus, a
reduction of 2 points or more on the St. Mark’s score was
considered clinically relevant in this sample of young,
healthy women.11 We expected improvements in AI symp-
toms in 50% of participants in the PFME group and 20%
in the control group. With a chosen significance level of
5%, and an 80% chance of detecting a significant difference
between the two groups, the sample size required to com-
plete the study was found to be 72, with 36 participants in
each group, given that our assumptions were correct.
Clinical assessments
All participants underwent clinical assessments including
St. Mark’s score,12 endoanal ultrasound (EAUS), manome-
try measurements, and evaluation of PFM function and
VPFMC at baseline and after the 6-month intervention per-
iod. The primary outcome measure was change in AI from
baseline to post-intervention, as measured on the St.
Mark’s score. AI was defined as reporting incontinence of
stool monthly or more, incontinence of flatus weekly or
more, or combinations of two or more incontinence symp-
toms including fecal urgency on the St. Mark’s score.10,13
Based on the previously documented correlation between
the St. Mark’s score and patients’ subjective improvement
following PFME for AI symptoms, a reduction of 2 points
or more on the St. Mark’s score was considered clinically
relevant in this sample of young, healthy women.11 At Øst-
fold Hospital Trust, two-dimensional (2D) and three-
dimensional EAUS was performed in a side-lying position
using a B-K Medical EAUS machine type 2050 with a
360° probe (B-K Medical Aps, Denmark). At St Olavs
Hospital, 2D imaging was performed in the supine position
using a Hitachi Hi-Vision EAUS machine (Hitachi EUB-
6500, Providian Medical Equipment, USA). Anal sphincter
defects were classified by one of two experienced colorectal
surgeons using the EAUS defect score.14 Other secondary
outcome measures included evaluations of anal sphincter
length and mean anal sphincter strength during VPFMC as
measured by way of anal manometry. Standardised
manometry measurements were performed in a side-lying
position using a Polygraf ID manometry machine with a
catheter puller and POLYGRAM NET software (Alpine
Biomed, USA) and an average of three VPFMC was calcu-
lated. All participants received individual instruction by a
specialist physiotherapist and/or continence nurse on how
to perform a correct VPFMC prior to clinical examinations

at both baseline and post-intervention tests. PFM function
and ability to perform VPFMC was graded as either,
absent, weak, normal or strong as advised by the Interna-
tional Continence Society (ICS).15 In most cases, VPFMC
was graded by two examiners (a combination of two expe-
rienced colorectal surgeons and/or physiotherapists) and a
consensus was reached. All baseline clinical examinations
and ultrasound scorings were performed by two experi-
enced colorectal surgeons. The clinical examinations at
post-intervention were performed by two colorectal sur-
geons and/or two specialist physiotherapists.
Intervention (PFME group)
The 6-month intervention period consisted of four to six
individual appointments with a specialist physiotherapist at
the local anorectal out-patient clinics or local community
specialist pelvic floor physiotherapist. The individually
adapted home PFME program followed principles for
increasing muscle strength16 and was based on exercise
programs with previously documented effect in reducing
postpartum UI.17 Women were encouraged to perform
three sets of 8–10 close to maximum PFM contractions
daily and to hold the contractions for more than 3 seconds.
Progression of the PFME included holding the VPFMC for
up to 10 or 12 seconds, adding three fast contractions at
the end of the VPFMC, and using progressively more chal-
lenging starting positions from supine lying, sitting, stand-
ing, and standing with legs apart to walking, jumping, etc.
Women in the PFME group with absent VPFMC were
offered electrical stimulation (Neurotrac ETSTM/PelvitoneTM,
Verity Medical Ltd, UK and AnuformTM hard anal probe,
Neen Mobilis Health Care Group, UK) for home use until
they were able to perform active VPFMC and commence
PFME treatment. Adherence to the protocol was defined as
performing PFME once a week or more. PFME frequency
was recorded by the participants using a personal training
diary, as well as through reports to the physiotherapists at
the individual appointments, and at the post-intervention
test.
Control group
At baseline and post-intervention testing, women in the
control group received individual instruction on how to
perform a VPFMC, and written information about PFME.
For ethical reasons, PFME was permitted in the control
group but was not actively encouraged, and all control
group participants were offered individualised follow up
and home exercises after the initial 6-month intervention
period.
Statistical considerations
Two-sided Student’s t-tests were performed for between-
group comparison of continuous variables with a normal
ª 2016 Royal College of Obstetricians and Gynaecologists
Exercise reduces postpartum anal incontinence
distribution, Mann–Whitney U-test was used for skewed
continuous variables, and chi-square test for comparison of
categorical variables. Paired t-tests were performed for
within-group comparison of repeated continuous variables.
Multiple linear regression analyses were used to assess the
independent strength of the association between post-inter-
vention St. Mark’s scores and baseline measures including
EAUS defect score and baseline St. Mark’s score, randomi-
sation and frequency of PFME. The imputed intention-to-
treat (ITT) analyses was performed using the multiple
imputation algorithms based on linear regression in SPSS.
Baseline St. Mark’s score was used to replace missing items
in the outcome variable the post-intervention St. Mark’s
score. No other variables were imputed and consequently
women with incomplete test results either at baseline
(n = 9) or at post-intervention (n = 2) were not included
in the multiple linear regression analyses. None of the vari-
ables in the multiple linear regression models was highly
correlated. A significance level of 5% was used throughout.
All statistical analyses were performed using IBM SPSS
Statistics version 21 (IBM, Armonk, NY, USA) and Micro-
soft EXCEL FOR WINDOWS 2010 (Microsoft Corpora-
tion, Redmond, WA, USA).
Results
Of the 109 women included, 54 were randomised to the
intervention group and 55 to the control group (Figure 1).
The mean age at inclusion was 30.1 years (SD: 4.1 years)
and the participants were included at approximately 1 year
postpartum (mean 382 days, SD 132). There were no sig-
nificant differences with regard to background or delivery-
related variables between the groups except for there being
more ethnic Norwegian women in the control group and
more women of European decent in the PFME group. In
the intervention group, 32 women performed PFME more
than once a week during the intervention period, including
two women unable to attend the post-intervention
appointment. In the PFME group, 12 women did not
attend the post-intervention appointment (Table 1). No
side-effects of the intervention were reported.
There was a significant difference in the reduction in St.
Mark’s score from baseline to post-intervention in favour
of the PFME group (P = 0.040) (Table 1). Furthermore,
the primary outcome measure, change in St. Mark’s score
from baseline to post-intervention, was significantly
reduced in both groups; however, only the reduction of 2.1
points in the PFME group was considered clinically rele-
vant. Significantly fewer women reported leakage of gas at
post-intervention than at baseline in both groups (PFME:
32 versus 11, P < 0.001; control group: 38 versus 14,
P = 0.001). More than twice as many women reported
urgency at baseline (39%) than at post-intervention (17%;
3
 Hølmo Johannessen et al.
Assessed for eligibility
(n = 570)
Randomized (n = 109)
Allocated to intervention (n = 54)
- Did not attend follow-up (n = 12)
- Unable to attend follow-up (n = 2)
Figure 1. Flow chart of the recruiting and inclusion process through each stage in the randomised controlled trial.
P = 0.033) in the PFME group. No other significant
changes in AI symptoms from baseline to post-intervention
were found. The measured mean anal sphincter length dur-
ing VPFMC was reduced from 2.1 to 1.7 cm in both
groups (P = 0.820); however, there were no differences in
change of mean anal sphincter strength (PFME group: 80
versus 68 mmHg; control group: 81 versus 77 mmHg,
P = 0.241) or VPFMC between groups (PFME group: 2.3
versus 2.5; control group: 2.6 versus 2.3; P = 0.132) from
baseline to post-intervention.
Baseline St. Mark’s score, randomisation group, EAUS
defect score, anal sphincter length, VPFMC evaluation
score and PFME frequency were used in a multiple linear
regression analysis to predict the primary outcome measure
change in St. Mark’s score from baseline to post-interven-
tion. There were only minor differences between the results
found in the multiple imputed ITT analysis and the un-
imputed analyses performed on data collected from the
women completing the intervention period (Tables 2 and
3). Both prediction models were statistically significant
(ITT: F7.92 = 4.359, P ≤ 0.001; Un-imputed: F7.73 = 4.000,
P = 0.001), and accounted for between 20 and 30% of the
variance of the St. Mark’s score (ITT: R² = 0.249, Adjusted
R² = 0.192. Un-imputed: R² = 0.277, Adjusted R² = 0.208,
Table 2). The difference in St. Mark’s score at post-inter-
vention between the PFME and control groups was signifi-
cant when analysing the imputed data for the ITT analysis
(P = 0.014; Table 3). This difference remained significant
when analysing the un-imputed data (P = 0.006; Table 3).
Baseline St. Mark’s score, randomisation group and EAUS
defect score predicted post-intervention St. Mark’s score in
4 ª 2016 Royal College of Obstetricians and Gynaecologists
Excluded (n = 159)
- Neurological disease/other disease (n = 12)
- Inadequate Norwegian (n = 32)
- Misinterpreted questions on St.Mark’s score (n = 44)
- Moved/geographical reasons (n = 30)
-Women who had started PFME due to severe postpartum
AI or pelvic floor dysfunction (n = 3)
-No AI on inclusion (n = 38)
Declined participation (n = 302)
- Work (n = 5)
- Did not attend initial appointment (n = 17)
- Cancelled initial appointment (n = 13)
- Unable to reach/address unknown (n = 53)
- Other (n = 214) (declined participation/embarrassment/
did not want to consult doctor/unknown)
Allocated to control group (n = 55)
- Did not attend follow-up (n = 12)
both models (Table 3). The association between PFME fre-
quency and reduction in post-intervention St. Mark’s score
was significant in the un-imputed analysis (P = 0.044;
Table 3), but did not remain significant in the ITT analysis
(P = 0.115).
Discussion
Main findings
This randomised controlled trial showed that the reduction
in postpartum AI among women in the PFME intervention
group was significantly larger than the reduction in post-
partum AI reported by women in the control group. Base-
line St. Mark’s score was the strongest predictor of the
primary outcome change in St. Mark’s score, followed by
randomisation to the PFME group, performing PFME
more than once a week, and anal sphincter defect score.
Women with lower baseline St. Mark’s and anal sphincter
defect scores had lower post-intervention scores. The pre-
sent results indicate that regular PFME may be an effective
first-line treatment when offered to women experiencing
postpartum AI symptoms.
Strengths and limitations
The main strengths in this study were that the intervention
consisted of a standardised and structured PFME program
based on the principles for effective strength training,16 and
all participants were instructed how to perform VPFMC
prior to anal manometry examinations. The ability to per-
form VPFMC was graded on the ICS score15 by experi-
enced colorectal surgeons and women’s health specialist
 Exercise reduces postpartum anal incontinence

Table 1. Maternal characteristics and pregnancy outcomes (n = 109)

Variables Intervention group (n = 54) Control group (n = 55) Between-group comparison

n (%) n (%) P-value

Age (years), mean (SD) [range] 29.7 (4.3) [20–38] 30.6 (3.8) [23–40] 0.241*
Inclusion (days postpartum), mean (SD) 389 (122) 375 (141) 0.599*
Ethnicity
Norwegian 42 (77.8) 51 (92.4) 0.032**
European 8 (14.8) 3 (5.5)
Asian 4 (7.4) 1 (1.8)
Civil status
Married or cohabitant 51 (94.4) 52 (94.5) 0.465**
Employed full- or part-time 47 (87.1) 47 (85.4) 0.254***
Education
Primary school 2 (3.7) 1 (1.8) 0.101***
High school 16 (29.6) 10 (18.2)
College or university 35 (64.8) 44 (80)
Mode of delivery
Vaginal delivery 33 (61.1) 40 (72.7) 0.613**
Vacuum delivery 18 (33.3) 12 (21.9)
Forceps delivery1 2 (3.8) 1 (1.8)
Caesarean section 1 (1.9) 2 (3.6)
Presentation
Occiput anterior 46 (85.2) 46 (83.6) 0.647**
Occiput posterior 2 (3.7) 8 (14.6)
Other 6 (11.1) 1 (1.8)
Parity
1 42 (77.8) 43 (78.2) 0.937***
2 9 (16.7) 10 (18.2)
3 3 (5.5) 2 (3.6)
Birthweight (g), mean (SD) 3678 (538) 3672 (540) 0.953*
Adherence
Weekly/daily PFME 32 (59.3) – n/a
Monthly PFME 5 (9.3) –
No PFME 5 (9.3)
Withdrew/dna 12 (22.2) 12 (21.9)
Control group – 43 (78.2)
Head circumference (cm), mean (SD) 35.3 (1.8) 35.2 (1.7) 0.751*
VPFMC baseline, mean (SD)18 2.3 (0.7) 2.6 (0.9) 0.113*
VPFMC post-intervention, mean (SD) 2.5 (0.6) 2.3 (0.6) 0.108*
EAUS defect score
0 points 19 (35.2) 11 (20.0) 0.884***
1 point 3 (5.6) 6 (10.9)
2 points 4 (7.4) 15 (27.3)
3 points 9 (16.7) 7 (12.7)
4 points 7 (13.0) 7 (12.7)
5 points 3 (5.6) 3 (5.6)
6 points 0 0
7 points 5 (9.3) 2 (3.6)
Missing 4 (7.4) 4 (7.3)
Imputed data (n = 109)
St. Mark’s score baseline, mean (SD) 5.4 (3.6)* 5.0 (3.2) 0.576*
St. Mark’s score post-intervention, mean (SD) 3.3 (3.5)* 4.2 (3.4) 0.188*

physiotherapists at baseline and post-intervention. Further, they were able to perform a correct VPFMC. Adherence
among the seven women with absent VPFMC at baseline in was recorded using exercise diaries and monitored by the
the intervention group, PFME was not commenced until treating physiotherapists.

ª 2016 Royal College of Obstetricians and Gynaecologists 5

 Hølmo Johannessen et al.

Table 1. (Continued)

Variables Intervention group (n = 54) Control group (n = 55) Between-group comparison

n (%) n (%) P-value

Change in St. Mark’s score, mean (SD) –2.1 (3.5)* –0.8 (2.7) 0.040*
Un-imputed data (n = 83)
St. Mark’s score baseline, mean (SD) 5.7 (3.8)* 5.1 (3.2) 0.437*
St. Mark’s score post-intervention, mean (SD) 3.3 (4.0)* 4.3 (3.7) 0.195*
Change in St. Mark’s score, mean (SD) –2.4 (4.0)* –0.7 (2.9) 0.029*

Data are presented as number (percentage) unless otherwise stated. Boldface numbers = P < 0.05; dna: did not attend follow-up appointment.
*Independent sample’s t-test.
**Chi-square test.
***Mann–Whitney U-test.
1
Forceps delivery including failed ventouse delivery.
EAUS, endo anal ultrasound; SD, standard deviation; VPFMC, voluntary pelvic floor muscle contraction.

Table 2. Correlation of the variables included in the multiple linear regression analyses evaluating the association between changes in St. Mark’s
score and clinical measures at baseline; Intention-to-treat analysis and analysis of un-imputed data

Variable 2 3 4 5 6 7 8

Imputed intention-to-treat analyses, changes in St. Mark’s score (n = 100)

1. Changes in St. Mark’s score 0.347 0.097 0.169 0.020 0.170 0.156 0.000
2. Baseline St. Mark’s score – 0.038 0.028 0.261 0.013 0.282 0.054
3. Baseline sphincter length – 0.421 0.203 0.022 0.194 0.006
4. Baseline sphincter strength – 0.291 0.014 0.362 0.037
5. Baseline VPFMC score 0.127 0.240 0.025
6. PFME versus control group – 0.019 0.550
7. EAUS defect score – 0.174
8. PFME frequency –
R2 = 0.249, Adjusted R2 = 0.192
Analyses of un-imputed data, changes in St. Mark’s score (n = 81)
1. Changes in St. Mark’s score 0.316 0.113 0.191 0.029 0.196 0.205 0.055
2. Baseline St. Mark’s score – 0.034 0.016 0.314 0.025 0.252 0.078
3. Baseline sphincter length – 0.457 0.307 0.113 0.246 0.042
4. Baseline sphincter strength – 0.347 0.115 0.372 0.150
5. Baseline VPFMC score – 0.063 0.264 0.101
6. PFME versus control group – 0.002 0.820
7. EAUS defect score – 0.034
8. PFME frequency –
R2 = 0.277, Adjusted R2 = 0.208

Adjusted R2, coefficient of determination adjusted for number of predictors in the model; EAUS, endoanal ultrasound; PFME, pelvic floor muscle
exercises; R2, coefficient of determination; VPFMC, voluntary pelvic floor contraction.

As in many randomised controlled trials, one of the
limitations in the present study was that a relatively high
proportion of eligible women declined to participate in the
study. Thus, to ensure a sufficient sample size, the partici-
pating women represent a heterogeneous group of parous
women with and without OASI. Based on a previous study,
a reduction in AI symptoms of 2 points or more on the St.
Mark’s score was defined as an improvement and as
clinically relevant in the present study.11 In comparison, a
difference in pre- and post-intervention scores of more
than 5 points has previously been suggested as clinically
meaningful.18 However, Bols et al.18 included patients with
a mean age of 59 years, of whom nearly half experienced
severe AI symptoms for more than 5 years (mean: 18,
SD: 2.8) and these results may thus not be comparable to
our results based on the young, healthy participants in the

6 ª 2016 Royal College of Obstetricians and Gynaecologists

 Exercise reduces postpartum anal incontinence

Table 3. Results from the multiple linear regression models evaluating the association between changes in St. Mark’s score and clinical measures
at baseline. Intention-to-treat analysis and analysis of un-imputed data

Model Unstandardized Standarised coefficients Pearson r P-value

coefficients

Intention-to-treat analyses (n = 100) B SE-b Beta

Constant 1.155 1.470 0.378

2. Baseline St. Mark’s score –0.407 0.094 –0.424 –0.347 <0.001
3. Baseline sphincter length –0.603 0.439 –0.019 –0.097 0.847
4. Baseline sphincter strength –0.007 0.010 –0.073 –0.169 0.500
5. Baseline VPFMC score –0.111 0.383 –0.033 0.020 0.739
6. PFME versus control group -1.642 0.666 –0.291 –0.170 0.014
7. EAUS defect score 0.403 0.153 0.265 0.156 0.012
8. PFME frequency 0.458 0.288 0.190 –0.007 0.105

Analyses of un-imputed data (n = 81) B SE-b Beta

Constant 10.467 10.736 0.301

2. Baseline St. Mark’s score –0.398 0.111 –0.389 –0.316 <0.001
3. Baseline sphincter length –0.374 0.555 –0.078 –0.113 0.503
4. Baseline sphincter strength –0.008 0.012 –0.082 –0.191 0.502
5. Baseline VPFMC score 0.121 0.470 0.029 0.029 0.798
6. PFME versus control group –30.337 10.177 –0.502 –0.196 0.006
7. EAUS defect score 0.394 0.179 0.247 0.205 0.031
8. PFME frequency 10.430 0.696 0.366 0.055 0.044

EAUS, endoanal ultrasound; PFME, pelvic floor muscle exercises; VPFMC, voluntary pelvic floor contraction; Boldface numbers indicate P < 0.05.

present study. In the present cohort of young women with
fairly low mean St. Mark’s scores (mean 5 points),
although truly bothersome, a reduction of AI symptoms by
2 points may be considered to be important for their qual-
ity of life.19
Another limitation of the present study was that due to
limited resources, the main assessors were not blinded.
However, all clinical assessments were performed according
to standardised procedures and it is therefore unlikely that
this affected the results.
Interpretation
Recent guidelines recommend physiotherapy follow up
after OASI and primary repair,20 and some suggest that
individual follow up may be more efficient when treating
women with severe AI symptoms, major PFM or anal
sphincter defects, and reduced ability to perform correct
VPFMC.21 Further, a recent Cochrane review concluded
that targeted PFME intervention in populations with AI
was more effective than mixed prevention and treatment
approaches.4 The success of PFME depends on the
patients’ ability to actively contract the pelvic floor
muscles, and Bø et al.22 demonstrated that 30% of the
participating women were unable to perform a correct
VPFMC on the first attempt. In the study where PFME
was taught in a general fitness class for pregnant women
without individual instruction or examination of VPFMC
ability, no effect of PFME in reducing incontinence symp-
toms was found.23 The participants in the present study
received individual follow up, and similar to previous
studies,11 the mean reduction in AI symptoms was higher
in the PFME group than the control group. The women
in the intervention group who performed PFME on a
weekly basis, reduced their AI symptoms more compared
with the women in the control group, and also compared
with the women in the intervention group performing
PFME less often than weekly. This concurs with previous
findings that efficacy of PFME is associated with high
adherence to the treatment protocol.4,21 Thus, our results
may indicate that PFME frequency and intensity has a
major role in reducing postpartum AI symptoms. One in
five women failed to attend their post-intervention
appointment and this may have resulted in the discrepan-
cies between the un-imputed and the intention-to-treat
analyses.
The duration of the VPFMC among AI sufferers has
been reported to be shorter than in healthy controls despite
similar manometric measurements of maximum

ª 2016 Royal College of Obstetricians and Gynaecologists 7

 Hølmo Johannessen et al.
VPFMC.24,25 Thus, the focus in the present study was on
mean VPFMC strength measurements, and we were unable
to compare our findings to the reported normal ranges of
maximum VPFMC in women of reproductive age.26 How-
ever, there is some evidence to suggest that combining
PFME and electrical stimulation may improve incontinence
symptoms and anal sphincter strength more than PFME or
electrical stimulation alone.6,8,9 Electrical stimulation was
only offered as a treatment aid to women in the PFME
group who were initially unable to perform a VPFMC. We
found no significant differences in manometric measures of
mean anal sphincter muscle strength between groups at
baseline or post-intervention. Moreover, the mean strength
measures in both groups at post-intervention were lower
than at baseline. Overall, the results in the present study
may reflect the multifactorial pathogenesis of AI. Anal con-
tinence is in large part maintained by the integrity of the
sphincter apparatus. It has been suggested that other fac-
tors, such as co-ordination of the pelvic floor muscles, the
ano-recto inhibitory reflex, and psychological factors may
play major roles in the development of postpartum AI,
adding to the potential effect of pregnancy-related hor-
monal and mechanical change as well as injury to the
pudendal nerve27 during pregnancy or at delivery. The
women in the present study were included approximately
1 year postpartum, with the majority of women having
returned to their normal hormonal and mechanical status.
Still, the present findings indicate that the spontaneous
improvements in AI symptoms postpartum28 may be
enhanced by PFME.
Some authors report that OASI may not always result
in functional compromise, and therefore recommend that
both manometry measurements and EAUS evaluations are
performed when examining AI patients.29 In the present
trial the anal sphincter defect scores were similar in the
two groups, and the results show that women with low
anal sphincter defect scores reported fewer post-interven-
tion AI symptoms. This concurs with the results from a
study by Roos et al.30 who found that women with OASI
grade 3c and 4 have more defecatory symptoms and
poorer quality of life compared with women with OASI
grade 3a or 3b. Previous studies suggest that a shorter
anal sphincter length following primary repair of OASI is
associated with poorer outcomes.31 In contrast, we found
no association between anal sphincter length as measured
by way of manometry and post-intervention AI symp-
toms.
Conclusion
The present results indicate that performing individually
adapted PFME on a regular basis may reduce postpartum
AI symptoms, and that PFME may be offered as a first-line
8 ª 2016 Royal College of Obstetricians and Gynaecologists
treatment for postpartum AI. Further research of the opti-
mal timing and frequency of PFME treatment is warranted.
Disclosure of interests
None declared. Completed disclosure of interests form
available to view online as supporting information.
Contribution to authorship
All authors contributed to drafting of the original study
protocol and contributed to the analysis and interpretation
of the data. HHJ drafted the paper. All authors commented
and approved the final version.
Details of ethics approval
Participants received written and/or verbal information and
written consent was obtained prior to inclusion in the
study. The study is registered at clinicaltrials.gov
(NCT00970320) and was approved by the Norwegian
Regional Committees for Medical and Health Research
Ethics (REC Central, No (6)2008.1318. Approved
04.07.2008) and the Norwegian Social Science Data Services
(NSD).
Funding
This study was funded by the Norwegian Women’s Public
Health Association/the Norwegian Extra Foundation for
Health and Rehabilitation through EXTRA funds, Østfold
Hospital Trust, St Olavs Hospital, Trondheim University
Hospital, the Norwegian University of Science and
Technology, and the Central Norway Regional Health
Authority.
Acknowledgements
The authors would like to thank Clara Karoliussen for
assembling the cohort, collecting background data and con-
tributing to the clinical investigations at St Olavs Hospital,
Julia Trevor, Hanne Roelfsen, Cathrine Bergum and Linda
Sørby for implementing the PFME routines into their clini-
cal practices, Negin Sadati for collection of background
data, and Stacey Haukeland-Parker for proofreading the
manuscript. &
References
1 Swift SE. The distribution of pelvic organ support in a population of
female subjects seen for routine gynecologic health care. Am
J Obstet Gynecol 2000;183:277–85.
2 Bump RC, Norton PA. Epidemiology and natural history of pelvic
floor dysfunction. Obstet Gynecol Clin 1998;25:723–46.
3 Haylen BT, de Ridder D, Freeman RM, Swift SE, Berghmans B, Lee J,
et al. An International Urogynecological Association (IUGA)/
International Continence Society (ICS) joint report on the
terminology for female pelvic floor dysfunction. Neurourol Urodyn
2010;29:4–20.

4 Boyle R, Hay-Smith EJ, Cody JD, Morkved S. Pelvic floor muscle
training for prevention and treatment of urinary and faecal
incontinence in antenatal and postnatal women. Cochrane Database
Syst Rev 2012;000:CD007471. doi: 10.1002/14651858. CD002111.
pub3. Review
5 Dumoulin C, Hay-Smith J. Pelvic floor muscle training versus no
treatment, or inactive control treatments, for urinary incontinence in
women. Cochrane Database Syst Rev 2010;000:CD005654.
6 Norton C, Cody JD. Biofeedback and/or sphincter exercises for the
treatment of faecal incontinence in adults. Cochrane Database Syst
Rev 2012;CD002111. doi: 10.1002/14651858.CD002111.pub3.
Review
7 Stafne SN, Salvesen KA, Romundstad PR, Torjusen IH, Morkved S.
Does regular exercise including pelvic floor muscle training prevent
urinary and anal incontinence during pregnancy? A randomised
controlled trial BJOG 2012;119:1270–80.
8 Mahony RT, Malone PA, Nalty J, Behan M, O’Connell PR, O’Herlihy
C. Randomized clinical trial of intra-anal electromyographic
biofeedback physiotherapy with intra-anal electromyographic
biofeedback augmented with electrical stimulation of the anal
sphincter in the early treatment of postpartum fecal incontinence.
Am J Obstet Gynecol 2004;191:885–90.
9 Fynes MM, Marshall K, Cassidy M, Behan M, Walsh D, O’Connell
PR, et al. A prospective, randomized study comparing the effect of
augmented biofeedback with sensory biofeedback alone on fecal
incontinence after obstetric trauma. Dis Colon Rectum
1999;42:753–8; discussion 8–61.
10 Johannessen HH, Wibe A, Stordahl A, Sandvik L, Backe B, Morkved
S. Prevalence and predictors of anal incontinence during pregnancy
and 1 year after delivery: a prospective cohort study. BJOG
2014;121:269–79.
11 Maeda Y, Pares D, Norton C, Vaizey CJ, Kamm MA. Does the St.
Mark’s incontinence score reflect patients’ perceptions? A review of
390 patients. Dis Colon Rectum 2008;51:436–42.
12 Vaizey CJ, Carapeti E, Cahill JA, Kamm MA. Prospective comparison
of faecal incontinence grading systems. Gut 1999;44:77–80.
13 Johannessen HH, Morkved S, Stordahl A, Sandvik L, Wibe A. Anal
incontinence and Quality of Life in late pregnancy: A cross-sectional
study. BJOG 2014;121:978–87.
14 Norderval S, Dehli T, Vonen B. Three-dimensional endoanal
ultrasonography: intraobserver and interobserver agreement using
scoring systems for classification of anal sphincter defects.
Ultrasound Obstet Gynecol 2009;33:337–43.
15 Messelink B, Benson T, Berghmans B, Bo K, Corcos J, Fowler C,
et al. Standardization of terminology of pelvic floor muscle function
and dysfunction: report from the pelvic floor clinical assessment
group of the International Continence Society. Neurourol Urodyn
2005;24:374–80.
16 American College of Sports M. Progression models in resistance
training for healthy adults. Med Sci Sports Exerc 2002;34:364–80.
17 Morkved S, Bo K. Effect of pelvic floor muscle training during
pregnancy and after childbirth on prevention and treatment of urinary
incontinence: a systematic review. Br J Sports Med 2014;48:299–310.
ª 2016 Royal College of Obstetricians and Gynaecologists
Exercise reduces postpartum anal incontinence
18 Bols EM, Hendriks EJ, Deutekom M, Berghmans BC, Baeten CG, de
Bie RA. Inconclusive psychometric properties of the Vaizey score in
fecally incontinent patients: a prospective cohort study. Neurourol
Urodyn 2010;29:370–7.
19 Bartlett L, Nowak M, Ho YH. Impact of fecal incontinence on quality
of life. World J Gastroenterol 2009;15:3276–82.
20 Royal College of Obstetricians and Gynaecologists UK. Third and
fourth degree perineal tears, management (Green top Guideline no
29) 2015 [updated 12 June 2015; cited 12 June 2015]. 3rd Edn:
(This guideline provides evidence-based guidance on the diagnosis,
management and treatment of third- and fourth-degree perineal
tears (obstetric anal sphincter injuries, referred to as OASIS).).
Available from: www.rcog.org.uk/en/guidelines-research-services/
guidelines/gtg29/.
21 Hilde G, Staer-Jensen J, Siafarikas F, Ellstrom Engh M, Bo K.
Postpartum pelvic floor muscle training and urinary incontinence: a
randomized controlled trial. Obstet Gynecol 2013;122:1231–8.
22 Bø KLS, Oseid S, Kvarstein B, Hagen R, Jørgensen J. Knowledge
about and ability to correct pelvic floor muscle exercises in
women with urinary stress incontinence. Neurourol Urodyn
1988;7:261–2.
23 Bo K, Haakstad LA. Is pelvic floor muscle training effective when
taught in a general fitness class in pregnancy? A randomised
controlled trial Physiotherapy 2011;97:190–5.
24 Chiarioni G, Scattolini C, Bonfante F, Vantini I. Liquid stool
incontinence with severe urgency: anorectal function and effective
biofeedback treatment. Gut 1993;34:1576–80.
25 Martinez-Puente Mdel C, Pascual-Montero JA, Garcia-Olmo D.
Customized biofeedback therapy improves results in fecal
incontinence. Int J Colorectal Dis 2004;19:210–4.
26 Chaliha C, Sultan AH, Emmanuel AV. Normal ranges for anorectal
manometry and sensation in women of reproductive age. Colorectal
Dis 2007;9:839–44.
27 Sultan AH, Nugent K. Pathophysiology and nonsurgical treatment of
anal incontinence. BJOG 2004;111(Suppl 1):84–90.
28 Nazir M, Stien R, Carlsen E, Jacobsen AF, Nesheim BI. Early
evaluation of bowel symptoms after primary repair of obstetric
perineal rupture is misleading: an observational cohort study. Dis
Colon Rectum 2003;46:1245–50.
29 Sultan AH. Re: Incontinence after primary repair of obstetric anal
sphincter tears is related to relative length of reconstructed
external sphincter: a case–control study. Ultrasound Obstet
Gynecol 2012;40:207–214. Ultrasound Obstet Gynecol 2012;40:
126–7.
30 Roos AM, Thakar R, Sultan AH. Outcome of primary repair of
obstetric anal sphincter injuries (OASIS): does the grade of tear
matter? Ultrasound Obstet Gynecol 2010;36:368–74.
31 Norderval S, Rossaak K, Markskog A, Vonen B. Incontinence after
primary repair of obstetric anal sphincter tears is related to the
relative length of the reconstructed external sphincter: a case
control study. Ultrasound Obstet Gynecol 2012;40:207–14.
9