Pelvic floor dysfunction after intervention, compared with expectant management, in prolonged second stage of labour: A population‐based questionnaire and cohort study

To investigate the effect of vacuum extraction (VE) or caesarean section (CS), compared with expectant management, on pelvic floor dysfunction (PFD) 1–2 years postpartum in primiparous women with a prolonged second stage of labour.


| I N TRODUC T ION
A prolonged second stage of labour, exceeding 3 h, [1][2][3] affects 10%-20% of primiparous women. 1,4,5It has been associated with maternal and neonatal short-term adverse outcomes. 1,4,6To what degree and how a prolonged second stage affects pelvic floor function in the long term is unclear, and the associations are complex.A prolonged second stage has been associated with urinary incontinence, 7 particularly after prolonged pushing or operative vaginal delivery, 8,9 whereas other studies show no association. 10,11nal incontinence was unrelated to second stage duration in some reports, 11,12 but a prolonged second stage may lead to subsequent anal incontinence mediated by operative vaginal delivery and obstetric anal sphincter injury (OASI). 13A prolonged second stage duration increases the risk of levator ani muscle injury, 14,15 which predisposes the development of pelvic organ prolapse (POP), 14,16 although POP has not been independently associated with second stage duration. 11,17Current guidelines regarding management of the second stage of labour lack recommendations on when and how to intervene to minimize the risk of maternal and neonatal complications. 3,18,19The objective of this study was to investigate the effect of intervention by caesarean section (CS) or vacuum extraction (VE) in a prolonged second stage, compared with expectant management, on patient-reported pelvic floor symptoms 1-2 years after delivery.The hypothesis was that in women with a second stage of 3 h or more, intervention at a specific hour, compared with expectant management, might reduce the risk of PFD.With this approach, we tried to mimic the clinical challenge of deciding when to intervene or to wait with respect to future maternal PFD.

| Study population and exposure
The source population was all women who gave birth in the Stockholm region, Sweden, from 1 December 2017 through to 30 November 2018, constituting approximately 25% of births in Sweden. 20During the study period, there were six labour ward hospitals in Stockholm using the same computerized medical records system (Obstetrix: Cerner Sweden, Oracle Cerner, Austin, TX, USA).Virtually no births occur outside of hospital.In Sweden, current management of the second stage of labour is to await the spontaneous urge to push after the passive descent of the fetus.Augmentation with oxytocin is initiated after 1 h without progress in the passive phase of the second stage or after 30 min of pushing without imminent delivery.Active pushing is usually initiated no later than after 3 h of passive descent.A maximum pushing duration of 1 h is recommended but could be extended if progress is seen.When intervention is needed, and the criteria for VE are fulfilled, such as no signs of disproportion and fetal station at or below the ischial spines, this is usually the delivery mode of choice.Operative vaginal delivery in Sweden is almost exclusively performed with VE, not forceps.
The inclusion criteria were primiparous women at term (≥37 weeks of gestation) with a live, singleton, cephalic birth, and with a second stage duration of ≥3 h.This cut-off was chosen as with a normal duration of <3 h, [1][2][3] the duration itself is not an indication for intervention.
Second stage duration was calculated as the duration between the first recording of full cervical dilation and birth.Eligible women were sent a study invitation letter, a consent form and a questionnaire on pelvic floor function at 12-24 months after delivery, 21 from 1 December 2019 to 30 June 2020.One postal reminder was sent after 2 weeks.For all consenting women, the characteristics and outcomes were obtained from computerized medical records held by the Stockholm Region Enterprise Data Warehouse.Women with a subsequent delivery were instructed to answer the questions based on symptoms before her second pregnancy.Patients were not involved in the design or performance of the study.
Exposure comprised two different interventions, compared with expectant management, and comparisons were made at two distinct time points during the second stage.Specifically, exposure was defined as intervention with CS or VE at a second stage duration of 3-4 h or 4-5 h.The comparison group was expectant management, i.e. with the second stage proceeding without intervention (with second stage durations of >4 h and >5 h, respectively).Women with spontaneous vaginal delivery (SVD) within the stated time periods (3-4 h and 4-5 h) were included in the expectant management group (no intervention needed).In women with a second stage duration of >5 h, exposure was intervention with CS or VE, and the comparison group was SVD.

| Outcome measures
At the start of this study, there was no published core outcome set for pelvic floor function after childbirth.We used the 1-year follow-up questionnaire from the Swedish National Perineal Laceration Register, containing questions about urinary incontinence, anal incontinence and symptoms of POP, that is normally only distributed to women after OASI. 21The primary outcome was a composite outcome of moderate-to-severe pelvic floor dysfunction (PFD) including at least weekly symptoms of urinary incontinence, anal incontinence or POP, as defined below.Secondary outcomes were each of the included symptoms.
Urinary incontinence was measured by three questions validated and used in previous studies, 22,23 covering the presence of urinary leakage or involuntary urination, stress incontinence or urgency incontinence.All questions had five alternative answers from 'never' to 'more than once a day'. 21he International Continence Society has defined urinary incontinence as a complaint of (any) involuntary loss of urine. 24We wanted a more clinically useful definition, and hence defined urinary incontinence as the leakage of urine (stress incontinence or urgency) at least weekly.
Anal incontinence was measured by the Wexner score. 25he score includes parameters for leakage of gas, liquid or solid stool, the use of pads and affected quality of life, with five alternative answers from 'never' to 'daily', and with a maximum score of 20. 26 To ensure that we identified anal incontinence that was affecting quality of life, we defined anal incontinence as a Wexner score of ≥4, as this level has been reported to affect quality of life in 86.6% of women with previous OASI. 27ymptoms of POP were measured by five questions validated and used in previous studies. 22,23The questions covered symptoms of vaginal bulging, usage of pessaries, difficulties emptying the bladder or bowel and the need of digital manipulation to defecate.All questions had five alternative answers from 'never' to 'daily', except for the question regarding pessary use, which included the answers 'never', 'have used a pessary, but not currently' and 'use pessary'. 21he International Urogynecological Association and the International Continence Society has published a joint report on the terminology of female POP, including a broad range of symptoms and signs of POP at physical examination and a method for quantification. 28We wanted a clinically useful definition based only on patient-reported symptoms and hence defined the outcome of POP as a sense of vaginal bulging at least weekly, current use of pessary, use of digital manipulation to defecate or difficulties in emptying the bladder or bowel at least weekly.

| Background characteristics
Maternal characteristics were registered by the midwife at the first antenatal visit.Gestational age at birth was determined by free routine dating ultrasound, performed at 12-14 or 18-19 weeks of gestation, and categorised in 2-week intervals.Delivery characteristics were registered by the attending midwife or physician.Fetal station was collected from the partograph.If the fetal station was below the ischial spines at any time point it was categorized as 'yes', otherwise it was listed as 'no/unknown'.'Unknown' included a lack of registration (i.e.no recorded vaginal examination or the fetus never reached this station).Delivery and neonatal outcomes were registered by the attending midwife or physician.Perineal and vaginal injuries were classified according to the Swedish version of the International Classification of Diseases, 10th Edition (ICD-10) after clinical examination by the attending midwife and/or physician.The degree of injury was collected from diagnostic and procedure codes.Women were categorized in groups according to their most severe registered injury: first-degree perineal tear (ICD-10 O70.0), including women without any diagnosis of perineal or vaginal injury and women with only a diagnosis of isolated vaginal tear (O71.4);second-degree perineal tear (O70.1),including women with only an episiotomy diagnosis (TMA00); and OASI, including third-(O70.2) and fourth-degree (O70.3)perineal tears, or the procedure code for surgical repair of a third-or fourth-degree perineal tear (MBC33).

| Statistical analyses
Statistical analyses were performed with SPSS 28.0 (IBM, Armonk, NY, USA).Prevalence and proportions of characteristics and outcomes were calculated and compared by chisquare test or with Fisher's exact test, for variables with rare events.P < 0.05 was considered statistically significant.The relative risk of PFD in the intervention groups compared with expectant management was calculated using Poisson regression with robust variance estimation. 29Results are presented as crude and adjusted relative risk (RR and aRR) with 95% confidence intervals (95% CIs).Covariates for the adjusted analyses were selected using a directed acyclic graph (DAG) for the exposure and the outcome: maternal age (< or ≥35 years), BMI (< or ≥30 kg/m 2 ), fetal position (occiput anterior, occiput posterior or other) and birthweight (< or ≥4000 g) (Figure S1).
As OASI is a marker of severe pelvic floor injury and is associated with PFD, a mediation analysis was performed to investigate the relationship between intervention with VE, the primary and secondary outcomes (PFD, anal incontinence, urinary incontinence and POP) and the mediator OASI.The DAG was used to select covariates for the mediation analyses (Figure S1).Mediation analyses were performed using the 'paramed' command from Stata 16.1 (StataCorp, College Station, Texas, USA) and the results are presented as RRs with 95% CIs for total effect, natural direct effect, natural indirect effect and proportion mediated. 30Due to the inappropriateness of the rare event assumption for the outcomes examined in this study a loglinear model was employed to fit a regression model for the various outcomes for the mediation analysis.Unfortunately, certain mediation analysis results could not be obtained due to a non-convergence issue.
Missing data are presented in tables and constituted a minor proportion (<3%), except for body mass index (BMI, 4.2%).We performed multiple imputation for the BMI variable and used the imputed version in the regression and mediation analyses.
We performed sensitivity analyses excluding women who answered the questionnaire later than 24 months or with a subsequent delivery.We also compared the study participants to the non-responders regarding second stage duration and mode of delivery.Prevalence and proportions were calculated and compared by chi-square test.Median second stage duration and range were calculated and compared with the Mann-Whitney U-test.
T A B L E 1 Background characteristics, birth characteristics and outcomes in women with a second stage duration of ≥3 h according to intervention after 3-4 h, compared with expectant management a .elective CS (n = 1030) and a second stage duration of <3 h or with a non-calculable duration (n = 1527) were excluded.For the remaining 3042 women, we obtained contact information and information regarding second stage duration and mode of delivery.Women with missing contact information (n = 190), duplicate registrations (n = 2), forceps delivery (n = 2), and outliers with a second stage exceeding 13 h (n = 3) were excluded, leaving 2845 eligible study participants (Figure S2).
The response rate was 45.8% (1303/2845 distributed questionnaires).In women with SVD, the response rate was 47.2% (923/1957); in women with VE, the response rate was 45.7% (280/613); and in women with CS, the response rate was 36.1% (99/274).One unintendedly included breech birth was excluded (Figure S2).The final study cohort included 1302 women (Figure S2).Compared with non-responders, the study cohort did not differ regarding the median second stage duration or distribution in time intervals.There was a higher proportion of SVD, a similar proportion of VE and a lower proportion of CS in the study cohort, compared with non-responders (Table S1).
In the study cohort, 659 (50.6%) women had a second stage duration of 3-4 h, 414 (31.8%) women had a duration of 4-5 h and 229 (17.6%) women had a duration that exceeded 5 h.The proportion of mode of delivery in each time interval is presented in Figure S3.The rates of CS and VE increased with prolonged duration, with 2.4% CS at 3-4 h, 7.7% CS at 4-5 h and 22.3% CS at >5 h, and with 14.9% VE at 3-4 h, 28.7% VE at 4-5 h and 27.5% VE at >5 h.The proportion of SVD simultaneously decreased from 82.7% at a second stage duration of 3-4 h, to 63.5% at 4-5 h and 50.2% at >5 h (Figure S3). Background characteristics did not differ between groups (Tables 1 and 2).The rate of patients with a recorded head station below the ischial spines was less common, whereas occiput posterior position and a birthweight of ≥4000 g were more common among women with CS at 3-4 h and 4-5 h, compared with expectant management (Tables 1 and 2).In VE at 3-4 h, second-degree perineal injury and episiotomy were more common than with expectant management, but OASI was not (Table 1).In VE at 4-5 h, episiotomy and OASI were more common than with expectant management (Table 2).The proportion of OASI was significantly higher after VE at 4-5 h, compared with VE at 3-4 h (P < 0.01, not in table).
The prevalence of symptoms of pelvic floor dysfunction are presented in Table 3.In total, PFD was reported by more than one in three women.The highest prevalence of PFD was

T A B L E 1 (Continued)
T A L E 2 Background characteristics, birth characteristics and outcomes in women with a second stage duration of ≥3 h according to intervention after 4-5 h, compared with expectant management a .reported after VE, both at 3-4 h and 4-5 h.The lowest prevalence of PFD was reported after CS at 3-4 h.With CS, the prevalence of PFD was higher at 4-5 h and at >5 h than at 3-4 h.After 5 h, no statistically significant difference was reported in the prevalence of PFD after CS, VE or SVD.For a Wexner score of ≥4, the highest prevalence was reported after VE at a second stage of >5 h and urinary incontinence was most common after VE at 4-5 h.No woman with CS at 3-4 h reported these outcomes.For prolapse symptoms, no significant differences in prevalence were reported between any of the groups.The aRR of PFD was increased after VE compared with expectant management, both at 3-4 h (aRR 1.33, 95% CI 1.06-1.65)and 4-5 h (aRR 1.34, 95% CI 1.05-1.70)(Table 4).After 5 h, there was no significant difference in the risk after VE or CS, compared with SVD (Table 4).The aRR for a Wexner score of ≥4 was significantly increased after VE after >5 h compared with SVD (aRR 3.04, 95% CI 1.40-6.63)(Table 4).The relative risk of urinary incontinence was significantly increased after VE at 4-5 h compared with expectant management (Table 4).Also, the aRR of prolapse symptoms was increased after VE at 3-4 h compared with expectant management (aRR 1.52, 95% CI 1.09-2.13)(Table 4).
In the mediation analyses, the increased risk of PFD after VE at 3-4 h was not mediated by OASI, as the natural indirect effect was 1.00 (95% CI 0.96-1.04)and the proportion mediated was 0% (Table S2).The secondary outcomes were not mediated by OASI either, possibly except for anal incontinence at 4-5 h, with a near significant natural indirect effect and with a proportion mediated of 26%.Nonconvergence obstructed some of the analyses.
Of the 1302 study participants, 69 (3.9%) women answered the questionnaire later than 24 months after delivery (25-27 months).Also, 147 (11.3%) women had a subsequent birth, 12 (12.1%) of the women with a first CS, 26 (9.3%) of the women with a first VE and 109 (11.8%) of the women with a first SVD.Sensitivity analyses testing the robustness of our findings excluding women who answered the questionnaire more than 24 months after delivery or with a subsequent birth did not depart from the entire-sample analyses.

| Main findings
This study showed that moderate-to-severe PFD affected more than one in three primiparous women with a prolonged second stage at 1-2 years after childbirth.The risk of PFD was  increased by with VE at second stage durations of 3-4 h and 4-5 h, whereas CS did not significantly decrease the risk, compared with expectant management.The increased risk after VE was not clearly mediated by OASI.The result for each PFD symptom (anal incontinence with a Wexner score of ≥4, at least weekly urinary incontinence or prolapse symptoms) was ambiguous, and should be interpreted with caution.Overall, an extended duration ≥3 h in this population of women with at a second stage of 3 h or more, did not affect the risk of PFD.PFD seemed rather affected by a forced passage of the fetus through the pelvic floor.

| Interpretation in light of other evidence
Our findings are consistent with previous reports of an overall PFD prevalence of around 25%-30% of women after childbirth, 31,32 and also with the few reports relating PFD to second stage duration, albeit with different definitions or shorter durations. 11,17,33PFD has also previously been associated with operative vaginal delivery, [34][35][36][37][38] and injury to pelvic floor muscles, ligaments and nerves. 11Still, we could not demonstrate that the increased risk of after VE, compared with expectant management, was mediated by OASI.A plausible explanation could be that PFD is dependent not only on OASI but also on other injuries.There was a trend for OASI mediating anal incontinence after VE at 4-5 h, as seen in our previous report. 12ntervention with CS avoids the passage of the fetus through the pelvic floor and thus some of the traumatic tissue injuries, although this was only observed if CS was performed at a second stage duration of 3-4 h.Pressure from the fetal head has been reported to contribute to PFD in women with CS during labour, 39,40 although this finding has been contradicted. 34In our study, over 40% of the women with CS at second stage durations of 3-4 h and 4-5 h had a recorded fetal head station below the ischial spines, implying that the head could exert pressure on the surrounding tissues.Women with CS at 3-4 h reported less PFD than all other groups, but with increasing duration the difference between CS and the other groups became smaller, supporting the idea of a duration of pressure effect.

| Strengths and limitations
The strengths of this study include the design mimicking the clinical challenge to intervene or to wait and a large population-based sample.In addition, the outcomes were patient reported, and did not rely on coding in medical records, and the questions that we used have been validated and are used in a Swedish national register.The main limitation of this study was the moderate response rate with a risk of selection bias.Women with symptoms could be more likely to respond, introducing an overestimated prevalence of PFD.Even though the second stage duration did not differ between responders and non-responders, there was an under-representation of women with CS among responders.Despite our sensitivity a second pregnancy or birth could also the prevalence of PFD.To avoid this, women with a second birth were instructed to answer the questionnaire based on the symptoms they had before their second pregnancy, but even so, this could be a source of recall bias.Unfortunately, we lacked information regarding continuing pregnancies.Other important limitations are the risk of a lack of power for the relatively rare events and the cross-sectional, observational design measuring the prevalence of PFD after childbirth.The mode and time point for delivery were not randomly allocated, but a result of clinical judgement.This should have selected women to a suitable treatment, which could have affected the differences, resulting in residual, unmeasured confounding.For secondary outcomes, the results diverged.For each intervention interval during the second stage, VE increased the risk of POP, urinary incontinence and anal incontinence, in that order, but also sequentially lost statistical significance for these outcomes.This could reflect an advancing co-selection of factors increasing the risk of posterior pelvic floor injury or a pressure duration effect, but it could also be a result of decreasing power.A larger sample size and/or a higher response rate would have improved the power and reduced the risk of selection bias.

| CONCLUSION
For patients and clinicians, the results from this study indicate that a prolonged duration of the second stage of labour of ≥3 h should not be an indication to end delivery to reduce the risk of PFD if the birth is accomplished by SVD or CS later.In this study, most of the PFD is driven by the need to use external force (VE) to deliver the fetus.However, the choice of delivery route when the delivery eventually must end should be governed by a multifactorial assessment of the likelihood of complications in second stage CS versus VE, and patients should be counselled for shared decision-making.

AU T HOR C ON T R I BU T ION S
SB, AS and SBW conceived the study.SB acquired the data and managed the data set.SB, AS, JMS and SBW planned the analyses.SB, SBW and HZ performed the analyses.All authors interpreted the results.SB wrote the first draft of the article, with critical and technical input from all other authors.All authors approved the final version for publication.

AC K NOW L E D GE M E N T S
The authors thank Bengt Eriksson, system developer at Stockholm Region Enterprise Data Warehouse, who made the data retrieval possible.We also thank all participating women for their contribution and Alexandra Spasojevic, MD, and Erika Lantz, MD, for assisting in the distribution and gathering of questionnaires.

F U N DI NG I N FOR M AT ION
No specific funding was received for this study.b Composite outcome, including a Wexner score of ≥4, urinary incontinence (leakage of urine ≥1-3 times/month, stress incontinence or urgency incontinency ≥1-4 times/ month) or prolapse symptoms (current use of pessary, digital manipulation to defecate ≥1-3 times/week, difficulties empty bladder or bowel or vaginal bulging ≥1-3 times/ week).c Adjusted for maternal age (< or ≥35 years), BMI (< or ≥30 kg/m 2 ), fetal position (occiput anterior, occiput posterior or other) and birthweight (< or ≥4000 g).
e Defined as current use of pessary, digital manipulation to defecate ≥1-3 times/week, difficulty in emptying bladder or bowel, or vaginal bulging ≥1-3 times/week.
Abbreviations: CS, caesarean section; OASI, obstetric anal sphincter injury; VE, vacuum extraction.a Expectant management includes women with spontaneous vaginal delivery between 4 and 5 h.b Including episiotomy.

T A B L E 3
Abbreviations: CS, caesarean section; SVD, spontaneous vaginal delivery; VE, vacuum extraction.a Expectant management includes women with spontaneous vaginal delivery at 3-4 h and at 4-5 h, respectively.b Composite outcome including a Wexner score of ≥4, urinary incontinence or prolapse symptoms, as defined below.

T A B L E 4
Association between pelvic floor dysfunction in women with a second stage duration of ≥3 h and intervention with caesarean section or vacuum extraction, compared with expectant management a .