Pelvic floor disorders 4 years after first delivery: a comparative study of restrictive versus systematic episiotomy


Dr X Fritel, Gynécologie & Obstétrique, CHD Félix-Guyon, F-97405 Saint-Denis cedex, France, Email


Objective  To compare two policies for episiotomy: restrictive and systematic.

Design  Quasi-randomised comparative study.

Setting  Two French university hospitals with contrasting policies for episiotomy: one using episiotomy restrictively and the second routinely.

Population  Seven hundred and seventy-four nulliparous women delivered during 1996 of a singleton in cephalic presentation at a term of 37–41 weeks.

Methods  A questionnaire was mailed 4 years after delivery. Sample size was calculated to allow us to show a 10% difference in the prevalence of urinary incontinence with 80% power.

Main outcome measures  Urinary incontinence, anal incontinence, perineal pain, and pain during intercourse.

Results  We received 627 responses (81%), 320 from women delivered under the restrictive policy, 307 from women delivered under the routine policy. In the restrictive group, 186 (49%) deliveries included mediolateral episiotomies and in the routine group, 348 (88%). Four years after the first delivery, there was no difference in the prevalence of urinary incontinence (26 versus 32%), perineal pain (6 versus 8%), or pain during intercourse (18 versus 21%) between the two groups. Anal incontinence was less prevalent in the restrictive group (11 versus 16%). The difference was significant for flatus (8 versus 13%) but not for faecal incontinence (3% for both groups). Logistic regression confirmed that a policy of routine episiotomy was associated with a risk of anal incontinence nearly twice as high as the risk associated with a restrictive policy (OR = 1.84, 95% CI: 1.05–3.22).

Conclusions  A policy of routine episiotomy does not protect against urinary or anal incontinence 4 years after first delivery.


Episiotomy has long been recommended to avoid perineal sequelae after delivery. Despite the absence of evidence of its efficacy, it is still very widely used.1 Nonetheless, its frequency varies greatly between hospitals and in different European countries. For example, rates are reported to be 10% in Upssala (Sweden) compared with 58% in Perugia (Italy).2,3 Little is known about its long-term sequelae or benefits. Our objective was to compare results of pelvic floor disorders several years after first delivery at hospitals with two different policies for episiotomy: one with a policy of routine episiotomy and the other with a policy of avoiding episiotomy as much as possible.


The study included nulliparous women who gave birth in 1996 at a term of 37–41 weeks to a liveborn singleton child in cephalic presentation and who had an up-to-date mail address in 2000. Examination of delivery registries allowed us to identify the mothers who met these criteria. Data about the mothers (age, height, and weight before conception), pregnancy (presentation), and delivery (epidural, mode of delivery, duration of the active-pushing second stage of labour, and child’s weight) were collected at delivery. Information about pelvic floor disorders was obtained from a questionnaire mailed 4 years after delivery. A second and even a third mailing went to the women from whom we received no response. The questionnaire asked about educational level, postpartum pelvic floor exercises, subsequent deliveries, and urinary symptoms during the preceding 4 weeks. Those women who answered ‘yes’ to the entry question ‘Do you have involuntary loss of urine?’ were considered to have urinary incontinence and were then asked further questions from a validated questionnaire4 about the frequency, amount, and circumstances of leakage and if incontinence was a problem for her. Stress incontinence was defined by any positive response (‘occasionally’, ‘sometimes’, ‘most of the time’, or ‘all of the time’) to ‘Does urine leak when you are physically active, exert yourself, cough or sneeze?’, urge incontinence by any positive response to ‘Does urine leak before you can get to the toilet?’, and mixed incontinence by a positive response to both the previous questions.4 Severity of urinary incontinence was measured with Sandvik’s score, which has been validated with pad-weighing tests.5 Additional items asked about urinary urgency, voiding difficulty, pain, and anal incontinence. Urinary urgency was assessed by ‘Do you have to rush to the toilet to urinate?’.4 Voiding difficulty was assessed by ‘Do you have difficulties in emptying your bladder?’. Perineal pain was defined as a ‘yes’ (versus ‘no’) response to ‘Do you have chronic perineal pain (perineum designates the skin and muscle around the vaginal and anal outlets)?’. Pain during intercourse was defined as a ‘yes’ (versus ‘no’ response or ‘no intercourse at present’) response to ‘Do you experience pain during sexual intercourse?’. Anal incontinence was defined by ‘yes’ (versus ‘no’) response to ‘Do you have involuntary loss of flatus or stool?’. The severity of anal incontinence was assessed as a function of the type of incontinence (gas only, liquid stool, or solid stool) and its frequency, according to Pescatori’s score.6 The questionnaire used for the study is available from the authors upon request.

The questionnaire was initially tested on a small sample (50 women in each maternity ward). This pilot test allowed us to estimate the prevalence of urinary incontinence (20%) among women with vaginal deliveries, the percentage of caesarean sections (10%), and the risk of nonresponse (20%) overall and in each facility. We determined that it would require 248 responses from each group to show a 10% difference in urinary incontinence (15% versus 25%, α= 0.05 and β= 0.20, bilateral test) among women with vaginal deliveries. Given the rates of caesarean delivery and nonresponse, we needed at least 345 women in each group. This number is approximately the number of nulliparous women meeting the inclusion criteria delivered in 1 year in each hospital.

Policies and protocols at the first hospital (A) recommended strongly against episiotomy, while in the second (B) it was strongly recommended for first deliveries. All episiotomies were mediolateral. In both cases, residents repaired episiotomies. Technical guidelines for episiotomy and its repair were similar in each hospital.7,8 Each obstetric department had previously published its arguments in favour of or against episiotomy.9,10

We first analysed the differences between hospitals A and B and between responders and nonresponders (Table 1). We also examined the variables associated with episiotomies in the restrictive-policy maternity ward (A).

Table 1.  Women’s characteristics according to maternity unit and response to questionnaire
CharacteristicsTotal mean, nMaternityPResponse to questionnaireP
A*, mean (±SD) or n (%)B**, mean (±SD) or n (%)Yes, mean (±SD) or n (%)No, mean (±SD) or n (%)
  • Chi-square test for nominal variables (n [%]), t test for continuous variables (mean [SD]).

  • *

    Unit A: restrictive episiotomy.

  • **

    Unit B: routine episiotomy.

  • ***

    Data available only for responders.

High school diploma
Yes 191 (61)220 (74)0.001411 (67)*** 
No 120 (39)79 (26)199 (33) 
Age at delivery (years)28.227.1 (±4.7)29.3 (±4.5)<0.000128.4 (±4.5)27.4 (±5.2)0.03
<30522288 (76)234 (59)<0.0001415 (66)107 (73)0.12
≥3025291 (24)161 (41)212 (34)40 (27)
Body mass index (kg/m2)21.421.5 (±3.1)21.4 (±3.0)0.5221.4 (±3.0)21.7 (±3.3)0.31
<25690340 (90)350 (89)0.53564 (90)126 (87)0.21
≥257936 (10)43 (11)60 (10)19 (13)
UI before pregnancy
Yes 17 (6)16 (5)0.8733 (6)*** 
No 283 (94)282 (95)565 (94) 
UI during pregnancy
Yes 65 (21)68 (23)0.69133 (22)*** 
No 238 (79)230 (77)468 (78) 
Gestational age (weeks)39.940.2 (±1.2)39.6 (±0.9)<0.000139.9 (±1.1)39.7 (±1.0)0.06
<40385147 (39)238 (60)<0.0001306 (49)79 (54)0.28
≥40389232 (61)157 (40)321 (51)68 (46)
Yes654277 (73)377 (95)<0.0001526 (84)128 (87)0.34
No120102 (27)18 (5)101 (16)19 (13)
Active second phase (minute)12.013.1 (±7.7)11.0 (±7.6)0.000212.1 (±7.7)11.5 (±7.7)0.41
<20694330 (89)364 (92)0.13561 (90)133 (91)0.82
≥207241 (11)31 (8)59 (10)13 (9)
Mode of delivery
Spontaneous435235 (62)200 (50)0.003368 (59)67 (46)0.002
Operative266117 (31)149 (38)209 (33)57 (39)
Caesarean7327 (7)46 (12)50 (8)23 (16)
Yes534186 (49)348 (88)<0.0001433 (69)101 (69)0.93
No240193 (51)47 (12)194 (31)46 (31)
Third-degree tear
Yes72 (0.5)5 (1.3)0.286 (1.0)1 (0.7)0.75
No767377 (99.5)395 (98.7)621 (99.0)146 (99.3) 
Birth weight (g)32663293 (±430)3240 (±394)0.083273 (±413)3239 (±413)0.37
<4000754362 (96)388 (98)0.03605 (96)145 (99)0.18
≥40002417 (4)7 (2)22 (4)2 (1)
Response to questionnaire
Yes627320 (84)307 (78)0.02 
No14759 (16)88 (22) 
Postpartum pelvic floor exercises
Yes 93 (31)147 (49)<0.0001240 (40)*** 
No 210 (69)151 (51)361 (60) 
Second delivery
Yes 195 (63)186 (63)0.93381 (63)*** 
No 117 (37)110 (37)227 (37) 

We used bivariate analysis to examine perineal disorders (urinary incontinence, urinary urgency, voiding difficulty, perineal pain, pain during intercourse, and anal incontinence) as a function of maternity ward episiotomy policy (routine versus restrictive). Analysis was performed on an intention-to-treat basis, as if patients had been randomised before delivery. In particular, women with caesareans were not excluded because each nulliparous woman included had a singleton term pregnancy in cephalic presentation, with no indication at inclusion for elective caesarean section, and each woman was exposed to the hospital episiotomy policy. The factors retained for the multivariable analysis were those that differed significantly between the two hospitals, even if they were not significantly associated with incontinence. Each statistical test was considered significant if P < 0.05. All analyses were performed with Statview (SAS Institute Inc., Cary, NC, USA).

Our study complied with French law about biomedical research. The department head of each department approved the study. Each respondent provided informed consent.


Of the women who gave birth in 1996, 774 met the inclusion criteria. We had 627 (81%) responses to the questionnaire. Table 1 reports the differences between those who gave birth in hospital A and hospital B and between respondents and nonrespondents. The women who gave birth in hospital B were significantly older and had a higher educational level, more epidurals, and more instrumental or caesarean deliveries. Respondents were significantly older, had more spontaneous deliveries, and came more often from hospital A. In the maternity ward with a restrictive policy (A), episiotomy in women with vaginal delivery was associated with epidural anaesthesia (53% episiotomies compared with 38% without epidural, P= 0.01), with instrumental intervention (71% episiotomy versus 39% for spontaneous delivery, P < 0.0001), and with active second stage pushing for longer than 20 minutes (78% episiotomies versus 45% for less than 20 minutes, P < 0.0001). The other variables (age, body mass index, and infant birthweight) were not associated with episiotomy.

The bivariate comparison between the two institutions showed no differences for urinary disorders, perineal pain, or pain during intercourse. Flatus incontinence, on the other hand, was more frequent in women who gave birth at the maternity ward with a routine episiotomy policy (Table 2). Anal incontinence was reported by 33 women in the restrictive group and 50 in the routine group. Its severity, measured by Pescatori’s score, was similar in both groups with a median score of 3 in both. Among women reporting only flatus incontinence, 21 (33%) considered it ‘a serious problem’, 10 (16%) ‘quite a problem’, and 31 (48%) ‘a bit of a problem’; these results did not differ significantly between the two hospitals (P= 0.79).

Table 2.  Pelvic floor disorders in 627 women 4 years after delivery according to episiotomy (bivariate analysis, chi-square test)
Pelvic floor disorders 4 years after first deliveryMaternity A, restrictive episiotomy, n (%)Maternity B, routine episiotomy, n (%)P
  1. UI, urinary incontinence; AI, anal incontinence.

No231 (74)207 (68)0.09
Yes82 (26)99 (32)
UI type (% among women with UI)
Stress24 (29)31 (31)0.67
Urge6 (7)6 (6)
Mixed51 (62)58 (59)
UI severity (Sandvik score)
Dry231 (74)207 (68)0.45
Slight48 (15)62 (20)
Moderate21 (7)21 (7)
Severe8 (3)8 (3)
UI bothersome (% among women with UI)
Not a problem7 (9)17 (18)0.33
A bit of a problem53 (67)54 (56)
Quite a problem12 (15)15 (16)
A serious problem7 (9)10 (10)
Never133 (43)114 (38)0.22
Occasionally102 (33)116 (39)
Sometimes54 (17)57 (19)
Often or all of the time23 (7)14 (5)
Difficult voiding
Never224 (71)217 (72)0.32
Occasionally44 (14)52 (17)
Sometimes37 (12)24 (8)
Often or all of the time9 (3)7 (2)
Perineal pain
No291 (94)272 (92)0.26
Yes17 (6)23 (8)
Pain during intercourse
No intercourse7 (2)9 (3)0.45
No pain247 (80)225 (76)
Yes54 (18)62 (21)
Anal incontinence
No276 (89)249 (84)0.04
Yes33 (11)49 (16)
AI bothersome (% among women with AI)
Not a problem0 (0)1 (2)0.65
A bit of a problem14 (42)22 (45)
Quite a problem7 (21)6 (12)
A serious problem12 (36)18 (37)
AI type
Flatus only24 (8)40 (13)0.02
Stool9 (3)9 (3)0.94

Among women with vaginal deliveries (spontaneous or instrumental), the prevalence of urinary incontinence was 27% (78/294) in the restrictive episiotomy group versus 33% (92/275) in the routine group (P= 0.07), and the prevalence of anal incontinence was 10% (29/291) in the restrictive and 17% (46/269) in the routine group (P= 0.01).

The comparison between the hospitals for urinary incontinence and anal incontinence was adjusted according to the known differences between the populations of each hospital. In the multivariable analysis, the episiotomy policy did not affect the risk of urinary incontinence 4 years after the first delivery. On the other hand, a routine episiotomy policy nearly doubled the risk of anal incontinence (Table 3).

Table 3.  Risk of urinary and anal incontinence 4 years after first delivery according to episiotomy policy (routine versus restrictive) adjusted for variables associated with maternity hospital (logistic regression)
CharactersNUrinary incontinenceAnal incontinence
n (%)Crude OR (95% CI)Adjusted OR (95% CI)n (%)Crude OR (95% CI)Adjusted OR (95% CI)
A restrictive32082 (26)1133 (11)11
B systematic30799 (32)1.35 (0.95–1.91)1.21 (0.80–1.83)49 (16)1.65 (1.03–2.64)1.84 (1.05–3.22)
High school diploma
No19957 (29)1126 (13)11
Yes411118 (29)0.98 (0.67–1.42)0.74 (0.49–1.10)54 (13)0.95 (0.58–1.58)0.80 (0.47–1.35)
Age at delivery (years)
<3041598 (24)1148 (12)11
≥3021283 (39)2.07 (1.45–2.97)2.13 (1.46–3.13)34 (16)1.45 (0.90–2.34)1.31 (0.79–2.17)
Gestational age (weeks)
<4030682 (27)1143 (14)11
≥4032199 (31)1.22 (0.86–1.72)1.51 (1.03–2.22)39 (12)0.84 (0.53–1.33)0.98 (0.60–1.61)
No10128 (28)1117 (17)11
Yes526153 (29)1.06 (0.66–1.70)0.88 (0.52–1.49)65 (12)0.69 (0.38–1.23)0.47 (0.24–0.91)
Active second phase (minute)
<20561161 (29)1168 (12)11
≥205919 (32)1.16 (0.65–2.06)1.00 (0.54–1.85)13 (22)1.97 (1.01–3.83)2.17 (1.07–4.43)
Mode of delivery
Spontaneous368102 (28)1145 (12)11
Operative20968 (33)1.25 (0.86–1.80)1.08 (0.73–1.61)30 (14)1.19 (0.72–1.96)1.13 (0.67–1.92)
Caesarean5011 (22)0.72 (0.35–1.46)0.63 (0.29–1.34)7 (14)1.21 (0.51–2.86)1.22 (0.49–3.00)
Birth weight (g)
<4000605175 (29)1181 (13)11
≥4000226 (27)0.90 (0.35–2.35)0.74 (0.26–2.07)1 (5)0.66 (0.15–2.88)0.34 (0.04–2.74)
Postpartum pelvic floor exercises
No36182 (23)1140 (11)11
Yes24095 (40)2.20 (1.54–3.14)2.12 (1.45–3.10)39 (17)1.54 (0.96–2.48)1.43 (0.86–2.36)


Our study found no benefits to routine mediolateral episiotomy during first deliveries. This result is consistent with the few studies that have compared restrictive to routine episiotomy.11 The West Berkshire perineal management trial is the only comparative trial that focused on late consequences of mediolateral episiotomy. It found no differences in urinary incontinence, perineal pain, or dyspareunia 3 years after delivery in the two groups, randomised to restrictive (11%) or liberal (52%) use of mediolateral episiotomy.12 Our work confirms that liberal use of episiotomy does not diminish the prevalence of urinary incontinence. It also confirms that episiotomy is not associated with an increased risk of perineal pain or dyspareunia.

Our questionnaire response rate (81%) 4 years after first delivery was similar to the rate observed 3 years after the West Berkshire trial (76% among 885 women with known addresses).12 The questionnaires were in French. While it is unlikely that any women who did not speak French responded, we do not think that there is any reason that language should have a differential effect on the prevalence of perineal sequelae between the two institutions.

The restrictive episiotomy rate may appear quite high, but it must be borne in mind that episiotomy rates are higher in France than elsewhere in Europe. In 1996, the date of first delivery for the women in our study, the episiotomy rate in France was 79% for nulliparous women; in 2002 it was still 68%.13

In our study flatus incontinence was more frequent in the women in the routine episiotomy group. The association between episiotomy and perineal trauma (third-degree lacerations) is complex. On the one hand, mediolateral episiotomy is associated with a lower risk of anal sphincter rupture at delivery.14,15 On the other hand, it has also been shown that the number of mediolateral episiotomies can be reduced without an increase in perineal trauma.9,16 Thus, the protective effect of mediolateral episiotomy may be limited to situations in which its use is inevitable, while its routine performance may increase the risk of anal incontinence. It is possible that performing an episiotomy when the anal sphincter is not in danger increases risk of direct scissors injury to the sphincter. Unfortunately, we do not know what episiotomy rate offers the best balance between benefits and risks for the anal sphincter. Similarly, we do not know at what moment of perineal dilatation during delivery or according to what clinical signs it is best to perform an episiotomy to protect the external anal sphincter.

The interest of our work is that we looked at late consequences of episiotomy, a subject for which data are sparse. We decided not to attempt a randomised trial, such as the West Berkshire perineal management trial, because it appears very difficult to us to ask each team trained according to particular practice to develop a different practice for the trial. Deliveries without episiotomy are practised differently than those with routine episiotomies. The method of comparison used in our study is called ‘quasi-randomised’ because the exposure (episiotomy policy) is controlled. If the populations are similar and adjustment is planned for the possible differences, this type of study produces robust findings.

Despite the adjustment for known differences between institutions (women’s age, educational level, gestational age, epidural, time of pushing, mode of delivery, birthweight, and postpartum pelvic floor exercises), it remains possible that our results are due to other differences in the populations or in medical practices between the two institutions. We did not take into account such risk factors for postpartum incontinence as smoking, bladder neck mobility, or prenatal pelvic floor exercises.17–19 However, it is very unlikely that taking these possible differences between the two institutions into account would lead to the conclusion that routine episiotomy has a protective effect.


A policy of routine episiotomy does not protect against urinary or anal incontinence 4 years after first delivery.

Contribution to authorship

XF has contributed to the conception, design, analysis, interpretation of data, and article writing; JPS and AP contributed to the conception, design, data interpretation, and revision of the manuscript; VB and CL contributed to data management; and AF contributed to analysis, data interpretation, and revision of the manuscript.

Disclosure of interests

We have no direct or indirect commercial financial incentive associated with publishing the article.

Ethical approval

Our work complied with French statutes and regulations, which authorise epidemiological surveys without advance approval of an ethics committee. Our survey involved no intervention and is thus excluded from the French statute on biomedical research (Loi Huriet-Sérusclat, dated 20 December 1998). We complied with all French statutes concerning data about the subjects, confidentiality, and restrictions (e.g. no religious or racial data). Informed consent was obtained from each responding woman.


We had no exterior funding for this work.


We thank Jo Ann Cahn and Liliane Cotte for editorial assistance.