Contribution of the second stage of labour to pelvic floor dysfunction: a prospective cohort comparison of nulliparous women


  • Presented at the American Urogynecology Society Annual Meeting, Chicago, IL, 6 October 2012.



Maternal expulsive efforts are thought to damage the pelvic floor. We aimed to compare pelvic floor function and anatomy between women who delivered vaginally (VB) versus those with caesarean delivery (CD) prior to the second stage of labour.


Prospective cohort.


University Hospital Midwifery practice.




Pregnant nulliparas were recruited during pregnancy and women who underwent CD prior to the 2nd stage of labour at birth were recruited immediately postpartum. Both groups were followed prospectively to 6 months postpartum.

Main outcome measures

POPQ, perineal ultrasound (U/S) and Paper Towel Test (PTT), an objective measure of stress incontinence; Incontinence Severity Index (ISI), Pelvic Floor Impact Questionnaire (PFIQ-7), Wexner Fecal Incontinence Scale (W) and Female Sexual Function Index (FSFI).


336/448 (75%) VB and 138/224 (62%) CD followed up. The VB group was younger (23.9 ± 4.9 versus 26.6 ± 6.1 years, P < 0.001) and less overweight/obese (38 versus 56%, < 0.001); baseline functional measures were similar (all P > 0.05). At follow-up, urinary incontinence (UI) (55 versus 46% ISI > 0, P = 0.10), fecal incontinence (FI) (8 versus 13% FI on W, P = 0.12), sexual activity rates (88 versus 92%, P = 0.18) and PFIQ-seven scores were similar. Positive PTT tests (17 versus 6%, P = 0.002) and ≥ Stage 2 prolapse (22 versus 15%, P = 0.03) were higher with VB; differences were limited to points Aa and Ba. U/S findings were not different between groups. Stepwise multivariate analyses controlling for age, body mass index, and non-Hispanic White race for prolapse of points Aa and Ba did not alter conclusions (all < 0.004).


VB resulted in prolapse changes and objective UI but not in increased self-report pelvic floor dysfunction at 6 months postpartum compared with women who delivered by CD prior to the second stage of labour. The second stage of labour had a modest effect on postpartum pelvic floor function.


Caesarean delivery (CD) is thought to diminish the risk of pelvic floor dysfunction following delivery, including urinary (UI) and anal incontinence (AI), pelvic organ prolapse (POP) and pelvic pain, as well as preserving sexual function.[1-3] Recent data confirm that operative vaginal delivery significantly increases the risk of pelvic floor dysfunction 5–10 years after first delivery; however, rates of forceps, vacuum and episiotomy have decreased significantly over the past decade and little is known about the impact of normal birth on the pelvic floor without these interventions.[3-5] In a large retrospective cohort study, caesarean delivery prior to the second stage of labour was not associated with an increased risk of UI, AI or POP when compared with CD prior to labour.[3] Although it is thought that delivery of the fetal head poses the greatest threat to pelvic floor function, prior studies have not focused their comparisons on groups who did and did not enter the second stage of labour, choosing instead to include all women randomised to caesarean delivery prior to labour, as in the Term Breech Trial,[6] or women had a caesarean for a variety of indications. This makes it challenging to describe the contribution of the second stage of labour to subsequent pelvic floor dysfunction.

Alterations in the Pelvic Floor in Pregnancy, Labor and the Ensuing Years (APPLE) is a prospective cohort study of nulliparous midwifery and CD from low risk obstetrical services. This study was designed to investigate postpartum pelvic floor changes among women who did and those who did not enter the second stage of labour, by comparing women who underwent caesarean delivery prior to the second stage of labour with women who gave birth vaginally. We aimed to evaluate the full range of pelvic floor function including UI and AI, sexual function, pain and anatomical changes. We recruited women who underwent CD prior to entry into the second stage of labour, as maternal expulsive efforts are thought to pose the greatest damage to the pelvic floor and offer the attendant the greatest opportunity to vary practice to decrease postpartum pelvic floor changes.[7] We aimed to compare 6-month postpartum pelvic floor function and anatomical changes between women who delivered by caesarean (CD group) prior to the second stage of labour with those who delivered vaginally (VB group) in order to better define the contributions of the second stage to pelvic floor dysfunction.


Nulliparous patients cared for by the Midwifery service and primiparous CD patients were recruited prospectively from December 2006 to January 2011 in Albuquerque, NM, USA. Eligibility criteria were age ≥18 years of age, ability to read either English or Spanish, singleton gestation, absence of serious medical problems, gestational age of ≤36 weeks and no late second trimester pregnancy losses. All midwife patients were eligible for midwifery care until birth. Midwife patients who delivered vaginally constituted the vaginal birth (VB) group and included women who underwent episiotomy and operative delivery. The CD group was recruited from the low risk obstetrical ward cared for by the Obstetrics and Gynecology or the Family Medicine service, and were eligible if they had not entered the second stage of labour. Caesarean delivery on maternal request is rare in our institution, and the women enrolled in this study planned to undergo vaginal delivery. Fetal malpresentation was not an indication for exclusion. Women who underwent caesarean section (CS) were recruited on the first postpartum day and followed prospectively until 6 months postpartum. Midwife patients who underwent CD without entering the second stage of labour were included in the CD group for analyses. Low risk women seeking care for pregnancy at the University of New Mexico can choose either to be followed by the midwifery service or to be cared for by physicians. Each service follows their patients during pregnancy, at delivery and postpartum. The study was approved by the Institutional Review Board of the University of New Mexico, and all women gave written informed consent.

Physical exam and pelvic floor functional data were assessed during early and late pregnancy and at 6 months postpartum for all midwife patients. Physical exam data collected included Pelvic Floor Quantification Exams (POPQ),[8] assessment of pelvic floor muscle strength using the Brinks scale,[9] and visual inspection of the perineum. All certified nurse-midwife (CNM) examiners underwent POPQ training with live models prior to the study. In addition, a subset of exams was repeated by a second CNM examiner to determine inter-rater reliability at the 6-month follow-up visit. Functional data were gathered using the following quality of life and symptom severity scales: the Incontinence Severity Index (ISI),[10] the Questionnaire for Urinary Incontinence Diagnosis (QUID),[11] the Pelvic Floor Impact Questionnaire (PFIQ-20),[12, 13] the Wexner Fecal Incontinence Scale (W)[14], the Present Pain Intensity Scale (PPI)[15] and the Female Sexual Function Index (FSFI).[16, 17] Any UI was defined as ISI scores >0; moderate to severe UI was defined as ISI scores ≥3.[10] Any AI (flatus, liquid or solid stool) was defined as a Wexner score >0; fecal incontinence was defined as an affirmative answer to the questions regarding leakage of liquid or solid stool.

Intrapartum data were collected at delivery by the attendant midwife, or shortly thereafter for the CD group, and included maternal, fetal and labour characteristics. Race/ethnicity was self-defined by participants. The VB group underwent a detailed mapping of perineal trauma. For perineal lacerations greater than or equal to a second degree laceration, a second observer assessed the perineum. Perineal trauma suturing was standardised as previously described[18] and midwives participated in annual laceration repair workshops.

Recruitment of the CD group was accomplished by daily review of labour and delivery records. Primiparous women who delivered by CD were identified, the patient's chart was reviewed to determine whether she had entered the second stage of labour and, if not, the woman was approached for participation. Once formal consent was obtained, women in the CD group were asked to provide pelvic floor functional data as experienced in the last 3 months of pregnancy. This group was then followed prospectively with the VB group.

At the 6-month postpartum visit, in addition to providing functional data and undergoing the POPQ examination, all women underwent transperineal ultrasound (U/S). Women were placed in the lithotomy position and were imaged immediately after voiding. Disruption of the internal or external anal sphincters at the proximal, mid and/or distal levels were counted as not intact. Women also completed a Paper Towel Test (PTT).[19] With a full bladder, women were asked to cough three times within a 10-second time span, with the paper towel applied to the perineum. Staining of the paper towel with fluid was recorded as a positive test. Physical exam of the perineum included observation of a ‘dovetail sign’, loss of perianal anterior radial folds and/or perianal dimpling, both of which are thought to represent separation of the anal sphincter. Postpartum ultrasound and physical exam examiners were blinded to mode of delivery and participating women were instructed not to inform the examiner of delivery mode. Women were compensated $75.00 at the 6-month visit for travel and babysitting costs. If women could not present for appointments, they were mailed questionnaires.

We aimed to recruit 765 women: 630 nulliparas from the Midwifery service and an additional 135 primiparous women delivered by caesarean from the low risk Obstetrical and Family Medicine services. We anticipated that 20% of women recruited from midwifery clinics would not deliver with the midwives secondary to early pregnancy loss, relocation, insurance changes or medical complications of pregnancy. In addition, 50 midwife patients were expected to deliver by caesarean. Since the majority of the CD group was recruited immediately after delivery, baseline comparisons between groups included those VB patients who gave functional data late in pregnancy (after 36 completed weeks of pregnancy) and functional data given by non-midwife CD patients at birth who were asked to recall functional symptoms during their third trimester.

Two of the authors performed statistical analyses (R.G.R., C.Q.). With an assumption of 75% follow-up, we estimated that 119 CD and 375 VB women would give data at 6 months postpartum. We also assumed that the prevalence of bothersome UI and/or AI would be low, and the proportion of women who reported sexual activity would be high. With 494 women and an alpha of 0.05, we had 80% power to detect any of the disorders where the two groups had a prevalence of UI or AI, or sexual inactivity that differed by at least 10% if prevalence was low (<10%) or high (>80%). Descriptive statistics were used to compare groups at baseline and follow-up, categorical data were compared using Fisher's Exact test or chi-square analyses, and POPQ stage was evaluated with the Jonckheere–Terpstra trend test. Continuous variables were compared using t-tests. To evaluate the impact of baseline differences between groups on POPQ measures and sexual function, a stepwise regression multivariate analysis was performed, which included variables found to be different at baseline between groups as well as known predictors of outcomes. All analyses were conducted using SAS version 9.3, SAS Institute, Inc., Cary, NC, USA.


In all, 782 women consented to participate: 627 midwife patients and 155 CD patients. Of the midwife group, 541 women delivered at the University of New Mexico; 448 women gave birth vaginally and 93 underwent CD. As anticipated, the 69 midwife patients who underwent CD prior to the second stage of labour were combined with the CD group recruited immediately postpartum; there were thus 224 women in the CD group. Study participation is outlined in Figure S1. Overall, 474/672 (71%) of eligible women gave data at 6 months postpartum, 138/224 (62%) women in the CD group and 336/448 (75%) in the VB group.

Women who followed up (= 474) were slightly older than women who did not (= 199) (25.2 ± 5.5 versus 23.9 ± 5.3, P = 0.006), had more years of education (14.1 ± 2.8 versus 13.1 ± 2.7, P < 0.001) and had slightly lower body mass index (BMI) (25.1 ± 5.6 versus 26.2 ± 6.2, P = 0.02). At baseline, follow-up groups did not differ in their rates of UI, AI or sexual activity, sexual function scores, BMI or race/ethnicity (Table S1). We were able to collect prospective data on a third of the CD group who were cared for by midwives during pregnancy (= 69); their reports of function in the third trimester of pregnancy did not differ from the functional reports of CD participants recruited postpartum, although these comparisons were limited to questionnaire responses and not physical exam findings (Table 1).The VB group was younger (23.9 ± 4.9 versus 26.6 ± 6.1 years, P < 0.001) and less overweight/obese (38 versus 56%, P < 0.001) but groups did not differ in years of education, tobacco use, ethnicity, fetal birth weight, weight gain during pregnancy or epidural and oxytocin use. Native American women, who represented 8% of the study population, were more likely to undergo CD than VB; however, the baseline functional assessment did not differ from those of other race/ethnicities (data not shown). Among women who gave birth vaginally, eight women (2%) underwent episiotomy and 22 (5%) sustained third or fourth degree lacerations. Of the women who sustained second degree lacerations or greater (n = 129), 69% had a second observer, all but one of whom agreed with the second degree diagnosis. One woman had a forceps delivery and 25 (6%) a vacuum delivery. Among women who delivered vaginally, the mean length of the second stage of labour was 72.2 ± 63.9 minutes (Table 1).

Table 1. Maternal, newborn and labour characteristics
 Vaginal birth = 448Caesarean delivery = 224 P
Age (mean ± SD)23.9 ± 4.926.6 ± 6.1<0.001
Years education (mean ± SD)13.8 ± 2.713.8 ± 3.10.87
BMI, kg/m2 (mean ± SD)24.6 ± 5.327.1 ± 6.3<0.001
Overweight/obesea (%)168 (38)126 (56)<0.001
Weight gain in pregnancy, in pounds (mean ± SD)16.1 ± 6.215.6 ± 7.20.30
Fetal birth weight, g (mean ± SD)3216 ± 4253211 ± 5330.91
Race (%)b
Non-Hispanic White193 (43)79 (35)0.05
Hispanic201 (45)104 (46)
Native American26 (6)25 (11)
Other28 (6)14 (6)
Tobacco use (%)30 (7)15 (7)1.00
Epidural (%)269 (60)139 (62)0.68
Oxytocin (%)212 (47)124 (55)0.07
Length of second stage, minutes (mean ± SD)72.2 ± 63.9Not applicable 
Perineal trauma
Intact210 (47)Not applicable 
First degree laceration87 (19)
Second degree laceration129 (29)
Third and fourth degree lacerations22 (5)
Episiotomy (%)8 (2)Not applicable 
Baseline functional data in the 3rd trimester
 = 412= 208 P
  1. a

    Overweight/obese defined as BMI > 25 kg/m2.

  2. b

    Race was self-defined; two women in the CD group failed to provide race/ethnicity data.

  3. c

    indicates Wexner Fecal Incontinence Scale.

Any urinary incontinence (ISIc > 0) (%)300 (73)143 (69)0.30
Midwife CD = 6941 (68)Physician CD = 155102 (69) = 1.0
Any anal incontinence, W*** ≥ 1 (%)279 (68)149 (74)0.19
Midwife CD = 6938 (66)Physician CD = 155111 (77) P = 0.11
% Women sexually active265 (76)131 (72)0.29
Midwife CD = 6936 (72)Physician CD = 15595 (72) P = 1.0
Female Sexual Function Index Scores (mean ± SD)26.4 ± 5.926.2 ± 6.00.98
Midwife CD = 6926.5 ± 5.6Physician CD = 15526.1 ± 6.0 = 0.65

At 6 months postpartum, although approximately half of the women in each group reported some UI, less than 10% of women in either group reported moderate to severe UI. The CD group did not report more UI compared with the VB group on self-report measures, whether evaluated by presence, severity scores or quality of life impact (all P = NS). Of those with UI, ISI and IIQ-7 mean scores were similarly low. Women who gave birth vaginally were more likely than those who delivered by CD to have QUID stress greater than urge scores (27 versus 16%, P = 0.02). The PTT was performed by 445/474 (94%) of women who followed up. Most women had no leakage of urine (86%). Women in the VB group were more likely to have a wet paper towel than those who delivered by caesarean (17% versus 6%, P = 0.002).

Many women reported some AI, including loss of flatus (50%), however, less than 10% of women reported fecal incontinence. The VB group was not more likely to report greater presence, severity or quality of life impact of AI symptoms than women who gave birth by CD (all P = NS) (Table 2). Dovetail signs were few and did not differ between groups, and no woman was observed to have perianal dimpling. Of women who agreed to U/S [= 410 (87%)], there were 34 internal and four external anal sphincter disruptions, with no differences between groups (all P = NS). Although the CD group was not more likely to report perineal pain or be sexually inactive, 6-month FSFI scores were significantly lower; this difference was explained by lower Desire domain scores (Table 2).

Table 2. Pelvic floor outcomes 6 months after delivery
Urinary incontinenceVaginal birth = 336Caesarean delivery = 138 P
Any urinary incontinence, ISI > 0 (%)183 (55)63 (46)0.08
Moderate/severe ISI scores (%)30 (9)8 (6)0.35
Paper Towel Test wet (%)56 (17)8 (6)0.002
Women with any urinary incontinence= 183= 63 
ISI scores among those with any incontinence (mean ± SD)1.76 ± 1.11.73 ± 1.00.30
IIQ scores among those with any urinary incontinence (mean ± SD)5.3 ± 11.36.6 ± 13.60.51
Questionnaire for Urinary Diagnosis
Stress > urge scores (%)88 (27)22 (16)0.02
Urge > stress scores (%)68 (21)34 (25)0.33
Anal incontinenceVaginal birth = 336Caesarean section = 138 
Any anal incontinence (Wexnera ≥ 1) (%)163 (50)76 (55)0.26
Fecal incontinence, positive response on Wexner scale (%)27 (8)18 (13)0.12
Anal sphincter separation on physical exam (%)11 (3)3 (2)0.77
Women with any anal incontinence= 163= 76 
Wexner Scores among those with any anal incontinence (mean ± SD)2.0 ± 1.81.9 ± 1.10.53
Colorectal anal impact questionnaire scores (mean ± SD)2.5 ± 9.72.0 ± 7.10.70
Pelvic organ prolapseVaginal birth = 336Caesarean section = 138 
POPQ Stage prolapse
046 (14)28 (22)0.03b
1209 (65)81 (63)
266 (21)19 (15)
31 (0)0 (0)
40 (0)0 (0)
Aa (mean ± SD)−1.9 ± 0.8−2.1 ± 0.7<0.001
Ba (mean ± SD)−1.9 ± 0.8−2.1 ± 0 .7<0.001
Ap (mean ± SD)−2.6 ± 0.5−2.7 ± 0.40.16
Bp (mean ± SD)−2.6 ± 0.5−2.7 ± 0.40.18
C (mean ± SD)−5.2 ± 1.4−5.1 ± 1.50.80
D (mean ± SD)−6.7 ± 1.6−6.6 ± 1.80.48
Pb (mean ± SD)3.4 ± 0.83.3 ± 0.80.06
GH, rest (mean ± SD)2.6 ± 0.92.7 ± 0.70.11
GH, strain (mean ± SD)3.0 ± 1.03.3 ± 0.8<0.001
Total vaginal length (mean ± SD)7.4 ± 1.47.1 ± 1.70.07
Pelvic organ prolapse impact questionnaire scores (mean ± SD)2.2 ± 8.71.7 ± 6.60.53
Women with stage 2 or greater prolapse= 67= 19 
Pelvic organ prolapse impact Scores among women with stage 2 or greater POP (mean ± SD)3.5 ± 11.81.3 ± 4.50.21
Transperineal ultrasound= 291= 119 
External anal sphincter separation (%)2 (<1)2 (<1)0.58
Internal anal sphincter separation (%)29 (10)5 (4)0.07
Sexual functionVaginal Birth= 336Caesarean Section= 138 
Sexually active (%)281 (88)123 (92)0.14
Female Sexual Function Index scores (mean ± SD)28.5 ± 5.426.6 ± 6.30.004
Desire (mean ± SD)3.8 ± 1.23.3 ± 1.2<0.001
Arousal (mean ± SD)4.1 ± 1.84.1 ± 1.50.89
Lubrication (mean ± SD)4.5 ± 2.04.4 ± 4.10.73
Orgasm (mean ± SD)4.2 ± 2.04.3 ± 1.70.52
Satisfaction (mean ± SD)4.7 ± 1.44.6 ± 1.30.49
Pain (mean ± SD)5.1 ± 1.24.9 ± 1.30.37
  1. IIQ, Incontinence Impact Questionnaire; ISI, Incontinence Severity Index; PPI, Present Pain Intensity.

  2. a

    Wexner Fecal Incontinence Scale.

  3. b

    Jonckheere–Terpstra trend test.

No perineal pain, PPI = none (%)301 (92)131 (95)0.27

On physical examination, the VB group was more likely to have a higher stage of prolapse than the CD group at 6 months postpartum; the POPQ differences were limited to the anterior vaginal wall (Table 2). Seventeen women underwent repeat exams for inter-rater reliability testing of the POPQ exam; agreement between examiners was high (82% complete agreement) for prolapse stage, and all POPQ points were within 1 cm of agreement for 76% of measures. Although differences were noted between groups on physical exam, prolapse quality of life impact scores did not differ between groups (P = NS).

Because our CD group was recruited from both the midwifery and physician services, we performed a comparison of outcomes between these two groups. No differences were found in urinary or anal incontinence, sexual activity, sexual function scores or perineal pain. The MD caesarean patients were more likely to have stage 2 prolapse, although all prolapse was at least 1 cm above the hymen as measured on all nine POPQ points. Given the very small differences in measurable prolapse, we feel that these results are not clinically significant (data not shown). When entered into a stepwise regression model including baseline variables that were different between groups (Age, BMI and weight gain), as well as non-Hispanic White race/ethnicity, a known predictor of prolapse, delivery mode remained a significant predictor of POPQ points Aa and Ba and FSFI scores (Table 3).

Table 3. Stepwise multivariable analysis of impact of mode of delivery on anterior vaginal wall descent and sexual function scores
Univariate analysisMultivariable analysis
OutcomeFactorStandardised Betaa P OutcomeFactorbStandardised BetaaAdjusted P
  1. FSFI, Female Sexual Function Index; NHW, non-Hispanic White.

  2. a

    Standardised beta is the log odds regression coefficient standardised by the standard deviation of the predictor variable. In plain language, standardised beta represents the effect of a 1 SD change in the predictor variable on log odds of the outcome variable. Standardised betas are equivalent to odds ratios, as exponentiated standardised beta is related to the odds ratios as a function of the ratio of standard deviations of the outcome to predictor variables.

  3. b

    Weight gain during pregnancy was also included but did not remain significant in the regression model.

  4. c

    Delivery mode: mode of delivery was coded as vaginal birth = 1, and caesarean delivery = 2.

AaDelivery modec−0.160.002AaDelivery modec−0.140.004
Age (years)−0.020.66
BMI (kg/m2)−0.130.007
BaDelivery modec−0.160.001BaDelivery modec−0.140.004
Age (years)−0.040.47
BMI (kg/m2)−0.130.006
FSFI total scoresDelivery modec−0.150.004FSFI total scoresDelivery modec−0.160.002
Age (years)−0.050.370.04
BMI (kg/m2)0.11 


Main findings

We found that pelvic floor dysfunction was uncommon among primiparous women following either a vaginal or caesarean birth without entry into the second stage of labour. In this study, use of interventions such as episiotomy and operative delivery, already known to have an adverse effect on pelvic floor function, were quite low. Self-reported rates of UI, AI, pain and sexual activity were similar between groups. Women who gave birth vaginally were more likely to have greater descent of the anterior vaginal wall; however, differences were small and POP quality of life impact did not differ between groups. In addition, women who gave birth vaginally were more likely to have objective loss of urine on PTT; however, most women were dry on objective testing and few women reported moderate to severe UI symptoms. Finally, women who underwent CD were more likely than women who delivered vaginally to report lower sexual function scores; again, these differences were small between groups and were confined to the Desire domain of the FSFI. Our findings support that the second stage of labour does not have a significantly negative impact on global postpartum pelvic floor function at 6 months postpartum in a low-episiotomy, low-operative delivery setting. Although our vaginal birth group was cared for by midwives, the low rates of episiotomy and operative delivery are reflective of the low rates of these interventions of all practitioners who attend births in our labour and delivery unit.[20, 21] In addition, our practices are reflective of US national trends in the performance of episiotomy and operative delivery and reflect evidence-based care.[22, 23]

We found that changes in pelvic floor function following birth in either group were small, unlikely to affect quality of life, and did not vary greatly by mode of delivery. This study is unique because we chose to focus on the impact of the second stage of labour rather than planned caesarean without labour. Although maternal choice caesarean delivery has gained support in some settings, entry into labour remains the norm for most primiparous women.[24] Our findings can reassure women that birth in a low-operative delivery and episiotomy setting results in a low incidence of postpartum pelvic floor dysfunction at 6 months. We found pelvic floor dysfunction present at low levels in both cohorts, indicating that there may be contributions to pelvic floor dysfunction from pregnancy or the postpartum period or changes that occur in the first stage of labour.

Strengths and limitations

There are several unique aspects that strengthened our study. We recruited women during pregnancy and immediately after CD and followed them prospectively postpartum. We evaluated the full range of pelvic floor function using validated questionnaires and pelvic floor examinations rather than focusing on a single aspect of pelvic floor function. Unlike prior studies, we included measures of sexual function, a pelvic floor outcome important to young healthy women. We took care to standardise perineal repair and the consistency of pelvic floor measurements between examiners and blinded examiners to mode of delivery. We also included multiple measures of each type of pelvic floor problem; for example, we evaluated the presence, severity, type and quality of life changes associated with UI, as well as an objective measure for UI with a PTT. The overall consistency between measures that we observed further supports our conclusions.

Our findings are limited to the short term following pregnancy. While a follow-up of 6 months is short, other studies have shown that women with persisting UI at 3 months following birth have a 92% risk of UI at 5 years.[25, 26] In addition, urethral needle EMG testing has been shown to improve and stabilise between 6 weeks and 6 months postpartum, findings supportive of pelvic floor stabilisation following birth at 6 months postpartum.[27] Whether this cohort will show differences in the longer term is unknown; other studies have found differences in as short as 18 months or as long as 5–10 years after delivery.[3, 28] Other weaknesses of this study include that the majority of the CD group was recruited immediately after delivery and these women were asked to recall their pelvic floor function in the third trimester. Asking women to recall function is subject to bias and may have led to under- or overestimation of pelvic floor problems during pregnancy. We were able to collect prospective data on a third of the CD group who were cared for by midwives during pregnancy (= 69); their reports of baseline function in the third trimester of pregnancy did not differ from the functional reports of CD participants recruited postpartum, although these comparisons were limited to questionnaire responses and not physical exam findings. In addition, functional outcomes at 6 months postpartum were not different between the women in the CD group who were followed through pregnancy and those who were recruited postpartum. The examiners at the 6-month follow-up were blinded to mode of delivery; however, experienced midwives may be able to ascertain mode of delivery through the appearance of the perineum. Our overall follow-up rate of 71% is slightly lower than anticipated. When we compared women who followed up with those lost to follow-up, we found differences in age, educational level and BMI; however, the differences were small in this largely homogeneous group of first time mothers, and multivariate analyses controlling for age and BMI did not affect our findings. Our follow-up rates were lower in the CD than in the VB group; however, baseline differences with the CD group between those who followed up and those who did not were limited to a small difference in years of education and thus were unlikely to alter results significantly. It is possible that there are unmeasured differences between midwife and physician care that may affect outcomes which we cannot account for. Finally, we cannot extrapolate our findings to women who undergo maternal choice caesarean prior to entry into labour because all of our patients anticipated giving vaginal birth.


A recent cohort survey study of nulliparous women followed through pregnancy and birth to 18 months postpartum in Australia found an increased risk of persistent UI among women who gave birth vaginally compared with those delivered by CD; a prolonged second stage of labour and operative delivery increased the odds of postpartum UI[17, 28, 29]. Another cohort study of US women recruited 5–10 years after delivery found an increased risk of UI in the VB group compared with the caesarean group without labour. In addition, the VB group in that study was more likely to have prolapse when compared with women delivered by CD, with the greatest difference seen in the group that underwent operative vaginal delivery[3, 4]. Both of these cohorts had high rates of operative delivery (31% in the Australian group and 39% in the US group) and may not be representative of current evidence-based obstetrical practice. In contrast to these previous reports, we did not find that women who gave birth vaginally were more likely to report any UI on self-report measures than were those delivered by CD at 6 months postpartum. We did find that UI type varied between groups and that those who had given birth vaginally were more likely to leak than those who underwent CD. Our lack of difference in self-report UI may be because we chose to include only women recruited from low risk services who had low exposures to operative delivery and episiotomy. Alternatively, our sample size, powered to observe differences of 10% or greater, was unable to detect smaller differences. We chose a difference of 10% between groups because we believe differences in functional outcomes less than 10% are unlikely to be clinically significant. Interestingly, both the Australian and US studies found that AI was not increased in women giving birth vaginally compared with CD.[30, 31] In agreement with these previous studies, we found that AI was not related to mode of delivery. Recent work has suggested that anal sphincter laceration risk with subsequent anal incontinence may be a result of inherited familial risk.[32]


The second stage of labour experienced by low risk women who had a midwife-managed vaginal birth with low exposure to operative vaginal delivery and episiotomy does not contribute to bothersome pelvic floor dysfunction 6 months after delivery when compared with women who undergo caesarean delivery prior to the second stage of labour.

Disclosure of interests

RGR is Data Safety Monitoring Chair for the TRANSFORM trial sponsored by AMS and receives royalties for writing a prolapse chapter for UPTODATE. No other authors have disclosures.

Contribution to authorship

Conception of study, planning, carrying out, analysing and manuscript writing: RGR, LML, RJH. Conception of study, conduct of trial, manuscript writing: NB, AMF, DT, LLA. Analysing and manuscript writing: CQ. Conception of study, analysing and manuscript writing: EB.

Details of ethics approval

Reviewed and approved by the University of New Mexico's Human Research Protections Office # 04-244.


Supported by NICHD 1R01HD049819-01A2 and National Center for Research Resources and the National Center for Advancing Translational Sciences of the National Institutes of Health through Grant Number 8UL1TR000041.

Pelvic floor troubles after vaginal birth, more fear than harm

  • Xavier Fritel

  • Department of Gynecology and Obstetrics, Poitiers University Hospital, Poitiers, France

Mini commentary on ‘Contribution of the second stage of labour to pelvic floor dysfunction: a prospective cohort comparison of nulliparous women’

The results of the study by Rogers et al. (BJOG 2013) are counter-intuitive but nevertheless confirm some previous observations showing that obstetrical pelvic floor damage is more feared than actually common (Fritel et al. Acta Obstet Gynecol Scand 2012;91:901–10). An interesting point in this work is the low rates of caesarean section and pelvic floor troubles in the group of low-risk nulliparous women followed by midwives. The minimal impacts observed of vaginal delivery on the pelvic floor were therefore not a consequence of early recourse to caesarean delivery in labour, and such findings should be broadly generalisable to low risk women planning vaginal delivery. This observation should, of course, be compared with the Cochrane analysis showing more spontaneous vaginal delivery in pregnant women attended by midwives (Sandal et al. Cochrane 2013 DOI: 10.1002/14651858.CD004667.pub3).

One of the weaknesses of the work (recognised by the authors) is the collection after delivery, in the caesarean group, of data about continence during pregnancy; although useful, this information cannot ascertain comparability in women from both groups with regard to pelvic floor function before delivery. As was expected, both groups (vaginal and caesarean) differ in terms of age and body mass index. The similar postnatal pelvic floor symptoms in women having given vaginal birth and those having given caesarean birth may quite possibly be explained largely by the effects of age and obesity in the latter group. Indeed, the risk of urinary or fecal incontinence increases with age and obesity.

One of the particular qualities of the work of Rogers et al. consists in its use of single-blind postnatal clinical examination. As expected, pelvic organ prolapse during clinical examination was slightly more common in the vaginal delivery group; multivariate analysis confirmed the association between vaginal delivery and anterior anatomical defect. However, it is important to note that this does not affect sexuality. In fact, this study confirms previous works showing that the mode of delivery has no impact on postpartum sexuality when assessed several months subsequent to childbirth (Fauconnier et al. Prog Urol 2012;22:225–32).

The pathophysiology of urinary and fecal incontinence is multifactorial and the role of birth trauma is probably low, with other factors such as pregnancy itself, aging or obesity probably being more contributory. As the authors point out, these results are important in terms of providing reassurance to women, and birth attendants, as to the limited consequences of vaginal delivery on pelvic floor function in low-risk nulliparous women.

Disclosure of interests

The author has no conflicts of interest to declare.